Background/Purpose: Expenditure on biologic therapies for rheumatoid arthritis (RA) accounts for the highest pharmaceutical spending in many western healthcare systems. Given their clinical benefit with marked reduction in symptoms and prevention of joint damage, biologics are expected to lead to reduced treatment related costs, such as joint surgeries or physician visits. Despite over a decade of routine use, there is currently no direct evidence supporting an association between biologic use and subsequent reductions in healthcare utilization.  The objective of this study was to estimate the impact of biologic treatment compared to DMARDs on other healthcare utilization using instrumental variable (IV) techniques to control for confounding by indication.

Methods: From a population-based cohort of all patients with a rheumatologist diagnosis of RA identified from administrative data, we selected patients that mimick  eligibility for reimbursement for biologics between 2003-2007 based on DMARD use history and having 3-year follow-up under the care of rheumatologists since eligibility. Patients receiving biologics in the first follow-up year form the treatment group whereas others who never received biologics throughout the follow-up form the control group. The data contains patient-level information on hospitalizations, physician visits (including investigations and procedures), and prescription medications. Since disease severity is known to be associated with treatment received and healthcare use, but is not directly captured in administrative data, conventional risk-adjustment approach might not adequately adjust for this confounding by indication.  We use an IV based on the prescribing rheumatologist’s preference for biologic use at the time of treatment assignment to address the issue of unobserved confounder.

Results: The final analysis included 314 and 486 patients in the biologic and DMARD groups, respectively. As expected, conventional multivariable regression adjusting for observed confounders, but not necessarily for confounding by indication, did not attribute biologic use with cost offsets. The IV analysis results suggest that RA-related costs in the biologic group, compared to the DMARD group, in the 2- and 3-year follow up periods , was 22.5% (95% CI: -21.0% – 90.0%) and 48.3% (2.0% – 115.6%) more for physician visits but 45.9% (-63.7% – 82.1%) and 32.7%  (-97.4% – 77.0%) less for medications excluding biologics or DMARD, whereas the odds ratios of hospitalization were 2.308 (0.44 – 12.00) and 3.64 (0.71 – 18.68), respectively. A similar pattern was observed for total resource utilization.

Conclusion: While RCTs are the gold standard for determining causal relationships, such studies would be infeasible to investigate this important question. Through the use of established econometric methods and a population-based administrative dataset, this study finds no signal that biologic use offsets health care costs in the short-term. Limitations include large confidence intervals common in IV studies, and follow-up of 3 years, which may be too short to observe cost savings from joint replacement prevention.

---

« Back to 2013 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/does-biologic-treatment-for-rheumatoid-arthritis-offset-health-care-costs-in-patients-with-rheumatoid-arthritis-an-instrumental-variable-approach-using-administrative-data/