Ammar Lin, Bird Kristin, Nian Hui, Maxwell-Horn Angela, Lee Rees, Ding Tan, Riddell Corinne, Gebretsadik Tebeb, Snyder Brittney, Hartert Tina, Wu Pingsheng
Division of Epidemiology, Vanderbilt University School of Medicine, Nashville, TN 37232, USA.
Department of Pediatrics, Vanderbilt University Medical Center, Nashville, TN 37232, USA.
Children (Basel). 2024 Oct 21;11(10):1271. doi: 10.3390/children11101271.
We aimed to develop an algorithm that accurately identifies children with Down syndrome (DS) using administrative data.
We identified a cohort of children born between 2000 and 2017, enrolled in the Tennessee Medicaid Program (TennCare), who either had DS coded on their birth certificate or had a diagnosis listed using an International Classification of Diseases (ICD) code (suspected DS), and who received care at Vanderbilt University Medical Center, a comprehensive academic medical center, in the United States. Children with suspected DS were defined as having DS if they had (a) karyotype-confirmed DS indicated on their birth certificate; (b) karyotype-pending DS indicated on their birth certificate (or just DS if test type was not specified) and at least two healthcare encounters for DS during the first 6 years of life; or (c) at least three healthcare encounters for DS, with the first and last encounter separated by at least 30 days, during the first six years of life. The positive predictive value (PPV) of the algorithm and 95% confidence interval (CI) were reported.
Of the 411 children with suspected DS, 354 (86.1%) were defined as having DS by the algorithm. According to medical chart review, the algorithm correctly identified 347 children with DS (PPV = 98%, 95%CI: 96.0-99.0%). Of the 57 children the algorithm defined as not having DS, 50 (97.7%, 95%CI: 76.8-93.9%) were confirmed as not having DS by medical chart review.
An algorithm that accurately identifies individuals with DS using birth certificate data and/or ICD codes provides a valuable tool to study DS using administrative data.
我们旨在开发一种算法，该算法可利用管理数据准确识别唐氏综合征（DS）患儿。
我们确定了一组2000年至2017年出生且参加田纳西医疗补助计划（TennCare）的儿童，这些儿童在出生证明上有DS编码，或使用国际疾病分类（ICD）编码列出了诊断结果（疑似DS），并且在美国一家综合性学术医疗中心范德堡大学医学中心接受治疗。疑似DS儿童若符合以下情况则被定义为患有DS：（a）出生证明上显示核型确诊的DS；（b）出生证明上显示核型待确认的DS（若未指定检测类型则仅显示DS），且在生命的前6年至少有两次因DS的医疗就诊；或（c）在生命的前6年至少有三次因DS的医疗就诊，且第一次和最后一次就诊间隔至少30天。报告了该算法的阳性预测值（PPV）及95%置信区间（CI）。
在411名疑似DS儿童中，354名（86.1%）被该算法定义为患有DS。根据病历审查，该算法正确识别出347名DS儿童（PPV = 98%，95%CI：96.0 - 99.0%）。在该算法定义为未患DS的57名儿童中，50名（97.7%，95%CI：76.8 - 93.9%）经病历审查确认未患DS。
一种利用出生证明数据和/或ICD编码准确识别DS个体的算法为利用管理数据研究DS提供了一种有价值的工具。
