METHODS: We performed a cross-sectional study of 314 adolescents aged 16 years from February 2015 to April 2015 in Putrajaya, Malaysia. The accuracy of self-reporting was assessed using a data collection sheet and was validated by the urine cotinine strip test. Three schools were chosen by the simple random method, where all Form 4 students constituted the sample unit. The kappa statistic was used for determining agreement between self-reporting and urine cotinine strip testing.

RESULTS: There was a substantial agreement between self-reporting and the urine cotinine strip test (kappa = 0.757, 95% confidence interval [CI]: 0.63, 0.88); there was 95.86% overall agreement. The prevalence of self-reported smoking was 8% (95% CI: 7.47, 8.53) and that of urine cotinine strip testing was 10.8% (95% CI: 10.20, 11.41). There was a discrepancy with the results of the urine cotinine strip test in 8% of self-reported smokers and 3.8% of self-reported nonsmokers. Self-reporting had 67.6% sensitivity and 99.3% specificity as compared to those of urine cotinine strip testing and had 92% positive predictive value and 96.2% negative predictive value.