Using expired air carbon monoxide to determine smoking status during pregnancy: preliminary identification of an appropriately sensitive and specific cut-point

Abstract

Background: Measurement of carbon monoxide in expired air samples (ECO) is a non-invasive, cost-effective biochemical marker for smoking. Cut points of 6ppm-10ppm have been established, though appropriate cut-points for pregnant woman have been debated due to metabolic changes. This study assessed whether an ECO cut-point identifying at least 90% of pregnant smokers, and misidentifying fewer than 10% of non-smokers, could be established.

Methods: Pregnant women (N=167) completed a validated self-report smoking assessment, a urine drug screen for cotinine (UDS), and provided an expired air sample twice during pregnancy.

Results: Half of women reported non-smoking status early (51%) and late (53%) in pregnancy, confirmed by UDS. Using a traditional 8ppm+cut-point for the early pregnancy reading, only 1% of non-smokers were incorrectly identified as smokers, but only 56% of all smokers, and 67% who smoked 5+ cigarettes in the previous 24h, were identified. However, at 4ppm+, only 8% of non-smokers were misclassified as smokers, and 90% of all smokers and 96% who smoked 5+ cigarettes in the previous 24h were identified. False positives were explained by heavy second hand smoke exposure and marijuana use. Results were similar for late pregnancy ECO, with ROC analysis revealing an area under the curve of .95 for early pregnancy, and .94 for late pregnancy readings.

Conclusions: A lower 4ppm ECO cut-point may be necessary to identify pregnant smokers using expired air samples, and this cut-point appears valid throughout pregnancy. Work is ongoing to validate findings in larger samples, but it appears if an appropriate cut-point is used, ECO is a valid method for determining smoking status in pregnancy.

Keywords: Biochemical verification of smoking; CO; COHb; ECO; ETS; Exhaled air carbon monoxide; NPV; PPV; Pregnancy smoking; ROC; Smoking assessment; UDS; carbon monoxide; carboxyhaemoglobin; environmental tobacco exposure; expired carbon monoxide; negative predictive value; parts per million; positive predictive value; ppm; receiver operating characteristic; urine drug screen.