Cherpitel C J
Western Consortium for Public Health, Alcohol Research Group, Berkeley, California 94709, USA.
J Stud Alcohol. 1999 Sep;60(5):705-11. doi: 10.15288/jsa.1999.60.705.
The purpose of this study was to compare the performance of two screening instruments for alcohol problems, the CAGE and the TWEAK, against ICD-10 and DSM-IV criteria for alcohol dependence by gender and ethnicity, and to evaluate whether characteristics associated with health services utilization may affect the performance of screening instruments, in a representative sample of the U.S. adult general population.
Data are from the Alcohol Research Group's 1995 National Alcohol Survey; these were weighted to account for the design effect inherent in multistage cluster sampling and oversampling of blacks and Hispanics. Effective sample size was 2,443: 797 blacks, 642 Hispanics and 1,004 whites and others (primarily Asian and Native American). Sensitivity and specificity were examined for both screening instruments. Logistic regression was used to evaluate the predictive value, separately, for the CAGE and TWEAK, controlling for gender, ethnicity, region of the country and service use (primary care and emergency room).
The TWEAK was more sensitive for men than for women, but no differences were found by service use, while the CAGE was more sensitive for men who had used the emergency room during the preceding year compared with those who had not. The TWEAK was more sensitive than the CAGE among white and Hispanic men, among men who had not used the emergency room and among whites who had no service use. Performance of neither the CAGE nor the TWEAK was found to vary by region of the country.
Data suggest that while the performance of screening instruments may vary across demographic subgroups in the general population, instruments may perform equally well for identifying problem drinkers in general populations as in clinical populations.
本研究旨在比较两种酒精问题筛查工具（CAGE和TWEAK）与国际疾病分类第10版（ICD - 10）及精神疾病诊断与统计手册第4版（DSM - IV）中酒精依赖标准在性别和种族方面的表现，并评估与卫生服务利用相关的特征是否会影响筛查工具的性能，研究对象为美国成年普通人群的代表性样本。
数据来自酒精研究小组1995年的全国酒精调查；对数据进行加权处理，以考虑多阶段整群抽样以及对黑人和西班牙裔进行过度抽样所固有的设计效应。有效样本量为2443人：797名黑人、642名西班牙裔以及1004名白人和其他人（主要是亚洲人和美洲原住民）。对两种筛查工具均进行了敏感性和特异性检测。使用逻辑回归分别评估CAGE和TWEAK的预测价值，同时控制性别、种族、所在地区以及服务利用情况（初级保健和急诊室）。
TWEAK对男性的敏感性高于女性，但在服务利用方面未发现差异，而CAGE对前一年使用过急诊室的男性比未使用过的男性更敏感。在白人和西班牙裔男性中、未使用过急诊室的男性中以及未使用任何服务的白人中，TWEAK比CAGE更敏感。未发现CAGE和TWEAK的性能因所在地区而有所不同。
数据表明，虽然筛查工具的性能在普通人群的不同人口亚组中可能存在差异，但这些工具在识别普通人群中的问题饮酒者方面可能与在临床人群中表现得同样出色。
