Impact of pay-for-performance for stroke unit access on mortality in Queensland, Australia: an interrupted time series analysis

BACKGROUND: Stroke unit care provides substantial benefits for all subgroups of patient with stroke, but consistent access has been difficult to achieve in many healthcare systems. Pay-for-performance incentives have been introduced widely in attempt to improve quality and efficiency in healthcare, but there is limited evidence of positive impact when they are targeted at hospitals. In 2012, a pay-for-performance program targeting stroke unit access was co-designed and implemented within a clinical quality improvement network across public hospitals in Queensland, Australia. We assessed the impact on access to specialist care and mortality following stroke. METHODS: We used interrupted time series analysis on linked hospital and death registry data to compare changes in level (absolute proportions) and trends in outcomes (stroke/coronary care unit admission, 6-month mortality) for stroke, and a control condition of myocardial infarction (MI) without pay-for-performance incentive, from 2009 before, to 2017 after introduction of the pay-for-performance scheme in 2012. FINDINGS: We included 23,572 patients with stroke and 39,511 with MI. Following pay-for-performance introduction, stroke unit access increased by an absolute 35% (95% CI 29, 41) more than historical trend prediction, with greater impact for regional/rural residents (41% vs major city 24%) where baseline access was lowest (18% vs major city residents 53%). Historical upward 6-month mortality trends following stroke (+0.11%/month) reversed to a downward slope (−0.05%/month) with pay-for-performance; difference −0.16%/month (95% CI −0.29, −0.03). In contrast, access to coronary care and mortality trends for MI controls were unchanged, difference-in-difference for mortality −0.18%, (95% CI −0.34, −0.02). INTERPRETATION: This clinician led pay-for-performance incentive stimulated significant improvements in stroke unit access, reduced regional disparities; and resulted in a sustained decline in 6-month mortality. As our findings contrast with lack of effect in most hospital directed pay-for-performance programs, differences in design and context provide insights for optimal program design. FUNDING: Queensland Advancing Clinical Research Fellowship, 10.13039/501100000925National Health and Medical Research Council Senior Research Fellowship.