1 Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina , Charleston, South Carolina.
J Womens Health (Larchmt). 2014 May;23(5):413-9. doi: 10.1089/jwh.2013.4599. Epub 2014 Apr 21.
Latent class analysis (LCA) is a useful statistical tool that can be used to enhance understanding of how various patterns of combined sexual behavior risk factors may confer differential levels of HIV infection risk and to identify subtypes among African American adolescent girls.
Data for this analysis is derived from baseline assessments completed prior to randomization in an HIV prevention trial. Participants were African American girls (n=701) aged 14-20 years presenting to sexual health clinics. Girls completed an audio computer-assisted self-interview, which assessed a range of variables regarding sexual history and current and past sexual behavior.
Two latent classes were identified with the probability statistics for the two groups in this model being 0.89 and 0.88, respectively. In the final multivariate model, class 1 (the "higher risk" group; n=331) was distinguished by a higher likelihood of >5 lifetime sexual partners, having sex while high on alcohol/drugs, less frequent condom use, and history of sexually transmitted diseases (STDs), when compared with class 2 (the "lower risk" group; n=370). The derived model correctly classified 85.3% of participants into the two groups and accounted for 71% of the variance in the latent HIV-related sexual behavior risk variable. The higher risk class also had worse scores on all hypothesized correlates (e.g., self-esteem, history of sexual assault or physical abuse) relative to the lower risk class.
Sexual health clinics represent a unique point of access for HIV-related sexual risk behavior intervention delivery by capitalizing on contact with adolescent girls when they present for services. Four empirically supported risk factors differentiated higher versus lower HIV risk. Replication of these findings is warranted and may offer an empirical basis for parsimonious screening recommendations for girls presenting for sexual healthcare services.
潜在类别分析(LCA)是一种有用的统计工具,可用于增强对各种组合性行为风险因素模式如何赋予不同水平的 HIV 感染风险的理解,并识别非裔美国少女中的亚组。
本分析的数据来自于一项 HIV 预防试验的随机分组前完成的基线评估。参与者是非裔美国少女(n=701),年龄在 14-20 岁,就诊于性健康诊所。女孩们完成了一个音频计算机辅助自我访谈,评估了有关性史以及当前和过去性行为的一系列变量。
使用概率统计方法识别出了两个潜在类别,该模型中这两个组的概率分别为 0.89 和 0.88。在最终的多变量模型中,第 1 类(“高风险”组;n=331)的特征是更高的可能性具有>5 个性伴侣,在醉酒/吸毒时发生性行为,较少频繁使用避孕套,以及性传播疾病(STD)史,与第 2 类(“低风险”组;n=370)相比。该衍生模型正确地将 85.3%的参与者分为两组,解释了 71%的潜在 HIV 相关性行为风险变量的方差。高风险组在所有假设的相关性(例如,自尊、性侵犯或身体虐待史)方面的得分都比低风险组差。
性健康诊所是通过在少女就诊时利用与她们接触的机会,提供与 HIV 相关的性行为风险行为干预的独特切入点。四个经验支持的风险因素区分了更高与更低的 HIV 风险。需要对这些发现进行复制,这可能为少女就诊性保健服务时进行简明的筛查建议提供实证依据。