An optimal cutoff point of expired-air carbon monoxide levels for detecting current smoking: in the case of a Japanese male population whose smoking prevalence was sixty percent.

An optimal cutoff point of expired-air carbon monoxide (Ex-CO) for detecting smokers should be determined in terms of its sensitivity and specificity and the prevalence of smoking in the target population. The purpose of this study is to determine the optimal cutoff point of Ex-CO for detecting smoking males in a Japanese community whose smoking prevalence was over 50%. Among free-living residents in a rural population, "true smokers" determined by presence of cotinine in serum were 61% (n = 94). When Ex-CO at 7 ppm or over differentiated "smokers" from "non-smokers", sensitivity and specificity for detecting smokers was 0.93 and 0.95, respectively, which comprised the best Youden's index. This setting also produced the minimum percentage of misclassified cases. In conclusion, 7 ppm of Ex-CO, which is exceptionally low value relative to the western standard, appears to be the most optimal cutoff point for a survey in a population with such high smoking prevalence.

To validate self-reported data on smoking, use of biochemical markers has been encouraged. Among some examined markers, cotinine is considered to be very good because of its high sensitivity and specificity to tobacco smoke exposure 14). However, it has not been widely used in usual medical practice or population surveys due to its high cost and the necessity of venepuncture. While using saliva or urine samples is much less invasive than using serum samples 3,5), the problem of cost would still be unresolved. In the 1970s, measuring expired-air carbon monoxide (Ex-CO) was shown to be valid as an alternative marker of current smoking6, 7 ). During this decade, a handy and inexpensive instrument to measure Ex-CO 8) was developed and gradually popularized. This is equipped with a sealed electrochemical sensor and provides an immediate display of subjects' Ex-CO levels, which are proportional to carboxyhemoglobin levels in blood. Using this instrument, 8 -11 ppm cutoff points were often used to discriminate smokers from non-smokers 6,[9][10][11][12][13][14] However, selection of the cutoff points was almost arbitrary. An optimal cutoff point of Ex-CO for detecting current smokers should be determined in terms of its sensitivity and specificity and the prevalence of smoking in the target population 15). Ideally the optimal cutoff point should be set for each target group and for each specific purpose.
Compared with other developed countries, the Japanese male population is generally characterized as having an extremely high prevalence of smoking, approximately 60% by self report 16,17). Studies of the reliability of biochemical markers such as Ex-CO and serum cotinine and the setting of an optimal cutoff point for Ex-CO to detect current smoking in such a field setting have been rare.

MATERIALS AND METHODS
The study area was a rural community in the west central Japan, Yamasaki Town, Shiso County in Hyogo Prefecture. The subjects were adult males living in this area, who participated in an annual health checkup by the local administration in October, 1990. As habitual smoking is not so common for adult females living in rural areas in Japan, the subjects were limited to males only. Consecutive health checkup participants on a day were invited to be involved in our study. We asked them to report their smoking status and measured their Ex-CO using a CO-detector, "Micro-Smokerlizer" (Bedfont Technical Instrument Ltd., Sittingbourne, Kent., UK) that was calibrated with the standard gas by the company just before the examination. This measurement was done in the morning in a large room with good ventilation and no equipment producing excess CO. The subjects were requested to follow the standard protocol described in the manual: they were asked to exhale fully, then inhale deeply and hold breath for 15-20 seconds before exhaling into "Micro-Smokerlizer" through a disposable mouthpiece. Immediately after the subjects exhaled into the instrument, the maximum readings were noted as their Ex-CO levels. Throughout the study, Ex-CO levels were measured by one operator using the same instrument. On the same day, we took their fasting venous blood samples and after centrifugation, serum cotinine concentrations were measured by gas-chromatography. The lowest detectable limit of this method was 0.06,uM/L (10 ng/ml).
Before the principal examination, reproducibility of measuring Ex-CO levels by "Micro-Smokerlizer" was evaluated with selected subjects. They were tested twice (the interval was supposed to be within almost three minutes), and obtained values were used for simple regression and Pearson's correlation analysis. The relationships of serum cotinine concentrations to Ex-CO levels and to self-reported numbers of cigarettes smoked before the test were examined by using Spearman's rank correlation coefficient.
To evaluate the reliability of Ex-CO for detecting smoking status, sensitivity and specificity were assessed for the various cutoff points. Predictive values and the number of subjects misclassified were also obtained for each characteristic.
According to the lowest limit of the gas-chromatographic technique that we adopted and the cutoff point of serum cotinine previously established (0.084,u MIL) 2,3.15) , we defined subjects whose serum cotinine was not detectable by this method as "non-smokers". Subjects, other than these "nonsmokers", were classified as "smokers" in the present study. Namely, serum cotinine concentration was regarded as the gold standard in this research.    Table  2). and the prevalence of smoking in the target population.