THE HEALTH-LOSS BURDEN OF OSTEOARTHRITIS IN NEW ZEALAND: RESULTS FROM A NEW COMPUTER-SIMULATION MODEL
Method: We developed the NZ-MOA model, a population-based state-transition micro-simulation model of the natural history of radiographic knee OA. Simulated individuals in the model—drawn from the 2013 NZ population aged 45–79—are followed through an annual sequence of health-state transitions based on radiographic disease progression. Health-related quality of life (HRQoL) was assigned to each health state using the six dimensions of the SF-6D. Input data were sourced from the NZ Census, the NZ Health Survey, the Osteoarthritis Initiative, and published research.
Lifetime prevalence and total QALYs lost due to radiographic knee OA were calculated for the cohort, stratified by sex and ethnicity (Maori vs. non-Maori). Preliminary model validation was conducted by running the model using the 2006 NZ population and comparing the results with those previously obtained from the Osteoarthritis Policy (OAPol) model, an established computer simulation model originally developed for the US population. Sensitivity analyses were run on the model by varying the method used to assign HRQoL values to health states: using either NZ EQ-5D population norms or Global Burden of Disease (GBD) study disability weights, both dependent on the severity of OA-related pain, in place of the multi-dimensional SF-6D instrument.
Result: Radiographic knee OA was estimated to affect 26% of the cohort over their lifetime, resulting in a total of 511,000 QALYs lost. QALYs lost per capita were higher for women than for men, and for non-Maori than Maori ethnicity. Results for the 2006 population were similar to previous OAPol estimates, although QALY losses for Maori and for older individuals were slightly higher than in the OAPol model. QALY losses were somewhat higher (579,000 QALYs) using EQ-5D population norms, and substantially lower using the GBD disability weights (322,000 QALYs).
Conclusion: The total disease burden of knee OA in NZ is substantial. The new NZ-MOA model provided estimates comparable to those from the established and well-validated OAPol model. Estimates were sensitive to varying assumptions on the HRQoL impacts of OA; the lower estimates obtained using the GBD disability weights may be due to the absence of mental and emotional health dimensions in the valuation of those disability weights.