BACKGROUND: In 2014, the government introduced elements of quality-based funding (pay-for-performance) for the hospital sector. Survival is included as a quality indicator. If such quality indicators are to be used for funding purposes, it must be established that the observed variations are caused by conditions that the hospital trusts are able to influence, and not by any underlying variables. The objective of this study was to investigate how the predicted mortality after myocardial infarction was influenced by various forms of risk adjustment. MATERIAL AND METHOD: Data from the Norwegian Patient Register on 10,717 patients who had been discharged with the diagnosis of myocardial infarction in 2009 were linked to data on socioeconomic status, comorbidity, travel distances and mortality. The predicted 30-day mortality after myocardial infarction was analysed at the hospital-trust level, using three different models for risk adjustment. RESULTS: Unadjusted 30-day mortality was highest in the catchment area of Førde Hospital Trust (12.5%) and lowest in Asker og Bærum (5.2%). Risk adjustment changed the estimates of mortality for many of the hospital trusts. In the model involving the most comprehensive risk adjustment, mortality was highest in the catchment area of Akershus University Hospital (10.9%) and lowest in the catchment areas of Sunnmøre Hospital Trust (5.2%) and Nordmøre og Romsdal Hospital Trust (5.2%). INTERPRETATION: The variation in treatment quality between the hospital trusts, as measured by predicted mortality after myocardial infarction, is influenced by the methods used for risk adjustment. If the quality-based funding scheme is to continue, well-documented models for risk adjustment of the quality indicators need to be established.
BACKGROUND: In 2014, the government introduced elements of quality-based funding (pay-for-performance) for the hospital sector. Survival is included as a quality indicator. If such quality indicators are to be used for funding purposes, it must be established that the observed variations are caused by conditions that the hospital trusts are able to influence, and not by any underlying variables. The objective of this study was to investigate how the predicted mortality after myocardial infarction was influenced by various forms of risk adjustment. MATERIAL AND METHOD: Data from the Norwegian Patient Register on 10,717 patients who had been discharged with the diagnosis of myocardial infarction in 2009 were linked to data on socioeconomic status, comorbidity, travel distances and mortality. The predicted 30-day mortality after myocardial infarction was analysed at the hospital-trust level, using three different models for risk adjustment. RESULTS: Unadjusted 30-day mortality was highest in the catchment area of Førde Hospital Trust (12.5%) and lowest in Asker og Bærum (5.2%). Risk adjustment changed the estimates of mortality for many of the hospital trusts. In the model involving the most comprehensive risk adjustment, mortality was highest in the catchment area of Akershus University Hospital (10.9%) and lowest in the catchment areas of Sunnmøre Hospital Trust (5.2%) and Nordmøre og Romsdal Hospital Trust (5.2%). INTERPRETATION: The variation in treatment quality between the hospital trusts, as measured by predicted mortality after myocardial infarction, is influenced by the methods used for risk adjustment. If the quality-based funding scheme is to continue, well-documented models for risk adjustment of the quality indicators need to be established.