Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, United States; Division of Rheumatology, Hanyang University Hospital for Rheumatic Diseases, Seoul, South Korea.
Division of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States.
Semin Arthritis Rheum. 2020 Aug;50(4):592-597. doi: 10.1016/j.semarthrit.2020.04.006. Epub 2020 May 20.
To develop and validate claims-based algorithms to identify interstitial lung disease (ILD) in patients with rheumatoid arthritis (RA) METHODS: Using Medicare claims data linked with the electronic medical records (2012-2014), we first selected RA patients based on ≥2 diagnostic codes for RA and ≥1 disease-modifying antirheumatic drugs.Then, to identify ILD in RA, we developed eight claims-based algorithms using a combination of ICD-9 diagnosis codes and procedure codes related to the diagnosis or management of ILD. We assessed the positive predictive value (PPV) for each of the eight algorithms relative to confirmed ILD cases using chest computerized tomography or lung biopsy as the gold standard.
A total of 5,214 RA patients were included in the study, and the ILD cases identified by each algorithm ranged from 181 to 993. The PPV of the diagnosis code-based algorithms ranged from 43.4% (≥1 diagnosis code by any physician) to 52.0% (≥2 diagnosis codes by any physician). When the algorithms further required ≥1 procedure code (e.g., imaging, bronchoscopy), the PPV did not improve. However, the algorithms that required ILD diagnosis codes by specialists (i.e., pulmonologist or rheumatologist) had PPVs of 61.5% with ≥1 code; 72.4% with ≥2 codes.
In a cohort of RA patients, our algorithm that required ≥2 ILD diagnosis codes by specialists demonstrated a PPV of 72.4% in ascertaining ILD. Our results support the utility of the claims-based algorithm to identify a population-based cohort of RA patients with ILD using large administrative claims data.
开发并验证基于索赔的算法，以识别类风湿关节炎（RA）患者中的间质性肺病（ILD）。
使用医疗保险索赔数据与电子病历（2012-2014 年）进行关联，我们首先根据≥2 个 RA 诊断代码和≥1 种疾病修饰抗风湿药物选择 RA 患者。然后，为了在 RA 中识别 ILD，我们开发了八种基于索赔的算法，使用 ICD-9 诊断代码和与 ILD 诊断或管理相关的程序代码的组合。我们使用胸部计算机断层扫描或肺活检作为金标准，评估了每种算法识别 ILD 的阳性预测值（PPV）。
共纳入 5214 名 RA 患者，每种算法识别的 ILD 病例数从 181 到 993 不等。基于诊断代码的算法的 PPV 从 43.4%（任何医生的≥1 个诊断代码）到 52.0%（任何医生的≥2 个诊断代码）不等。当算法进一步要求≥1 个程序代码（例如，成像，支气管镜检查）时，PPV 并没有提高。然而，要求ILD 诊断代码由专家（即肺科医生或风湿病专家）开具的算法的 PPV 为 61.5%，至少有 1 个代码；≥2 个代码时为 72.4%。
在 RA 患者队列中，我们要求≥2 个由专家开具的 ILD 诊断代码的算法确定 ILD 的 PPV 为 72.4%。我们的结果支持使用大型管理索赔数据，基于索赔的算法来识别基于人群的 RA 患者 ILD 队列的有效性。
