Epidemiology and clinical features of interstitial lung disease in patients with rheumatoid arthritis from the JointMan database
Data source
Patient demographics and disease characteristics were retrospectively analyzed after data extraction from the Discus Analytics JointMan database, a large US electronic health record-based dataset initiated in March 2009. The JointMan database includes > 17,000 rheumatology patients covered by commercial, Medicare, or Medicaid insurance. plan. Practices in the following eight states are included: Washington, New York, Oregon, Florida, Georgia, California, Wisconsin and Kentucky. Patient data were collected in rheumatology centers and anonymized prior to analysis. In addition to electronic medical record data, the JointMan user interface collects clinical results recorded by physicians at the time of the encounter.
Patient population
Patients were included if they were ≥ 18 years old at first visit to a rheumatologist participating in the JointMan network, had a provider-selected diagnosis of RA between January 1, 2009 and September 20, 2019, and had ≥ 1 visit after the first visit. visit date. Patients were excluded if their first encounter occurred after RA diagnosis or if they experienced a drug-induced ILD diagnosis [International Classification of Disease, Tenth Revision, Clinical Modification (ICD-10-CM) codes J70.2 and J70.4] at any time during the study period. Patients were assigned to the RA cohort (patients with confirmed RA but no diagnosis of ILD during the study period) or the RA-ILD cohort (patients with a diagnosis of unmedicated ILD on or after the initial diagnosis date of RA). ). The RA index date was defined as the first RA diagnosis date recorded in the JointMan database (provided by the rheumatologist).
The total study population consisted of patients followed from the day after the RA index date until the patient’s last encounter date or the end of the study (September 20, 2019), whichever came first. RA was diagnosed according to the ICD, Ninth Revision, CM (ICD-9-CM) code 714.0 and ICD-10-CM codes M05 and M06. ILD was identified by ICD diagnosis codes (ICD-9-CM codes: 516.0, 516.2, 516.3, 516.4, 516.5, 516.8, and 516.9; ICD-10-CM codes: J84.0, J84.1, J84.2, J84 .81, J84.82, J84.83, J84.89 and J84.9) or as indicated by the provider.
A subanalysis was performed on a series of patients grouped according to ILD diagnosis. For the subanalysis population, the ILD diagnosis index was defined as the first date of ILD diagnosis recorded in the JointMan database (for patients in the RA-ILD cohort), and patient characteristics were described for the 90-day periods before and after the ILD. diagnosis index. For patients without ILD, the index date was based on the distribution of the number of days between RA diagnosis and ILD diagnosis in the RA-ILD cohort; characteristics were described for the 90-day periods before and after the index date (Supplementary Figure S1).
Primary endpoints
The primary endpoints, assessed in the total study population, were the prevalence and time to onset of ILD. Prevalence was defined as the proportion of patients with RA and a diagnosis of ILD divided by the total number of patients with RA during the study period. Time to onset of ILD was defined as the time from the initial diagnosis of RA to the first observed non-drug ILD diagnosis.
Exploratory endpoints
Exploratory endpoints, assessed in the exploratory analysis population, included baseline demographics, comorbidities, RA characteristics, and overall RA disease activity in the RA cohort compared with the RA-ILD cohort. RA features include joint stiffness, erosions, extra-articular disease, anti-CCP antibodies, joint swelling, ESR, C-reactive protein (CRP), and Clinical Disease Activity Index (CDAI). The CDAI remission score was defined as ≤ 2.8; CDAI low, moderate, and high disease activity scores were defined as >2.8–10, >10–22, and >22, respectively19. The Simplified Disease Activity Index (SDAI) remission score was defined as ≤ 3.3; SDAI low, moderate, and high disease activity scores were defined as > 3.3 to 11, > 11 to 26, and > 26, respectively19. Disease activity score in 28 joints using CRP (DAS28 [CRP]) remission score was defined as ≤ 2.3; DAS28 (CRP) low, moderate and high disease activity scores were defined as > 2.3 to 2.7, > 2.7 to < 4.1 and ≥ 4.1, respectively20. DAS28 (ESR) remission score was defined as <2.6; DAS28 (ESR) low, moderate, and high disease activity scores were defined as 2.6 to 2.6, respectively < 3,2, 3,2–5,1 en > 5.1.19 Routine Assessment of Patient Index Data 3 (RAPID3) remission score was defined as ≤ 3; RAPID3 low, moderate, and high disease activity scores were defined as >3 to 6, >6 to 12, and >12, respectively21. Variables were assessed as potential predictors of RA-ILD.
Subanalysis endpoints
For patients included in the subanalysis population, CDAI and RAPID3 scores, number of swollen and swollen28 joints, number of encounters with rheumatologists, and treatment use before and after the ILD diagnosis index were also assessed. The number of swollen and swollen28 joints is part of the DAS/DAS28 score: the number of swollen joints is an assessment of 28 or more (maximum 44) joints, while the number of swollen28 joints is an assessment of only 28 pre-selected joints22.
static analysis
The prevalence (95% confidence intervals [CIs]) of the first observed ILD diagnosis during follow-up was calculated. Time to ILD diagnosis was examined using unadjusted Kaplan-Meier survival curves. Descriptive statistics for continuous baseline variables were compared using Student’s Ttest and percentages for categorical and binary basic variables were compared using the Chi-square test.
Potential predictors of RA-ILD were analyzed with a Cox regression model. Patient demographics and comorbidities were collected at baseline and controlled in the Cox model. RA features were identified during and after initial RA diagnosis and were controlled as time-varying covariates in the Cox model. The final covariate lists were based on clinical rationale and model fit; Hazard Ratios, 95% Confidence Intervals, and P Values were provided for each covariate. Statistical significance for model inclusion was set at P<0.05.
The number and percentage of patients with visits to a rheumatologist, treatment utilization, and each disease activity score in the pre- and post-index periods were calculated. P-values for the disease activity score category compared pre- and post-index periods and correspond to Fisher’s exact test or Chi-square test with statistical significance set at P<0.05.
Ethical approval
This study was conducted in accordance with the International Society for Pharmacoepidemiology Guidelines for Good Pharmacoepidemiology Practices and applicable regulatory requirements23. The study protocol was reviewed by the internal BMS Observational Protocol Review Committee (OPRC). No identifiable protected health information was retrieved from or accessed from the database during the study. Therefore, the BMS OPRC confirmed that this analysis did not require ethical oversight. In addition, the study did not involve the collection, use, or transmission of individually identifiable data, and the data was collected in the setting for the patient’s usual care. Informed consent from the study participants was not required because the dataset used in this observational study consisted of anonymized secondary data released for research purposes.
