目的與背景： 實務上針對年輕軍人使用呼氣一氧化碳測量儀(carbon monoxide
breathalyzer,COB)時，常參考廠商建議的最佳檢驗切點值(optimal cut-off level)來評估吸煙習慣，偶爾會出現數據與真實習慣不符的情形。因此本研究透過對年輕軍人吸煙習慣的檢
測，探討適用之最佳檢驗切點值，並對準確性作驗證與分析，以提供較正確有力之依據，亦有利後續戒菸之追蹤評估。方法：利用軍營戒菸宣導活動，採「立意取樣」對來自台灣各營區的軍人共117位，實施問卷調查並以COB檢測每人之呼氣ㄧ氧化碳濃度。吸煙習慣之黃金診斷標準以問卷中回答每日吸菸量小於10支者，定義為”A組”，大於10支者為”B組”；分別以0-10 ppm為檢驗切點，計算各點所具有之敏感度及專一度，並以ROC曲線（receiver operating characteristic curve）進行準確性分析。結果：問卷結果分為A組31人，B組86人；樣本全體平均年齡25.2±6.2歲；呼氣一氧化碳平均值，樣本全體:12.6±10.8 ppm，A組:3.3±2.2
ppm，B組:16.0±10.1 ppm，達明顯統計差異(p<0.001)。最佳檢驗切點值顯示為6 ppm，其敏
感度86％，專一度94％，陽性相似比（positive likelihood ratio,LR+）:13.3。ROC曲線下面積（area under curve,AUC）為0.923（95％信賴區間:0.86-0.985）。結論：運用ROC曲線分析，得知呼氣一氧化碳濃度在本研究中，區分兩種吸菸習慣的最佳檢驗切點值應為6 ppm，也再次驗證COB仍是評估年輕軍人吸煙習慣之優良診斷工具。

Background: The optimal cut-off level recommended by manufacturers of carbon monoxide breathalyzer(COB) is often adopted in soldiers’smoking cessation; however, the data obtained doesn’t always correspond to the real smoking habits.This research aimed at evaluating the optimal cut-off level of COB and further analyzing its accuracy when assesseing smoking
habit in young soldiers, which shall offer a more reliable basis for following programs.Methods: Self-reported questionnaires and exhaled carbon monoxide(CO) detection by COB were performed by 117 soldiers from various camps all over Taiwan. Based on the self-reported smoking status, soldiers who smoked less than 10 cigarettes a day were defined as “Group A”, while more than 10 cigarettes as“Group B”.Under this gold standard, 0-10 ppm of carbon monoxide concentration was taken as cut-off level and the sensitivity, specificity of each level was calculated.The diagnostic accuracy was analysed with receiver operating characteristic (ROC) curve.
Results: Of the total population (average age 25.2±6.2 years), 31 and 86 subjects were classified to Group A and B,respectively.The mean value of exhaled CO was 12.6±10.8 ppm,constituted by Group A at 3.3±2.2 ppm and Group B at 16.0±10.1 ppm(p<0.001). The optimal cut-off level was 6 ppm with a sensitivity of 86%, specificity of 94% and positive likelihood ratio(LR+) as 13.3. The area under ROC curve(AUC) was 0.923(95％C.I.＝0.860-0.985).Conclusion: After the analysis of ROC curve, we found that 6 ppm should be the optimal cut-off level of exhaled CO for distinguishing between two smoking habits in our population and that COB is still an excellent diagnostic tool for assessing smoking habits in young soldiers.