Comparison of self-reported and observed prevalence of safety belt and helmet use in Florence.

Abstract

Safety belt and helmet use was estimated from PASSI data and measured through Ulisse observations. Between 2008 and 2012 a total of 2,081 cars and motorcycle users were interviewed in the LHU of Florence and a total of 59,787 drivers (11,870 front passengers, 1,129 rear passengers and 16,816 motorcyclists) were observed. The comparison between self-reported and observed prevalences was performed by calculating the over-reporting factor (ORF), defined as the ratio of the self-reported to the observed prevalence of seat belt or helmet use. The time trend of the prevalence (both from self-reported and observed data) and of the ORF was assessed by using linear regression and Poisson's regression, respectively. The correlation between self-reported and observed prevalence is high, with a Pearson's correlation coefficient of 0.95 (p <0.05). Regarding front seat belt use rates, the difference between self-reported and observed data increases over time and the ORF range varies from 1.12 to 1.32. Rear seat belt data show a great variability, and the ORF varies from 0.67 to 1.37. In 2011 and 2012, the observed prevalence was higher than the self-reported one (ORF <1). Helmet use rates are very high, close to 100% with both methods; ORF has very small oscillations and ranges from 0.98 to 1, showing a good correlation between self-reported and observational data. There are no significant temporal variations both for the prevalences of use and for the ORF. The reasonable accuracy of self-reported data makes this method fit in the routinary assessment of safety belts and helmet usage, in order to limit the observations of the Ulisse system at predetermined time intervals. However, self-reported estimates need to be adjusted using an appropriate over-reporting factor.