BACKGROUND
Decision aids are interventions designed to help people make specific and deliberative choices among options (including status quo) by providing (at a minimum) information on the options and outcomes relevant to a person's health status. A systematic review is needed to summarize what is known about their efficacy.
OBJECTIVES
1. To create an inventory of existing decision aids. 2. To conduct a systematic review of randomized controlled trials of decision aids aimed at improving people's decision making and outcomes.
SEARCH STRATEGY
The following electronic databases were searched: MEDLINE, EMBASE, PsycINFO, CINAHL, Aidsline, Cancerlit, the Cochrane Controlled Trials Register. In addition, individuals known to be active in the field of decision support were contacted.
SELECTION CRITERIA
We screened titles and abstracts for all interventions providing structured, detailed, and specific information on treatment or screening options and outcomes to aid decision making. Information about the decision aids was compiled in an inventory and those evaluated in a randomized controlled trial were reviewed in detail.
DATA COLLECTION AND ANALYSIS
Using Cochrane review methods, two reviewers independently screened and extracted data on several evaluation criteria. Results of randomized trials were pooled using weighted mean differences and relative risks with all data analysed using a random effects model.
MAIN RESULTS
Eighty-seven decision aids were identified. Twenty-three of them had been evaluated in 24 randomized controlled trials. Among the trials comparing decision aids to usual care interventions, decision aids performed better in terms of: a) greater knowledge of options in seven of seven studies (weighted mean difference (WMD = 19 out of 100, 95% confidence interval (CI): 13,25); b) more realistic expectations in two of three studies (RR=1.48 95%CI 1.02, 2.14); c) lower decisional conflict related to feeling informed in three of three studies (WMD = -9.0 of 100 95%CI: -15, -3); d) reducing the proportion of people who were passive in decision making in three of four studies (RR = 0.63, 95% CI: 0.5, 0.8). When simpler decision aids were compared to more detailed ones, the relative improvement was significant in: a) improved knowledge in two of seven studies (WMD = 4 out of 100, 95% CI: 2, 5); b) realistic expectations in two of two studies (relative risk (RR ranged from 1.5 95%CI: 1.3,1.8); and c) decisional conflict in one of two studies (WMD= -4 95% CI -8,-0.2). There was a consistent trend for decision aids to do no better than comparison interventions in affecting: satisfaction with the decision in five of six studies, satisfaction with the decision making process in four of five studies, and anxiety in four of five studies. Decision aids had a variable effect on which screening or treatment options were selected. Exposure to decision aids relative to controls resulted in a consistent non-significant trend toward reduced preference for major surgery over conservative options by 21 to 42 per cent in five of five studies; however, they had no impact on circumcision in two of two studies. For colon and prostate cancer screening choices, the results were inconsistent. Two studies have shown no impact on persistence with chosen option and few studies have shown significant benefits on general health outcomes (two of three studies), or disease-specific health outcomes (one of four studies).
REVIEWER'S CONCLUSIONS
The number of decision aids is expanding, but there is considerable overlap in some areas leaving gaps in others. Trials of decision aids indicate that they are superior to usual care interventions in improving knowledge and realistic expectations of the benefits and harms of options; reducing passivity in decision making; and lowering decisional conflict stemming from feeling uninformed. When simpler versions of decision aids are compared to more detailed aids, the differences in knowledge are marginal but there are other benefits in terms of creating realistic expectations and in reducing decisional conflict. To date, decision aids have had little effect on anxiety or satisfaction with the decision making process or satisfaction with the decision. Their effects on choices vary with the decision. The effects on persistence with chosen therapies and health outcomes require further evaluation. The essential elements in decision aids for different groups and different types of decisions need to be established. Consensus needs to be reached regarding standards for developing and evaluating decision aids.