We used data from previous applications of the confidante method in Côte d\'Ivoire, Ethiopia, Ghana, Java (Indonesia), Nigeria, Uganda, and Rajasthan (India). We estimated one-year induced abortion incidence rates for confidantes in each context, attempting to adjust for selection, reporting and transmission bias in a standardized manner.
In each setting, majority of the foundational confidante method assumptions were violated. Adjusting for transmission bias using self-reported abortions consistently yielded the highest incidence estimates compared with other published approaches. Differences in analytic decisions and bias assessments resulted in the incidence estimates from our standardized analysis varying widely from originally published rates.
We recommend that future studies clearly state which biases were assessed, if associated assumptions were violated, and how violations were adjusted for. This will improve the utility of confidante method estimates for national-level decision making and as inputs for global or regional model-based estimates of abortion.
方法:我们使用了科特迪瓦以前应用的红颜知己方法的数据,埃塞俄比亚,加纳,Java(印度尼西亚),尼日利亚,乌干达,拉贾斯坦邦(印度)。我们估计了每种情况下知己一年的人工流产发生率,尝试调整选择,以标准化的方式报告和传播偏差。
结果:在每个设置中,大多数的基本知己方法的假设被违反。与其他已发表的方法相比,使用自我报告的流产调整传输偏差始终产生最高的发生率估计值。分析决策和偏倚评估的差异导致我们的标准化分析得出的发病率估计与最初发表的比率相差很大。
结论:我们建议未来的研究清楚地说明评估的偏见,如果违反了相关假设,以及如何调整违规行为。这将提高红颜知己方法估计在国家一级决策中的实用性,并作为基于全球或区域模型的堕胎估计的投入。