Introduction and aims: A significant level of misreport or error occurs during questionnaire-based assessment of smoking behaviour. Failure to measure environmental tobacco smoke, and participant's inclination to under-report their smoking raise questions as to the accuracy of assessment. In order to establish an estimation of the possible error associated with such assessment, the accuracy of self-reported smoking status among a group of pregnant Aboriginal and Torres Strait Islander women was examined.
Design and methods: Women attending two Aboriginal Medical services in Far North Queensland for antenatal care were invited to participate. Women completed an interviewer assisted questionnaire relating to their smoking status and a 24 h diary of their exposure to nicotine and consumption of alcohol. Urine samples were analysed for cotinine using an Enzyme Linked Immunosorbent Assay.
Results: Cotinine analysis indicated that 17% of women who reported that they were non-smokers were likely to have misreported this status, or be exposed to high levels of passive smoke. The only significant predictors of cotinine level were self-reported nicotine exposure (including passive smoke) and number of cigarettes smoked in the previous 24 h. Other individual and environmental variables had no significant influence on cotinine level using this analysis technique.
Discussion and conclusions: The level of potential error in smoking assessment among this group was substantial. Exposure to environmental tobacco smoke might explain part of this error, but the reasons for misreport can only be speculated. This rate of misclassification should be taken into consideration in routine screening of antenatal women in primary health care.