Validity of Smoking Measurements during Pregnancy: Specificity, Sensitivity and Cut-Off Points

Abstract

Background: Non disclosure by pregnant women smokers of their smoking habit questions the validity of self-declarations. The purpose of this research is to determine the rate of Non disclosure and to establish the validity of exhaled CO as a method of biochemical validation. Method: Data obtained in a Randomised Clinical Trial in 12 Health Centres in Malaga. 454 pregnant women smokers, of whom 104 declared that they had stopped smoking at the start. Measurement of the habit: self-declaration, carbon monoxide (CO) and cotinine in the urine in those subjects who declared that they no longer smoked. The ROC curve was obtained for the CO, calculating the area under the curve and the sensitivity and specificity for different cut-off points. Using the CO as the gold standard, the validity of the selfdeclarations was determined in terms of sensitivity and specificity. Results: Cotinine/self-declaration comparison: rate of Non disclosure 15.4% (IC 95% 9.3-24.1). Cotinine/CO comparison: Area under the ROC curve of 0.838 (IC 95% 0.740-0.935). For a cut-off point of 9, recommended in the relevant bibliography, we achieved 100% specificity with 12.5% sensitivity. CO/self-declaration comparison (cut-off point 9): False negations 0.8%, prevalence of spontaneous abandonment of smoking habit 58.1%. Conclusions: A percentage of Non disclosure similar to other studies and the validity of CO as a method for the identification of women smokers are confirmed. At cut-off point 9, the validity of CO and self-declaration would be similar, with the prevalence of spontaneous abandonment increasing up to unreal figures. At the present time, the evidence is not sufficient in order to recommend 9 as the CO cut-off point. The results suggest that this should be lower. More extensive research is necessary in pregnant women who state that they do not smoke, using an appropriate methodology.