Several interventions have been developed to promote informed consent for participants in clinical trials. However, many of these interventions focus on the content and structure of information (e.g. enhanced information or changes to the presentation format) rather than the process of decision making. Patient decision aids support a decision making process about medical options. Decision aids support the decision process by providing information about available options and their associated outcomes, alongside information that enables patients to consider what value they place on particular outcomes, and provide structured guidance on steps of decision making. They have been shown to be effective for treatment and screening decisions but evidence on their effectiveness in the context of informed consent for clinical trials has not been synthesised.
To assess the effectiveness of decision aids for clinical trial informed consent compared to no intervention, standard information (i.e. usual practice) or an alternative intervention on the decision making process.
We searched the following databases and to March 2015: Cochrane Central Register of Controlled Trials (CENTRAL), The Cochrane Library; MEDLINE (OvidSP) (from 1950); EMBASE (OvidSP) (from 1980); PsycINFO (OvidSP) (from 1806); ASSIA (ProQuest) (from 1987); WHO International Clinical Trials Registry Platform (ICTRP) (http://apps.who.int/trialsearch/); ClinicalTrials.gov; ISRCTN Register (http://www.controlled-trials.com/isrctn/). We also searched reference lists of included studies and relevant reviews. We contacted study authors and other experts. There were no language restrictions.
We included randomised and quasi-randomised controlled trials comparing decision aids in the informed consent process for clinical trials alone, or in conjunction with standard information (such as written or verbal) or alongside alternative interventions (e.g. paper-based versus web-based decision aids). Included trials involved potential trial participants, or their guardians, being asked to consider participating in a real or hypothetical clinical trial.
At least two authors independently assessed studies for inclusion, extracted reported data and assessed risk of bias. Findings were pooled where appropriate. We used GRADE to assess the quality of the evidence for each outcome.
We identified one study (290 randomised participants) that investigated the effectiveness of decision aids compared to standard information in the informed consent process for clinical trials. This study reported two separate decision aid randomised controlled trials (RCTs). The decision aid trials were nested within two different parent trials focusing on breast cancer in postmenopausal women. One trial focused on informed consent for treatment in women who had previously had surgery for ductal carcinoma in situ (DCIS), the other on informed consent for prevention in women at high risk for breast cancer. Two different decision aids were used in these RCTs, and were compared with standard information.The pooled findings highlight the uncertainty surrounding most reported outcomes, including knowledge, decisional conflict, anxiety, trial participation and attrition. There was very low quality evidence that decision aids lower levels of decisional regret to a small degree (MD -5.53, 95% CI -10.29 to -0.76). No data were identified on several prespecified primary outcomes, including accurate risk perception, values-based decision, or whether potential participants recognised that a decision needed to be made, were able to identify features of options that matter most to individuals, or were involved in the decision.
There was insufficient evidence to determine whether decision aids to support the informed consent process for clinical trials are more effective than standard information. Additional well designed, adequately powered clinical trials in more diverse clinical and social populations are needed to strengthen the results of this review. More generally, future research on which outcomes are most relevant for assessment in this context would be helpful.