文献速递其他结缔组织病

多机构电子幼年特发性关节炎表型的开发与验证

Development and validation of a multi-institutional electronic juvenile idiopathic arthritis phenotype.
PubMed 文献 · Semin Arthritis Rheum · 2026-02-16

本研究利用三个机构的报销和电子健康记录数据开发可计算的幼年特发性关节炎(JIA)表型算法,纳入ICD编码且年龄≤20岁者并排除其他风湿病。手动审查253例(3%)初始PPV为0.74,当要求≥4次JIA编码时PPV升至0.95,估计识别5377例,验证了算法在多机构间的一致性能。

要点

  • 基于ICD编码和排除标准构建可移植的JIA表型算法。
  • 增加JIA编码次数要求可提高阳性预测值至0.95。
  • 算法在多机构回顾性数据中表现一致,支持协作研究。

核心结论

  • 开发并复现了可计算、可移植的JIA表型算法,在多机构结构化回顾性数据中验证并显示一致性能。

研究设计与样本

  • 多机构研究,使用三个站点的报告资源,纳入有JIA和/或银屑病性幼年关节病ICD-9和ICD-10-CM编码且首次合格就诊年龄≤20岁的受试者,排除有系统性红斑狼疮、皮肌炎和多发性肌炎多个ICD编码者,各站点手动审查部分病历以验证诊断。

主要发现

  • 共识别8093例潜在JIA受试者。
  • 手动审查253例(3%)受试者,初始队列(一次编码计数加排除标准)的阳性预测值(PPV)为0.74。
  • 随着纳入所需的JIA编码计数增加,识别的JIA受试者数量减少,PPV增加。
  • 使用≥4次JIA编码计数、年龄≤20岁并排除其他疾病时,PPV为0.95,估计有5377例受试者。

关键数值与推荐

  • 初始队列(一次编码计数加排除标准)PPV为0.74。
  • ≥4次JIA编码计数、年龄≤20岁并排除其他疾病时,PPV为0.95,估计5377例受试者。

临床怎么用

  • 该算法可用于在结构化回顾性数据中识别JIA患者,支持协作研究网络中更广泛采用以提高JIA研究准确性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Ascertaining a true cohort of juvenile idiopathic arthritis (JIA) patients using claims and electronic health record (EHR) data is challenging due to clinical complexities. This study aims to develop a computable JIA phenotype using large databases with multi-site validation and replication. METHODS: This multi-institutional study used reporting resources at three sites identified through the Maternal and Pediatric Precision in Therapeutics Hub. Subjects with ICD-9 and ICD-10-CM codes for JIA and/or psoriatic juvenile arthropathy and age ≤20 at first eligible encounter were included with exclusion of subjects with multiple ICD codes for systemic lupus erythematosus, dermatomyositis, and polymyositis. Each site manually reviewed a subset of charts to validate diagnosis. Summary statistics on variations of encounter counts with a JIA diagnosis are reported. RESULTS: There were 8093 subjects identified with potential JIA. Manual review of 253 (3 %) subjects provided a positive predictive value (PPV) of 0.74 in an initial cohort of one code count with exclusion criteria. As the number of JIA code counts required for inclusion increased, the number of subjects with JIA identified decreased, and the PPV increased. When using ≥4 code counts for JIA, age ≤20 years old, and excluding other conditions, the PPV was 0.95, with an estimated 5377 subjects. CONCLUSION: We developed and reproduced a computable and portable phenotype algorithm to identify JIA patients in structured, retrospective data, validated this across multiple institutions, and demonstrated its consistent performance across sites. This approach supports broader adoption in collaborative research networks to enhance the accuracy of JIA studies.

来源:Semin Arthritis Rheum,查看原文。版权归原作者所有。

引用

  1. Taxter A, Oberle E, Rakestraw AL, et al. Development and validation of a multi-institutional electronic juvenile idiopathic arthritis phenotype. Semin Arthritis Rheum. 2026. PMID: 41722162. doi:10.1016/j.semarthrit.2026.152952

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。