Background Many injury studies use samples of hospital patients. If a study requires further contact, consent is usually required but the sample may be biased if many fail to consent.

Objective To determine whether wording requiring ‘active’ consent resulted in more refusals than wording that was ‘passive’—that is, where consent is implied in the absence of direct refusal.

Methods Subjects were injured children seen in the emergency departments in five hospitals where the Canadian Hospital Injury Reporting and Prevention Program (CHIRPP) operates. For CHIRPP, parents or older children complete a one-page questionnaire to describe the injury; one question seeks consent to follow-up for research. Three of the hospitals use passive wording for this question and two use active wording. All cases where CHIRPP coordinators completed forms using the medical record were treated as refusals. It was hypothesised that there would be a significant association between the form of wording (active or passive) and the rate of consent, and that this association would be affected by socio-demographic and clinical variables.

Results On average, 64.5% of parents gave consent for follow-up at hospitals using passive consent wording versus 42% where active consent wording was used. Passive wording consistently yielded a higher percentage of consents for all variables. The differences were greater than 5% for families living in census tracts with low median household incomes and greater than 10% for those age 15–19 years. For parent completed forms the adjusted OR for active wording was 0.48 (95% CI 0.43 to 0.54).

Conclusion Caution is needed when interpreting results from studies using hospital samples, especially when wording of consent for follow-up requires direct affirmation.