Selection effect of needle exchange in Anchorage, Alaska.
Selection effect of needle exchange in Anchorage, Alaska.
复制标题
阿拉斯加安克雷奇针交换的选择效应。
DOI：
10.1093/jurban/79.1.128
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
Harbke,ColinR
中科院分区：
文献类型：
--
作者：
Fisher,DennisG;Reynolds,GraceL;Harbke,ColinR
参与偏倚（选择偏倚）可能是一个问题，在研究中，试图评估针头交换计划（NEPs）的影响。目前的研究只关注那些注射吸毒者（IDUs），他们被随机放置在针头交换条件下，进行了一项双臂随机临床试验。实验NEP条件（n=296，中位数=261 d）与药房销售条件（n=304，中位数=256 d）随访时间差异无统计学意义[χ2(1, n= 600)=0.42， P= 0.52]。在NEP条件下，对使用NEP的idu （n=65，中位数=199天）和未使用NEP的idu （n=231，中位数=286天）的随访时间进行类似分析，结果非常显著，χ2(1, n=296)=17.3， P= 0.0001。预测NEP使用的最终逻辑回归模型是最近30天内注射海洛因次数的log 10转换（优势比[OR]=4.9，置信区间[CI] 2.8, 8.9），最近30天内注射过安非他明（OR=4.9, CI1.09, 22.5），以及最近30天内曾经共用注射设备（OR=2.9, CI1.5, 5.5）。在NEP条件下，使用NEP和不使用NEP的人的随访率不同。在被随机分配到新经济政策的吸毒者中，那些真正使用新经济政策的人吸毒水平更高。使用NEP的预测指标与Hagen等人的研究一致。本研究表明，选择偏差发生在使用nep的注射吸毒者中。
Participation bias (selection bias) may be a problem in studies that attempt to evaluate the effects of needle-exchange programs (NEPs). The present study looked at only those injection drug users (IDUs) who were randomly placed in the needle-exchange condition in a two-arm randomized clinical trial of needle exchange. Time to follow-up between the experimental NEP condition (n=296; median=261 days) and pharmacy sales condition (n=304; median=256 days) was not statistically different [χ2(1, N=600)=0.42, P=.52]. Within the NEP condition, a similar analysis comparing time to follow-up between IDUs who used the NEP (n=65; median=199 days) and those who refrained from using the NEP (n=231; median=286 days) was highly significant, χ2(1, n=296)=17.3, P=.0001. The final logistic regression model predicting use of the NEP was the log 10 transformation of the number of times injected heroin in the last 30 days (odds ratio [OR]=4.9, confidence interval [CI] 2.8, 8.9), ever injected amphetamine in the last 30 days (OR=4.9, CI1.09, 22.5), and ever shared injection equipment in the last 30 days (OR=2.9, CI1.5, 5.5). Within the NEP condition, follow-up rates differed between those who used the NEP and those who did not use it. Of drug users randomly assigned to an NEP, the ones who actually used the NEP had higher levels of drug use. Predictors of who used the NEP were consistent with the studies of Hagen et al. This study demonstrates that selection bias occurs among IDUs who use NEPs.