The clinical status and economic savings associated with remission among patients with rheumatoid arthritis: leveraging linked registry and claims data for synergistic insights

Abstract

Introduction: Treat to target guidelines recommend achieving remission or low disease activity in rheumatoid arthritis (RA). However, the reduction in adverse events and costs associated with lower disease activity is unclear.

Methods: We used Corrona linked to Medicare data to identify RA patients. Time varying disease activity was measured using Clinical Disease Activity Index (CDAI); outcomes included all-cause hospitalization, a composite of hospitalization or emergency department (ED) visits, mortality, and medical costs. Outcome-specific Cox proportional models evaluated the adjusted hazard ratios between disease activity and outcomes, controlling for potential confounders including comorbidities grouped into four patient phenotypes. Costs were analyzed with mixed models using a Gaussian distribution with log transformation.

Results: Depending on outcome, 4593 RA patients contributed up to 12 001 person years. Median age was 71 years, 75% women. At baseline, approximately 50-60% of patients were in remission or low disease activity. There was a dose-response relationship between RA disease activity (remission, low, moderate, and high) and the incidence of hospitalizations (13.1, 17.8, 21.2, 27.5 per 100 py, respectively); all adjusted hazard ratios were significant: 0.68 (remission), 0.87 (low), and 1.24 (high) compared with moderate disease activity. Similar trends were observed for ED visits and mortality. The crude difference in annual medical costs between remission ($11 145) and moderate disease activity ($17 646) was $-6 500; the adjusted difference (95%CI) was $-3133 (-4737.72, -1528.43).

Conclusion: Leveraging the benefits of linking registry and administrative data together, lower disease activity in RA was associated with incrementally reduced risks of all-cause hospitalization, ED visits, mortality, and medical costs in a dose-dependent fashion. Copyright © 2016 John Wiley & Sons, Ltd.

Keywords: Medicare; administrative claims; disease activity; pharmacoepidemiology; registry; rheumatoid arthritis.

Figure 1

Comorbidity-related patient phenotypes derived from k-means clustering plotted on first 3 principal component vectors Cluster numbers refer to the categories in Table 1: Cluster 1 (black), Cluster 2 (red), Cluster 3 (blue), Cluster 4 (green)

Figure 2

Rate of Hospitalization Associated with RA Disease Activity Category and Comorbidity Cluster