Cost-Effectiveness of Family Conferences to Reduce Polypharmacy in Frail Older Adults

Cost-effectiveness of family conferences on deprescribing with joint prioritization of treatment goals in primary care has not been investigated so far. We assessed cost-effectiveness in the cluster-randomized controlled COFRAIL trial conducted with general practitioners and 521 older frail patients with polypharmacy cared for at home in Germany.

Hospital admissions averted and quality-adjusted life years (QALYs) gained were associated with costs from the German Social Insurance perspective. We applied adjusted GLM regressions with specified distributions to estimate group differences on imputed data, plotted bootstrap cost-outcome pairs by simulated resampling of the study population to illustrate uncertainty and calculate the probability of cost-effectiveness given a willingness-to-pay threshold, and assessed robustness in sensitivity analyses.

Intervention-related costs were €391 (US$459) per capita. On 100 people, the COFRAIL intervention had about 7 more hospital admissions (95% CI: -12; 26), 2 QALYs gained (95% CI: -1; 6), and additional costs of €117,681 (95% CI: -28,838; 264,201)/US$138,027 (95% CI: -33,824; 309,880) or €124,866 (95% CI: -12,649; 262,380)/US$146,455 (95% CI: -14,836; 307,745) without or with hospital costs, respectively, compared to usual care. By bootstrapping, we observed the COFRAIL intervention to have higher costs and more hospital admissions with a relative frequency of 28%-78%, or in terms of QALYs 57%-91%. The COFRAIL intervention had additional costs of €50,966 (US$59.778) per QALY gained with a 46% probability of being cost-effective at a willingness to pay of €45,000/QALY (≈US$50,000/QALY).

The COFRAIL intervention affected QALYs rather than hospital admissions after 12 months. The intervention tended to be associated with higher costs and QALYs but was less likely to be cost-effective than usual care at commonly used willingness-to-pay thresholds. Long-term cost-effectiveness should be assessed.