Dixon Brian E, Rahurkar Saurabh, Ho Yenling, Arno Janet N
Department of Epidemiology, Indiana University Richard M Fairbanks School of Public Health, Indianapolis, Indiana, USA
Center for Biomedical Informatics, Regenstrief Institute, Indianapolis, Indiana, USA.
BMJ Health Care Inform. 2019 Aug;26(1). doi: 10.1136/bmjhci-2019-100074.
International Classification of Diseases (ICD) codes in administrative health data are used to identify cases of disease, including sexually transmitted infections (STIs), for population health research. The purpose of this review is to examine the extant literature on the reliability of ICD codes to correctly identify STIs.
We conducted a systematic review of empirical articles in which ICD codes were validated with respect to their ability to identify cases of chlamydia, gonorrhoea, syphilis or pelvic inflammatory disease (PID). Articles that included sensitivity, specificity and positive predictive value of ICD codes were the target. In addition to keyword searches in PubMed and Scopus databases, we further examined bibliographies of articles selected for full review to maximise yield.
From a total of 1779 articles identified, only two studies measured the reliability of ICD codes to identify cases of STIs. Both articles targeted PID, a serious complication of chlamydia and gonorrhoea. Neither article directly assessed the validity of ICD codes to identify cases of chlamydia, gonorrhoea or syphilis independent of PID. Using ICD codes alone, the positive predictive value for PID was mixed (range: 18%-79%).
While existing studies have used ICD codes to identify STI cases, their reliability is unclear. Further, available evidence from studies of PID suggests potentially large variation in the accuracy of ICD codes indicating the need for primary studies to evaluate ICD codes for use in STI-related public health research.
行政卫生数据中的国际疾病分类（ICD）编码用于识别疾病病例，包括性传播感染（STIs），以进行人群健康研究。本综述的目的是研究现有文献中关于ICD编码正确识别性传播感染的可靠性。
我们对实证文章进行了系统综述，其中对ICD编码识别衣原体、淋病、梅毒或盆腔炎（PID）病例的能力进行了验证。目标是包含ICD编码敏感性、特异性和阳性预测值的文章。除了在PubMed和Scopus数据库中进行关键词搜索外，我们还进一步检查了被选作全文综述的文章的参考文献，以最大限度地提高文献产量。
在总共识别出的1779篇文章中，只有两项研究测量了ICD编码识别性传播感染病例的可靠性。这两篇文章均以PID为目标，PID是衣原体和淋病的一种严重并发症。两篇文章均未独立于PID直接评估ICD编码识别衣原体、淋病或梅毒病例的有效性。仅使用ICD编码时，PID的阳性预测值不一（范围：18%-79%）。
虽然现有研究使用ICD编码来识别性传播感染病例，但其可靠性尚不清楚。此外，来自PID研究的现有证据表明，ICD编码的准确性可能存在很大差异，这表明需要进行初步研究，以评估ICD编码在性传播感染相关公共卫生研究中的应用。
