Document Type : Original Article
Authors
1
Department of Health Economics, School of Business and Economics, Vrije Universiteit, Amsterdam, The Netherlands
2
Amsterdam University Medical Center, Amsterdam, The Netherlands
3
Department of Geriatrics, Amsterdam University Medical Center, Amsterdam, The Netherlands
4
Leading the Change, Zorgevaluatie Nederland, The Netherlands
5
National Healthcare Institute, Diemen, The Netherlands
6
ONVZ Health Insurance, Houten, The Netherlands
7
Zilveren Kruis Health Insurance, Zeist, The Netherlands
Abstract
Background
Reducing low-value care (LVC) remains a significant challenge in healthcare systems, despite efforts for de-implementation. To facilitate de-implementation payers may use provider Audit & Feedback (A&F). This study aims to examine trends in LVC following a payer-initiated A&F program, and to explore whether this association varies across hospital payment models.
Methods
Multilevel logistic regression was used to examine the trend in LVC in hospitals, its correlation with an A&F intervention, and whether this varied with payment models including fee-for-service (FFS), global budget (GB), cost ceiling (CC) and gradual tariff (GT). To assess this, we used hospital level claims data from a large nationwide Dutch health insurer on 70 hospitals in the period 2017-2022. The primary outcome was the yearly proportion of LVC per hospital. Independent variables were time, market share and payment models. We hypothesized that payer-initiated A&F had a stronger impact in hospitals were the payer had a larger market share, and that the effect would be stronger still for hospitals with financial incentives to reduce production.
Results
LVC showed a small, but significant decline over time (odds ratio [OR]: 0.98; 95% confidence level [CI]: 0.98-0.99). LVC tended to be higher in hospitals where the payer had a higher market share (OR: 1.03; 95% CI: 1.00-1.05). The CC model was associated with an increase in LVC (OR: 1.11; 95% CI: 0.99–1.03). The other payment models were associated with a decrease in LVC. The FFS model exhibited the smallest reduction in LVC (OR: 0.98; 95% CI: 0.98-0.99), and the GT model (OR: 1.01; 95% CI: 0.99-1.03) showed similar patterns in LVC. The GB model was associated with the largest reduction in LVC (OR: 0.94; 95% CI: 0.91-0.97).
Conclusion
By itself A&F did not clearly contribute to the decline of the proportion of LVC; combined with volumeconstraining financial incentives, it was associated with larger reductions.
Keywords
