Sensitivity and specificity of a self-administered questionnaire of tobacco use; including the Fagerström test

Abstract

Background: Preventing tobacco consumption and promoting cessation among health professionals are of great significance as their habits can influence their patients' attitudes. Knowledge of the prevalence and characteristics of tobacco use in a specific population is important for the design of efficient strategies for preventing people from acquiring the habit and persuading them to stop. Self-administered questionnaires are a very common method for determining tobacco use, but assessment is needed of their validity for specific groups, such as occasional smokers or students of health sciences.

Objectives: The aim of this study was to determine the sensitivity and specificity of a self-administered questionnaire for determining tobacco use among a population of young female students of health sciences. As a gold standard, we used a measure of the concentration of cotinine in saliva with different cut-off points. We also analysed the influence on sensitivity and specificity of the questionnaire of nicotine dependence and exposure to secondary smoke.

Design: This is an internal validity study (sensitivity and specificity) of a self-administered questionnaire.

Settings: The study was carried out in the School of Health Sciences of the University of León, at its centres in León and Ponferrada.

Participants: Data were collected on 432 of a total of 480 women aged under 25 who were studying Health Sciences at the University of León in 2007.

Methods: The self-administered questionnaire included data concerning demography, education, tobacco habits and exposure to environmental smoke. Saliva cotinine concentration was determined with an EIA kit.

Results: The questionnaire used showed high values of sensitivity and specificity (85.3% and 95.3%) and a very good correlation (Kappa=81.0%) with cotinine values of 10ng/ml or higher. Discrepancies in the test results for those who declared themselves to be smokers only occurred in women with a low nicotine dependency. Among those who did not declare themselves smokers, those exposed to second-hand smoke were 7 times more frequently classified erroneously as smokers than those who had not been exposed.

Conclusion: The self-administered questionnaire used showed a very good internal validity and a good correlation with cotinine levels of 10ng/mg, and therefore seems to be a good instrument for measuring tobacco use in this population. The availability of information on passive smoking and nicotine dependence is essential for the correct interpretation of the discrepancies.