Interventions for improving the adoption of shared decision making by healthcare professionals

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies.

Summary of findings for the main comparison.

Interventions targeting patients compared with usual care for improving the adoption of shared decision making by healthcare professionals
Outcomes*	Type of outcome	Median of the standardized mean difference or median of the risk difference (range)	No of measures (studies**)	Quality of the evidence (GRADE)
Observer‐based SDM measures	Continuous measure	Unavailable data	3 (1)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.21 (0.04 to 0.50)	6 (4)	Very low ^1,2,3
	Categorical measure	‐0.02 (‐0.28 to ‐0.01)	5 (4)	Very low ^1,2,3
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting patients compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.13 (1.04 to 1.21)	2 (2)	Very low ^1,2,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.29 (‐0.05 to 0.63)	6 (2)	Very low ^1,2,3
	Categorical measure	0.04 (‐0.21 to 0.12)	11 (8)	Low ^1,2
	Qualitative quote	0 significant study on 3	3 (3)	Very low ^1,2,4
Interventions targeting healthcare professionals compared with usual care for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.08 (0.38 to 2.07)	4 (3)	Very low ^1,2,3,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.11	1 (1)	Very low ^1,2
	Categorical measure	0.05 (0.00 to 0.09)	3 (2)	Low ^2,3
	Qualitative quote	0 significant study on 1	1 (1)	Very low ^2,4
Interventions targeting healthcare professionals compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	‐0.12	1 (1)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting healthcare professionals compared with another intervention targeting healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	‐0.3	1 (1)	Very low ^2,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.20 (‐0.09 to 0.48)	7 (2)	Very low ^1,2,3
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting both patients and healthcare professionals compared with usual care for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	2.83	4 (2)	Very low ^1,2,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant stdy on 1	1 (1)	Very low ^2,4
Patient‐reported SDM measures	Continuous measure	0.16	3 (3)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant study on 2	2 (2)	Very low ^1,2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.42	1 (1)	Very low ^2,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.09 (‐0.06 to 0.73)	5 (3)	Very low ^1,2,3
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant measure on 2	2 (1)	Very low ^2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.06	1 (1)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant study on 1	1 (1)	Very low ^1,2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting both patients and healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	‐0.04	1 (1)	Very low ^1,2
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
* Where studies reported more than one measure for each endpoint, the primary measure (as defined by the authors of the study) or the median measure was abstracted. For categorical measures, we calculated the risk difference between the intervention of interest and the control intervention across various outcomes. Forcontinuous endpoints, we calculated standardized mean difference by dividing the mean score difference of the intervention and comparison groups in each study by the pooled standard deviation estimate for the two groups across various outcomes ** Three studies reported results in more than one type of measure
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Important risk of bias according to EPOC checklist ² Indirectness of evidence ³ Heterogeneity ⁴ Imprecision of the observed effect ⁵ Publication bias

Background

Description of the condition

Shared decision making (SDM) is defined as a process by which a healthcare choice is made by the patient (or significant others, or both) together with one or more healthcare professionals (Charles 1997; Légaré 2011; Towle 1999) and is said to be the crux of patient‐centred care (Weston 2001). Briefly, SDM rests upon knowing and understanding the best available evidence on the risks and benefits across all available options while ensuring that the patient's values are taken into account (Charles 1997; Elwyn 1999; Towle 1999). Although SDM represents a complex set of behaviours that must be achieved by both members of the patient‐healthcare professional dyad (LeBlanc 2009), it is possible to specify behaviours that both parties must adopt for SDM to occur in clinical practice (Frosch 2009; Légaré 2007a). A systematic review of SDM as a concept identified 161 definitions and summarized the key elements in one integrative model of SDM in medical encounters (Makoul 2006). This model identifies nine essential elements that can be translated into various SDM‐related specific behaviours for healthcare professionals during consultations with patients:

define and explain the healthcare problem,
present options,
discuss pros and cons (benefits, risks, costs),
clarify patient values and preferences,
discuss patient ability and self‐efficacy,
present what is known and make recommendations,
check and clarify the patient's understanding,
make or explicitly defer a decision,
arrange follow up.

The notion that the healthcare professional is the only party requiring access to evidence is no longer credible. Instead, SDM assumes that both healthcare professional(s) and patient require access to information about the evidence informing a decision, while understanding and respecting both the patient's values and the healthcare professional's recommendations.

Description of the intervention

A variety of interventions have been designed to change healthcare professionals' behaviour. Based on the Effective Practice and Organisation of Care (EPOC) taxonomy of interventions (EPOC 2008), these may include but are not limited to the distribution of printed educational materials, educational meetings, audit and feedback, reminders, educational outreach visits and patient‐mediated interventions (that is any intervention aimed at changing the performance of healthcare professionals through interactions with patients, or information provided by or to patients). Additionally, in the context of SDM it is possible to identify three overarching categories of implementation intervention: 1) interventions targeting patients, 2) interventions targeting healthcare professionals, and 3) interventions targeting both.

How the intervention might work

Theoretical and empirical evidence about behaviour change in healthcare professionals (Godin 2008) and complex behaviour change (Michie 2009) allows us to make certain hypotheses regarding the mechanisms by which interventions might promote SDM. For example, the distribution of printed educational materials may improve professionals' attitudes towards adopting SDM‐related behaviours by reinforcing the underlying salient beliefs associated with their intention to adopt SDM (Giguère 2012). The training of professionals in SDM through educational meetings may increase professionals' perceptions of self‐efficacy, one of the key determinants of behaviour (Godin 2008). Patient‐mediated interventions such as decision aids have been shown to improve patient knowledge (Stacey 2011), and this may provide patients with more resources with which to engage in the decision‐making process. In turn, the engagement of patients in the decision making process may change the habits of healthcare professionals by enhancing their knowledge of emerging evidence within their area of expertise and by increasing their use of this evidence (Brouwers 2010).

Why it is important to do this review

Policy makers perceive SDM as desirable (Shafir 2012) because: a) patient involvement is accepted as a right (Straub 2008); b) it may reduce the overuse of options not clearly associated with benefits for all; c) it may enhance the use of options clearly associated with benefits for the vast majority of the concerned population; d) it may reduce unwarranted healthcare practice variations (Mulley 2012; Wennberg 2004); and e) it may foster the sustainability of the healthcare system by increasing patient ownership of their own healthcare (Coulter 2006). Nonetheless, SDM has not yet been widely implemented in clinical practice. A systematic review of 33 studies using the Observing Patient Involvement in Decision Making instrument (OPTION) showed low levels of patient‐involving behaviours (Couët 2013).

Objectives

The objective of this review was to determine the effectiveness of interventions to improve healthcare professionals’ adoption of SDM.

To address this objective, we compared each of the three categories of targeted intervention (targeting patients, targeting healthcare professionals, and targeting both) to the same category of targeted intervention, to each of the other categories of targeted intervention, and to usual care. Thus there were nine comparison categories.

Group 1. Interventions targeting patients compared to usual care.

Group 2. Interventions targeting patients compared to other interventions targeting patients.

Group 3. Interventions targeting healthcare professionals compared to usual care.

Group 4. Interventions targeting healthcare professionals compared to interventions targeting patients.

Group 5. Interventions targeting healthcare professionals compared to other interventions targeting healthcare professionals.

Group 6. Interventions targeting both patients and healthcare professionals compared to usual care.

Group 7. Interventions targeting both patients and healthcare professionals compared to interventions targeting patients alone.

Group 8. Interventions targeting both patients and healthcare professionals compared to interventions targeting healthcare professionals alone.

Group 9. Interventions targeting both patients and healthcare professionals compared to other interventions targeting both patients and healthcare professionals.

Methods

Criteria for considering studies for this review

Types of studies

This review considered randomised controlled trials (RCTs) and non‐randomised controlled trials (NRCTs), controlled before and after studies (CBAs) and interrupted time series (ITS) analyses (EPOC 2008). To be included as a CBA, we required the study to have a minimum of two intervention sites and two control sites. For ITS studies, there needed to be a clearly defined point in time when the intervention occurred and at least three data points before and three after the intervention. We considered publications in English and French only for eligible studies that needed data extraction.

Types of participants

In this review, there were two main types of participants. The first type were healthcare professionals, including professionals in training who were responsible for patient care (residents, fellows, and other pre‐licensure healthcare professionals). We defined professionals as having licensure or, in the case of professionals in training, basic pre‐licensure education (for example residents who had a medical degree). The second type were patients, including healthcare consumers and standardized patients. Standardized patients were only deemed to be acceptable participants if the outcome was observer‐reported.

Types of interventions

We included in this review studies that evaluated an intervention designed to increase healthcare professionals' adoption of SDM. We organized interventions into categories using the EPOC taxonomy of interventions (EPOC 2008). Patient decision aids were considered a patient‐mediated intervention since one of their purposes is to foster patients' participation in decisions during the clinical encounter (Stacey 2011).

We considered studies that evaluated patient‐mediated interventions (for example patients' use of patient decision aids in preparation for their consultation or during their consultation with a healthcare professional) only if these studies directly assessed the healthcare professional‐related outcome of interest, that is the professional's adoption of SDM (see Types of outcome measures).

In keeping with the EPOC taxonomy of interventions, we sorted interventions into three categories: interventions targeting patients (for example patient‐mediated interventions), interventions targeting healthcare professionals (for example distribution of printed educational material, an educational meeting, audit and feedback, reminders and educational outreach visits) and interventions targeting both patients and healthcare professionals (for example a patient‐mediated intervention combined with an intervention targeting healthcare professionals). Usual care was the fourth category. This gave us nine comparison categories in total (see Objectives).

Types of outcome measures

In this updated review, we considered not only observer‐based findings but also findings by the patients themselves, presenting a more complete portrait of the impact of interventions on adoption of SDM. We specifically avoided inclusion of healthcare professionals' self‐reported SDM behaviours given that they tend to over‐rate their personal behaviours.

Thus the primary outcomes evaluated by this review were observer‐based outcome measures (OBOM) or patient‐reported outcome measures (PROM) of healthcare professionals' adoption of SDM.

For each eligible study that included the primary outcome of interest, whether OBOM or PROM, we also extracted secondary outcomes. These were measures of patient health outcomes (for example results of a blood test, health‐related quality of life) and other measures reported by healthcare professionals or patients (for example knowledge, attitudes, or satisfaction).

We also extracted potential harms of interventions: a) measures of patient anxiety (from patient health outcomes); b) longer duration of consultations; and c) costs.

Search methods for identification of studies

Electronic searches

An information specialist (S Ratté) developed the search strategies in consultation with the authors.

The SDM component of the search strategy was based on the search strategy developed for a previous systematic review on barriers and facilitators for implementing SDM in clinical practice as perceived by healthcare professionals (Légaré 2008a). Given that the implementation of SDM in clinical practice is a relatively new area of research, we favoured a broad search strategy with high sensitivity as opposed to a very specific search. Searches were conducted at the beginning of August 2012; exact search dates for each database are included in Appendix 1 to Appendix 11.

All databases were searched from their inception to March 2009 for the first review. This update searched for additional literature from 15 March 2009 to August 2012. In addition to our database searches in August 2012, we contacted experts in the field and conducted brief searches in PubMed. By doing so, we identified a number of studies published later than August 2012. We included articles published in English and French only.

The following electronic databases were searched for primary studies:

Cochrane Central Register of Controlled Trials (CENTRAL), part of The Cochrane Library (www.thecochranelibrary.com) (August 2012);
Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register;
MEDLINE via Pubmed (1950 to August Week 1, August 2012) using OvidSP;
EMBASE (1980 to Week 29 2012) via OvidSP;
CINAHL (Cumulative Index to Nursing and Allied Health Literature) (1981 to August 2012) via EBSCOhost;
NHS Economic Evaluation Database, Centre for Reviews and Dissemination (CRD);
Health Technology Assessment Database, CRD;
PsycINFO (1806 to Week 1 August2012).

Our database searches, in all the databases above, were limited by publication year and month (March 2009 to August 2012). For PsycINFO, we were unable to place strict date limitations and manually excluded citations retrieved outside this date range.

Searching other resources

Trial registries

ClinicalTrials.gov, US National Institutes of Health (NIH) at http://clinicaltrials.gov/ (Week 2 January 2013).

Others

We also:

handsearched proceeding so the a) International Conference on Shared Decision Making (years 2003, 2005, 2007, 2009, 2011) and b) the annual meetings of the Society for Medical Decision Making (years 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011). Although we intended to search the European Association for Communication in Healthcare (EACH), we were unable to obtain detailed information either online or as a paper copy;

reviewed reference lists of all included studies, relevant systematic reviews and primary studies;

contacted authors of relevant studies or reviews to clarify reported published information and to seek unpublished data. Through this process we identified a number of papers published after August 2012;
included results from searches conducted for a review focused on patient‐reported outcomes (Légaré 2012a).

Data collection and analysis

Selection of studies

At least two review authors (MJC, MS, PZ, ST) independently screened each title and abstract to find studies that met the inclusion criteria. We retrieved full text copies of all studies that might be relevant or for which the inclusion criteria were not clear in the title or abstract. In this update, when more than one publication described the same study but each presented new and complementary data we included them all. Any disagreements on the selection were resolved by discussion among the review authors (FL, DS).

Data extraction and management

To extract data, we designed a form derived from the EPOC Review Group data collection checklist (EPOC 2008). At least two review authors (MJC, MS, PZ, ST) independently extracted data from eligible studies. We reached consensus about discrepancies, and any disagreement was adjudicated by FL and DS. We entered data into Review Manager software (RevMan 5) and checked for accuracy. When information regarding any of the above was unclear, we attempted to contact the authors of the original reports to provide further details.

In addition to EPOC's standardized data collection checklist, we extracted the following characteristics of the settings and interventions.

Level of care: primary or specialized care (as defined by the type of provider).

Setting of care: ambulatory or non‐ambulatory care (i.e. hospitalised patients in acute‐care or long‐term care facilities).

Conceptual or theoretical underpinnings of the intervention (i.e. authors stated in their paper that the intervention was based on a theory or at least referred to a theory).

Barriers assessment (i.e. authors stated in their paper that a barriers assessment was conducted and the intervention was designed to overcome identified barriers).

Number of components included in the intervention based on the EPOC taxonomy (when a barriers assessment was mentioned, such as the one above, it was considered a component of the intervention).

For ongoing studies, when available we described the primary outcome, the research question(s), the methods and the outcome (see Ongoing studies).

Assessment of risk of bias in included studies

At least two review authors (MJC, MS, PZ, ST) independently assessed the risk of bias in each included study using the criteria outlined in the EPOC Review Group data collection checklist for studies with a separate control group (EPOC 2008) and the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) for ITS designs. We resolved any disagreement by discussion with FL. We assessed each quality criterion as 'Done', 'Not done', or 'Unclear', as recommended by the EPOC Review Group. Then we transformed these three scores into 'Low risk', 'High risk', and 'Unclear' when we entered the data into RevMan 5. The seven standard criteria as suggested for all RCTs and CBA studies are listed below.

1) Concealment of allocation (protection against selection bias).

2) Follow‐up of professionals (protection against exclusion bias).

3) Follow‐up of patients or episodes of care.

4) Blinded assessment of primary outcome(s) (protection against detection bias).

5) Baseline measurement.

6) Reliable primary outcome measure(s).

7) Protection against contamination.

For PROM measures, the criterion 'reliable primary outcome' was not applicable because of the nature of the outcome.

Measures of treatment effect

We structured data analysis using statistical methods developed for EPOC by Grimshaw and colleagues (Grimshaw 2004). For each study, we reported results for categorical and continuous primary outcomes separately and in natural units. For categorical measures, we calculated the difference in risk between the intervention of interest and the control intervention. We calculated standardized mean difference for continuous measures by dividing the mean score difference of the intervention and comparison groups in each study by the pooled estimate standard deviation for the two groups. When possible, for categorical and continuous outcomes we constructed 95% confidence intervals (CIs) to compare groups before and after the intervention, according to the recommendations in RevMan 5. The absence of a '0' value in the CI indicated that the baselines differed or that the intervention had a statistically significant positive effect compared to the control intervention or to usual care. When the baseline was different in the two groups, we used the size of the difference and its associated standard error to compare them. If information was not available for the standard error, we extracted a qualitative quote from the primary study on the effectiveness of the intervention and on confounding factors, if available. When no baseline was reported, we considered groups to be similar prior to the intervention. For the analysis, the studies were divided into nine categories of intervention, which were applied to both PROM and OBOM outcomes (that is nine categories for each). Where studies reported more than one primary outcome in the same category, the median measure was abstracted. For each category of intervention and outcome for which a significant effect on our main outcome of interest (healthcare professionals' adoption of SDM) was observed, we reported the median of the standardized mean difference (or risk difference) and a range. We considered a standardized mean difference of 0.2 as small, 0.5 as medium, and 0.8 as large (Cohen 1988). For studies in which the quantitative data were absent or insufficient to make the calculation, and if no replies were obtained from the authors, we reproduced the qualitative data as presented in the article. A meta‐analysis would have been performed if the nature of the primary outcome of the various comparisons had been similar.

Summary of findings table

The quality of evidence was evaluated according to GRADE for the 18 categories of intervention and outcome. For each category, conclusions were categorized into four ratings: high quality (further research is very unlikely to change our confidence in the estimate of effect), moderate quality (further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate), low quality (further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate), and very low quality (we are very uncertain about the estimate). This rating was downgraded if it met one of the five following criteria.

1) Important risk of bias according to the EPOC checklist: quality of evidence downgraded if the EPOC 'unclear risk' or 'high risk' risk of bias criteria were applicable.

2) Indirectness of evidence: quality of evidence was downgraded if it met one of four further criteria, i) a difference between the population of interest and participants in the studies (applicability); ii) a difference between the intervention of interest and interventions in the studies (applicability); iii) the use of surrogate endpoints to measure SDM (PROM and OBOM are each prone to particular biases and have their own strengths and weaknesses, we can thus rate PROM and OBOM as being of even quality in the context of a process experienced by the patient); and iv) no head‐to‐head comparisons were made or comparisons between two or more interventions of interest (e.g. multifaceted intervention compared to another multifaceted intervention).

3) Inconsistency: quality of evidence was downgraded according to the heterogeneity index (I² > 30%). This criterion was evaluated separately for categorical and continuous measures. It was not appropriate for qualitative statements.

4) Imprecision of the observed effect: quality of evidence was downgraded if the sample size in a study was insufficient or if there was a qualitative statement.

5) Publication bias: publication bias was tested using a funnel plot.

Quality of evidence was upgraded in three cases: 1) demonstration of a strong association in a well‐executed observational study; 2) all plausible biases from observational or randomised studies may have been working to underestimate an apparent intervention effect; and 3) there was evidence of a gradient.

Unit of analysis issues

We included cluster‐randomised trials in the analyses along with individually randomised trials. Comparisons that randomise or allocate clusters (groups of healthcare professionals or organizations) but do not account for clustering during the analysis have potential unit of analysis errors that can produce artificially significant P values and overly narrow CIs (Ukoumunne 1999). Therefore, when possible, we contacted primary authors for missing information and attempted to re‐analyse studies with potential unit of analysis errors. When missing information was unavailable from the study authors, we only reported the point estimate.

Assessment of heterogeneity

To explore heterogeneity, we designed tables that compared the studies' standardized mean differences and their risk differences. We considered the following variables as potential sources of heterogeneity to explain variations in the results of the included studies: type of intervention; characteristics of the intervention (for example duration); clinical setting (primary care versus specialized care); type of healthcare professional (physicians versus other healthcare professionals); level of training of healthcare professionals (for example healthcare professionals in training versus those in practice); and type of outcome (continuous or categorical).

Results

Description of studies

Results of the search

For this update, we found 11,757 potentially relevant citations; for previous versions of this review, we screened 9035 citations (Légaré 2012a). This provided a total of 20,792 potentially relevant citations that we considered, of which 7119 were excluded prior to review of the full publications (4374 were duplicates and 2745 did not match the date range for our update). Of the remaining citations, we retrieved 436 full publications for a more detailed screening. From these, we excluded another 397 citations based on the identified inclusion criteria. This resulted in 39 studies. For more details, see Figure 1.

Figure 1

Flow diagram of Cochrane update on interventions for improving the adoption of shared decision making by healthcare professionals (up to 31 December 2012).

Included studies

We included 39 studies in this review. This current version updates our 2010 version (Légaré 2010), which included five OBOM studies and another systematic review of 21 PROM studies (Légaré 2012). Two studies (Butow 2004; Elwyn 2004) were in both reviews. Three studies were excluded: one was excluded because it reported "preferred role during the consultation" (that is the role the patient would like to play) and not 'assumed role' (the role actually played, the outcome relevant to our review) (Brown 2004). Two more were excluded because they reported the more vaguely‐worded "active patient" but not 'assumed role' (Kopke 2009; Whelan 2003).

This updated search added 18 new studies to the 21 original studies that were included, for a total of 39 studies (Bernhard 2011; Cooper 2011; Deen 2012; Deinzer 2009; Fossli 2011; Hess 2012; Landrey 2012; Légaré 2012; Leighl 2011; Montori 2011; Mullan 2009; Murray 2010; Myers 2011; Raynes‐Greenow 2010; Roter 2012; Schroy 2011; Shepherd 2011; van Peperstraten 2010).

We identified a further 20 RCTs as ongoing studies (see Characteristics of ongoing studies).

All studies in this review were RCTs except for one, which was a non‐randomised controlled trial (NRCT) (Deinzer 2009). Among the RCTs, seven were cluster‐randomised trials (Elwyn 2004; Hamann 2007; Haskard 2008; Légaré 2012; Loh 2007; O'Cathain 2002; Wetzels 2005).

Characteristics of settings and participants

Interventions targeting patients (18 studies)

Of the 18 studies of interventions targeting patients, eight were conducted in the United States (Deen 2012; Dolan 2002; Krist 2007; Landrey 2012; Montori 2011; Nannenga 2009; Schroy 2011; Street 1995), three in Canada (Davison 1997; Deschamps 2004; Lalonde 2006), two in Germany (Kasper 2008; Vodermaier 2009), two in the Netherlands (Stiggelbout 2008; van Peperstraten 2010), two in Australia (Butow 2004; Raynes‐Greenow 2010) and one in the United Kingdom (Murray 2001). With regard to care settings, eight out of 18 trials were conducted in primary care (Deschamps 2004; Dolan 2002; Krist 2007; Lalonde 2006; Landrey 2012; Montori 2011; Murray 2001; Schroy 2011) and nine in specialized care (Butow 2004; Davison 1997; Kasper 2008; Nannenga 2009; Raynes‐Greenow 2010; Stiggelbout 2008; Street 1995; van Peperstraten 2010; Vodermaier 2009). One study was carried out both in primary and specialized care (Deen 2012). All studies were conducted and recruited patients in an ambulatory setting except one, which was in non‐ambulatory care (Vodermaier 2009).

Although there was a total of 236 reported participating healthcare professionals, this number under‐represented the total number of professionals as eight studies did not report the total number of healthcare professionals involved in the study (Deen 2012; Deschamps 2004; Kasper 2008; Lalonde 2006; Murray 2001; Raynes‐Greenow 2010; van Peperstraten 2010; Vodermaier 2009). The minimum number of healthcare professionals reported was two (Davison 1997) and the maximum number was 60 (Montori 2011).

All studies reported the number of patients involved in the study. A total of 4055 patients were enrolled in the interventions, with a minimum of 26 (Lalonde 2006) and a maximum of 666 (Schroy 2011). The most common clinical condition was cancer (seven studies) (Butow 2004; Davison 1997; Dolan 2002; Krist 2007; Schroy 2011; Street 1995; Vodermaier 2009).

Interventions targeting healthcare professionals (eight studies)

Of the eight studies of interventions targeting healthcare professionals, two were conducted in Canada (Légaré 2012; Stacey 2006), two in the United Kingdom (Elwyn 2004; O'Cathain 2002), one in Australia (Shepherd 2011), one in Germany (Krones 2008 (ARRIBA‐Herz)) and one in Norway (Fossli 2011). One study was conducted with international collaboration, specifically Australia, New Zealand, Switzerland, Germany and Austria (Bernhard 2011). Seven studies were conducted in primary care (Elwyn 2004; Fossli 2011; Krones 2008 (ARRIBA‐Herz); Légaré 2012; O'Cathain 2002; Shepherd 2011; Stacey 2006) and one in specialized care (Bernhard 2011). All eight trials recruited patients in ambulatory care settings.

Although a total of 593 participating healthcare professionals were reported, this number under‐represented the total number of professionals as one study did not report the total number of healthcare professionals involved in the study (O'Cathain 2002). The minimum number of healthcare professionals reported was 21 (Elwyn 2004) and the maximum number was 270 (Légaré 2012).

Two studies (Shepherd 2011; Stacey 2006) used simulated patients facing different clinical situations: depression (Shepherd 2011), gall bladder disorders, attention deficit hyperactivity disorder, amniocentesis, and allergy (Stacey 2006). Among the six studies without standardized patients, one did not report the number of patients in the study (Fossli 2011) and five studies (Bernhard 2011; Elwyn 2004; Krones 2008 (ARRIBA‐Herz); Légaré 2012; O'Cathain 2002) had a total of 13,707 patients enrolled (minimum 694 (Bernhard 2011) and maximum 10,070 (O'Cathain 2002) patients per study). The five studies that reported numbers of patients involved diverse clinical conditions: breast cancer (Bernhard 2011), cardiovascular disease (Krones 2008 (ARRIBA‐Herz)), acute respiratory infection (Légaré 2012), maternity care (O'Cathain 2002), and multi‐clinical conditions of non‐valvular atrial fibrillation or prostatism or menorrhagia or menopausal symptoms (Elwyn 2004). Most interventions enrolled both male and female patients, except for two studies (Bernhard 2011; O'Cathain 2002) which involved females only.

Interventions targeting both patients and healthcare professionals (13 studies)

Of the 13 studies of interventions targeting both patients and healthcare professionals, six were conducted in the United States (Cooper 2011; Haskard 2008; Hess 2012; Mullan 2009; Myers 2011; Roter 2012), four in Germany (Bieber 2006; Deinzer 2009; Hamann 2007; Loh 2007), one in the Netherlands (Wetzels 2005) and one in Canada (Murray 2010). One study was conducted with international collaboration, specifically Australia and Canada (Leighl 2011). Care settings were divided between primary care (seven studies) (Cooper 2011; Haskard 2008; Loh 2007; Mullan 2009; Myers 2011; Roter 2012; Wetzels 2005) and specialized care (six studies) (Bieber 2006; Deinzer 2009; Hamann 2007; Hess 2012; Leighl 2011; Murray 2010). Ten trials were conducted in ambulatory care settings (Bieber 2006; Cooper 2011; Haskard 2008; Hess 2012; Leighl 2011; Loh 2007; Mullan 2009; Myers 2011; Roter 2012; Wetzels 2005), two in non‐ambulatory care settings (Deinzer 2009; Hamann 2007) and one was set in both ambulatory and non‐ambulatory care settings (Murray 2010). A total of 571 healthcare professionals took part in these studies, ranging from 10 (Bieber 2006) to 156 (Haskard 2008) per study.

One study (Murray 2010) used five simulated patients facing care related to end of life treatment. Among the 12 studies without standardized patients, a total of 5474 patients were enrolled, with a minimum of 85 (Mullan 2009) and a maximum of 2196 (Haskard 2008). The most common clinical condition was hypertension (two studies) (Cooper 2011; Deinzer 2009), and multi‐clinical conditions (two studies) (Haskard 2008; Wetzels 2005). Most interventions enrolled both male and female patients, except for one study (Myers 2011) which involved males only.

In summary, of the 39 studies included in the review, the three most represented countries were the United States (14 studies), Germany (seven studies) and Canada (six studies). Only two of the 39 studies were conducted with international collaborations: Canada and Australia; and Australia; New Zealand, Switzerland, Germany and Austria. The setting was primary care in 22 studies, with only one in both primary and specialized care. More than half (53.8%) of the healthcare professionals involved in the studies were licensed and the three most frequent clinical conditions studied were cancer (nine studies), cardiovascular disease (eight studies) and multiple conditions (four studies).

Characteristics of interventions and comparisons

Characteristics of interventions

For details, see Characteristics of included studies.

Several studies had more than two arms (Cooper 2011; Deen 2012; Haskard 2008; Krist 2007; Raynes‐Greenow 2010; Schroy 2011). One study presented a RCT with two‐by‐two factorial design (Cooper 2011) and four arms: 1) a patient‐mediated intervention and an educational meeting; 2) an educational meeting; 3) a patient‐mediated intervention; and 4) control (patients and providers receiving minimal intervention). One study presented an RCT with four arms (Deen 2012): 1) a decision aid and patient activation; 2) a decision aid; 3) patient activation; and 4) control (doctor's visit). One study presented a cluster‐RCT (Haskard 2008) with four arms. The first arm (training of healthcare professional and patient) consisted of a multifacted intervention (an educational meeting, distribution of educational materials, and a patient‐mediated intervention). The second arm (training of healthcare professional only) consisted of a multifaceted intervention (an educational meeting and the distribution of educational materials). The third arm (patient training only) consisted of a single intervention (patient‐mediated intervention). The fourth arm (control group) consisted of usual care. One study presented an RCT (Krist 2007) with three arms: 1) mailed paper version of a decision aid; 2) Internet‐based decision aid; and 3) control. One study presented an RCT (Raynes‐Greenow 2010) with three arms: 1) a decision aid (booklet and audio); 2) a decision aid (booklet); and 3) a pamphlet. One study presented an RCT (Schroy 2011) with three arms: 1) a decision aid and decision guidance; 2) a decision aid only; and 3) control decision aid. Thus there was an overlap of studies between comparison types (objective).

Interventions targeting patients

Eight studies compared interventions targeting patients with usual care (Cooper 2011; Deen 2012; Haskard 2008; Krist 2007; Landrey 2012; Murray 2001; van Peperstraten 2010; Vodermaier 2009). Of these, three studies compared single interventions to usual care (Landrey 2012; Murray 2001; Vodermaier 2009), one compared multifaceted interventions to usual care (van Peperstraten 2010), and four studies (Cooper 2011; Deen 2012; Haskard 2008; Krist 2007) compared patient‐mediated interventions to usual care (RCTs with several arms).

Fourteen studies presented comparisons of interventions targeting the patient with other interventions targeting the patient (Butow 2004; Davison 1997; Deen 2012; Deschamps 2004; Dolan 2002; Kasper 2008; Krist 2007; Lalonde 2006; Montori 2011; Nannenga 2009; Raynes‐Greenow 2010; Schroy 2011; Stiggelbout 2008; Street 1995). Of these, eight studies compared a single intervention to another single intervention (Butow 2004; Davison 1997; Dolan 2002; Kasper 2008; Montori 2011; Nannenga 2009; Stiggelbout 2008; Street 1995), one study compared a multifaceted intervention to a single intervention (Deschamps 2004), one study compared a multifaceted intervention to another multifaceted intervention (Lalonde 2006), and four studies had arms comparing a patient‐mediated intervention to another patient‐mediated intervention (Deen 2012; Krist 2007; Raynes‐Greenow 2010; Schroy 2011).

Interventions targeting healthcare professionals

Seven studies compared interventions targeting the healthcare professional with usual care (Bernhard 2011; Cooper 2011; Fossli 2011; Légaré 2012; O'Cathain 2002; Shepherd 2011; Stacey 2006). Of these, two studies presented interventions containing educational meetings, audit and feedback, and distribution of educational materials (Bernhard 2011; Fossli 2011); two studies presented interventions using educational meetings and distribution of educational materials (Légaré 2012; O'Cathain 2002); and one presented the distribution of educational materials with educational meetings, audit and feedback, and barriers assessment, as part of a multifaceted intervention (Stacey 2006). We also found one study that compared a single intervention (educational outreach visit) to usual care (Shepherd 2011), and one study had an arm that compared an educational meeting to usual care (Cooper 2011).

One study compared an intervention targeting the healthcare professional with one targeting the patient (Cooper 2011). This study presented an arm comparing a educational meeting with a patient‐mediated intervention.

Two studies compared interventions targeting the healthcare professional with other interventions targeting the healthcare professional (Elwyn 2004; Krones 2008 (ARRIBA‐Herz)). Of these, one study compared a multifaceted intervention (educational meeting and audit and feedback focusing on SDM skills) to another multifaceted intervention (educational meetings and audit and feedback focusing on risk communication skills) (Elwyn 2004), and one study compared a multifaceted intervention (educational meeting, audit and feedback, distribution of educational material, and an educational outreach component) to a single intervention (educational meeting) (Krones 2008 (ARRIBA‐Herz)).

Interventions targeting both patients and healthcare professionals

Eight studies compared an intervention targeting patients and healthcare professionals with usual care (Cooper 2011; Hamann 2007; Haskard 2008; Hess 2012; Leighl 2011; Loh 2007; Murray 2010; Wetzels 2005). Of these, four studies presented interventions that used educational meetings and patient‐mediated interventions (Hamann 2007; Hess 2012; Leighl 2011; Loh 2007); one study presented an intervention that used educational meetings, distribution of educational materials, audit and feedback, barriers assessment, and educational outreach visits (Murray 2010); and one study presented a patient‐mediated intervention using educational outreach visits (Wetzels 2005). One study presented an arm with an intervention that used a combination of a patient‐mediated intervention, distribution of educational material and educational meetings (Haskard 2008); and one study presented a patient‐mediated intervention and an educational meeting (Cooper 2011).

Four studies compared interventions targeting both patients and healthcare professionals with interventions targeting patients alone (Bieber 2006; Cooper 2011; Deinzer 2009; Mullan 2009). Of these, three studies compared educational meetings and patient‐mediated interventions with patient‐mediated interventions alone (Bieber 2006; Deinzer 2009; Mullan 2009), and one study presented an arm comparing an educational meeting and patient‐mediated intervention with a patient‐mediated intervention alone (Cooper 2011).

Two studies compared interventions targeting both patients and healthcare professionals with interventions targeting healthcare professionals alone (Cooper 2011; Roter 2012). Of these, one study compared patient‐mediated interventions and the distribution of educational materials with the distribution of educational materials alone (Roter 2012), and one study presented an arm comparing educational meetings and patient‐mediated interventions with educational meetings alone (Cooper 2011).

One study compared an intervention targeting both patients and healthcare professionals with another intervention targeting both patients and healthcare professionals (Myers 2011). This study compared a multifaceted intervention including a patient‐mediated intervention and reminders with another multifaceted intervention also including a patient‐mediated intervention and reminders.

Conceptual framework and barriers assessment

Interventions targeting patients (18 studies)

Among the studies of interventions targeting patients, six studies explicitly referred to a conceptual framework or a theory to justify their intervention (Butow 2004; Davison 1997; Raynes‐Greenow 2010; Schroy 2011; Stiggelbout 2008; van Peperstraten 2010). Three studies (Raynes‐Greenow 2010; Schroy 2011; van Peperstraten 2010) referred to the Ottawa Decision Support Framework, one (Davison 1997) referred to the Empowerment Model by Conger and Kanungo, one (Stiggelbout 2008) to the Markov Model, and one (Butow 2004) did not provide detailed information.

One of the studies of interventions targeting patients reported performance of a barriers assessment (van Peperstraten 2010).

Interventions targeting healthcare professionals (eight studies)

Among the studies of interventions targeting healthcare professionals, four studies explicitly referred to a conceptual framework or a theory to justify their intervention (Elwyn 2004; Fossli 2011; Légaré 2012; Stacey 2006). One study (Stacey 2006) referred to the Ottawa Decision Support Framework, one (Elwyn 2004) referred to a model of interpersonal interaction, one (Fossli 2011) referred to the Four Habit Model, and one study (Légaré 2012) referred to the Theory of Planned Behaviour.

Of the eight studies of interventions targeting healthcare professionals, one (Stacey 2006) reported the performance of a barriers assessment and based its interventions on identified barriers.

Interventions targeting both patient and healthcare professionals (13 studies)

Five of the studies of interventions targeting both patients and healthcare professionals (Haskard 2008; Loh 2007; Murray 2010; Roter 2012; Wetzels 2005) referred to a conceptual framework or a theory to justify their interventions. One study (Murray 2010) referred to the Ottawa Decision Support Framework, one (Haskard 2008) referred to the 4E Model (Engage, Empathize, Educate and Enlist), one study (Roter 2012) referred to the LEAPS (Listen, Educate, Assess, Partner and Support) framework, one (Wetzels 2005) to the SWOT analysis (Strengths, Weaknesses, Opportunities and Threats), and one (Loh 2007) did not provide detailed information.

Of these studies, one (Murray 2010) reported the performance of a barriers assessment and based its interventions on identified barriers.

In summary, 15 studies out of the 39 included in this review used a conceptual framework. The Ottawa Decision Support Framework was the most cited framework. Lastly, only three based their interventions on barriers assessments.

Characteristics of outcomes

Characteristics of primary outcomes

Patient‐reported outcome measures (PROM)

Among the 16 PROM studies, 14 unique scales or subscales were used to measure the adoption of SDM by healthcare professionals from a patient perspective. Patient‐reported outcomes were predominantly represented by the 'perceived level of control in decision making' or 'assumed role during the consultation' (adaptation of the Control Preference Scale) in 15 studies (Butow 2004; Davison 1997; Deschamps 2004; Dolan 2002; Kasper 2008; Krist 2007; Landrey 2012; Légaré 2012; Leighl 2011; Murray 2001; O'Cathain 2002; Raynes‐Greenow 2010; Stiggelbout 2008; Street 1995; Vodermaier 2009). Other tools used were: COMRADE (Deinzer 2009; Elwyn 2004; Hamann 2007; Wetzels 2005), and the Man‐Son‐Hing Instrument or the Patient Participation Satisfaction scale (PPS) (Krones 2008 (ARRIBA‐Herz); Loh 2007; Vodermaier 2009). There were also 11 unique scales or subscales used in the studies analysed. For more details, see Characteristics of included studies.

Observer‐based outcome measures (OBOM)

Among the three OBOM studies, nine unique scales or subscales were used to measure the adoption of SDM by healthcare professionals from an observer‐based perspective. The observer‐based outcomes were predominantly represented by the OPTION scale in six studies (Elwyn 2004; Hess 2012; Montori 2011; Mullan 2009; Nannenga 2009; Shepherd 2011), and the Decision Support Analysis Tool (DSAT) in two studies (Murray 2010; Stacey 2006). There were also seven unique scales or subscales used in the studies analysed. For more details, see Characteristics of included studies.

It was noteworthy that the primary outcome of only five out of the 39 studies included in this review was the same as the primary outcome of this review, that is a measure of healthcare professionals' adoption of SDM (Dolan 2002; Elwyn 2004; Krist 2007; O'Cathain 2002; Wetzels 2005).

Characteristics of secondary outcomes

Risk of bias in included studies

Interventions targeting patients compared with usual care

Among the seven PROM studies (Cooper 2011; Deen 2012; Krist 2007; Landrey 2012; Murray 2001; van Peperstraten 2010; Vodermaier 2009), all had at least one unclear risk out of the seven risk of bias criteria. Four (Deen 2012; Krist 2007; Murray 2001; van Peperstraten 2010) studies had one high‐risk bias and three (Cooper 2011; Landrey 2012; Vodermaier 2009) had two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of the evidence, in three studies information reported about participants was inadequate (Deen 2012; Murray 2001; Vodermaier 2009) and in one study the participants were couples (van Peperstraten 2010) and therefore not comparable to the other study populations. The interventions varied from one study to another. In one study (Cooper 2011) comparisons were indirect. In the four studies using continuous measures of SDM (Cooper 2011; Deen 2012; van Peperstraten 2010; Vodermaier 2009) the results reported were inconsistent. In the four studies using categorical measures of SDM (Krist 2007; Landrey 2012; Murray 2001; Vodermaier 2009) results reported were inconsistent.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary for each included study.

There was one OBOM study (Haskard 2008) which had at least one unclear risk out of the seven risk of bias criteria, and no high‐risk bias (see Figure 2 and Figure 3). Regarding indirectness of the evidence, the only problematic criterion was intervention variability. There was publication bias in the OBOM studies with continuous outcomes.

Interpretation of results for this comparison needed to consider the heterogeneity across studies and the fact that all studies had potential bias from inadequate protection against contamination.

Interventions targeting patients compared with other interventions targeting patients

Among the 12 PROM studies (Butow 2004; Davison 1997; Deen 2012; Deschamps 2004; Dolan 2002; Kasper 2008; Krist 2007; Lalonde 2006; Raynes‐Greenow 2010; Schroy 2011; Stiggelbout 2008; Street 1995), 10 (Butow 2004; Davison 1997; Deen 2012; Deschamps 2004; Krist 2007; Lalonde 2006; Schroy 2011; Stiggelbout 2008; Street 1995) had at least one unclear risk out of the seven risk of bias criteria. Eight studies (Davison 1997; Deen 2012; Dolan 2002; Kasper 2008; Krist 2007; Lalonde 2006; Stiggelbout 2008; Street 1995) had one high‐risk bias and four (Butow 2004; Deschamps 2004; Raynes‐Greenow 2010; Schroy 2011) had two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of the evidence, in four studies there was inadequate information about participants (Deen 2012; Kasper 2008; Lalonde 2006; Raynes‐Greenow 2010). The interventions varied from one study to another. In two studies (Deschamps 2004; Lalonde 2006) comparisons reported were indirect. Two studies (Deen 2012; Schroy 2011) used continuous measures of SDM and their results were inconsistent. Eight studies (Butow 2004; Davison 1997; Deschamps 2004; Dolan 2002; Kasper 2008; Krist 2007; Raynes‐Greenow 2010; Stiggelbout 2008) used categorical measures of SDM and their results were consistent. Three studies reported qualitative statements (Butow 2004; Lalonde 2006; Street 1995) and were imprecise as to the observed effect.

Of the two OBOM studies (Montori 2011; Nannenga 2009), one (Nannenga 2009) had at least one unclear risk out of the seven risk of bias criteria. Both studies had one high‐risk bias (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, the only problematic criterion was that the intervention varied from other studies. The two studies used continuous measures of SDM and the results reported were consistent. There was publication bias in the OBOM studies with continuous outcomes.

Interpretation of results for this comparison needed to consider the heterogeneity across the types of patient‐mediated interventions and the fact that all studies had potential bias from inadequate protection against contamination.

Interventions targeting healthcare professionals compared with usual care

Among the four PROM studies (Bernhard 2011; Cooper 2011; Légaré 2012; O'Cathain 2002) all reported at least one unclear risk out of the seven risk of bias criteria. One study (Cooper 2011) reported two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, all studies reported on similar populations, but the intervention varied from one study to another. In one study (Cooper 2011) the comparisons were indirect. One study (Cooper 2011) used a continuous measure of SDM. Two studies (Légaré 2012; O'Cathain 2002) used categorical measures of SDM and the results were inconsistent. One study reported qualitative statements (Bernhard 2011) and was imprecise as to the observed effect.

Among the three OBOM studies (Fossli 2011; Shepherd 2011; Stacey 2006) all had at least one unclear risk out of the seven risk of bias criteria. One study had one high‐risk bias (Stacey 2006) and one study had two high‐risk biases (Fossli 2011) (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, two studies (Shepherd 2011; Stacey 2006) used standardized patients and in one study (Fossli 2011) there was inadequate information about the participants. The interventions varied from one study to another. There were no indirect comparisons in these studies. The three studies used continuous measures, their results were inconsistent, and they were imprecise as to the observed effect because of small sample size. There was publication bias in these studies.

Interpretation of results for this comparison needed to consider that half of the studies were small and there was heterogeneity across the types of population included.

Interventions targeting healthcare professionals compared with another interventions targeting patients

One study used PROM (Cooper 2011). This study had at least one unclear risk out of the seven risk of bias criteria and two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, the quality of evidence was downgraded because: 1) the intervention varied from one study to another, and 2) the comparisons were indirect.

Interpretation of results for this comparison needed to recognize that findings were based on only one highly biased study.

Interventions targeting healthcare professionals compared with other interventions targeting healthcare professionals

In both PROM studies (Elwyn 2004; Krones 2008 (ARRIBA‐Herz)) there was at least one unclear risk out of the seven risk of bias criteria. One study (Krones 2008 (ARRIBA‐Herz)) had two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, both studies reported on similar populations. The intervention varied between studies. There were indirect comparisons in one study (Elwyn 2004). Both studies used continuous measures and results were inconsistent.

In the one OBOM study (Elwyn 2004) there was least one unclear risk out of the seven risk of bias criteria and no high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, the quality of evidence was downgraded because: 1) the intervention varied from one study to another, and 2) the comparisons were indirect. This study used continuous measures of SDM and results were imprecise as to the observed effect because of the small sample size. There was publication bias in the OBOM studies with continuous outcomes.

Interpretation of results for this comparison needed to consider the significant findings from one highly biased study due to problems with follow‐up of professionals and baseline measurement.

Interventions targeting both patients and healthcare professionals compared with usual care

All five PROM studies (Cooper 2011; Hamann 2007; Leighl 2011; Loh 2007; Wetzels 2005) had at least one unclear risk out of the seven risk of bias criteria. Three studies (Hamann 2007; Leighl 2011; Wetzels 2005) had one high‐risk bias, and two studies (Cooper 2011; Loh 2007) had two high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, in two studies there was inadequate information about participants (Hamann 2007; Wetzels 2005). The intervention varied from one study to another. Comparisons in one study (Cooper 2011) were indirect. Three studies (Cooper 2011; Hamann 2007; Wetzels 2005) used continuous measures of SDM and their results were consistent. Two studies reported qualitative statements (Leighl 2011; Loh 2007) and were imprecise as to the observed effect.

All three OBOM studies (Haskard 2008; Hess 2012; Murray 2010) had at least one unclear risk out of the seven risk of bias criteria. One study (Hess 2012) had one high‐risk bias (see Figure 2 and Figure 3). Regarding criteria for evaluating the indirectness of evidence, the intervention varied from one study to another. There was publication bias in the two OBOM studies with continuous outcomes (Haskard 2008; Hess 2012). One study (Murray 2010) reported qualitative statements and was imprecise as to the observed effect.

This comparison group had the most homogenous studies. However, interpretation of results needed to consider the small number of studies and the presence of some methodological bias.

Interventions targeting both patients and healthcare professionals compared with interventions targeting patients

All four PROM studies (Bieber 2006; Cooper 2011; Deinzer 2009; Mullan 2009) had at least one unclear risk out of the seven risk of bias criteria. One study (Bieber 2006) had one high‐risk bias and one had two (Cooper 2011) (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, all studies reported on similar populations. The intervention varied from one study to another. Comparisons in all studies were indirect. Three studies (Bieber 2006; Cooper 2011; Mullan 2009) used continuous measures of SDM and their results were inconsistent. One study (Deinzer 2009) reported qualitative statements and was imprecise as to the observed effect.

The one OBOM study (Mullan 2009) had at least one unclear risk out of the seven risk of bias criteria and no high‐risk biases (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, two criteria were problematic: 1) the interventions varied, and 2) comparisons were indirect. This study used continuous measures of SDM and had a small sample size. There was publication bias in the OBOM studies with continuous outcomes.

Interpretation of results for this comparison needed to consider the heterogeneity across studies and the fact that most studies had multiple arms.

Interventions targeting both patients and healthcare professionals compared with interventions targeting only healthcare professionals

Both studies using patient‐reported outcome measures (Cooper 2011; Roter 2012) reported at least one unclear risk out of the seven risk of bias criteria. There was one high‐risk bias in one study (Roter 2012) and two in the other (Cooper 2011) (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, only two criteria were problematic: the interventions varied, and comparisons were indirect. One study (Cooper 2011) reported continuous measures of SDM. One study (Roter 2012) reported qualitative statements and was imprecise as to the observed effect.

Interpretation of results for this comparison needed to consider that findings were based on only two highly biased studies.

Interventions targeting both patients and healthcare professionals compared with other interventions targeting both patients and healthcare professionals

One OBOM study (Myers 2011) had at least one unclear risk out of the seven risk of bias criteria and one high‐risk bias (see Figure 2 and Figure 3). Regarding evaluation of the indirectness of evidence, two criteria were problematic: 1) the interventions varied, and 2) comparisons were indirect.

None of the included studies were exempt from bias and there was a publication bias for OBOM and PROM studies with continuous data; there appeared to be a lack of published studies with negative results on a continuous score. No publication bias was found in PROM studies with categorical measures (only one OBOM study used categorical measures). For more details, see Figure 4; Figure 5 and Figure 6. As the funnel plot showed there were few negative OBOM studies, therefore positive OBOM studies could have been over‐represented in our review.

Figure 4

Patient‐reported outcome (categorical measure).

Figure 5

Observer‐based outcome (continuous measure).

Figure 6

Patient‐reported outcome measure (continuous measure).

Interpretation of results for this comparison needed to consider that findings were based on only one highly biased study.

Effects of interventions

See: Summary of findings for the main comparison

Primary outcome

Interventions targeting patients compared with usual care

For more details, see Table 1.

Table 1. Effect of interventions: Intervention targeting patients compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Control	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Physician informative and participatory	NA	‐0,04 (0,36)	NA	0,09 (0,38)	Unit of error analysis
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Patient active	NA	0,00 (0,30)	NA	0,05 (0,35)	Unit of error analysis
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Physician‐patient interaction	NA	‐0,01 (0,43)	NA	0,03 (0,46)	Unit of error analysis
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Control	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Control	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Control	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Deen 2012	Patient mediated intervention (Decision aid) (n=69)	Usual Care (n=69)	Patient Activation Measure (PAM)	41,78 (5,42)	43,68 (5,28)	42,21 (5,22)	44,06 (5,66)	‐0,07 (‐0,40 to 0,26)	0,04 (‐0,07 to 0,09)
Deen 2012	Patient mediated intervention (Patient Activation) (n=73)	Usual Care (n=69)	Patient Activation Measure (PAM)	42,31 (6,35)	44,57 (6,16)	42,21 (5,22)	44,06 (5,66)	0,09 (‐0,24 to 0,41)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Usual Care (n=69)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	42,21 (5,22)	44,06 (5,66)	0,04 (‐0,29 to 0,38)
van Peperstraten 2010	Patient mediated intervention (n=124)	Usual Care (n=128)	Decision Evaluation scale	NA	4,1 (0,56)	NA	3,8 (0,57)	0,50 (0,25 to 0,75)
Vodermaier 2009	Patient mediated intervention	Usual Care	Man‐Son‐Hing Instrument	No data
Cooper 2011	Patient mediated intervention (n=40)	Usual Care (n=43)	Participatory Decision making (PDM)	70,94 (24,67)	74,17 (23,25)	74,61 (21,59)	69,38 (21,50)	0,21 (‐0,22 to 0,64)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Control	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Krist 2007	Patient mediated intervention (Decision aid brochure) (n=174)	Usual Care (n=63)	Modified Control Preference Scale	NA	63/174	NA	23/63	0,00 (‐0,14 to 0,14)	‐0,01 (‐0,01 to 0,00)
Krist 2007	Patient mediated intervention (Decision aid web) (n=198)	Usual Care (n=63)	Modified Control Preference Scale	NA	71/198	NA	23/63	‐0,01 (‐0,14 to 0,13)
Landrey 2012	Patient mediated intervention (n=74)	Usual Care (n=78)	Modified Control Preference Scale	NA	29/74	NA	33/78	‐0,03 (‐0,19 to 0,12)	‐0,03
Murray 2001	Patient mediated intervention (n=57)	Usual Care (n=48)	Modified Control Preference Scale	NA	34/57	NA	42/48	‐0,28 (‐0,44 to ‐0,12)	‐0,28
Vodermaier 2009	Patient mediated intervention (n=53)	Usual Care (n=54)	Modified Control Preference Scale	NA	35/53	NA	36/54	‐0,01 (‐0,19 to 0,17)	‐0,01
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Control	Outcome	Qualitative quote
No study

Data from six continuous PROMs in four RCTs were evaluated (Cooper 2011; Deen 2012; van Peperstraten 2010; Vodermaier 2009). Data from three studies (Cooper 2011; Deen 2012; van Peperstraten 2010) were available for re‐analysis. The median of the standardized mean difference was 0.21 (range 0.04 to 0.50) indicating a small improvement for the group that received the intervention targeting patients.

Data from five categorical PROMs in four RCTs were evaluated (Krist 2007; Landrey 2012; Murray 2001; Vodermaier 2009). We calculated a 0.02 reduction in the median of the risk difference for these outcomes (range ‐0.28 to ‐0.01) indicating no evidence of a difference for the group that received the intervention targeting patients.

Data from three continuous OBOMs in one RCT were evaluated (Haskard 2008). A unit of analysis error was observed in this study, and so we could not estimate the statistical significance of the effects reported.

Interventions targeting patients compared with other interventions targeting patients

For more details, see Table 2.

Table 2. Effect of interventions: Intervention targeting patients compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Montori 2011	Patient mediated intervention (n=52)	Patient mediated intervention (n=48)	OPTION	NA	49,80 (21,40)	NA	27,30 (14,70)	1,21 (0,78 to 1,64)	1,21
Nannenga 2009	Patient mediated intervention (n=48)	Patient mediated intervention (n=43)	OPTION	NA	7,13 (6,63)	NA	1,74 (2.53)	1,04 (0,60 to 1,48)	1,04
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Deen 2012	Patient mediated intervention (Decision aid) (n=69)	Patient mediated intervention (Patient Activation) (n=73)	Patient Activation Measure (PAM)	41,78 (5,42)	43,68 (5,28)	42,31 (6,35)	44,57 (6,16)	‐0,15 (‐0,48 to 0,18)	‐0,05 (‐0,15 to 0,11)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Patient mediated intervention (Patient Activation) (n=73)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	42,31 (6,35)	44,57 (6,16)	‐0,05 (‐0,38 to 0,28)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Patient mediated intervention (Decision aid) (n=69)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	41,78 (5,42)	43,68 (5,28)	0,11 (‐0,22 to 0,45)
Schroy 2011	Patient mediated intervention (Decision aid) (n=205)	Patient mediated intervention (Educational material) (n=217)	Satisfaction with the decision making process	NA	50,70 (6,20)	NA	46,00 (7,90)	0,66 (0,46 to 0,85)	0,63 (‐0,03 to 0,66)
Schroy 2011	Patient mediated intervention (Decision aid + YDR) (n=214)	Patient mediated intervention (Educational material) (n=217)	Satisfaction with the decision making process	NA	50,50 (6,20)	NA	46,00 (7,90)	0,63 (0,44 to 0,83)
Schroy 2011	Patient mediated intervention (Decision aid + YDR) (n=214)	Patient mediated intervention (Decision aid) (n=205)	Satisfaction with the decision making process	NA	50,50 (6,20)	NA	50,70 (6,20)	‐0,03 (‐0,22 to 0,16)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Butow 2004	Patient mediated intervention (n=69)	Patient mediated intervention (n=62)	Modified Control Preference Scale	NA	22/69	NA	17/62	0,04 (‐0,11 to 0,20)	0,04
Davison 1997	Patient mediated intervention (n=30)	Patient mediated intervention (n=30)	Modified Control Preference Scale	NA	10/30	NA	15/30	‐0,17 (‐0,41 to 0,08)	‐0,17
Deschamps 2004	Patient mediated intervention (n=42)	Patient mediated intervention (n=48)	Modified Control Preference Scale	NA	24/42	NA	22/48	0,11 (‐0,09 to 0,32)	0,11
Dolan 2002	Patient mediated intervention (n=43)	Patient mediated intervention (n=43)	Modified Control Preference Scale	NA	27/43	NA	22/43	0,12 (‐0,09 to 0,32)	0,12
Kasper 2008	Patient mediated intervention (n=136)	Patient mediated intervention (n=142)	Modified Control Preference Scale	NA	55/136	NA	53/142	0,03 (‐0,20 to 0,27)	0,03
Krist 2007	Patient mediated intervention (Decision aid web) (n=198)	Patient mediated intervention (Decision aid brochure) (n=174)	Modified Control Preference Scale	NA	71/198	NA	63/174	0,00 (‐0,10 to 0,09)	0
Raynes‐Greenow 2010	Patient mediated intervention (Decision Aid (Audio)) (n=176)	Pamphlet (n=175)	Modified CPS ‐ First Follow‐up	NA	39/176	NA	31/175	0,04 (‐0,04 to 0,13)	0,04 (0,04 to 0,07)
Raynes‐Greenow 2010	Patient mediated intervention (Decision aid) (n=168)	Pamphlet (n=175)	Modified CPS ‐ First Follow‐up	NA	37/168	NA	31/175	0,04 (‐0,04 to 0,13)
Raynes‐Greenow 2010	Patient mediated intervention (Decision Aid (Audio)) (n=141)	Pamphlet (n=136)	Modified CPS ‐ Second Follow‐up	NA	26/141	NA	19/136	0,04 (‐0,04 to 0,13)
Raynes‐Greenow 2010	Patient mediated intervention (Decision aid) (n=150)	Pamphlet (n=136)	Modified CPS ‐ Second Follow‐up	NA	31/150	NA	19/136	0,07 (‐0,02 to 0,13)
Stiggelbout 2008	Patient mediated intervention (n=31)	Patient mediated intervention (n=33)	Modified Control Preference Scale	NA	16/31	NA	24/33	‐0,21 (‐0,44 to 0.02)	‐0,21
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Lalonde 2006	Patient mediated intervention	Patient mediated intervention	Decision satisfaction inventory	No statistically significant differences in patient satisfaction with the decision‐making process were detected between the study groups. Page 55
Street 1995	Patient mediated intervention	Patient mediated intervention	Perceived Decision Control Instrument	The experimental manipulation (computer program versus brochure) had very little effect on the dependent variables. Page 2280
Butow 2004	Patient mediated intervention	Patient mediated intervention	Physician behaviours facilitating patient involvement	On average, oncologists demonstrated about 7.5 of the 12 behaviours, with no significant differences between the groups (cancer consiltation preparation package (CCPP) versus control booklet). Page 4406

Data from six continuous PROM in two RCTs were evaluated (Deen 2012; Schroy 2011). The median standardized mean difference was 0.29 (‐0.05 to 0.63), indicating a small improvement for the group that received a multifaceted patient‐mediated intervention (Schroy 2011) compared to the group that received only educational material (Schroy 2011).

Data from 11 categorical PROMs in eight RCTs were evaluated (Butow 2004; Davison 1997; Deschamps 2004; Dolan 2002; Kasper 2008; Krist 2007; Raynes‐Greenow 2010; Stiggelbout 2008). We calculated a 0.04 improvement in the median of the risk difference for these outcomes (range ‐0.21 to 0.12) indicating no evidence of a difference between the two interventions targeting patients.

Three outcomes from three studies (Butow 2004; Lalonde 2006; Street 1995) could not be included in this analysis because of incomplete data sets. None of the authors of the three studies reported any improvement after exposure of study participants to the intervention targeting patients.

Data from two continuous OBOMs in two RCTs were evaluated (Montori 2011; Nannenga 2009). The median of the standardized mean difference was 1.13 (range 1.04 to 1.21) indicating a large improvement for the group that received a patient decision aid (Montori 2011) compared to the group that received a booklet (Montori 2011).

Interventions targeting healthcare professionals compared with usual care

For more details, see Table 3.

Table 3. Effect of interventions: Intervention targeting healthcare professionals compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Fossli 2011	Educational meeting, audit and feedback, distribution of educational material (n=26)	Usual Care (n=25)	Fours Habits Coding Scheme (4HCS)	59,66 (8,78)	63,57 (11,96)	60,87 (11,08)	58,85 (12,19)	0,38 (‐0,17 to 0,94)	0,38
Shepherd 2011	Educational outreach visit (n=18)	Usual Care (n=18)	Assessing Communication about Evidence and Patient Preferences (ACEPP)	NA	21,30 (3,58)	NA	16,70 (3,63)	0,90 (0,21 to 1,58)	1,08 (0,90 to 1,25)
Shepherd 2011	Educational outreach visit (n=18)	Usual Care (n=18)	OPTION	NA	36,60 (12,62)	NA	25,00 (12,72)	1,25 (0,53 to 1,97)
Stacey 2006	Distribution of educational materials, educational meeting, audit and feedback and barriers assessment (n=18)	Usual Care (n=20)	Decision Support Analysis Tool (DSAT)	0,53 (0,18)	0,81 (0,17)	0,43 (0,17)	0,44 (0,18)	2,07 (1,26 to 2,87)	2,07
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Educational meeting (n=51)	Usual Care (n=43)	Participatory Decision making (PDM)	68,46 (22,81)	71,57 (19,94)	74,61 (21,59)	69,38 (21,50)	0,11 (‐0,30 to 0,51)	0,11
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Légaré 2012	Educational meeting and distribution of educational material (n=176)	Usual Care	Modified Control Preference Scale (n=177)	71/182	79/176	59/171	64/177	0,09 (‐0,01 to 0,19)	0,09
O'Cathain 2002	Educational meeting and distribution of educational material (Pre: n=1526; Post: n=1531)	Usual Care	Modified Control Preference Scale (antenatal sample) (Pre: n=1219; Post: n=1206)	345/1526	263/1531	287/1219	235/1206	‐0,02 (‐0,05 to 0,01)	0,00 (‐0,02 to 0,02)
O'Cathain 2002	Educational meeting and distribution of educational material (Pre: n=1490; Post: n=1515)	Usual Care	Modified Control Preference Scale (postnatal sample) (Pre: n=1666; Post: n=1698)	369/1490	354/1515	426/1666	358/1698	0,02 (‐0,01 to 0,05)
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Bernhard 2011	Educational meeting, audit and feedback, distribution of educational material	Usual Care	Patient involvement preference and actual involvement	There was considerable variation in patient outcomes between the SGA and ANZ cohorts and no substantial training effect. Page 6

Data from one continuous PROM in one RCT were evaluated (Cooper 2011). The standardized mean difference was 0.11.

Data from three categorical PROMs in two RCTs were evaluated (Légaré 2012; O'Cathain 2002). The median of the risk difference was 0.05 (range 0.00 to 0.09) indicating a small improvement for the group that received the healthcare professional targeted intervention.

One outcome from one study (Bernhard 2011) could not be included in this analysis because of incomplete data sets. Study authors reported no improvement after exposure of study participants to the intervention targeting healthcare professionals.

Data from four continuous OBOMs in three RCTs were evaluated (Fossli 2011; Shepherd 2011; Stacey 2006). The median of the standardized mean difference was 1.08 (range 0.38 to 2.07) indicating a significant improvement for the group that received the intervention targeting healthcare professionals.

Interventions targeting healthcare professionals compared with interventions targeting patients

For more details, see Table 4.

Table 4. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Educational meeting	Patient mediated intervention (n=51)	Participatory Decision making (PDM) (n=40)	68,46 (22,81)	71,57 (19,94)	70,94 (24,67)	74,17 (23,25)	‐0,12 (‐0,53 to 0,29)	‐0,12
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Data from one continuous PROM in one RCT were evaluated comparing an intervention targeting healthcare professionals with an intervention targeting patients (Cooper 2011). The standardized mean difference was ‐0.12.

Interventions targeting healthcare professionals compared with other interventions targeting healthcare professionals

For more details, see Table 5.

Table 5. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Elwyn 2004	Educational Meeting and Audit and feedback (n=9)	Educational Meeting and Audit and feedback (n=11)	OPTION	27,00 (14,00)	39,00 (11,80)	32,00 (13,80)	43,00 (13,60)	‐0,30 (‐1,19 to 0,59)	‐0,3
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=79; Post: n=139)	Educational Meeting and Audit and feedback (Pre: n=108; Post: n=188)	COMRADE (communication) ‐ Time 1	63,50 (18,60)	67,30 (14,10)	66,30 (13,50)	68,30 (14,10)	‐0,07 (‐0,29 to 0,15)	‐0,09 (‐0,18 to 0,05)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=69; Post: n=121)	Educational Meeting and Audit and feedback (Pre: n=94; Post: n=169)	COMRADE (communication) ‐ Time 2	62,10 (18,10)	62,40 (17,00)	63,30 (16,20)	64,20 (16,30)	‐0,11 (‐0,34 to 0,13)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=79; Post: n=139)	Educational Meeting and Audit and feedback (Pre: n=108; Post: n=188)	COMRADE (confidence) ‐ Time 1	72,00 (9,90)	74,20 (9,40)	72,00 (9,80)	73,70 (9,20)	0,05 (‐0,17 to 0,27)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=69; Post: n=121)	Educational Meeting and Audit and feedback (Pre: n=94; Post: n=169)	COMRADE (confidence) ‐ Time 2	70,00 (10,80)	70,00 (13,10)	71,80 (9,30)	72,20 (11,00)	‐0,18 (‐0,42 to 0,05)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=582)	Educational Meeting (n=550)	PPS (Man Son‐Hing) : I made the decision jointly (Score inversé pour respecter le sens de l'échelle)	NA	1,36 (0,25)	NA	1,24 (0,25)	0,48 (0,36 to 0,60)	0,48 (0,40 to 6,11)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=550)	Educational Meeting (n=582)	Shared Decision Making Q (SDM‐Q)	NA	9,18 (4,08)	NA	7,46 (4,5)	0,40 (0,28 to 0,52)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=539)	Educational Meeting (n=513)	PPS (Man‐Son‐Hing)	NA	7,69 (0,16)	NA	6,87 (0,1)	6,11 (5.82 to 6.40)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Seven continuous PROMs in two RCTs were evaluated (Elwyn 2004; Krones 2008 (ARRIBA‐Herz)). The median of the standardized mean difference was 0.20 (range ‐0.09 to 0.48) indicating some improvement in the group that received a multifaceted intervention (that is an educational meeting, audit and feedback, distribution of educational materials, and educational outreach visit) (Krones 2008 (ARRIBA‐Herz)) compared to the group that received a single intervention (for example an educational meeting on an alternative topic) (Krones 2008 (ARRIBA‐Herz)).

Data from one continuous OBOM in one RCT were evaluated (Elwyn 2004). The standardized mean difference for this study was ‐0.30.

Interventions targeting both patients and healthcare professionals compared with usual care

For more details, see Table 6.

Table 6. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Physician informative and participatory (n=66)	NA	0,02 (0,39)	NA	‐0,10 (0,41)	Unit of error analysis
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Patient active (n=66)	NA	‐0,02 (0,32)	NA	‐0,08 (0,37)	Unit of error analysis
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Physician‐patient interaction (n=66)	NA	‐0,03 (0,46)	NA	‐0,06 (0,50)	Unit of error analysis
Hess 2012	Patient mediated intervention + educational meeting (n=100)	Usual Care	OPTION (n=100)	NA	26,60 (8,10)	NA	7,00 (5,50)	2,83 (2,44 to 3,22)	2,83
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Murray 2010	Educational meeting, audit and feedback, distribution of educational materials, educational outreach, barriers assement	Usual Care	Decision Support Analysis Tool (DSAT)	"The mean score change from baseline in the intervention group 3.75 (95% CI 2.46 to 5.03) was significantly greater than the mean score change in the control group ‐0.667 (95% CI ‐1.57 to 0.24) using the two sided t‐test (P < 0.0001)" Page 116
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Usual Care	Participatory Decision making (PDM) (n=43)	66,67 (23,98)	72,84 (21,19)	74,61 (21,59)	69,38 (21,50)	0,16 (‐0,23 to 0,56)	0,16
Hamman 2007	Patient mediated intervention + Educational meeting (n=33)	Usual Care	Combined Outcome Measure for Risk Communication and Treatment (COMRADE) (n=49)	NA	76,8 (20,9)	NA	73,5 (19,3)	0,16 (‐0,28 to 0,61)	0,16
Wetzels 2005	Patient mediated Intervention + educational outreach visit (n=121)	Usual Care	Combined Outcome Measure for Risk Communication and Treatment (COMRADE) ‐ 4 items (n=142)	1,82 (NA)	1,83 (NA)	1,89 (NA)	1,80 (NA)	Unable to calculate. No differences between groups were detected.	NA
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Leighl 2011	Patient mediated intervention and educational meeting	Usual Care	Modified CPS	There was no difference after the intervention: the mean score of the item on the CPS scale in the intervention group was: 2.86 (0.92), it was 2.87 (1.04) in the control group. See Figure 4, page 2082. Data are from the authors.
Loh 2007	Patient mediated intervention and educational meeting	Usual Care	PPS (Man‐Son‐Hing)	In the intervention group, significantly higher patient participation from pre‐ to post‐intervention was found for … the Man‐Son‐Hing patient participation scale, P = 0.10. Page 329

Data from three continuous PROMs in three RCTs were evaluated (Cooper 2011; Hamann 2007; Wetzels 2005). Data from two studies (Cooper 2011; Hamann 2007) were available for re‐analysis. The median of the standardized mean difference was 0.16 (range 0.16 to 0.16) indicating no evidence of a difference for the group that received the intervention targeting patients and healthcare professionals.

Two outcomes from two studies (Leighl 2011; Loh 2007) could not be included in this analysis because of incomplete data sets. Authors of one of these studies reported that outcomes improved after exposure of study participants to interventions targeting both patients and healthcare professionals (Loh 2007).

Data from four continuous OBOMs in two RCTs were evaluated (Haskard 2008; Hess 2012). A unit of analysis error was observed in one study (Haskard 2008) and so we could not estimate the statistical significance of the effects reported. The standardized mean difference for the other study was 2.83, indicating significant improvement for the group that received the intervention targeting both patients and healthcare professionals.

One outcome from one study (Murray 2010) could not be included in this analysis because of incomplete data sets. Study authors reported significant improvement after exposure of study participants to an intervention targeting both patients and healthcare professionals.

Interventions targeting both patients and healthcare professionals compared with interventions targeting patients

For more details, see Table 7.

Table 7. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Mullan 2009	Patient mediated intervention + Education meeting (n=21)	Patient mediated intervention (n=19)	OPTION	NA	49,70 (17,74)	NA	27,70 (11,75)	1,42 (0,72 to 2,12)	1,42
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (first consultation) (n=33)	NA	4,11 (0,7)	NA	3,59 (0,7)	0,73 (0,24 to 1,23)	0,73 (0,50 to 0,88)
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (3 months) (n=33)	NA	4,05 (0,7)	NA	3,67 (0,8)	0,50 (0,01 to 0,99)
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (6 months) (n=33)	NA	3,8 (0,8)	NA	3,13 (0,7)	0,88 (0,38 to 1,38)
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Patient mediated intervention	Participatory Decision making (PDM) (n=40)	66,67 (23,98)	72,84 (21,19)	70,94 (24,67)	74,17 (23,25)	‐0,06 (‐0,46 to 0,34)	‐0,06
Mullan 2009	Patient mediated intervention + Educational meeting (n=47)	Patient mediated intervention	Validated pictorial instrument (n=36)	NA	4,8 (1,1)	NA	4,7 (1,1)	0,09 (‐0,34 to 0,52)	0,09
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Deinzer 2009	Patient mediated intervention + Educational meeting	Patient mediated intervention	Combined Outcome Measure for Risk Communication and Treatment (COMRADE)	The degree of SDM was significantly higher in the SDM group at basline and after 1‐year visits. Both groups showed an increase in SDM (both P = 0.001). Page 268
Deinzer 2009	Patient mediated intervention + Educational meeting	Patient mediated intervention	Autonomy Preference Index (API)	The preference for SDM as assessed by the API (Figure 2) showed no differences between the SDM and control group at baseline (P = 0.60) and did not change after 1 year (P = 0.83). Page 268

Data from five continuous PROMs were evaluated in three RCTs (Bieber 2006; Cooper 2011; Mullan 2009). The median of the standardized mean difference was 0.09 (range ‐0.06 to 0.73) indicating no evidence of a difference for the group that received the intervention targeting patients and healthcare professionals.

Data from two outcomes from one study (Deinzer 2009) could not be included in this analysis because of incomplete data sets. Study authors reported significant improvement for one outcome after exposure of study participants to an intervention targeting both patients and healthcare professionals.

Data from one continuous OBOM in one RCT were evaluated (Mullan 2009). The standardized mean difference was 1.42, indicating significant improvement for the group that received the intervention targeting both patients and healthcare professionals.

Interventions targeting both patients and healthcare professionals compared with interventions targeting healthcare professionals only

For more details, see Table 8.

Table 8. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Educational meeting (n=51)	Participatory Decision making (PDM)	66,67 (23,98)	72,84 (21,19)	68,46 (22,81)	71,57 (19,94)	0,06 (‐0,32 to 0,44)	0,06
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Roter 2012	Patient mediated intervention and distribution of educational materials	Distribution of educational materials	LEAPS	The study interventions led to significant and parallel increases in both patient and physician reported use of patient‐centered communication skills, and an increase in patient satisfaction with communication‐related visit goals. For patients, the intervention was associated with a positive change in reported skills in five of the six communication areas. Page 412

Data from one continuous PROM in one RCT were evaluated (Cooper 2011). The standardized mean difference for this study was 0.06 indicating no evidence of a difference between groups.

One outcome from one study (Roter 2012) could not be included in this analysis because of incomplete data sets. The authors reported that outcomes improved after exposure of study participants to interventions targeting both patients and healthcare professionals.

Interventions targeting both patients and healthcare professionals compared with other interventions targeting both patients and healthcare professionals

For more details, see Table 9.

Table 9. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to intervention targeting both patients and healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Myers 2011	Patient mediated intervention + reminders (n=74)	Patient mediated intervention + reminders (n=60)	Informed decision making scale (IDM)	NA	3/74	NA	5/60	‐0,04 (‐0,13 to 0,04)	‐0,04
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Data from one categorical OBOM in one RCT were evaluated (Myers 2011). The risk difference for this study was ‐0.04, indicating no evidence of a difference between the two interventions targeting both patients and healthcare professionals.

Heterogeneity

While the goal of this review was not to conduct a meta‐analysis, we did briefly explore causes of heterogeneity. Given that we observed heterogeneity in comparison groups with enough studies, the positive effect found in some studies could not be explained by study characteristics only.

Secondary outcomes

Additional data were available in 'Additional tables': Table 10, Table 11, Table 12 and Table 13.

Table 10. Secondary outcome: patient health measures (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Elwyn 2004	Anxiety (short form of Spielberger) Time 1	Pre: 79 Post: 138	11.33 (3.74)	10.00 (3.55)	Pre: 107 Post: 187	11.62 (3.67)	9.86 (3.78)	Pre: ‐0.08 (‐0.37 to 0.21) Post: 0.04 (‐0.18 to 0.26)
Elwyn 2004	Anxiety (short form of Spielberger) Time 2	Pre: 73 Post: 117	9.94 (3.42)	11.25 (4.28)	Pre: 92 Post: 164	10.36 (3.59)	10.23 (3.79)	Pre: ‐0.12 (‐0.43 to 0.19) Post: 0.25 (0.02 to 0.49)
Elwyn 2004	Anxiety (short form of Spielberger) Time 3	Pre: 61 Post: 101	10.15 (3.24)	10.51 (3.93)	Pre: 75 Post: 136	10.87 (3.55)	9.99 (3.23)	Pre: ‐0.21 (‐0.55 to 0.13) Post: 0.15 (‐0.11 to 0.40)
Elwyn 2004	Health status (SF‐1220) mental subscale Time 1	Pre: 101 Post: 171	48.65 (10.26)	50.41 (10.90)	Pre: 68 Post: 124	50.31 (9.66)	47.77 (11.21)	Pre: ‐0.16 (‐0.47 to 0.14) Post: 0.24 (0.00 to 0.47)
Elwyn 2004	Health status (SF‐1220) mental subscale Time 2	Pre: 79 Post: 149	49.11 (11.14)	51.16 (10.41)	Pre: 68 Post: 108	50.16 (10.73)	49.23 (11.98)	Pre: ‐0.09 (‐0.42 to 0.23) Post: 0.17 (‐0.07 to 0.42)
Elwyn 2004	Health status (SF‐1220) physical subscale Time 1	Pre: 101 Post: 171	41.16 (13.05)	42.47 (11.76)	Pre: 68 Post:124	43.01 (12.48)	41.90 (13.08)	Pre: ‐0.14 (‐0.45 to 0.16) Post: 0.05 (‐0.18 to 0.27)
Elwyn 2004	Health status (SF‐1220) physical subscale Time 2	Pre: 79 Post: 149	39.71 (12.35)	40.81 (12.14)	Pre: 68 Post: 108	43.34 (11.46)	40.91 (11.81)	Pre: ‐0.30 (‐0.63 to 0.02) Post: ‐0.01 (‐0.26 to 0.24)
Hamann 2007	Clinical global impression scale	35	NA	4.0 (1.5)	40	NA	4.1 (1.4)	‐0.07 (‐0.52 to 0.38)
Hamann 2007	Global assessment of function scale	30	NA	54.7 (16.5)	37	NA	51.0 (18.5)	0.21 (‐0.27 to 0.69)
Légaré 2012	Quality of life physical scale	181	49.30 (8.80)	49.40 (7.50)	178	47.70 (8.90)	48.20 (7.80)	0.16 (‐0.05 to 0.36)
Van Peperstraten 2010	Level of anxiety	Pre:150 Post: 127	35.60 (10.60)	36.40 (10.20)	Pre: 154 Post: 135	34.60 (9.50)	34.70 (8.20)	0.18 (‐0.06 to 0.43)
Categorical data (n/N)
Hamann 2007	Patient hospitalised within 6 mo after discharge	36	NA	8/36	37	NA	8/37	0.01 (‐0.18 to 0.20)
Hamann 2007	Patient hospitalised within 18 mo after discharge	38	NA	20/38	41	NA	19/41	0.06 (‐0.16 to 0.28)
Hamann 2007	Patient with drug switches (main antipsychotic) within 6 mo after discharge	36	NA	12/36	40	NA	16/40	‐0.07 (‐0.28 to 0.15)
Hess 2012	Admitted to hospital	101	NA	6	103	NA	6	0 (‐0.06 to 0.07)
Hess 2012	Repeat emergency department visit	101	NA	3	103	NA	0	0.03 (‐0.01 to 0.07)
Hess 2012	Rehospitalization	101	NA	2	103	NA	0	0.02 (‐0.01 to 0.05)
Hess 2012	Acute myocardial infarction	101	NA	1	103	NA	0	0.01 (‐0.02 to 0.04)
Légaré 2012	Proportion of use of antibiotics	Pre: 182 Post: 180	75	49	Pre :171 Post:178	67	93	‐0.25 (‐0.35 to ‐0.15)
Van Peperstraten 2010	Subclinical depression	Pre:147 Post: 126	16	16	Pre: 151 Post: 136	13	5	0.09 (0.02 to 0.16)
Qualitative data
Butow 2004	Spielberger State Trait Anxiety Scale	"In both groups, anxiety decreased by 3 points after the consultation, and there was no significant difference between the groups immediately after the consultation and one month later." Page 4407
Butow 2004	Beck Depression Inventory (short form)	"No significant differences between groups were observed in raw or change scores on depression immediately after the consultation or one month later." Page 4407
Mullan 2009	Adherence	" ... adherence to diabetes medications were near perfect in both groups and significantly better in the control group." Page 1565
Mullan 2009	HbA	"The decision aid did not affect glycemic control or patient‐reported health status at six months" Page 1565
Krones 2008 (ARRIBA‐Herz)	Framingham Scoring system	Non significant (P = 0.31)
Bernhard 2011	Anxiety (State Trait Anxiety Inventory)	"Anxiety slightly decreased over time for all cohorts. Patients in the SGA (Figure 4a) and ANZ (Figure 4b) cohorts reported comparable anxiety levels at each time point. The quality of life indicators showed similar findings (data not shown)." Page 6
Bernhard 2011	Quality of life	"Anxiety slightly decreased over time for all cohorts. Patients in the SGA (Figure 4a) and ANZ (Figure 4b) cohorts reported comparable anxiety levels at each time point. The quality of life indicators showed similar findings (data not shown)." Page 6
Bieber 2006	Center for epidemiological studies depression scale ‐ CES‐D	Non significant: P = 0.26 (table 4). Page 363
Bieber 2006	Visual analogue scale for pain intensity	Non significant: P = 0.45 (table 4). Page 363
Bieber 2006	Health status and physical function SF‐36	Non significant: P = 0.89 (table 4). Page 363
Bieber 2006	Hannover Functional Questionnaire FFbH	Non significant: P = 0.81 (table 4). Page 363
Cooper 2011	Blood Pressure control	"Improvements in patient adherence and BP control did not differ across groups for the overall patient sample" p1; "In the overall sample, changes in systolic and diastolic BP at 12 months did not differ for any of the intervention groups when compared to the patient+physician minimal intervention group" p1300; "Changes in patient‐reported adherence to medications at 12 months did not differ for any of the intervention groups compared to the patient+physician minimal intervention group."Page 1300
Davison 1997	Spielberger State Trait Anxiety Scale	"There was no evidence trait scores were different among groups, among measurement times, or between groups and measurement times" Page 195
Davison 1997	Center for epidemiological studies depression scale ‐ CES‐D	"No significant differences in mean depression scores were found among the groups, among measurement times, or between groups and measurement times" Page 196
Deinzer 2009	Self measurement of systolic and diastolic blood pressure	"Thus in both groups BP decreased but there were no significant differences between the 2 groups (systolic P = 0.24 and diastolic P = 0.16 respectively)." Page 268
Hamann 2007	Severity of illness (PANSS)	NA "... there were no differences between ... PANSS score at discharge" Page 994
Hamann 2007	Plasma level of antipsychotic	Not reported
Hamann 2007	Medication at discharge	Not reported
Hess 2012	Major adverse cardiac event	"Excluding the index presentation, there were no deaths or major adverse cardiac events within 30 days" Page 256
Leighl 2011	Functional Assessment of Cancer Therapy–General (FACT‐G)	Patients completed the physical, emotional, and functional subscales of the Functional Assessment of Cancer Therapy ‐ General (FACT‐G) and had similar scores in both arms comparable with those of other patients with advanced cancer. Page 2080
Loh 2007	Brief PHQ‐D ‐ Depression severity	Non significant (P = 0.236)
Murray 2001	Health status and physical function SF‐36	" ... no difference in score was observed between the two groups" Page 5
Murray 2001	Health states and valuation of health states EQ‐SD	" ... no difference in score was observed between the two groups" Page 5
Murray 2001	Spielberger state of trait anxiety inventory short form	"The Spielberger scores were similar in the final assessment in the two groups" Page 5
Murray 2001	Prostatic symptoms (American Urological Association symptom scale)	"The amount of change was not significantly different in the two groups" Page 5
Raynes‐Greenow 2010	Mode of delivery	There were no differences between labour and birth outcomes between the groups P = 0.97 (table 4). See page 10
Raynes‐Greenow 2010	Labour Type	There were no differences between labour and birth outcomes between the groups P = 0.97 (table 4). See page 10
Raynes‐Greenow 2010	Analgesia used	There were no significant differences between groups in regards to analgesia use (P = 0.18‐0.84). See page 7
Raynes‐Greenow 2010	Apgar score	P = 0.12 (1 minute) and P = 0.68 ( 5 minutes) (table 4). See page 10
Stiggelbout 2008	Quality of life (HADS)	"Patients' quality of life was stable over time, in both groups. No effects were observed in the repeated measures for the anxiety and depression scales of the HADS, nor on the quality of life scales" Page 757
Stiggelbout 2008	100 mm visual analogue	"Patients' quality of life was stable over time, in both groups. No effects were observed in the repeated measures for the anxiety and depression scales of the HADS, nor on the quality of life scales" (100 mm visual analogue scale) Page 757

Table 11. Secondary outcome: duration of consultation (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Stacey 2006	Call length	Pre: 18 Post: 18	17.80 (4.50)	18.50 (6.30)	Pre: 20 Post: 20	16.70 (7.70)	16.70 (6.50)	Pre: 0.17 (‐0.47 to 0.81) Post: 0.27 (‐0.36 to 0.91)
Qualitative data
Butow 2004	Consultation length	"Consultation length was similar between groups ‐ on average, 36 minutes per consultation." Page 4407
Elwyn 2004	Consultation length	"There was no difference in the mean consultation lengths at baseline, phase 1 and phase 2 (overall consultation mean duration was 12.5 minutes)" Page 342
Fossli 2011	Consultation length	"There was a non significant difference between both groups ( RD: ‐1:03 CI ‐6:13;4:07) P = 0.69" Page 4
Krist 2007	Consultation length	"These [discussion times] patient‐physician differences did not differ significantly across the control, brochure, and Web groups." Page 116
Loh 2007	Consultation length	Non significant differences between the groups (Table 2) Page 329
Montori 2011	Consultation length	"The median (range)duration of osteoporosis discussions was 12.4 minutes (2.3‐27.4) in the decision aid arm compared with 9.4 minutes (2.1‐58) in the usual care arm (P .045)" Page 552‐553
Murray 2001	Consultation length	Not reported
Murray 2010	Consultation length	"At baseline there was no significant difference. However, in the post‐calls, the mean call duration was longer in the intervention group at 13,47 minutes (95% confidence interval 11.8;14.21), than in the control group at 10.29 minutes (95% CI 8.79 to 11.79 P = 0.004)" Page 117
Nannenga 2009	Consultation time	"We found no significant difference in face‐to‐face consultation duration with the staff endocrinologist (mean difference 3.8 min longer with the decision aid, 95% CI ‐ 2.9 to 10.5)." Page 42
Shepherd 2011	Consultation length	"These effects occurred without any significant difference in consultation length, mean consultation lengths were 26 minutes for control and intervention visits." Page 381
Vodermaier 2009	Consultation time	"No time differences emerged in the length of the treatment decision consultation with the physicians on patient self‐reports. The mean time for the treatment decision making appointment was about 15 minutes" Page 593
Wetzel 2005	Consultation time	No differences between intervention and control groups were detected, consultations was between 12.2 and 13 minutes for all groups (Table 4) Page 292

Table 12. Secondary outcome: other measurement reported by the healthcare professional (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Haskard 2008	Physician satisfaction questionnaire	61	NA	74.82 (5.47)	66	NA	74.60 (6.47)	Unit of error analysis
Haskard 2008	Satisfaction with the management and functioning of their office practice	61	NA	3.20 (0.65)	66	NA	3.08 (0.58)	Unit of error analysis
Haskard 2008	Overall quality of life	63	NA	3.00 (0.83)	63	NA	2.82 (0.73)	Unit of error analysis
Haskard 2008	Stress	61	NA	2.68 (0.69)	66	NA	2.78 (0.60)	Unit of error analysis
Mullan 2009	Acceptability amount of information	21	NA	6.59 (0.91)	19	NA	6.37 (1.14)	0.20 (‐0.41 to 0.83)
Mullan 2009	Acceptability clarity of information	21	NA	6.20 (0.96)	19	NA	6.20 (0.80)	0.00 (‐0.62 to 0.62)
Mullan 2009	Helpfulness of the information	21	NA	6.15 (0.94)	19	NA	5.74 (1.04)	0.41 (‐0.22 to 1.03)
Mullan 2009	Would recommend to others	21	NA	6.16 (1.51)	19	NA	5.89 (1.82)	0.16 (‐0.46 to 0.78)
Mullan 2009	Would want to use for other decisions	21	NA	6.04 (1.55)	19	NA	5.69 (1.75)	0.21 (‐0.44 to 0.84)
Murray 2010	Knowledge	35	NA	69.30 (2.98)	35	NA	60.50 (2.27)	3.28 (2.55 to 4.02)
Krones 2008 (ARRIBA‐Herz)	Patient participation scale, physician rating	19	NA	1.66 (0.45)	26	NA	1.65 (0.48)	0.02 (‐0.57 to 0.61)
Bieber 2006 (first consultation)	Difficult doctor patient questionnaire	34	NA	29.40 (5.80)	33	NA	33.50 (10.00)	‐0.50 (‐0.98 to ‐0.02)
Bieber 2006 (month follow up)	Difficult doctor patient questionnaire	34	NA	28.90 (6.70)	33	NA	32.20 (6.50)	‐0.49 (‐0.98 to ‐0.01)
Légaré 2012	Physician quality of decision	Pre: 172 Post: 166	8.20 (1.10)	8.20 (1.30)	Pre: 162 Post: 170	8.20 (1.40)	8.40 (1.00)	‐0.17 (‐0.39 to 0.04)
Légaré 2012	Physician intention to follow CPG	Pre: 151 Post: 132	1.60 (0.80)	1.70 (0.90)	Pre: 108 Post: 98	1.60 (0.90)	1.80 (0.70)	‐0.12 (‐0.38 to 0.14)
Loh 2007	Physician's assessment of treatment adherence	96	4.20 (1.10)	4.30 (1.10)	191	4.30 (0.90)	4.80 (0.60)	Intracluster correlation error
Categorical data: (n/N)
Légaré 2012	Physician Decisional Conflict (Proportion who had a value of 2.5 or more )	Pre: 178 Post: 175	8	8	Pre: 166 Post: 176	5	2	0.03 (‐0.00 to 0.07)
Murray 2001	Perceived role in decision making: shared role	48	NA	25/48	49	NA	32/49	‐0.13 (‐0.33 to 0.06)
Vodermaier 2009	Chose Breast‐conserving therapy	39	NA	37/39	41	NA	36/41	0.07 (‐0.05 to 0.19)
Vodermaier 2009	Chose Chemotherapy	35	NA	11/35	39	NA	11/39	0.03 (‐0.18 to 0.24)
Vodermaier 2009	Chose pre‐operative chemotherapy	16	NA	10/16	15	NA	7/15	0.16 (‐0.19 to 0.50)
Qualitative data
Butow 2004	Physician satisfaction with the decision making process	"Physicians were also equally satisfied with decision making whether or not their patients had received the CCPP or the control booklet" Page 4407
Elwyn 2004	Clinician perception of the level of clinician agreement	"Clinicians showed significant differences between the RC and SDM arms (see Table S3). Doctors receiving the risk communication tools and training first perceived significantly higher doctor–patient agreement on treatment (P 0.001), patient satisfaction with information (P = 0.01), doctor satisfaction with decision (P = 0.01) and general overall satisfaction (P = 0.001) with the consultation than those who were exposed to SDM training. The latter group of doctors showed lower scores after the interventions. The differences were largely maintained in the second intervention phase, i.e. even when provided with the risk communication training and tools, the group of doctors who had received SDM training first still reported lower levels of satisfaction, agreement, etc. In contrast, doctors who had received risk communication training first maintained their higher levels of satisfactions and agreement, even when later given the SDM training which appeared less beneficial (to doctors) in the first phase." Page 343
Elwyn 2004	Clinician satisfaction with the decision and overall consultation	"Clinicians showed significant differences between the RC and SDM arms (see Table S3). Doctors receiving the risk communication tools and training first perceived significantly higher doctor–patient agreement on treatment (P 0.001), patient satisfaction with information (P 0.01), doctor satisfaction with decision (P 0.01) and general overall satisfaction (P 0.001) with the consultation than those who were exposed to SDM training. The latter group of doctors showed lower scores after the interventions. The differences were largely maintained in the second intervention phase, i.e. even when provided with the risk communication training and tools, the group of doctors who had received SDM training first still reported lower levels of satisfaction, agreement, etc. In contrast, doctors who had received risk communication training first maintained their higher levels of satisfactions and agreement, even when later given the SDM training which appeared less beneficial (to doctors) in the first phase." Page 343
Elwyn 2004	Patient satisfaction with information provided (as described by clinicians)	"Doctors receiving risk communication tool and training first perceived significantly higher doctor‐patient agreement on treatment (P < 0.001), patient satisfaction with information (P < 0.01), doctor satisfaction with decision (P < 0.01) and general overall satisfaction (P < 0.001)" Page 343
Mullan 2009	Decision aid acceptability	Not reported
Murray 2010	Acceptability of the instrument	"In all, 37 members of the intervention group (97%) commented on the acceptability of the skills building workshop. … The 31 (81%) agreed that the PtDA would be acceptable to patients, while 24 (63%) agreed that it would be acceptable to practitioners." Page 117
Murray 2010	Utility of the intervention PtDA	"All 36 who participated in the educational outreach call indicated an interested in using the POC PtDa and express frustration that it was not available for use in their clinical practice setting." Page 117
Murray 2010	Intention to engage	"All participants, regardless of group assignment, saw patient decision support as helpful to patients (n = 32 [100 percent] interventions; n = 38 [98 percent] control) While 27 members of the intervention group (87%) and 34 members of the control group (84%) indicated a positive intention to engage in decision support, 16 members of the intervention group (50%) strongly agreed that they could provide decision support compared to 11 members of the control group (28%)" Page 117
Stacey 2006	Nurses' knowledge	"The nurses in the intervention group (n = 19) had a mean knowledge score of 74% and the mean score in the control group (n = 20) was 60%. The difference between the groups was significant (P = 0.007)." Page 413
Stacey 2006	Nurses' perception of factors influencing use of the coaching protocol	"Most of the 19 nurses in the intervention group agreed that the protocol was compatible with their practice (n = 15), provided a logical approach (n = 17), was easy to try (n = 15) and helped with exploring the benefits and harms of the options available to callers (n = 16). Another advantage of using the protocol, as reported by one nurse, was that it ... increases focus on caller's needs rather than just giving information." Page 413
Bernhard 2011	Maslach Burnout Inventory	"When doctors’ stress and burnout factors were accounted for in the mixed effects models for decisional conflict, the ESs became slightly larger in the SGA cohort but remained low. There was no influence by these factors on the ESs in the ANZ cohort (data not shown)." Page 5
Hamann 2007	Doctor patient relationship	"Doctor‐patient relationship (WAI) and PANSS scores did not prove to be independent significant prognostic factors" Page 996
Hamann 2007	Physicians satisfaction with treatment results	Not reported
Hess 2012	Clinician satisfaction with and acceptability of the DA	"Of the 51 clinicians who used the decision aid, 50 (98%) considered it helpful, and 32 (63%) indicated their desire to use the decision aid again if given the opportunity. Most clinicians indicated a desire to use a decision aid for other clinical management decisions" Page 255
Krist 2007	Physician perception of the decision making process	"Physicians tended to reports that they had greater control over the decision than did the patients, as measured by the CPS" Page 116
Krist 2007	Number of test ordered	Not reported
Leighl 2011	Physician satisfaction with decision‐making score	"Australian medical oncologists were surveyed regarding their satisfaction with the decision‐making process after each consultation;scores were generally high and similar in both arms" Page 2080
Murray 2001	Evaluation of the intervention	"General practitioners were positive about the decision aid; of 50 follow up consultation with patients in the intervention group they said that the decision aid had helped in 46, made no difference in three, and hindered one." Page 5
Roter 2012	Time management	The area in which there was no significant difference in reported skill use was in relation to time management. p.412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Roter 2012	Treatment adherence	The area in which there was no significant difference in reported skill use was in relation to time management. page 412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Roter 2012	Interpersonal rapport	The area in which there was no significant difference in reported skill use was in relation to time management. page 412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Stiggelbout 2008	Surgeon's perceptions	"No differences were seen between the arms of the trial in the surgeons' reply to the question whether and how they presented probabilities; nor to the question on the risk that were discussed, the total number of risks that were discussed, or the understanding of the information by the patients; nor to the question whether much discussion had taken place during the consultation." Page 757
Street 1995	Physician facilitation	Not reported
Van Peperstraten 2010	Cost evaluation of the empowerment strategy	"The mean total savings in the intervention group were calculated to be €169.75 per couple included from the waiting list for in vitro fertilisation" Page 5

Table 13. Secondary outcomes: other measures reported by patients (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Bieber 2006	Satisfaction with decision scale	34	NA	4.11 (0.40)	33	NA	4.02 (0.60)	0.17 (‐0.30 to 0.65)
Bieber 2006	Satisfaction with decision scale	34	NA	4.10 (0.60)	33	NA	4.07 (0.60)	0.05 (‐0.43 to 0.53)
Bieber 2006	Desicional conflict scale	34	NA	12.90 (4.20)	33	NA	12.40 (3.70)	0.12 (‐0.35 to 0.60)
Bieber 2006	Desicional conflict scale	34	NA	12.80 (3.00)	33	NA	12.50 (3.40)	0.09 (‐0.39 to 0.57)
Deen 2012	Decision self‐efficacy (DSE)	17	73.52 (19.13)	79.55 (12.79)	15	76.97 (17.95)	77.42 (19.29)	0.13 (‐0.57 to 0.82)
Deen 2012	Decision self‐efficacy (DSE)	21	71.54 (25.57)	79.55 (12.79)	15	76.97(17.96)	77.42 (19.30)	0.13 (‐0.53 to 0.80)
Deen 2012	Decision self‐efficacy (DSE)	17	77.27 (16.13)	83.82 (15.56)	15	76.97(17.97)	77.42 (19.31)	0.36 (‐0.34 to 1.06)
Dolan 2002	Decisional conflict scale	45	NA	1.83 (0.52)	43	NA	2.03 (0.81)	‐0.30 (‐0.71 to 0.27)
Haskard 2009	Patient perceived decision‐making	61	NA	2.94 (0.43)	66	NA	2.85 (0.46)	Unit of error Analysis
Haskard 2009	Patient choice	61	NA	4.15 (0.55)	66	NA	3.96 (0.68)	Unit of error Analysis
Krones 2008 (ARRIBA‐Herz)	Decisional regret	372	NA	14.69 (NA)	372	NA	18.08 (NA)	Unable to calculate
Krones 2008 (ARRIBA‐Herz)	Knowledge	535	NA	2.03 (NA)	576	NA	1.92 (NA)	Unable to calculate
Lalonde 2006	Decisional conflict scale	26	2.49 (0.53)	2.36 (0.30)	24	2.50 (0.39)	2.33 (0.30)	Pre: ‐0.02 (‐0.58 to 0.53) Post: 0.0.10 (‐0.46 to 0.65)
Landrey 2012	Knowledge of prostate cancer screening	71	NA	3.50 (1.50)	77	NA	3.30 (1.40)	0.14 (‐0.19 to 0.46)
Légaré 2012	Patients' quality of decision	Pre: 158 Post: 162	8.70 (1.50)	8.50 (1.60)	Pre: 151 Post: 159	8.70 (1.50)	8.50 (1.50)	0 (‐0.22 to 0.22)
Légaré 2012	Intention to engage in shared decision‐making	Pre: 165 Post: 163	1.90 (1.20)	2.10 (1.10)	Pre: 164 Post: 165	2.00 (1.20)	1.90 (1.20)	0.17 (‐0.04 to 0.39)
Légaré 2012	Regret over decision	Pre: 165 Post: 162	10.50 (15.40)	12.40 (19.10)	Pre: 164 Post: 164	10.80 (20.80)	7.60 (13.70)	0.29 (0.07 to 0.51)
Loh 2007	Doctor facilitation (PICS‐DF)	191	15.40 (3.50)	17.40 (3.10)	96	14.70 (3.70)	14.50 (3.30)	Pre: 0.20 (‐0.05 to 0.44) Post: 0.91 (0.66 to 1.17)
Loh 2007	Information seeking (PICS‐IS)	191	12.30 (2.70)	12.30 (3.40)	96	11.30 (2.90)	10.30 (2.90)	Pre: 0.36 (0.11 to 0.61) Post: 0.61 (0.36 to 0.87)
Loh 2007	Treament adherence	191	4.30 (0.80)	4.30 (0.90)	96	3.90 (0.80)	3.90 (1.00)	Pre: 0.50 (0.25 to 0.75) Post: 0.43 (0.18 to 0.67)
Loh 2007	Patients satisfaction (ZUF8)	191	NA	29.80 (2.70)	96	NA	27.00 (3.60)	0.92 (0.66 to 1.18)
Mullan 2009	Acceptability clarity of information	NA	NA	6.20 (0.96)	NA	NA	6.20 (0.80)	‐0.01 (‐0.38 to 0.36)
Mullan 2009	Acceptability helpfulness of the information	NA	NA	6.15 (0.94)	NA	NA	5.74 (1.04)	0.38 (0.04 to 0.72)
Mullan 2009	Acceptability; would recommend to others	NA	NA	6.16 (1.51)	NA	NA	5.89 (1.82)	0.38 (‐0.28 to 1.05)
Mullan 2009	Acceptability; would want to use for other decisions	NA	NA	6.04 (1.55)	NA	NA	5.69 (1.75)	0.34 (‐0.39 to 1.08)
Mullan 2009	Decisional conflict scale	NA	NA	14.10 (17.89)	NA	NA	14.95 (12.68)	‐0.89 (‐5.37 to 3.59)
Mullan 2009	Informed subscale of DCS (knowledge)	NA	NA	13.65 (19.84)	NA	NA	15.28 (15.49)	‐2.49 (‐7.21 to 2.23)
Mullan 2009	Trust in Physician scale	NA	NA	94.69 (7.14)	NA	NA	93.06 (9.58)	2.06 (‐1.78 to 5.89)
Mullan 2009	Acceptable amount of information	NA	NA	6.59 (0.91)	NA	NA	6.37 (1.14)	0.2 (‐0.41 to 0.83)
Murray 2001	Decisional conflict score	57	NA	2.30 (0.40)	48	NA	2.60 (0.50)	‐0.66 (‐1.06 to ‐0.27)
Murray 2001	Prosectomy rates and referrals	57	NA	0.11 (0.31)	48	NA	0.02 (0.14)	0.36 (‐0.03 to 0.75)
Myers 2010	Knowledge change	142	NA	0.80 (1.90)	144	NA	1.50 (2.10)	‐0.35 (‐0.58 to ‐0.11)
Myers 2010	Decisional conflict	142	NA	0.32 (0.49)	144	NA	0.29 (0.34)	0.07 (‐0.16 to 0.30)
Raynes‐Greenow 2010	Decisional conflict at primary follow up	395	31.40 (12.80)	23.90 (10.60)	201	31.20 (13.40)	24.90 (12.90)	Pre: 0.02 (‐0.15 to 0.19) Post: ‐0.09 (‐0.25 to 0.08)
Raynes‐Greenow 2010	Decisional conflict at second follow up	395	31.40 (12.80)	19.90 (12.30)	201	31.20 (13.40)	20.20 (14.10)	Pre: 0.01 (‐0.15 to 0.18) Post: ‐0.02 (‐0.19 to 0.15)
Raynes‐Greenow 2010	Anxiety first follow up	395	33.90 (10.10)	33.30 (9.30)	201	34.30 (11.80)	34.30 (11.00)	Pre:‐0.04 (‐0.21 to 0.13) Post: ‐0.10 (‐0.27 to 0.07)
Raynes‐Greenow 2010	Anxiety second follow up	395	33.90 (10.10)	29.40 (8.50)	201	34.30 (11.00)	29.00 (9.50)	Pre: ‐0.04 (‐0.21 to 0.13) Post: 0.04 (‐0.12 to 0.21)
Raynes‐Greenow 2010	Satisfaction with decision making first follow up	395	NA	81.50 (10.30)	201	NA	80.70 (11.70)	0.07 (‐0.10 to 0.24)
Raynes‐Greenow 2010	Satisfaction with decision making second follow up	395	NA	84.40 (12.90)	201	NA	82.80 (16.10)	0.11 (‐0.06 to 0.28)
Raynes‐Greenow 2010	Knowledge of analgesia first follow up	395	53.40 (21.90)	65.10 (29.50)	201	54.40 (20.90)	56.50 (27.40)	Pre: 0.05 (‐0.22 to 0.12) Post: 0.30 (0.13 to 0.47)
Stiggelbout 2008	Active participation of the patient	31	NA	1.40 (0.90)	33	NA	1.00 (0.20)	0.61 (0.11 to 1.18)
Van Peperstraten 2010	Knowledge experienced	Pre: 150 Post: 127	5.70 (2.50)	7.70 (0.60)	Pre: 154 Post: 135	5.80 (2.50)	7.20 (1.20)	0.52 (0.27 to 0.77)
Van Peperstraten 2010	Knowledge actual	127	NA	6.20 (2.85)	135	NA	4.30 (1.76)	0.74 (0.49 to 0.99)
Vodermaier 2009	Decisional conflict scale	53	NA	1.82 (0.59)	54	NA	1.99 (0.62)	‐0.28 (‐0.66 to 0.10)
Vodermaier 2009	Perceived involvement in care doctor facilitation (1‐4)	53	NA	2.65 (0.66)	54	NA	2.72 (0.67)	‐0.10 (‐0.48 to 0.27)
Vodermaier 2009	Perceived involvement in care patient information	53	NA	3.04 (0.74)	54	NA	3.09 (0.73)	‐0.10 (‐0.40 to 0.36)
Vodermaier 2009	ZUF‐8	53	NA	29.08 (2.99)	54	NA	28.67 (2.86)	0.14 (‐0.24 to 0.52)
Categorical data (n/N)
Dolan 2002	Annual fecal occult blood test	45	NA	11/23	43	NA	6/17	0.12 (‐0.18 to 0.43)
Dolan 2002	No test (wait and see)	45	NA	8/8	43	NA	15/16	0.06 (‐0.14 to 0.26)
Dolan 2002	Annual fecal occult blood test and flexible sigmoidoscopy every five years	45	NA	2/6	43	NA	7/8
Dolan 2002	Flexible sigmoidoscopy every five years	45	NA	4/6	43	NA	1/2	0.17 (‐0.15 to 0.48)
Dolan 2002	Double contrast barium enema every five years	45	NA	0/1	43	NA	0/0	NA
Dolan 2002	Colonoscopy every ten years	45	NA	1/1	43	NA	0/0	NA
Hess 2012	The proportion of patients who decided to undergo observation, unit admission, and cardiac stress testing	100	NA	58	100	NA	77	‐0.19 (‐0.32 to ‐0.41)
Krist 2007	PSA test ordered	196	NA	163/196	75	NA	64/75	‐0.02 (‐0.1 to ‐0.07)
Krist 2007	PSA test ordered	226	NA	194/226	75	NA	64/75	0.01 (‐0.09 to 0.10)
O'Cathain 2002	More anxious (antenatal)	Pre: 600 Post: 803	69/600	96/803	Pre: 595 Post: 724	77/595	87/724	Pre: ‐0.01 (‐0.05 to 0.02) Post: 0 (‐0.03 to 0.03)
O'Cathain 2002	More anxious (postnatal)	Pre: 879 Post: 846	99/879	86/846	Pre: 772 Post: 630	89/772	64/630	Pre: ‐0 (‐0.03 to 0.03) Post: 0 (‐0.03 to 0.03)
O'Cathain 2002	Drank less (antenatal)	Pre: 599 Post: 796	474/599	623/796	Pre: 595 Post: 696	443/592	551/696	Pre: 0.04 (0.00 to 0.10) Post:‐0.10 (‐0.03 to 0.03)
O'Cathain 2002	Planned hospitals birth (antenatal)	Pre: 619 Post: 826	608/619	799/826	Pre: 620 Post: 743	604/620	725/743	Pre: 0.01 (0.01 to 0.02) Post:‐0.01 (‐0.02 to 0.01)
O'Cathain 2002	Had screening test (antenatal)	Pre: 619 Post: 824	518/619	653/824	Pre: 619 Post:827	619/619	826/827	Pre: ‐0.16 (‐0.19 to 0.13) Post: ‐0.21 (‐0.23 to ‐0.18)
O'Cathain 2002	Partner/family present during labour (postnatal)	Pre: 922 Post: 886	867/922	836/886	Pre: 819 Post: 661	777/819	619/661	Pre: ‐0.01 (‐0.03 to 0.01) Post: 0.01 (‐0.02 to 0.03)
O'Cathain 2002	Stayed in bed during labour (postnatal)	Pre: 888 Post: 847	420/888	428/847	Pre: 796 Post: 635	409/796	319/635	Pre: ‐0.04 (‐0.09 to 0.01) Post: 0 (‐0.05 to 0.05)
O'Cathain 2002	Continuous monitory (postnatal)	Pre: 922 Post: 886	451/922	397/886	Pre: 819 Post: 661	387/819	319/661	Pre: 0.02 (‐0.03 to 0.06) Post: ‐0.03 (‐0.08 to 0.02)
O'Cathain 2002	Had epidural (postnatal)	Pre: 922 Post: 886	216/922	223/886	Pre: 819 Post: 661	177/819	160/661	Pre: 0.02 (‐0.02 to 0.06) Post: 0.01 (‐0.03 to 0.05)
O'Cathain 2002	Breast fed (postnatal)	Pre: 921 Post: 883	518/921	511/883	Pre: 818 Post: 660	482/818	389/660	Pre: ‐0.03 (‐0.07 to 0.02) Post: ‐0.01 (‐0.06 to 0.04)
O'Cathain 2002	Satisfied with amount of information	Pre: 891 Post: 855	619/891	635/855	Pre: 780 Post: 637	536/780	458/637	Pre: 0.01 (‐0.04 to 0.05) Post: 0.02 (‐0.02 to 0.069)
O'Cathain 2002	Satisfied with way choices were made	Pre: 886 Post: 855	683/886	656/855	Pre: 780 Post: 633	600/780	502/633	Pre: 0 (‐0.04 to 0.04) Post: ‐0.03 (‐0.07 to 0.02)
O'Cathain 2002	Enough discussion	Pre: 883 Post: 847	570/883	548/847	Pre: 774 Post: 636	481/774	414/636	Pre: 0.02 (‐0.02 to 0.07) Post: ‐0 (‐0.05 to 0.04)
Raynes‐Greenow 2010	Enough information to make decision	395	NA	352/395	201	NA	160	0.10 (0.03 to 0.16)
Raynes‐Greenow 2010	Analgesia used:support	395	NA	258	201	NA	120	0.06 (‐0.03 to 0.14)
Raynes‐Greenow 2010	Analgesia used: bath use	395	NA	143	201	NA	65	0.04 (‐0.04 to 0.12)
Raynes‐Greenow 2010	Analgesia used: epidural used	395	NA	133	201	NA	66	0.01 (‐0.07 to 0.09)
Van Peperstraten 2010	Fully empowered couples, decision empowerment	Pre: 150 Post: 127	116	116	Pre: 154 Post: 99	112	99	0.18 (0.09 to 0.27)
Qualitative data
Butow 2004	Satisfaction with the consultation and decision	"No significant differences were found between the groups in satisfaction with either the consultation or treatment decision" Page 4407
Butow 2004	Satisfaction with the booklet	"No significant differences were found between groups in terms of reported anxiety provoked, perceived utility, or ease of understanding of materials. … There was significant reported usefulness of the CCPP and control booklet for the family (P = 0.004)." Page 4405
Butow 2004	Information subscale of the Krantz Health Opinion Survey	"No significant results were obtained" Page 4407
Deinzer 2009	Hypertension Questionnaire	"Only in the SDM group was there an increase in knowledge after 1 year (P = 0.006). After 1 year both groups showed similar levels of knowledge" Page 269
Deinzer 2009	Short Form 36 Item Health Survey (SF‐36)	"There were no differences between the 2 groups concerning health‐related quality of life measured with the 8 scales of SF‐36" Page 269
Deinzer 2009	Difficult Doctor Patient Relationship Questionnaire (DDPRQ)	"Doctor‐patient relationship … was better in the SDM group than the control at the beginning … and after 1 year (p.0016). In the control group … an improvement occurred … (P = 0.045) that did not occur in the SDM group (P = 0.16)" Page 269
Deinzer 2009	Autonomy Preference Index	"Preference for SDM as assessed by the API showed no differences between the SDM and control group at baseline (P = 0.60) and did not change after 1 year (P = 0.83)" Page 268
Deschamps 2004	Decision conflict score and the informed subscale items	The differences between groups were non‐significant (Table 2), page 25
Deschamps 2004	Satisfaction with preparation for decision making	The differences between groups were non‐significant (Table 3), page 25
Deschamps 2004	Satisfaction with decision	"Women in the pharmacist and decision‐aid groups had mean SWD scores of 4.3 and 4.4 respectively (scale range: 1 to 5) with no significant differences being reported between groups. Page 26
Deschamps 2004	Adherence to HRT	"There was no statistically significant difference in adherence between the study groups" Page 26
Elwyn 2004	Intention to adhere to chosen treatment	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes ... However, significant effects of the research clinic (i.e. mainly the provision of more time)did lead to improvement (0.7 increase, 95% CI 0.04 to 1.36, P < 0.05)" Page 351
Elwyn 2004	Patient's satisfaction with information provided	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Enablement	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Satisfaction with decision made	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Patient's perceived support in decision	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Fossli 2009	Patient global satisfaction	Non significant P = 0.38
Hamann 2007	Autonomy preference index (API)	Differences between groups not reported
Hamann 2007	Patient's satisfaction with overall care	Differences between groups not reported
Hamann 2007	The medication adherence rating scale	Differences between groups not reported
Hamann 2007	Patient knowledge of disease and treatment (7‐item multiple choice)	Differences between groups not reported
Hamann 2007	Compliance with drug regime	Overall compliance was "good" for 42 (49%) of the patients at 6 months and 40 (59%) at 18 months
Hess 2012	Knowledge	Knowledge (P < 0.0001) Table 2. Page 6
Hess 2012	DCS	DCS (MD=‐13.6 (‐19.1 to ‐8.1)) Table 2. Page 6
Hess 2012	Trust in physician	Trust in physician (MD=4.1 (‐1.4 to 9.6)), Table 2. Page 6
Hess 2012	Patient satisfaction with the decision‐making process	Patients who used the decision aid reported greater satisfaction with the decision‐making process (strongly agree, 61% versus 40%; absolute difference, 21%; 95% CI 7% to 33%). Page 5
Kasper 2008	Treatment decision	"Pearson's chi square P‐value for this table is not significant for patients already on immunotherapy at baseline and patients not yet on immunotherapy at baseline, compared to patients in the CG." Page 1350
Kasper 2008	Patients evaluation of the decision	"Six months after randomization, the two groups did not show any significant differences in their evaluation of their decisions" Page 1350
Kasper 2008	Measure of the decision making process	"Both groups progressed significantly in making their decision. However they did not show differences in the course of progress over the three measurement points" Page 1349
Krist 2007	Prostate cancer screening knowledge	"… the percentage of correct answers on the knowledge scale was 54% in the control group (P < 0.001) vs 69% in the brochure group (P < 0.001)" Page 115
Krist 2007	Decisional conflict score	"DCS scores among all 3 groups were equally low and did not differ significantly … " Page 115
Krist 2007	Patients and physicians topics covered in the discussion	"The decision aids did not appear to alter … the number of prostate cancer screening topics that patients or physicians recalled addressing" Page 115
Lalonde 2006	Risk perception	"No statistically significant improvements were observed after the intervention" p55 No mention of between‐group differences
Lalonde 2006	Knowledge of hypertension	"However, knowledge of the estimated benefits of treatment tended to improve after the intervention (29% versus 58%; P = 0.06)" No mention of differences between group" Page 55
Landrey 2012	Flyer acceptability	"Among patients who reported receiving the flyer, 86.4% felt the content was clearly presented, 86.4% felt it contained about the right amount of information, 45.5% felt the information was completely balanced, and 43.2% viewed it as biased against PSA testing; 88.6% would recommend it to others." Page 5
Leighl 2011	Decisional conflict score	Decision satisfaction and decisional conflict scores were similar in both arms. Page 2080
Leighl 2011	Patient satisfaction with decision	Decision satisfaction and decisional conflict scores were similar in both arms. Page 2080
Leighl 2011	Patient satisfaction with consultation	"Patients in both arms were highly satisfied with the consultation" Page 2080
Montori 2011	Knowledge: DA specific	Knowledge DA specific (P = 0.001) Table 2, page 553
Montori 2011	Knowledge: Not in the DA	Knowledge not in the DA (P = 0.35) Table 2, page 553
Montori 2011	Decisional conflict scale	Decisional conflict scale (P = 0.72) Table 2, page 553
Montori 2011	Trust	Trust (P = 0.46) Table 2, page 553
Murray 2001	Acceptability of decision aid	"Patients reacted positively to the decision aid" Page 5
Murray 2001	Satisfaction	Not reported
Murray 2001	Choice of treatment	The choice in treatment did not vary significantly from one group to another. For more details, see page 5.
Myers 2010	Screening use	"Screening use was lower in EI Group than in SI Group (63% versus 71%), but this difference was not statistically significant (odds ratio= 0.67; 95% confidence interval, CI: 0.41‐1.08; P = 0.102)" Page 4
Raynes‐Greenow 2010	Stages of decision making	"Even distribution among stages.... A small proportion of women in both groups were not considering their choices …, or had made up their mind and were 'unlikely to change mind' … A large proportion of women … were amenable to change or were in active deliberation stages ... the largest proportion ... were women who 'had made some choices but were willing to reconsider" Page 6
Raynes‐Greenow 2010	Choice predisposition towards analgesia	"Overall, higher proportions of women in both groups intended to use non‐pharmacological methods for labour pain relief rather than pharmacological methods." Page 6
Raynes‐Greenow 2010	Adherence and acceptability	"Most women had read all of the intervention (decision aid 98% compared to pamphlet group 95%, chi‐square = 2.782, df=1, P = 0.061), and equally both groups would recommend the intervention they received to a pregnant friend (decision aid group 94% compared to pamphlet group 93%, chi‐square, df=1, P = 0.57)" Page 7
Raynes‐Greenow 2010	Source of information	"Both groups equally relied on family and friends, books and antenatal classes" Page 7
Raynes‐Greenow 2010	Labour, Mode of delivery, Birth Weight, Apgar score	All information can be found in Table 4, page 10. There were no significant differences between groups
Roter 2012	Patient satisfaction:identification of problems and concerns	Patient satisfaction: identification of problems and concerns (P = 0.25) Table 6, page 411
Roter 2012	Patient satisfaction:information exchange	Patient satisfaction: information exchange (P = 0.01) Table 6, page 411
Roter 2012	Patient satisfaction:shared decision‐making	Patient satisfaction: shared decision‐making (P = 0.03) Table 6, page 411
Schroy 2011	Screening intentions	"Differences in intention to schedule or complete a screening test for the 2 intervention groups versus control corresponded to moderate effect sizes ranging between 0.36 and 0.44. Scores were comparable for the 2 intervention groups." Page 9
Stiggelbout 2008	Understanding	"The only difference that was seen for the items related to understanding was a difference in favour of the IB group in the stated understanding of the issues that were important in the treatment decision: 84% (n = 32) of the IB group felt that due to the brochure they had better understanding, v. 62% (n = 21) of the GB group (chi‐square test P = 0.004)" Page 756
Stiggelbout 2008	Consultation with the surgeon	"A main difference between the 2 groups was seen in satisfaction with the duration of the consultation … (chi‐square test P = 0.04). … For patients' impression whether the surgeon perceived them more as a medical problem that as a person with a problem, an interaction effect was observed F (1.68)=4.31, P = 0.04." Page 757
Street 1995	Patient knowledge	"The effect for method of communication approached significance (F = 3.30, P = 0.07) as patients in the computer group tended to learn more (mean, 75.5%; SD 13.64%) than did patients in the brochure group (mean, 71.4%; SD, 15.7%)" Page 2279
Street 1995	Patient optimism	"Optimism scores were not affected by … the educational intervention (F = 0.95, P = 0.93)" Page 2279
Street 1995	Patients' behavioural measures	Differences between groups not reported
Street 1995	Perceived involvement in decision making	Differences between groups not reported
Wetzels 2005	Point in time of decision making	The points in time of decision making were not statistically significant (p‐value = 0.93) Table 4, page 595
Wetzels 2005	Patient enablement index	Significant effect size difference: ‐0.232 (‐0.444; ‐0.021) P = 0.03, table 3, page 292
Wetzels 2005	Satisfaction with their care‐EUROPEP	Non significant; effect size difference ‐0.056 (‐0.302; 0.192) P = 0.66, table 3, page 292
Wetzels 2005	Use of leaflet	"Sub‐analyses showed that the scores for these 47 patients did not differ significantly on the outcomes measures from those of the control group or the intervention group non‐users" Page 290
Wetzels 2005	Discussion of one of the eight known underreported health problems	None of the discussion topics were shown to be statistically significant. Table 4, page 292

There was no significant effect detected for most secondary outcomes. No evidence of harms to patients was found following these interventions. We have present outcomes that were statistically significant; however, given that the majority of the outcomes had no effect, caution was needed in determining if the measure was relevant.

Patient health measures

Two studies reported an effect related to patient health (Elwyn 2004; van Peperstraten 2010). The Elwyn 2004 study reported two continuous measures of patient health with a small effect size. The authors nevertheless felt it was not clinically significant. A statistically significant standardized effect size of 0.25 (95% CI 0.02 to 0.49) was reported for one measure of anxiety (lower anxiety) when healthcare professionals received an SDM intervention compared to when they received a risk communication intervention. A statistically significant standardized effect size of 0.24 (95% CI 0.00 to 0.47) was also reported for one measure of mental health status when healthcare professionals received a risk communication intervention compared to when they received an SDM intervention. The van Peperstraten 2010 study reported one categorical measure of patient health with a risk difference of 0.09 (95% CI 0.02 to 0.16) for subclinical depression.

Duration of consultation

An effect related to the duration of the consultation was observed in two studies (Montori 2011; Murray 2010).

Other measurements reported by the healthcare professionals

An effect related to measures reported by the healthcare professionals was observed in five studies (Elwyn 2004; Murray 2010; Roter 2012; Stacey 2006; van Peperstraten 2010) with eight measures. Two studies (Murray 2010; Stacey 2006) showed that the knowledge of the healthcare professional was significantly higher in the intervention group than in the control group. One study (Elwyn 2004) using three measures reported that, according to the healthcare professionals, patients in the intervention group had greater agreement with their provider, satisfaction with the decision making and overall consultation, and satisfaction with the information reported. One study (Roter 2012) using two measures reported better treatment adherence and interpersonal rapport in the intervention group. Economic evaluation was only performed in one of the studies included in this review (van Peperstraten 2010); the patient‐mediated intervention effectively reduced the cost of clinical in vitro fertilization by increasing single (versus multiple) embryo transfers.

Other measurements reported by the patients

Details of these results are presented in Table 13.

Discussion

Summary of main results

This updated search added 34 new studies to the five studies included in the original Cochrane review for a total of 39 studies. It should be noted that 1400 professionals were enrolled in the 39 studies, with a minimum enrolment of two (Davison 1997) and a maximum of 270 (Légaré 2012), and there were 23,236 patients overall.

The countries most represented in this review were the United States, Germany and Canada. Only two of the 39 included studies were conducted with international collaborations (Bernhard 2011; Leighl 2011). Primary care was the setting of the majority of included studies and only one study was conducted in both primary and specialized care (Deen 2012). It is noteworthy that the primary outcome of only five out of the 39 studies was the same as the primary outcome of this review, that is a measure of healthcare professionals' adoption of SDM (Dolan 2002; Elwyn 2004; Krist 2007; O'Cathain 2002; Wetzels 2005).

For categorical measures of SDM, we observed no effect.

For continuous measures of SDM, we observed three main types of results: 1) slight significant effect, 2) dose‐response pattern with no conclusive effect, and 3) non‐significant effect. More specifically, for studies using continuous PROMs we observed a slight significant effect in three categories of comparison: 1) interventions targeting patients compared to usual care, 2) interventions targeting patients compared to other interventions targeting patients, and 3) interventions targeting healthcare professionals compared to other interventions targeting healthcare professionals. For studies using continuous OBOMs we observed a slight significant effect in two categories of comparisons: 1) interventions targeting patients compared to other interventions targeting patients, and 2) interventions targeting healthcare professionals compared to usual care.

We observed a non‐significant effect for studies using continuous PROMs in three categories of comparison: 1) interventions targeting both patients and healthcare professionals compared to usual care, 2) interventions targeting both patients and healthcare professionals compared to interventions targeting patients alone, and 3) interventions targeting both patients and healthcare professionals compared to interventions targeting healthcare professionals alone. There was no study reporting a continuous measure of SDM for the last category of comparison, interventions targeting both patients and healthcare professionals compared to interventions targeting both patients and healthcare professionals.

There was no significant effect detected for most of the secondary outcomes either, even for outcomes that could be impacted by adoption of SDM: duration of consultation, patient's health, and cost of the intervention.

Overall, our main results lead us to make the following observations.

First, while one precise intervention cannot be recommended over another, this review suggests that SDM interventions that actively target patients, health professionals, or both, are better than no intervention at all. Also, these results suggest that interventions targeting health professionals may achieve more than interventions targeting patients when each of these single‐target interventions are compared to usual care. In addition, they indicate that among interventions targeting patients some types perform better than others (for example a patient decision aid compared to a booklet (Montori 2011)). Although limited by the number of studies included in each category of comparison, our update does tell us something about whom the intervention should target. Targeting both members of the decision‐making dyad (patient and healthcare professional) may be more likely to be effective than those targeting solely the healthcare professional or solely the patient. SDM represents a complex set of behaviours in which both members of the patient‐healthcare professional dyad, and preferably the whole patient healthcare team, must engage (LeBlanc 2009). Future studies may consider both participants simultaneously to account for the impact of interaction, reciprocity and interdependence on the process (Guerrier 2013).

Second, among the 39 included studies only three targeted more than one type of healthcare professional, but all were positive. Although this appears promising, the lack of studies addressing the interprofessional approach is clearly a major limitation to understanding the implementation of SDM in clinical practice. Many healthcare systems are moving towards an interprofessional healthcare team‐based approach to patient care that will require this approach to decision making (Légaré 2008b). An interprofessional approach to SDM is an emerging field of research (Légaré 2011) and the reporting of an interprofessional approach to SDM is not yet standardized. In this review, authors only needed to report that the intervention involved more than one type of professional to be identified as taking an interprofessional approach to SDM. Therefore, more studies are needed to inform policy makers about the content, definition and effectiveness of an interprofessional approach to SDM.

Third, although the study of the implementation of SDM in healthcare professionals' practice is growing exponentially, we still need more international collaboration. Studies by international collaborations are starting to be published but these international collaborations do not involve low‐income countries, which are still under‐represented in the list of countries in which SDM is on the policy makers' agenda (Härter 2011). One international collaboration involves Australia and Canada, for example; another involves Austria, Australia, Canada, Germany, New Zealand and Switzerland. Multi‐country approaches permit the sharing of expertise and experiences regarding interventions in a range of settings. It would be important to expand this valuable knowledge base by including middle‐ and low‐income countries (International Shared Decision Making 2013). Specialized care clinical settings were also to some extent under‐represented in the studies included in this updated review, with only one study targeting both primary and specialized care. However, only four studies reflected the clinical heterogeneity that is the norm in primary care by focusing on a set of diverse clinical conditions (Elwyn 2004; Haskard 2008; Stacey 2006; Wetzels 2005), indicating that research is still slow in taking this basic characteristic of primary care into account. Most studies included in this updated review focused on licensed healthcare professionals, demonstrating the need for further implementation studies involving healthcare professionals in training as well (Stacey 2009). In terms of the clinical conditions targeted in the included studies, cancer and cardiovascular diseases were the most common. Implementation studies in SDM are thus addressing the diseases that healthcare professionals are most likely to encounter in their practice; these diseases have also been identified as the two most important causes of the global burden of disease (Institute for Helath Metrics and Evaluation 2013). However, more implementation studies in the area of multi‐morbidity are needed (Smith 2012).

Fourth, three of the secondary outcomes were worthy of note, but the results of the secondary outcomes must be interpreted with caution because most of the included studies did not show that the intervention had a statistically significant effect on healthcare professionals' adoption of SDM. First, the impact of SDM on length of consultations is still unclear. Second, in this review, 58 patient health measures were used to describe the impact of interventions on patient health outcomes, and all but two of these measurements (measures of anxiety and measure of mental health status) were non‐significant. Lastly, an economic evaluation was undertaken in only one of the 39 studies included in this review, although this was effective and resulted in a reduction of the cost of the intervention (van Peperstraten 2010). It should be noted that no evidence of harms to patients was found following these interventions.

Quality of the evidence

Overall, when reviewing studies assessing the impact of any interventions to improve the adoption of SDM by healthcare professionals, we observed that the evidence was of low quality. First, there is still no consensus on which type of measure (OBOM or PROM) is most accurate. However, there were differences between studies based on the type of measure they used. Each kind of study used different scales to capture SDM. In OBOM studies, the most commonly used instrument was OPTION, and in PROM studies the 'perceived level of control in decision making' scale (adapted from the Control Preference Scale) was most common. As for studies not using either of these two scales there were as many instruments as studies. These findings confirm that there is still no standardized instrument for assessing the adoption of SDM by healthcare professionals. However, we observed that studies that had coded SDM behaviours into categories that matched the eight essential elements of Makoul’s definition of SDM (Makoul 2006) had the most significant results, and most of these were OBOM studies. This line of inquiry needs to be pursued with a systematic analysis. Finally, it is important to highlight that in only five out of the 39 studies included in this review was the primary outcome of the study the same as the primary outcome of this review (Dolan 2002; Elwyn 2004; Krist 2007; O'Cathain 2002; Wetzels 2005), that is the adoption of SDM by healthcare professionals. This could explain the lack of positive effect in the majority of the studies. As the implementation of SDM in clinical practice was not their primary outcome of interest, they may not have been sufficiently powered to accurately assess its adoption by healthcare professionals.

Second, it is important to note that in line with the EPOC taxonomy of interventions we refer to patient‐mediated interventions as single entities and we have not disentangled the effectiveness of various elements of multifaceted patient‐mediated interventions. However, this information is contained in the tables. Moreover, we included a number of EPOC intervention types in the same intervention category. It would be important to consider the distinctions between EPOC intervention types in a further update that includes more studies.

In conclusion, due to the heterogeneity of interventions that were used, primary outcomes assessed, and the risks of bias that were observed, we cannot draw a robust conclusion regarding the objectives of our review, that is about the most effective types of intervention for increasing the adoption of SDM by healthcare professionals. The message of the study is nevertheless that SDM interventions that actively target patients, health professionals, or both, are better than no intervention at all. Also, it appears more promising to use interventions that target both the patient and the health professional together than those that target either the patient or the health professional alone. The overall quality of the evidence for the outcomes, assessed with the GRADE tool, ranged from low to very low.

Potential biases in the review process

We observed a potential publication bias in studies reporting a continuous OBOM measure of SDM. There appeared to be a lack of negative, continuous OBOM studies, implying that positive continuous OBOM studies might be over‐represented.

The adoption of SDM by healthcare professionals translates into the performance of a number of SDM‐related behaviours by both the patient and the healthcare professional (Frosch 2009; Légaré 2007a). We acknowledge that the assessment of this complex behaviour in healthcare professionals, and even more so in dyads of patients and healthcare professionals, is challenging and may suffer from many measurement biases (Butow 2009).

Overall, we were unable to extract much information regarding the general context of the included studies. We relied on published and publicly available material and contacted authors of included studies to obtain more information when needed. However, we were not able to always get an answer from them.

Agreements and disagreements with other studies or reviews

The Dwamema update (Dwamena 2012) of a Cochrane systematic review (Lewin 2001) on the effects of interventions targeting healthcare professionals that aim to promote patient‐centered care approaches in clinical consultations concluded that some interventions, such as training activities, are effective across studies in transferring patient‐centered skills to providers. The new finding of the Dwamema review was that short‐term training (less than 10 hours) is as successful as longer training for promoting patient‐centered care within clinical consultations. All the studies included in Dwamema's review that identified shared decision making as an aim of patient‐centered care (Bieber 2008; Krones 2008 (ARRIBA‐Herz); Loh 2007; Longo 2006) were also included in the present review, some as primary studies (Krones 2008 (ARRIBA‐Herz); Loh 2007) and others as complementary studies (Bieber 2008; Longo 2006).

Our review sought studies on all the types of intervention suggested by the EPOC taxonomy, including patient‐mediated interventions, while the Dwamema review focused solely on interventions targeting healthcare professionals in training. We believe that together the reviews add to the knowledge base and can inform policy makers on important implementation strategies regarding SDM in healthcare professionals' practices.

We also identified a recently published Cochrane review on the effects of interventions to promote SDM with children aged four to 18 years who are suffering from cancer (Coyne 2011). This review did not find any eligible studies.

Finally, the idea that effective interventions for changing clinical practice must target patients as well as healthcare professionals is gaining interest outside the SDM community. A recent systematic review on factors that differentiate between effective and ineffective computerized clinical decision support systems in improving the process of care or improving patient outcomes indicated that the likelihood of success was greater with systems that provided advice to patients and practitioners concurrently (Roshanov 2013).

Figure 1

Flow diagram of Cochrane update on interventions for improving the adoption of shared decision making by healthcare professionals (up to 31 December 2012).

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary for each included study.

Figure 4

Patient‐reported outcome (categorical measure).

Figure 5

Observer‐based outcome (continuous measure).

Figure 6

Patient‐reported outcome measure (continuous measure).

Interventions targeting patients compared with usual care for improving the adoption of shared decision making by healthcare professionals
Outcomes*	Type of outcome	Median of the standardized mean difference or median of the risk difference (range)	No of measures (studies**)	Quality of the evidence (GRADE)
Observer‐based SDM measures	Continuous measure	Unavailable data	3 (1)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.21 (0.04 to 0.50)	6 (4)	Very low ^1,2,3
	Categorical measure	‐0.02 (‐0.28 to ‐0.01)	5 (4)	Very low ^1,2,3
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting patients compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.13 (1.04 to 1.21)	2 (2)	Very low ^1,2,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.29 (‐0.05 to 0.63)	6 (2)	Very low ^1,2,3
	Categorical measure	0.04 (‐0.21 to 0.12)	11 (8)	Low ^1,2
	Qualitative quote	0 significant study on 3	3 (3)	Very low ^1,2,4
Interventions targeting healthcare professionals compared with usual care for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.08 (0.38 to 2.07)	4 (3)	Very low ^1,2,3,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.11	1 (1)	Very low ^1,2
	Categorical measure	0.05 (0.00 to 0.09)	3 (2)	Low ^2,3
	Qualitative quote	0 significant study on 1	1 (1)	Very low ^2,4
Interventions targeting healthcare professionals compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	‐0.12	1 (1)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting healthcare professionals compared with another intervention targeting healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	‐0.3	1 (1)	Very low ^2,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.20 (‐0.09 to 0.48)	7 (2)	Very low ^1,2,3
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Interventions targeting both patients and healthcare professionals compared with usual care for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	2.83	4 (2)	Very low ^1,2,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant stdy on 1	1 (1)	Very low ^2,4
Patient‐reported SDM measures	Continuous measure	0.16	3 (3)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant study on 2	2 (2)	Very low ^1,2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting patients for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	1.42	1 (1)	Very low ^2,4,5
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.09 (‐0.06 to 0.73)	5 (3)	Very low ^1,2,3
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant measure on 2	2 (1)	Very low ^2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	0.06	1 (1)	Very low ^1,2
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	1 significant study on 1	1 (1)	Very low ^1,2,4
Interventions targeting both patients and healthcare professionals compared with another intervention targeting both patients and healthcare professionals for improving the adoption of shared decision making by healthcare professionals
Observer‐based SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	‐0.04	1 (1)	Very low ^1,2
	Qualitative quote	Unavailable data	0 (0)
Patient‐reported SDM measures	Continuous measure	Unavailable data	0 (0)
	Categorical measure	Unavailable data	0 (0)
	Qualitative quote	Unavailable data	0 (0)
* Where studies reported more than one measure for each endpoint, the primary measure (as defined by the authors of the study) or the median measure was abstracted. For categorical measures, we calculated the risk difference between the intervention of interest and the control intervention across various outcomes. Forcontinuous endpoints, we calculated standardized mean difference by dividing the mean score difference of the intervention and comparison groups in each study by the pooled standard deviation estimate for the two groups across various outcomes ** Three studies reported results in more than one type of measure
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Important risk of bias according to EPOC checklist ² Indirectness of evidence ³ Heterogeneity ⁴ Imprecision of the observed effect ⁵ Publication bias

Table 1. Effect of interventions: Intervention targeting patients compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Control	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Physician informative and participatory	NA	‐0,04 (0,36)	NA	0,09 (0,38)	Unit of error analysis
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Patient active	NA	0,00 (0,30)	NA	0,05 (0,35)	Unit of error analysis
Haskard 2008	Patient mediated intervention (n=67)	Usual Care (n=80)	Physician‐patient interaction	NA	‐0,01 (0,43)	NA	0,03 (0,46)	Unit of error analysis
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Control	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Control	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Control	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Deen 2012	Patient mediated intervention (Decision aid) (n=69)	Usual Care (n=69)	Patient Activation Measure (PAM)	41,78 (5,42)	43,68 (5,28)	42,21 (5,22)	44,06 (5,66)	‐0,07 (‐0,40 to 0,26)	0,04 (‐0,07 to 0,09)
Deen 2012	Patient mediated intervention (Patient Activation) (n=73)	Usual Care (n=69)	Patient Activation Measure (PAM)	42,31 (6,35)	44,57 (6,16)	42,21 (5,22)	44,06 (5,66)	0,09 (‐0,24 to 0,41)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Usual Care (n=69)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	42,21 (5,22)	44,06 (5,66)	0,04 (‐0,29 to 0,38)
van Peperstraten 2010	Patient mediated intervention (n=124)	Usual Care (n=128)	Decision Evaluation scale	NA	4,1 (0,56)	NA	3,8 (0,57)	0,50 (0,25 to 0,75)
Vodermaier 2009	Patient mediated intervention	Usual Care	Man‐Son‐Hing Instrument	No data
Cooper 2011	Patient mediated intervention (n=40)	Usual Care (n=43)	Participatory Decision making (PDM)	70,94 (24,67)	74,17 (23,25)	74,61 (21,59)	69,38 (21,50)	0,21 (‐0,22 to 0,64)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Control	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Krist 2007	Patient mediated intervention (Decision aid brochure) (n=174)	Usual Care (n=63)	Modified Control Preference Scale	NA	63/174	NA	23/63	0,00 (‐0,14 to 0,14)	‐0,01 (‐0,01 to 0,00)
Krist 2007	Patient mediated intervention (Decision aid web) (n=198)	Usual Care (n=63)	Modified Control Preference Scale	NA	71/198	NA	23/63	‐0,01 (‐0,14 to 0,13)
Landrey 2012	Patient mediated intervention (n=74)	Usual Care (n=78)	Modified Control Preference Scale	NA	29/74	NA	33/78	‐0,03 (‐0,19 to 0,12)	‐0,03
Murray 2001	Patient mediated intervention (n=57)	Usual Care (n=48)	Modified Control Preference Scale	NA	34/57	NA	42/48	‐0,28 (‐0,44 to ‐0,12)	‐0,28
Vodermaier 2009	Patient mediated intervention (n=53)	Usual Care (n=54)	Modified Control Preference Scale	NA	35/53	NA	36/54	‐0,01 (‐0,19 to 0,17)	‐0,01
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Control	Outcome	Qualitative quote
No study

Table 1. Effect of interventions: Intervention targeting patients compared to usual care

Table 2. Effect of interventions: Intervention targeting patients compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Montori 2011	Patient mediated intervention (n=52)	Patient mediated intervention (n=48)	OPTION	NA	49,80 (21,40)	NA	27,30 (14,70)	1,21 (0,78 to 1,64)	1,21
Nannenga 2009	Patient mediated intervention (n=48)	Patient mediated intervention (n=43)	OPTION	NA	7,13 (6,63)	NA	1,74 (2.53)	1,04 (0,60 to 1,48)	1,04
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Deen 2012	Patient mediated intervention (Decision aid) (n=69)	Patient mediated intervention (Patient Activation) (n=73)	Patient Activation Measure (PAM)	41,78 (5,42)	43,68 (5,28)	42,31 (6,35)	44,57 (6,16)	‐0,15 (‐0,48 to 0,18)	‐0,05 (‐0,15 to 0,11)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Patient mediated intervention (Patient Activation) (n=73)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	42,31 (6,35)	44,57 (6,16)	‐0,05 (‐0,38 to 0,28)
Deen 2012	Patient mediated intervention (Decision aid + Patient Activation) (n=68)	Patient mediated intervention (Decision aid) (n=69)	Patient Activation Measure (PAM)	41,67 (5,68)	44,29 (5,47)	41,78 (5,42)	43,68 (5,28)	0,11 (‐0,22 to 0,45)
Schroy 2011	Patient mediated intervention (Decision aid) (n=205)	Patient mediated intervention (Educational material) (n=217)	Satisfaction with the decision making process	NA	50,70 (6,20)	NA	46,00 (7,90)	0,66 (0,46 to 0,85)	0,63 (‐0,03 to 0,66)
Schroy 2011	Patient mediated intervention (Decision aid + YDR) (n=214)	Patient mediated intervention (Educational material) (n=217)	Satisfaction with the decision making process	NA	50,50 (6,20)	NA	46,00 (7,90)	0,63 (0,44 to 0,83)
Schroy 2011	Patient mediated intervention (Decision aid + YDR) (n=214)	Patient mediated intervention (Decision aid) (n=205)	Satisfaction with the decision making process	NA	50,50 (6,20)	NA	50,70 (6,20)	‐0,03 (‐0,22 to 0,16)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Butow 2004	Patient mediated intervention (n=69)	Patient mediated intervention (n=62)	Modified Control Preference Scale	NA	22/69	NA	17/62	0,04 (‐0,11 to 0,20)	0,04
Davison 1997	Patient mediated intervention (n=30)	Patient mediated intervention (n=30)	Modified Control Preference Scale	NA	10/30	NA	15/30	‐0,17 (‐0,41 to 0,08)	‐0,17
Deschamps 2004	Patient mediated intervention (n=42)	Patient mediated intervention (n=48)	Modified Control Preference Scale	NA	24/42	NA	22/48	0,11 (‐0,09 to 0,32)	0,11
Dolan 2002	Patient mediated intervention (n=43)	Patient mediated intervention (n=43)	Modified Control Preference Scale	NA	27/43	NA	22/43	0,12 (‐0,09 to 0,32)	0,12
Kasper 2008	Patient mediated intervention (n=136)	Patient mediated intervention (n=142)	Modified Control Preference Scale	NA	55/136	NA	53/142	0,03 (‐0,20 to 0,27)	0,03
Krist 2007	Patient mediated intervention (Decision aid web) (n=198)	Patient mediated intervention (Decision aid brochure) (n=174)	Modified Control Preference Scale	NA	71/198	NA	63/174	0,00 (‐0,10 to 0,09)	0
Raynes‐Greenow 2010	Patient mediated intervention (Decision Aid (Audio)) (n=176)	Pamphlet (n=175)	Modified CPS ‐ First Follow‐up	NA	39/176	NA	31/175	0,04 (‐0,04 to 0,13)	0,04 (0,04 to 0,07)
Raynes‐Greenow 2010	Patient mediated intervention (Decision aid) (n=168)	Pamphlet (n=175)	Modified CPS ‐ First Follow‐up	NA	37/168	NA	31/175	0,04 (‐0,04 to 0,13)
Raynes‐Greenow 2010	Patient mediated intervention (Decision Aid (Audio)) (n=141)	Pamphlet (n=136)	Modified CPS ‐ Second Follow‐up	NA	26/141	NA	19/136	0,04 (‐0,04 to 0,13)
Raynes‐Greenow 2010	Patient mediated intervention (Decision aid) (n=150)	Pamphlet (n=136)	Modified CPS ‐ Second Follow‐up	NA	31/150	NA	19/136	0,07 (‐0,02 to 0,13)
Stiggelbout 2008	Patient mediated intervention (n=31)	Patient mediated intervention (n=33)	Modified Control Preference Scale	NA	16/31	NA	24/33	‐0,21 (‐0,44 to 0.02)	‐0,21
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Lalonde 2006	Patient mediated intervention	Patient mediated intervention	Decision satisfaction inventory	No statistically significant differences in patient satisfaction with the decision‐making process were detected between the study groups. Page 55
Street 1995	Patient mediated intervention	Patient mediated intervention	Perceived Decision Control Instrument	The experimental manipulation (computer program versus brochure) had very little effect on the dependent variables. Page 2280
Butow 2004	Patient mediated intervention	Patient mediated intervention	Physician behaviours facilitating patient involvement	On average, oncologists demonstrated about 7.5 of the 12 behaviours, with no significant differences between the groups (cancer consiltation preparation package (CCPP) versus control booklet). Page 4406

Table 2. Effect of interventions: Intervention targeting patients compared to another intervention targeting patients

Table 3. Effect of interventions: Intervention targeting healthcare professionals compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Fossli 2011	Educational meeting, audit and feedback, distribution of educational material (n=26)	Usual Care (n=25)	Fours Habits Coding Scheme (4HCS)	59,66 (8,78)	63,57 (11,96)	60,87 (11,08)	58,85 (12,19)	0,38 (‐0,17 to 0,94)	0,38
Shepherd 2011	Educational outreach visit (n=18)	Usual Care (n=18)	Assessing Communication about Evidence and Patient Preferences (ACEPP)	NA	21,30 (3,58)	NA	16,70 (3,63)	0,90 (0,21 to 1,58)	1,08 (0,90 to 1,25)
Shepherd 2011	Educational outreach visit (n=18)	Usual Care (n=18)	OPTION	NA	36,60 (12,62)	NA	25,00 (12,72)	1,25 (0,53 to 1,97)
Stacey 2006	Distribution of educational materials, educational meeting, audit and feedback and barriers assessment (n=18)	Usual Care (n=20)	Decision Support Analysis Tool (DSAT)	0,53 (0,18)	0,81 (0,17)	0,43 (0,17)	0,44 (0,18)	2,07 (1,26 to 2,87)	2,07
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Educational meeting (n=51)	Usual Care (n=43)	Participatory Decision making (PDM)	68,46 (22,81)	71,57 (19,94)	74,61 (21,59)	69,38 (21,50)	0,11 (‐0,30 to 0,51)	0,11
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Légaré 2012	Educational meeting and distribution of educational material (n=176)	Usual Care	Modified Control Preference Scale (n=177)	71/182	79/176	59/171	64/177	0,09 (‐0,01 to 0,19)	0,09
O'Cathain 2002	Educational meeting and distribution of educational material (Pre: n=1526; Post: n=1531)	Usual Care	Modified Control Preference Scale (antenatal sample) (Pre: n=1219; Post: n=1206)	345/1526	263/1531	287/1219	235/1206	‐0,02 (‐0,05 to 0,01)	0,00 (‐0,02 to 0,02)
O'Cathain 2002	Educational meeting and distribution of educational material (Pre: n=1490; Post: n=1515)	Usual Care	Modified Control Preference Scale (postnatal sample) (Pre: n=1666; Post: n=1698)	369/1490	354/1515	426/1666	358/1698	0,02 (‐0,01 to 0,05)
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Bernhard 2011	Educational meeting, audit and feedback, distribution of educational material	Usual Care	Patient involvement preference and actual involvement	There was considerable variation in patient outcomes between the SGA and ANZ cohorts and no substantial training effect. Page 6

Table 3. Effect of interventions: Intervention targeting healthcare professionals compared to usual care

Table 4. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Educational meeting	Patient mediated intervention (n=51)	Participatory Decision making (PDM) (n=40)	68,46 (22,81)	71,57 (19,94)	70,94 (24,67)	74,17 (23,25)	‐0,12 (‐0,53 to 0,29)	‐0,12
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Table 4. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting patients

Table 5. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Elwyn 2004	Educational Meeting and Audit and feedback (n=9)	Educational Meeting and Audit and feedback (n=11)	OPTION	27,00 (14,00)	39,00 (11,80)	32,00 (13,80)	43,00 (13,60)	‐0,30 (‐1,19 to 0,59)	‐0,3
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=79; Post: n=139)	Educational Meeting and Audit and feedback (Pre: n=108; Post: n=188)	COMRADE (communication) ‐ Time 1	63,50 (18,60)	67,30 (14,10)	66,30 (13,50)	68,30 (14,10)	‐0,07 (‐0,29 to 0,15)	‐0,09 (‐0,18 to 0,05)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=69; Post: n=121)	Educational Meeting and Audit and feedback (Pre: n=94; Post: n=169)	COMRADE (communication) ‐ Time 2	62,10 (18,10)	62,40 (17,00)	63,30 (16,20)	64,20 (16,30)	‐0,11 (‐0,34 to 0,13)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=79; Post: n=139)	Educational Meeting and Audit and feedback (Pre: n=108; Post: n=188)	COMRADE (confidence) ‐ Time 1	72,00 (9,90)	74,20 (9,40)	72,00 (9,80)	73,70 (9,20)	0,05 (‐0,17 to 0,27)
Elwyn 2004	Educational Meeting and Audit and feedback (Pre: n=69; Post: n=121)	Educational Meeting and Audit and feedback (Pre: n=94; Post: n=169)	COMRADE (confidence) ‐ Time 2	70,00 (10,80)	70,00 (13,10)	71,80 (9,30)	72,20 (11,00)	‐0,18 (‐0,42 to 0,05)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=582)	Educational Meeting (n=550)	PPS (Man Son‐Hing) : I made the decision jointly (Score inversé pour respecter le sens de l'échelle)	NA	1,36 (0,25)	NA	1,24 (0,25)	0,48 (0,36 to 0,60)	0,48 (0,40 to 6,11)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=550)	Educational Meeting (n=582)	Shared Decision Making Q (SDM‐Q)	NA	9,18 (4,08)	NA	7,46 (4,5)	0,40 (0,28 to 0,52)
Krones 2008	Educational meeting, audit and feedback, educational material and educational outreach visit (n=539)	Educational Meeting (n=513)	PPS (Man‐Son‐Hing)	NA	7,69 (0,16)	NA	6,87 (0,1)	6,11 (5.82 to 6.40)
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Table 5. Effect of interventions: Intervention targeting healthcare professionals compared to another intervention targeting healthcare professionals

Table 6. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to usual care

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Physician informative and participatory (n=66)	NA	0,02 (0,39)	NA	‐0,10 (0,41)	Unit of error analysis
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Patient active (n=66)	NA	‐0,02 (0,32)	NA	‐0,08 (0,37)	Unit of error analysis
Haskard 2008	Patient mediated intervention + Distribution of educational material + education meeting (n=61)	Usual Care	Physician‐patient interaction (n=66)	NA	‐0,03 (0,46)	NA	‐0,06 (0,50)	Unit of error analysis
Hess 2012	Patient mediated intervention + educational meeting (n=100)	Usual Care	OPTION (n=100)	NA	26,60 (8,10)	NA	7,00 (5,50)	2,83 (2,44 to 3,22)	2,83
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Murray 2010	Educational meeting, audit and feedback, distribution of educational materials, educational outreach, barriers assement	Usual Care	Decision Support Analysis Tool (DSAT)	"The mean score change from baseline in the intervention group 3.75 (95% CI 2.46 to 5.03) was significantly greater than the mean score change in the control group ‐0.667 (95% CI ‐1.57 to 0.24) using the two sided t‐test (P < 0.0001)" Page 116
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Usual Care	Participatory Decision making (PDM) (n=43)	66,67 (23,98)	72,84 (21,19)	74,61 (21,59)	69,38 (21,50)	0,16 (‐0,23 to 0,56)	0,16
Hamman 2007	Patient mediated intervention + Educational meeting (n=33)	Usual Care	Combined Outcome Measure for Risk Communication and Treatment (COMRADE) (n=49)	NA	76,8 (20,9)	NA	73,5 (19,3)	0,16 (‐0,28 to 0,61)	0,16
Wetzels 2005	Patient mediated Intervention + educational outreach visit (n=121)	Usual Care	Combined Outcome Measure for Risk Communication and Treatment (COMRADE) ‐ 4 items (n=142)	1,82 (NA)	1,83 (NA)	1,89 (NA)	1,80 (NA)	Unable to calculate. No differences between groups were detected.	NA
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Leighl 2011	Patient mediated intervention and educational meeting	Usual Care	Modified CPS	There was no difference after the intervention: the mean score of the item on the CPS scale in the intervention group was: 2.86 (0.92), it was 2.87 (1.04) in the control group. See Figure 4, page 2082. Data are from the authors.
Loh 2007	Patient mediated intervention and educational meeting	Usual Care	PPS (Man‐Son‐Hing)	In the intervention group, significantly higher patient participation from pre‐ to post‐intervention was found for … the Man‐Son‐Hing patient participation scale, P = 0.10. Page 329

Table 6. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to usual care

Table 7. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting patients

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Mullan 2009	Patient mediated intervention + Education meeting (n=21)	Patient mediated intervention (n=19)	OPTION	NA	49,70 (17,74)	NA	27,70 (11,75)	1,42 (0,72 to 2,12)	1,42
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (first consultation) (n=33)	NA	4,11 (0,7)	NA	3,59 (0,7)	0,73 (0,24 to 1,23)	0,73 (0,50 to 0,88)
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (3 months) (n=33)	NA	4,05 (0,7)	NA	3,67 (0,8)	0,50 (0,01 to 0,99)
Bieber 2006	Patient mediated intervention and educational meeting (n=34)	Patient mediated intervention	Questionnaire on the Quality of Physician‐Patient Interaction (QQPPI) (6 months) (n=33)	NA	3,8 (0,8)	NA	3,13 (0,7)	0,88 (0,38 to 1,38)
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Patient mediated intervention	Participatory Decision making (PDM) (n=40)	66,67 (23,98)	72,84 (21,19)	70,94 (24,67)	74,17 (23,25)	‐0,06 (‐0,46 to 0,34)	‐0,06
Mullan 2009	Patient mediated intervention + Educational meeting (n=47)	Patient mediated intervention	Validated pictorial instrument (n=36)	NA	4,8 (1,1)	NA	4,7 (1,1)	0,09 (‐0,34 to 0,52)	0,09
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Deinzer 2009	Patient mediated intervention + Educational meeting	Patient mediated intervention	Combined Outcome Measure for Risk Communication and Treatment (COMRADE)	The degree of SDM was significantly higher in the SDM group at basline and after 1‐year visits. Both groups showed an increase in SDM (both P = 0.001). Page 268
Deinzer 2009	Patient mediated intervention + Educational meeting	Patient mediated intervention	Autonomy Preference Index (API)	The preference for SDM as assessed by the API (Figure 2) showed no differences between the SDM and control group at baseline (P = 0.60) and did not change after 1 year (P = 0.83). Page 268

Table 7. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting patients

Table 8. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
Cooper 2011	Patient mediated intervention + Educational meeting (n=58)	Educational meeting (n=51)	Participatory Decision making (PDM)	66,67 (23,98)	72,84 (21,19)	68,46 (22,81)	71,57 (19,94)	0,06 (‐0,32 to 0,44)	0,06
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
Roter 2012	Patient mediated intervention and distribution of educational materials	Distribution of educational materials	LEAPS	The study interventions led to significant and parallel increases in both patient and physician reported use of patient‐centered communication skills, and an increase in patient satisfaction with communication‐related visit goals. For patients, the intervention was associated with a positive change in reported skills in five of the six communication areas. Page 412

Table 8. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to another intervention targeting healthcare professionals

Table 9. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to intervention targeting both patients and healthcare professionals

Observer‐based outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Observer‐based outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
Myers 2011	Patient mediated intervention + reminders (n=74)	Patient mediated intervention + reminders (n=60)	Informed decision making scale (IDM)	NA	3/74	NA	5/60	‐0,04 (‐0,13 to 0,04)	‐0,04
Observer‐based outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study
Patient reported outcome measure ‐ Continous Data
Study	Intervention	Intervention	Outcome	Pre mean (SD)	Post mean (SD)	Pre mean (SD)	Post mean (SD)	SMD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Categorical Data
Study	Intervention	Intervention	Outcome	Pre n/N	Post n/N	Pre n/N	Post n/N	RD	Median (Range) by Study
No study
Patient reported outcome measure ‐ Qualitative statement
Study	Intervention	Intervention	Outcome	Qualitative quote
No study

Table 9. Effect of interventions: Intervention targeting both patients and healthcare professionals compared to intervention targeting both patients and healthcare professionals

Table 10. Secondary outcome: patient health measures (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Elwyn 2004	Anxiety (short form of Spielberger) Time 1	Pre: 79 Post: 138	11.33 (3.74)	10.00 (3.55)	Pre: 107 Post: 187	11.62 (3.67)	9.86 (3.78)	Pre: ‐0.08 (‐0.37 to 0.21) Post: 0.04 (‐0.18 to 0.26)
Elwyn 2004	Anxiety (short form of Spielberger) Time 2	Pre: 73 Post: 117	9.94 (3.42)	11.25 (4.28)	Pre: 92 Post: 164	10.36 (3.59)	10.23 (3.79)	Pre: ‐0.12 (‐0.43 to 0.19) Post: 0.25 (0.02 to 0.49)
Elwyn 2004	Anxiety (short form of Spielberger) Time 3	Pre: 61 Post: 101	10.15 (3.24)	10.51 (3.93)	Pre: 75 Post: 136	10.87 (3.55)	9.99 (3.23)	Pre: ‐0.21 (‐0.55 to 0.13) Post: 0.15 (‐0.11 to 0.40)
Elwyn 2004	Health status (SF‐1220) mental subscale Time 1	Pre: 101 Post: 171	48.65 (10.26)	50.41 (10.90)	Pre: 68 Post: 124	50.31 (9.66)	47.77 (11.21)	Pre: ‐0.16 (‐0.47 to 0.14) Post: 0.24 (0.00 to 0.47)
Elwyn 2004	Health status (SF‐1220) mental subscale Time 2	Pre: 79 Post: 149	49.11 (11.14)	51.16 (10.41)	Pre: 68 Post: 108	50.16 (10.73)	49.23 (11.98)	Pre: ‐0.09 (‐0.42 to 0.23) Post: 0.17 (‐0.07 to 0.42)
Elwyn 2004	Health status (SF‐1220) physical subscale Time 1	Pre: 101 Post: 171	41.16 (13.05)	42.47 (11.76)	Pre: 68 Post:124	43.01 (12.48)	41.90 (13.08)	Pre: ‐0.14 (‐0.45 to 0.16) Post: 0.05 (‐0.18 to 0.27)
Elwyn 2004	Health status (SF‐1220) physical subscale Time 2	Pre: 79 Post: 149	39.71 (12.35)	40.81 (12.14)	Pre: 68 Post: 108	43.34 (11.46)	40.91 (11.81)	Pre: ‐0.30 (‐0.63 to 0.02) Post: ‐0.01 (‐0.26 to 0.24)
Hamann 2007	Clinical global impression scale	35	NA	4.0 (1.5)	40	NA	4.1 (1.4)	‐0.07 (‐0.52 to 0.38)
Hamann 2007	Global assessment of function scale	30	NA	54.7 (16.5)	37	NA	51.0 (18.5)	0.21 (‐0.27 to 0.69)
Légaré 2012	Quality of life physical scale	181	49.30 (8.80)	49.40 (7.50)	178	47.70 (8.90)	48.20 (7.80)	0.16 (‐0.05 to 0.36)
Van Peperstraten 2010	Level of anxiety	Pre:150 Post: 127	35.60 (10.60)	36.40 (10.20)	Pre: 154 Post: 135	34.60 (9.50)	34.70 (8.20)	0.18 (‐0.06 to 0.43)
Categorical data (n/N)
Hamann 2007	Patient hospitalised within 6 mo after discharge	36	NA	8/36	37	NA	8/37	0.01 (‐0.18 to 0.20)
Hamann 2007	Patient hospitalised within 18 mo after discharge	38	NA	20/38	41	NA	19/41	0.06 (‐0.16 to 0.28)
Hamann 2007	Patient with drug switches (main antipsychotic) within 6 mo after discharge	36	NA	12/36	40	NA	16/40	‐0.07 (‐0.28 to 0.15)
Hess 2012	Admitted to hospital	101	NA	6	103	NA	6	0 (‐0.06 to 0.07)
Hess 2012	Repeat emergency department visit	101	NA	3	103	NA	0	0.03 (‐0.01 to 0.07)
Hess 2012	Rehospitalization	101	NA	2	103	NA	0	0.02 (‐0.01 to 0.05)
Hess 2012	Acute myocardial infarction	101	NA	1	103	NA	0	0.01 (‐0.02 to 0.04)
Légaré 2012	Proportion of use of antibiotics	Pre: 182 Post: 180	75	49	Pre :171 Post:178	67	93	‐0.25 (‐0.35 to ‐0.15)
Van Peperstraten 2010	Subclinical depression	Pre:147 Post: 126	16	16	Pre: 151 Post: 136	13	5	0.09 (0.02 to 0.16)
Qualitative data
Butow 2004	Spielberger State Trait Anxiety Scale	"In both groups, anxiety decreased by 3 points after the consultation, and there was no significant difference between the groups immediately after the consultation and one month later." Page 4407
Butow 2004	Beck Depression Inventory (short form)	"No significant differences between groups were observed in raw or change scores on depression immediately after the consultation or one month later." Page 4407
Mullan 2009	Adherence	" ... adherence to diabetes medications were near perfect in both groups and significantly better in the control group." Page 1565
Mullan 2009	HbA	"The decision aid did not affect glycemic control or patient‐reported health status at six months" Page 1565
Krones 2008 (ARRIBA‐Herz)	Framingham Scoring system	Non significant (P = 0.31)
Bernhard 2011	Anxiety (State Trait Anxiety Inventory)	"Anxiety slightly decreased over time for all cohorts. Patients in the SGA (Figure 4a) and ANZ (Figure 4b) cohorts reported comparable anxiety levels at each time point. The quality of life indicators showed similar findings (data not shown)." Page 6
Bernhard 2011	Quality of life	"Anxiety slightly decreased over time for all cohorts. Patients in the SGA (Figure 4a) and ANZ (Figure 4b) cohorts reported comparable anxiety levels at each time point. The quality of life indicators showed similar findings (data not shown)." Page 6
Bieber 2006	Center for epidemiological studies depression scale ‐ CES‐D	Non significant: P = 0.26 (table 4). Page 363
Bieber 2006	Visual analogue scale for pain intensity	Non significant: P = 0.45 (table 4). Page 363
Bieber 2006	Health status and physical function SF‐36	Non significant: P = 0.89 (table 4). Page 363
Bieber 2006	Hannover Functional Questionnaire FFbH	Non significant: P = 0.81 (table 4). Page 363
Cooper 2011	Blood Pressure control	"Improvements in patient adherence and BP control did not differ across groups for the overall patient sample" p1; "In the overall sample, changes in systolic and diastolic BP at 12 months did not differ for any of the intervention groups when compared to the patient+physician minimal intervention group" p1300; "Changes in patient‐reported adherence to medications at 12 months did not differ for any of the intervention groups compared to the patient+physician minimal intervention group."Page 1300
Davison 1997	Spielberger State Trait Anxiety Scale	"There was no evidence trait scores were different among groups, among measurement times, or between groups and measurement times" Page 195
Davison 1997	Center for epidemiological studies depression scale ‐ CES‐D	"No significant differences in mean depression scores were found among the groups, among measurement times, or between groups and measurement times" Page 196
Deinzer 2009	Self measurement of systolic and diastolic blood pressure	"Thus in both groups BP decreased but there were no significant differences between the 2 groups (systolic P = 0.24 and diastolic P = 0.16 respectively)." Page 268
Hamann 2007	Severity of illness (PANSS)	NA "... there were no differences between ... PANSS score at discharge" Page 994
Hamann 2007	Plasma level of antipsychotic	Not reported
Hamann 2007	Medication at discharge	Not reported
Hess 2012	Major adverse cardiac event	"Excluding the index presentation, there were no deaths or major adverse cardiac events within 30 days" Page 256
Leighl 2011	Functional Assessment of Cancer Therapy–General (FACT‐G)	Patients completed the physical, emotional, and functional subscales of the Functional Assessment of Cancer Therapy ‐ General (FACT‐G) and had similar scores in both arms comparable with those of other patients with advanced cancer. Page 2080
Loh 2007	Brief PHQ‐D ‐ Depression severity	Non significant (P = 0.236)
Murray 2001	Health status and physical function SF‐36	" ... no difference in score was observed between the two groups" Page 5
Murray 2001	Health states and valuation of health states EQ‐SD	" ... no difference in score was observed between the two groups" Page 5
Murray 2001	Spielberger state of trait anxiety inventory short form	"The Spielberger scores were similar in the final assessment in the two groups" Page 5
Murray 2001	Prostatic symptoms (American Urological Association symptom scale)	"The amount of change was not significantly different in the two groups" Page 5
Raynes‐Greenow 2010	Mode of delivery	There were no differences between labour and birth outcomes between the groups P = 0.97 (table 4). See page 10
Raynes‐Greenow 2010	Labour Type	There were no differences between labour and birth outcomes between the groups P = 0.97 (table 4). See page 10
Raynes‐Greenow 2010	Analgesia used	There were no significant differences between groups in regards to analgesia use (P = 0.18‐0.84). See page 7
Raynes‐Greenow 2010	Apgar score	P = 0.12 (1 minute) and P = 0.68 ( 5 minutes) (table 4). See page 10
Stiggelbout 2008	Quality of life (HADS)	"Patients' quality of life was stable over time, in both groups. No effects were observed in the repeated measures for the anxiety and depression scales of the HADS, nor on the quality of life scales" Page 757
Stiggelbout 2008	100 mm visual analogue	"Patients' quality of life was stable over time, in both groups. No effects were observed in the repeated measures for the anxiety and depression scales of the HADS, nor on the quality of life scales" (100 mm visual analogue scale) Page 757

Table 10. Secondary outcome: patient health measures (Positive studies are in italics)

Table 11. Secondary outcome: duration of consultation (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Stacey 2006	Call length	Pre: 18 Post: 18	17.80 (4.50)	18.50 (6.30)	Pre: 20 Post: 20	16.70 (7.70)	16.70 (6.50)	Pre: 0.17 (‐0.47 to 0.81) Post: 0.27 (‐0.36 to 0.91)
Qualitative data
Butow 2004	Consultation length	"Consultation length was similar between groups ‐ on average, 36 minutes per consultation." Page 4407
Elwyn 2004	Consultation length	"There was no difference in the mean consultation lengths at baseline, phase 1 and phase 2 (overall consultation mean duration was 12.5 minutes)" Page 342
Fossli 2011	Consultation length	"There was a non significant difference between both groups ( RD: ‐1:03 CI ‐6:13;4:07) P = 0.69" Page 4
Krist 2007	Consultation length	"These [discussion times] patient‐physician differences did not differ significantly across the control, brochure, and Web groups." Page 116
Loh 2007	Consultation length	Non significant differences between the groups (Table 2) Page 329
Montori 2011	Consultation length	"The median (range)duration of osteoporosis discussions was 12.4 minutes (2.3‐27.4) in the decision aid arm compared with 9.4 minutes (2.1‐58) in the usual care arm (P .045)" Page 552‐553
Murray 2001	Consultation length	Not reported
Murray 2010	Consultation length	"At baseline there was no significant difference. However, in the post‐calls, the mean call duration was longer in the intervention group at 13,47 minutes (95% confidence interval 11.8;14.21), than in the control group at 10.29 minutes (95% CI 8.79 to 11.79 P = 0.004)" Page 117
Nannenga 2009	Consultation time	"We found no significant difference in face‐to‐face consultation duration with the staff endocrinologist (mean difference 3.8 min longer with the decision aid, 95% CI ‐ 2.9 to 10.5)." Page 42
Shepherd 2011	Consultation length	"These effects occurred without any significant difference in consultation length, mean consultation lengths were 26 minutes for control and intervention visits." Page 381
Vodermaier 2009	Consultation time	"No time differences emerged in the length of the treatment decision consultation with the physicians on patient self‐reports. The mean time for the treatment decision making appointment was about 15 minutes" Page 593
Wetzel 2005	Consultation time	No differences between intervention and control groups were detected, consultations was between 12.2 and 13 minutes for all groups (Table 4) Page 292

Table 11. Secondary outcome: duration of consultation (Positive studies are in italics)

Table 12. Secondary outcome: other measurement reported by the healthcare professional (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Haskard 2008	Physician satisfaction questionnaire	61	NA	74.82 (5.47)	66	NA	74.60 (6.47)	Unit of error analysis
Haskard 2008	Satisfaction with the management and functioning of their office practice	61	NA	3.20 (0.65)	66	NA	3.08 (0.58)	Unit of error analysis
Haskard 2008	Overall quality of life	63	NA	3.00 (0.83)	63	NA	2.82 (0.73)	Unit of error analysis
Haskard 2008	Stress	61	NA	2.68 (0.69)	66	NA	2.78 (0.60)	Unit of error analysis
Mullan 2009	Acceptability amount of information	21	NA	6.59 (0.91)	19	NA	6.37 (1.14)	0.20 (‐0.41 to 0.83)
Mullan 2009	Acceptability clarity of information	21	NA	6.20 (0.96)	19	NA	6.20 (0.80)	0.00 (‐0.62 to 0.62)
Mullan 2009	Helpfulness of the information	21	NA	6.15 (0.94)	19	NA	5.74 (1.04)	0.41 (‐0.22 to 1.03)
Mullan 2009	Would recommend to others	21	NA	6.16 (1.51)	19	NA	5.89 (1.82)	0.16 (‐0.46 to 0.78)
Mullan 2009	Would want to use for other decisions	21	NA	6.04 (1.55)	19	NA	5.69 (1.75)	0.21 (‐0.44 to 0.84)
Murray 2010	Knowledge	35	NA	69.30 (2.98)	35	NA	60.50 (2.27)	3.28 (2.55 to 4.02)
Krones 2008 (ARRIBA‐Herz)	Patient participation scale, physician rating	19	NA	1.66 (0.45)	26	NA	1.65 (0.48)	0.02 (‐0.57 to 0.61)
Bieber 2006 (first consultation)	Difficult doctor patient questionnaire	34	NA	29.40 (5.80)	33	NA	33.50 (10.00)	‐0.50 (‐0.98 to ‐0.02)
Bieber 2006 (month follow up)	Difficult doctor patient questionnaire	34	NA	28.90 (6.70)	33	NA	32.20 (6.50)	‐0.49 (‐0.98 to ‐0.01)
Légaré 2012	Physician quality of decision	Pre: 172 Post: 166	8.20 (1.10)	8.20 (1.30)	Pre: 162 Post: 170	8.20 (1.40)	8.40 (1.00)	‐0.17 (‐0.39 to 0.04)
Légaré 2012	Physician intention to follow CPG	Pre: 151 Post: 132	1.60 (0.80)	1.70 (0.90)	Pre: 108 Post: 98	1.60 (0.90)	1.80 (0.70)	‐0.12 (‐0.38 to 0.14)
Loh 2007	Physician's assessment of treatment adherence	96	4.20 (1.10)	4.30 (1.10)	191	4.30 (0.90)	4.80 (0.60)	Intracluster correlation error
Categorical data: (n/N)
Légaré 2012	Physician Decisional Conflict (Proportion who had a value of 2.5 or more )	Pre: 178 Post: 175	8	8	Pre: 166 Post: 176	5	2	0.03 (‐0.00 to 0.07)
Murray 2001	Perceived role in decision making: shared role	48	NA	25/48	49	NA	32/49	‐0.13 (‐0.33 to 0.06)
Vodermaier 2009	Chose Breast‐conserving therapy	39	NA	37/39	41	NA	36/41	0.07 (‐0.05 to 0.19)
Vodermaier 2009	Chose Chemotherapy	35	NA	11/35	39	NA	11/39	0.03 (‐0.18 to 0.24)
Vodermaier 2009	Chose pre‐operative chemotherapy	16	NA	10/16	15	NA	7/15	0.16 (‐0.19 to 0.50)
Qualitative data
Butow 2004	Physician satisfaction with the decision making process	"Physicians were also equally satisfied with decision making whether or not their patients had received the CCPP or the control booklet" Page 4407
Elwyn 2004	Clinician perception of the level of clinician agreement	"Clinicians showed significant differences between the RC and SDM arms (see Table S3). Doctors receiving the risk communication tools and training first perceived significantly higher doctor–patient agreement on treatment (P 0.001), patient satisfaction with information (P = 0.01), doctor satisfaction with decision (P = 0.01) and general overall satisfaction (P = 0.001) with the consultation than those who were exposed to SDM training. The latter group of doctors showed lower scores after the interventions. The differences were largely maintained in the second intervention phase, i.e. even when provided with the risk communication training and tools, the group of doctors who had received SDM training first still reported lower levels of satisfaction, agreement, etc. In contrast, doctors who had received risk communication training first maintained their higher levels of satisfactions and agreement, even when later given the SDM training which appeared less beneficial (to doctors) in the first phase." Page 343
Elwyn 2004	Clinician satisfaction with the decision and overall consultation	"Clinicians showed significant differences between the RC and SDM arms (see Table S3). Doctors receiving the risk communication tools and training first perceived significantly higher doctor–patient agreement on treatment (P 0.001), patient satisfaction with information (P 0.01), doctor satisfaction with decision (P 0.01) and general overall satisfaction (P 0.001) with the consultation than those who were exposed to SDM training. The latter group of doctors showed lower scores after the interventions. The differences were largely maintained in the second intervention phase, i.e. even when provided with the risk communication training and tools, the group of doctors who had received SDM training first still reported lower levels of satisfaction, agreement, etc. In contrast, doctors who had received risk communication training first maintained their higher levels of satisfactions and agreement, even when later given the SDM training which appeared less beneficial (to doctors) in the first phase." Page 343
Elwyn 2004	Patient satisfaction with information provided (as described by clinicians)	"Doctors receiving risk communication tool and training first perceived significantly higher doctor‐patient agreement on treatment (P < 0.001), patient satisfaction with information (P < 0.01), doctor satisfaction with decision (P < 0.01) and general overall satisfaction (P < 0.001)" Page 343
Mullan 2009	Decision aid acceptability	Not reported
Murray 2010	Acceptability of the instrument	"In all, 37 members of the intervention group (97%) commented on the acceptability of the skills building workshop. … The 31 (81%) agreed that the PtDA would be acceptable to patients, while 24 (63%) agreed that it would be acceptable to practitioners." Page 117
Murray 2010	Utility of the intervention PtDA	"All 36 who participated in the educational outreach call indicated an interested in using the POC PtDa and express frustration that it was not available for use in their clinical practice setting." Page 117
Murray 2010	Intention to engage	"All participants, regardless of group assignment, saw patient decision support as helpful to patients (n = 32 [100 percent] interventions; n = 38 [98 percent] control) While 27 members of the intervention group (87%) and 34 members of the control group (84%) indicated a positive intention to engage in decision support, 16 members of the intervention group (50%) strongly agreed that they could provide decision support compared to 11 members of the control group (28%)" Page 117
Stacey 2006	Nurses' knowledge	"The nurses in the intervention group (n = 19) had a mean knowledge score of 74% and the mean score in the control group (n = 20) was 60%. The difference between the groups was significant (P = 0.007)." Page 413
Stacey 2006	Nurses' perception of factors influencing use of the coaching protocol	"Most of the 19 nurses in the intervention group agreed that the protocol was compatible with their practice (n = 15), provided a logical approach (n = 17), was easy to try (n = 15) and helped with exploring the benefits and harms of the options available to callers (n = 16). Another advantage of using the protocol, as reported by one nurse, was that it ... increases focus on caller's needs rather than just giving information." Page 413
Bernhard 2011	Maslach Burnout Inventory	"When doctors’ stress and burnout factors were accounted for in the mixed effects models for decisional conflict, the ESs became slightly larger in the SGA cohort but remained low. There was no influence by these factors on the ESs in the ANZ cohort (data not shown)." Page 5
Hamann 2007	Doctor patient relationship	"Doctor‐patient relationship (WAI) and PANSS scores did not prove to be independent significant prognostic factors" Page 996
Hamann 2007	Physicians satisfaction with treatment results	Not reported
Hess 2012	Clinician satisfaction with and acceptability of the DA	"Of the 51 clinicians who used the decision aid, 50 (98%) considered it helpful, and 32 (63%) indicated their desire to use the decision aid again if given the opportunity. Most clinicians indicated a desire to use a decision aid for other clinical management decisions" Page 255
Krist 2007	Physician perception of the decision making process	"Physicians tended to reports that they had greater control over the decision than did the patients, as measured by the CPS" Page 116
Krist 2007	Number of test ordered	Not reported
Leighl 2011	Physician satisfaction with decision‐making score	"Australian medical oncologists were surveyed regarding their satisfaction with the decision‐making process after each consultation;scores were generally high and similar in both arms" Page 2080
Murray 2001	Evaluation of the intervention	"General practitioners were positive about the decision aid; of 50 follow up consultation with patients in the intervention group they said that the decision aid had helped in 46, made no difference in three, and hindered one." Page 5
Roter 2012	Time management	The area in which there was no significant difference in reported skill use was in relation to time management. p.412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Roter 2012	Treatment adherence	The area in which there was no significant difference in reported skill use was in relation to time management. page 412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Roter 2012	Interpersonal rapport	The area in which there was no significant difference in reported skill use was in relation to time management. page 412 Treatment adherence (P = 0.03); Interpersonal rapport (P = 0.004) Table 7, page 412
Stiggelbout 2008	Surgeon's perceptions	"No differences were seen between the arms of the trial in the surgeons' reply to the question whether and how they presented probabilities; nor to the question on the risk that were discussed, the total number of risks that were discussed, or the understanding of the information by the patients; nor to the question whether much discussion had taken place during the consultation." Page 757
Street 1995	Physician facilitation	Not reported
Van Peperstraten 2010	Cost evaluation of the empowerment strategy	"The mean total savings in the intervention group were calculated to be €169.75 per couple included from the waiting list for in vitro fertilisation" Page 5

Table 12. Secondary outcome: other measurement reported by the healthcare professional (Positive studies are in italics)

Table 13. Secondary outcomes: other measures reported by patients (Positive studies are in italics)

Study	Instrument	Intervention			Control			Std. effect size (CI 95%)
		N	Pre	Post	N	Pre	Post
Continuous data: mean (SD)
Bieber 2006	Satisfaction with decision scale	34	NA	4.11 (0.40)	33	NA	4.02 (0.60)	0.17 (‐0.30 to 0.65)
Bieber 2006	Satisfaction with decision scale	34	NA	4.10 (0.60)	33	NA	4.07 (0.60)	0.05 (‐0.43 to 0.53)
Bieber 2006	Desicional conflict scale	34	NA	12.90 (4.20)	33	NA	12.40 (3.70)	0.12 (‐0.35 to 0.60)
Bieber 2006	Desicional conflict scale	34	NA	12.80 (3.00)	33	NA	12.50 (3.40)	0.09 (‐0.39 to 0.57)
Deen 2012	Decision self‐efficacy (DSE)	17	73.52 (19.13)	79.55 (12.79)	15	76.97 (17.95)	77.42 (19.29)	0.13 (‐0.57 to 0.82)
Deen 2012	Decision self‐efficacy (DSE)	21	71.54 (25.57)	79.55 (12.79)	15	76.97(17.96)	77.42 (19.30)	0.13 (‐0.53 to 0.80)
Deen 2012	Decision self‐efficacy (DSE)	17	77.27 (16.13)	83.82 (15.56)	15	76.97(17.97)	77.42 (19.31)	0.36 (‐0.34 to 1.06)
Dolan 2002	Decisional conflict scale	45	NA	1.83 (0.52)	43	NA	2.03 (0.81)	‐0.30 (‐0.71 to 0.27)
Haskard 2009	Patient perceived decision‐making	61	NA	2.94 (0.43)	66	NA	2.85 (0.46)	Unit of error Analysis
Haskard 2009	Patient choice	61	NA	4.15 (0.55)	66	NA	3.96 (0.68)	Unit of error Analysis
Krones 2008 (ARRIBA‐Herz)	Decisional regret	372	NA	14.69 (NA)	372	NA	18.08 (NA)	Unable to calculate
Krones 2008 (ARRIBA‐Herz)	Knowledge	535	NA	2.03 (NA)	576	NA	1.92 (NA)	Unable to calculate
Lalonde 2006	Decisional conflict scale	26	2.49 (0.53)	2.36 (0.30)	24	2.50 (0.39)	2.33 (0.30)	Pre: ‐0.02 (‐0.58 to 0.53) Post: 0.0.10 (‐0.46 to 0.65)
Landrey 2012	Knowledge of prostate cancer screening	71	NA	3.50 (1.50)	77	NA	3.30 (1.40)	0.14 (‐0.19 to 0.46)
Légaré 2012	Patients' quality of decision	Pre: 158 Post: 162	8.70 (1.50)	8.50 (1.60)	Pre: 151 Post: 159	8.70 (1.50)	8.50 (1.50)	0 (‐0.22 to 0.22)
Légaré 2012	Intention to engage in shared decision‐making	Pre: 165 Post: 163	1.90 (1.20)	2.10 (1.10)	Pre: 164 Post: 165	2.00 (1.20)	1.90 (1.20)	0.17 (‐0.04 to 0.39)
Légaré 2012	Regret over decision	Pre: 165 Post: 162	10.50 (15.40)	12.40 (19.10)	Pre: 164 Post: 164	10.80 (20.80)	7.60 (13.70)	0.29 (0.07 to 0.51)
Loh 2007	Doctor facilitation (PICS‐DF)	191	15.40 (3.50)	17.40 (3.10)	96	14.70 (3.70)	14.50 (3.30)	Pre: 0.20 (‐0.05 to 0.44) Post: 0.91 (0.66 to 1.17)
Loh 2007	Information seeking (PICS‐IS)	191	12.30 (2.70)	12.30 (3.40)	96	11.30 (2.90)	10.30 (2.90)	Pre: 0.36 (0.11 to 0.61) Post: 0.61 (0.36 to 0.87)
Loh 2007	Treament adherence	191	4.30 (0.80)	4.30 (0.90)	96	3.90 (0.80)	3.90 (1.00)	Pre: 0.50 (0.25 to 0.75) Post: 0.43 (0.18 to 0.67)
Loh 2007	Patients satisfaction (ZUF8)	191	NA	29.80 (2.70)	96	NA	27.00 (3.60)	0.92 (0.66 to 1.18)
Mullan 2009	Acceptability clarity of information	NA	NA	6.20 (0.96)	NA	NA	6.20 (0.80)	‐0.01 (‐0.38 to 0.36)
Mullan 2009	Acceptability helpfulness of the information	NA	NA	6.15 (0.94)	NA	NA	5.74 (1.04)	0.38 (0.04 to 0.72)
Mullan 2009	Acceptability; would recommend to others	NA	NA	6.16 (1.51)	NA	NA	5.89 (1.82)	0.38 (‐0.28 to 1.05)
Mullan 2009	Acceptability; would want to use for other decisions	NA	NA	6.04 (1.55)	NA	NA	5.69 (1.75)	0.34 (‐0.39 to 1.08)
Mullan 2009	Decisional conflict scale	NA	NA	14.10 (17.89)	NA	NA	14.95 (12.68)	‐0.89 (‐5.37 to 3.59)
Mullan 2009	Informed subscale of DCS (knowledge)	NA	NA	13.65 (19.84)	NA	NA	15.28 (15.49)	‐2.49 (‐7.21 to 2.23)
Mullan 2009	Trust in Physician scale	NA	NA	94.69 (7.14)	NA	NA	93.06 (9.58)	2.06 (‐1.78 to 5.89)
Mullan 2009	Acceptable amount of information	NA	NA	6.59 (0.91)	NA	NA	6.37 (1.14)	0.2 (‐0.41 to 0.83)
Murray 2001	Decisional conflict score	57	NA	2.30 (0.40)	48	NA	2.60 (0.50)	‐0.66 (‐1.06 to ‐0.27)
Murray 2001	Prosectomy rates and referrals	57	NA	0.11 (0.31)	48	NA	0.02 (0.14)	0.36 (‐0.03 to 0.75)
Myers 2010	Knowledge change	142	NA	0.80 (1.90)	144	NA	1.50 (2.10)	‐0.35 (‐0.58 to ‐0.11)
Myers 2010	Decisional conflict	142	NA	0.32 (0.49)	144	NA	0.29 (0.34)	0.07 (‐0.16 to 0.30)
Raynes‐Greenow 2010	Decisional conflict at primary follow up	395	31.40 (12.80)	23.90 (10.60)	201	31.20 (13.40)	24.90 (12.90)	Pre: 0.02 (‐0.15 to 0.19) Post: ‐0.09 (‐0.25 to 0.08)
Raynes‐Greenow 2010	Decisional conflict at second follow up	395	31.40 (12.80)	19.90 (12.30)	201	31.20 (13.40)	20.20 (14.10)	Pre: 0.01 (‐0.15 to 0.18) Post: ‐0.02 (‐0.19 to 0.15)
Raynes‐Greenow 2010	Anxiety first follow up	395	33.90 (10.10)	33.30 (9.30)	201	34.30 (11.80)	34.30 (11.00)	Pre:‐0.04 (‐0.21 to 0.13) Post: ‐0.10 (‐0.27 to 0.07)
Raynes‐Greenow 2010	Anxiety second follow up	395	33.90 (10.10)	29.40 (8.50)	201	34.30 (11.00)	29.00 (9.50)	Pre: ‐0.04 (‐0.21 to 0.13) Post: 0.04 (‐0.12 to 0.21)
Raynes‐Greenow 2010	Satisfaction with decision making first follow up	395	NA	81.50 (10.30)	201	NA	80.70 (11.70)	0.07 (‐0.10 to 0.24)
Raynes‐Greenow 2010	Satisfaction with decision making second follow up	395	NA	84.40 (12.90)	201	NA	82.80 (16.10)	0.11 (‐0.06 to 0.28)
Raynes‐Greenow 2010	Knowledge of analgesia first follow up	395	53.40 (21.90)	65.10 (29.50)	201	54.40 (20.90)	56.50 (27.40)	Pre: 0.05 (‐0.22 to 0.12) Post: 0.30 (0.13 to 0.47)
Stiggelbout 2008	Active participation of the patient	31	NA	1.40 (0.90)	33	NA	1.00 (0.20)	0.61 (0.11 to 1.18)
Van Peperstraten 2010	Knowledge experienced	Pre: 150 Post: 127	5.70 (2.50)	7.70 (0.60)	Pre: 154 Post: 135	5.80 (2.50)	7.20 (1.20)	0.52 (0.27 to 0.77)
Van Peperstraten 2010	Knowledge actual	127	NA	6.20 (2.85)	135	NA	4.30 (1.76)	0.74 (0.49 to 0.99)
Vodermaier 2009	Decisional conflict scale	53	NA	1.82 (0.59)	54	NA	1.99 (0.62)	‐0.28 (‐0.66 to 0.10)
Vodermaier 2009	Perceived involvement in care doctor facilitation (1‐4)	53	NA	2.65 (0.66)	54	NA	2.72 (0.67)	‐0.10 (‐0.48 to 0.27)
Vodermaier 2009	Perceived involvement in care patient information	53	NA	3.04 (0.74)	54	NA	3.09 (0.73)	‐0.10 (‐0.40 to 0.36)
Vodermaier 2009	ZUF‐8	53	NA	29.08 (2.99)	54	NA	28.67 (2.86)	0.14 (‐0.24 to 0.52)
Categorical data (n/N)
Dolan 2002	Annual fecal occult blood test	45	NA	11/23	43	NA	6/17	0.12 (‐0.18 to 0.43)
Dolan 2002	No test (wait and see)	45	NA	8/8	43	NA	15/16	0.06 (‐0.14 to 0.26)
Dolan 2002	Annual fecal occult blood test and flexible sigmoidoscopy every five years	45	NA	2/6	43	NA	7/8
Dolan 2002	Flexible sigmoidoscopy every five years	45	NA	4/6	43	NA	1/2	0.17 (‐0.15 to 0.48)
Dolan 2002	Double contrast barium enema every five years	45	NA	0/1	43	NA	0/0	NA
Dolan 2002	Colonoscopy every ten years	45	NA	1/1	43	NA	0/0	NA
Hess 2012	The proportion of patients who decided to undergo observation, unit admission, and cardiac stress testing	100	NA	58	100	NA	77	‐0.19 (‐0.32 to ‐0.41)
Krist 2007	PSA test ordered	196	NA	163/196	75	NA	64/75	‐0.02 (‐0.1 to ‐0.07)
Krist 2007	PSA test ordered	226	NA	194/226	75	NA	64/75	0.01 (‐0.09 to 0.10)
O'Cathain 2002	More anxious (antenatal)	Pre: 600 Post: 803	69/600	96/803	Pre: 595 Post: 724	77/595	87/724	Pre: ‐0.01 (‐0.05 to 0.02) Post: 0 (‐0.03 to 0.03)
O'Cathain 2002	More anxious (postnatal)	Pre: 879 Post: 846	99/879	86/846	Pre: 772 Post: 630	89/772	64/630	Pre: ‐0 (‐0.03 to 0.03) Post: 0 (‐0.03 to 0.03)
O'Cathain 2002	Drank less (antenatal)	Pre: 599 Post: 796	474/599	623/796	Pre: 595 Post: 696	443/592	551/696	Pre: 0.04 (0.00 to 0.10) Post:‐0.10 (‐0.03 to 0.03)
O'Cathain 2002	Planned hospitals birth (antenatal)	Pre: 619 Post: 826	608/619	799/826	Pre: 620 Post: 743	604/620	725/743	Pre: 0.01 (0.01 to 0.02) Post:‐0.01 (‐0.02 to 0.01)
O'Cathain 2002	Had screening test (antenatal)	Pre: 619 Post: 824	518/619	653/824	Pre: 619 Post:827	619/619	826/827	Pre: ‐0.16 (‐0.19 to 0.13) Post: ‐0.21 (‐0.23 to ‐0.18)
O'Cathain 2002	Partner/family present during labour (postnatal)	Pre: 922 Post: 886	867/922	836/886	Pre: 819 Post: 661	777/819	619/661	Pre: ‐0.01 (‐0.03 to 0.01) Post: 0.01 (‐0.02 to 0.03)
O'Cathain 2002	Stayed in bed during labour (postnatal)	Pre: 888 Post: 847	420/888	428/847	Pre: 796 Post: 635	409/796	319/635	Pre: ‐0.04 (‐0.09 to 0.01) Post: 0 (‐0.05 to 0.05)
O'Cathain 2002	Continuous monitory (postnatal)	Pre: 922 Post: 886	451/922	397/886	Pre: 819 Post: 661	387/819	319/661	Pre: 0.02 (‐0.03 to 0.06) Post: ‐0.03 (‐0.08 to 0.02)
O'Cathain 2002	Had epidural (postnatal)	Pre: 922 Post: 886	216/922	223/886	Pre: 819 Post: 661	177/819	160/661	Pre: 0.02 (‐0.02 to 0.06) Post: 0.01 (‐0.03 to 0.05)
O'Cathain 2002	Breast fed (postnatal)	Pre: 921 Post: 883	518/921	511/883	Pre: 818 Post: 660	482/818	389/660	Pre: ‐0.03 (‐0.07 to 0.02) Post: ‐0.01 (‐0.06 to 0.04)
O'Cathain 2002	Satisfied with amount of information	Pre: 891 Post: 855	619/891	635/855	Pre: 780 Post: 637	536/780	458/637	Pre: 0.01 (‐0.04 to 0.05) Post: 0.02 (‐0.02 to 0.069)
O'Cathain 2002	Satisfied with way choices were made	Pre: 886 Post: 855	683/886	656/855	Pre: 780 Post: 633	600/780	502/633	Pre: 0 (‐0.04 to 0.04) Post: ‐0.03 (‐0.07 to 0.02)
O'Cathain 2002	Enough discussion	Pre: 883 Post: 847	570/883	548/847	Pre: 774 Post: 636	481/774	414/636	Pre: 0.02 (‐0.02 to 0.07) Post: ‐0 (‐0.05 to 0.04)
Raynes‐Greenow 2010	Enough information to make decision	395	NA	352/395	201	NA	160	0.10 (0.03 to 0.16)
Raynes‐Greenow 2010	Analgesia used:support	395	NA	258	201	NA	120	0.06 (‐0.03 to 0.14)
Raynes‐Greenow 2010	Analgesia used: bath use	395	NA	143	201	NA	65	0.04 (‐0.04 to 0.12)
Raynes‐Greenow 2010	Analgesia used: epidural used	395	NA	133	201	NA	66	0.01 (‐0.07 to 0.09)
Van Peperstraten 2010	Fully empowered couples, decision empowerment	Pre: 150 Post: 127	116	116	Pre: 154 Post: 99	112	99	0.18 (0.09 to 0.27)
Qualitative data
Butow 2004	Satisfaction with the consultation and decision	"No significant differences were found between the groups in satisfaction with either the consultation or treatment decision" Page 4407
Butow 2004	Satisfaction with the booklet	"No significant differences were found between groups in terms of reported anxiety provoked, perceived utility, or ease of understanding of materials. … There was significant reported usefulness of the CCPP and control booklet for the family (P = 0.004)." Page 4405
Butow 2004	Information subscale of the Krantz Health Opinion Survey	"No significant results were obtained" Page 4407
Deinzer 2009	Hypertension Questionnaire	"Only in the SDM group was there an increase in knowledge after 1 year (P = 0.006). After 1 year both groups showed similar levels of knowledge" Page 269
Deinzer 2009	Short Form 36 Item Health Survey (SF‐36)	"There were no differences between the 2 groups concerning health‐related quality of life measured with the 8 scales of SF‐36" Page 269
Deinzer 2009	Difficult Doctor Patient Relationship Questionnaire (DDPRQ)	"Doctor‐patient relationship … was better in the SDM group than the control at the beginning … and after 1 year (p.0016). In the control group … an improvement occurred … (P = 0.045) that did not occur in the SDM group (P = 0.16)" Page 269
Deinzer 2009	Autonomy Preference Index	"Preference for SDM as assessed by the API showed no differences between the SDM and control group at baseline (P = 0.60) and did not change after 1 year (P = 0.83)" Page 268
Deschamps 2004	Decision conflict score and the informed subscale items	The differences between groups were non‐significant (Table 2), page 25
Deschamps 2004	Satisfaction with preparation for decision making	The differences between groups were non‐significant (Table 3), page 25
Deschamps 2004	Satisfaction with decision	"Women in the pharmacist and decision‐aid groups had mean SWD scores of 4.3 and 4.4 respectively (scale range: 1 to 5) with no significant differences being reported between groups. Page 26
Deschamps 2004	Adherence to HRT	"There was no statistically significant difference in adherence between the study groups" Page 26
Elwyn 2004	Intention to adhere to chosen treatment	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes ... However, significant effects of the research clinic (i.e. mainly the provision of more time)did lead to improvement (0.7 increase, 95% CI 0.04 to 1.36, P < 0.05)" Page 351
Elwyn 2004	Patient's satisfaction with information provided	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Enablement	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Satisfaction with decision made	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Elwyn 2004	Patient's perceived support in decision	"No significant effects of the risk communication or SDM intervention were seen on the whole range of patient‐based outcomes" Page 351
Fossli 2009	Patient global satisfaction	Non significant P = 0.38
Hamann 2007	Autonomy preference index (API)	Differences between groups not reported
Hamann 2007	Patient's satisfaction with overall care	Differences between groups not reported
Hamann 2007	The medication adherence rating scale	Differences between groups not reported
Hamann 2007	Patient knowledge of disease and treatment (7‐item multiple choice)	Differences between groups not reported
Hamann 2007	Compliance with drug regime	Overall compliance was "good" for 42 (49%) of the patients at 6 months and 40 (59%) at 18 months
Hess 2012	Knowledge	Knowledge (P < 0.0001) Table 2. Page 6
Hess 2012	DCS	DCS (MD=‐13.6 (‐19.1 to ‐8.1)) Table 2. Page 6
Hess 2012	Trust in physician	Trust in physician (MD=4.1 (‐1.4 to 9.6)), Table 2. Page 6
Hess 2012	Patient satisfaction with the decision‐making process	Patients who used the decision aid reported greater satisfaction with the decision‐making process (strongly agree, 61% versus 40%; absolute difference, 21%; 95% CI 7% to 33%). Page 5
Kasper 2008	Treatment decision	"Pearson's chi square P‐value for this table is not significant for patients already on immunotherapy at baseline and patients not yet on immunotherapy at baseline, compared to patients in the CG." Page 1350
Kasper 2008	Patients evaluation of the decision	"Six months after randomization, the two groups did not show any significant differences in their evaluation of their decisions" Page 1350
Kasper 2008	Measure of the decision making process	"Both groups progressed significantly in making their decision. However they did not show differences in the course of progress over the three measurement points" Page 1349
Krist 2007	Prostate cancer screening knowledge	"… the percentage of correct answers on the knowledge scale was 54% in the control group (P < 0.001) vs 69% in the brochure group (P < 0.001)" Page 115
Krist 2007	Decisional conflict score	"DCS scores among all 3 groups were equally low and did not differ significantly … " Page 115
Krist 2007	Patients and physicians topics covered in the discussion	"The decision aids did not appear to alter … the number of prostate cancer screening topics that patients or physicians recalled addressing" Page 115
Lalonde 2006	Risk perception	"No statistically significant improvements were observed after the intervention" p55 No mention of between‐group differences
Lalonde 2006	Knowledge of hypertension	"However, knowledge of the estimated benefits of treatment tended to improve after the intervention (29% versus 58%; P = 0.06)" No mention of differences between group" Page 55
Landrey 2012	Flyer acceptability	"Among patients who reported receiving the flyer, 86.4% felt the content was clearly presented, 86.4% felt it contained about the right amount of information, 45.5% felt the information was completely balanced, and 43.2% viewed it as biased against PSA testing; 88.6% would recommend it to others." Page 5
Leighl 2011	Decisional conflict score	Decision satisfaction and decisional conflict scores were similar in both arms. Page 2080
Leighl 2011	Patient satisfaction with decision	Decision satisfaction and decisional conflict scores were similar in both arms. Page 2080
Leighl 2011	Patient satisfaction with consultation	"Patients in both arms were highly satisfied with the consultation" Page 2080
Montori 2011	Knowledge: DA specific	Knowledge DA specific (P = 0.001) Table 2, page 553
Montori 2011	Knowledge: Not in the DA	Knowledge not in the DA (P = 0.35) Table 2, page 553
Montori 2011	Decisional conflict scale	Decisional conflict scale (P = 0.72) Table 2, page 553
Montori 2011	Trust	Trust (P = 0.46) Table 2, page 553
Murray 2001	Acceptability of decision aid	"Patients reacted positively to the decision aid" Page 5
Murray 2001	Satisfaction	Not reported
Murray 2001	Choice of treatment	The choice in treatment did not vary significantly from one group to another. For more details, see page 5.
Myers 2010	Screening use	"Screening use was lower in EI Group than in SI Group (63% versus 71%), but this difference was not statistically significant (odds ratio= 0.67; 95% confidence interval, CI: 0.41‐1.08; P = 0.102)" Page 4
Raynes‐Greenow 2010	Stages of decision making	"Even distribution among stages.... A small proportion of women in both groups were not considering their choices …, or had made up their mind and were 'unlikely to change mind' … A large proportion of women … were amenable to change or were in active deliberation stages ... the largest proportion ... were women who 'had made some choices but were willing to reconsider" Page 6
Raynes‐Greenow 2010	Choice predisposition towards analgesia	"Overall, higher proportions of women in both groups intended to use non‐pharmacological methods for labour pain relief rather than pharmacological methods." Page 6
Raynes‐Greenow 2010	Adherence and acceptability	"Most women had read all of the intervention (decision aid 98% compared to pamphlet group 95%, chi‐square = 2.782, df=1, P = 0.061), and equally both groups would recommend the intervention they received to a pregnant friend (decision aid group 94% compared to pamphlet group 93%, chi‐square, df=1, P = 0.57)" Page 7
Raynes‐Greenow 2010	Source of information	"Both groups equally relied on family and friends, books and antenatal classes" Page 7
Raynes‐Greenow 2010	Labour, Mode of delivery, Birth Weight, Apgar score	All information can be found in Table 4, page 10. There were no significant differences between groups
Roter 2012	Patient satisfaction:identification of problems and concerns	Patient satisfaction: identification of problems and concerns (P = 0.25) Table 6, page 411
Roter 2012	Patient satisfaction:information exchange	Patient satisfaction: information exchange (P = 0.01) Table 6, page 411
Roter 2012	Patient satisfaction:shared decision‐making	Patient satisfaction: shared decision‐making (P = 0.03) Table 6, page 411
Schroy 2011	Screening intentions	"Differences in intention to schedule or complete a screening test for the 2 intervention groups versus control corresponded to moderate effect sizes ranging between 0.36 and 0.44. Scores were comparable for the 2 intervention groups." Page 9
Stiggelbout 2008	Understanding	"The only difference that was seen for the items related to understanding was a difference in favour of the IB group in the stated understanding of the issues that were important in the treatment decision: 84% (n = 32) of the IB group felt that due to the brochure they had better understanding, v. 62% (n = 21) of the GB group (chi‐square test P = 0.004)" Page 756
Stiggelbout 2008	Consultation with the surgeon	"A main difference between the 2 groups was seen in satisfaction with the duration of the consultation … (chi‐square test P = 0.04). … For patients' impression whether the surgeon perceived them more as a medical problem that as a person with a problem, an interaction effect was observed F (1.68)=4.31, P = 0.04." Page 757
Street 1995	Patient knowledge	"The effect for method of communication approached significance (F = 3.30, P = 0.07) as patients in the computer group tended to learn more (mean, 75.5%; SD 13.64%) than did patients in the brochure group (mean, 71.4%; SD, 15.7%)" Page 2279
Street 1995	Patient optimism	"Optimism scores were not affected by … the educational intervention (F = 0.95, P = 0.93)" Page 2279
Street 1995	Patients' behavioural measures	Differences between groups not reported
Street 1995	Perceived involvement in decision making	Differences between groups not reported
Wetzels 2005	Point in time of decision making	The points in time of decision making were not statistically significant (p‐value = 0.93) Table 4, page 595
Wetzels 2005	Patient enablement index	Significant effect size difference: ‐0.232 (‐0.444; ‐0.021) P = 0.03, table 3, page 292
Wetzels 2005	Satisfaction with their care‐EUROPEP	Non significant; effect size difference ‐0.056 (‐0.302; 0.192) P = 0.66, table 3, page 292
Wetzels 2005	Use of leaflet	"Sub‐analyses showed that the scores for these 47 patients did not differ significantly on the outcomes measures from those of the control group or the intervention group non‐users" Page 290
Wetzels 2005	Discussion of one of the eight known underreported health problems	None of the discussion topics were shown to be statistically significant. Table 4, page 292

Table 13. Secondary outcomes: other measures reported by patients (Positive studies are in italics)