THE COST-EFFECTIVENESS OF A DECISION AID FOR PATIENTS CONSIDERING TOTAL HIP OR KNEE ARTHROPLASTY: SUB-ANALYSIS OF A RANDOMIZED CONTROLLED TRIAL

Purpose: To establish the cost-effectiveness of a decision aid (plus surgeon preference report) for patients considering total hip or knee arthroplasty.

Methods: The economic evaluation was undertaken using data from a randomized controlled trial with two years follow-up. 343 patients with osteoarthritis were recruited from two orthopedic screening clinics in Ottawa, Canada. Patients were randomized to either the control group (standard education materials) or decision aid plus surgeon preference report (intervention). The intervention was a decision aid and a one-page preference report that documented patients' clinical and decision data for the surgeon. Patient-related data were collected at baseline and every 6 months up to two years. Primary outcomes were health system costs (in 2014 CAD$) and quality-adjusted life years (QALYs), with comparative results presented as an incremental cost-effectiveness ratio (ICER). Costs were calculated using self-reported resource utilization from patient diaries and associated unit costs. The cost of the intervention included the decision aid itself, the time required to compile the preference report, and an additional consultation with the surgeon. QALYs were calculated using EQ-5D health utilities, derived from the Western Ontario and McMaster Osteoarthritis Index (WOMAC) data. Both costs and QALYs were discounted at 5%. A combined multiple imputation and bootstrapping procedure was used to handle missing data and to estimate uncertainty in cumulative costs and QALYs over the study period.

Results: The sample comprised 174 intervention and 169 control group patients. The typical patient was 66 years old, retired and living with someone. The decision aid arm had fewer surgeries over the 2-year period (76.6% vs 80.8%), thereby incurring a negative incremental cost of –$367 (95% CI -$1,242 to $499) while also providing 0.065 (95% CI -0.007 to 0.154) additional QALYs. As a consequence the decision aid arm was dominant. The probability of the decision aid arm being dominant was 74% (see Figure).

Conclusion: Even with the assumption that the use of the decision aid would result in an additional surgical consultation, its use was dominant, resulting in lower costs and more QALYs. Given there are over 100,000 arthroplasties performed in Canada each year, implementing a decision aid could have important health policy implications. Future research will link trial and administrative data to allow for longer-term (5-year) evaluation of cost-effectiveness.