Objective: To compare characteristics and restraint use between a population-based and fitting service sample of child restraint users.
Method: Characteristics of the two samples were compared using chi-squared tests. Differences in errors in restraint use observed in the two samples were modeled using logistic regression.
Results: There were significant differences in child age (p < 0.001), and restraint types (p < 0.001) between the two samples, with more younger children in the fitting service sample. Controlling for differences in restraint type, the odds that adult participants were female were 61% less in the fitting service sample than in the population-based sample (OR 0.39, 95%CI 0.21-0.71). The odds that adult participants perceived a large risk associated with restraint misuse (OR 3.62, 95%CI 1.33-9.84), had a household income in the highest bracket (OR 3.89, 95%CI 1.20-12.62) and were living in areas of highest socioeconomic advantage (OR 2.72, 95%CI 1.22-6.06) were approximately three times higher in the fitting service sample. Overall, more participants had errors in restraint use in the population-based sample (p = 0.021). However, after controlling for restraint type, securing errors were three times more likely (OR 3.34, 95%CI 1.12-10.2), and serious installation errors were almost twice as likely (OR 1.91, 95% CI 1.09-3.39) in the fitting service sample.
Conclusions: While less resource intensive, convenience and/or fitting service samples may be less representative than population-based samples. Given the need for efficiency, methods that combine randomized population-based invitations to participate in restraint fitting check day events across geographically representative areas may be useful for ongoing surveillance of child restraint use.
Keywords: Child passenger; child car seat; misuse; observation; restraint.