Modest time-dependent misclassification in recalling age at first use of alcohol and tobacco

Abstract

To investigate whether recall of age at first use of alcohol and tobacco is time dependent. Respondents aged 14-89 years from the 2001 and 2007 waves of the Australian National Drug Strategy Household Survey were considered as independent samples of the Australian population at two different time points. The means of reported age at first use of alcohol and age at first use of tobacco from the two surveys were compared for each birth cohort. The mean differences were generally small (< 1 year) across birth cohorts. Overall, the increase in mean age at first use from 2001 to 2007 was 0.31 years for alcohol and 0.33 years for tobacco, after controlling for birth cohort and gender effects. The time lag between actual onset of first use and subsequent survey recall can introduce a small systematic bias, but the error is unlikely to pose a threat to retrospective studies that rely on recall information to estimate cumulative exposure of alcohol and tobacco use.