Public reporting of HAI rates: What we (mostly don't) know

Purpose: Health-care associated infections (HAIs) kill about 100,000 people annually; many are preventable. In response, 18 states currently require hospitals to publicly report their infection rates and national reporting is planned. Yet there is limited evidence on the effects of public reporting on HAI rates, and none on what elements of a reporting plan affect its impact on HAI rates. The author aims to review here what little we know, emphasizing his own case study of Pennsylvania. Design/methodology/approach: The paper contains a narrative description of empirical challenges in attributing changes in infection rates to the introduction of public reporting, and the author's own research findings from a case study of Pennsylvania using both infection rates estimated from administrative (billing) data ("inpatient rates") and public reported rates. Findings: Hospitals, faced with public HAI reporting, may respond both by reducing infection rates and through time-inconsistent reporting ("gaming"). Both effects are likely to be stronger at hospitals with high reported rates, relative to peers. From 2003-2008, Pennsylvania inpatient CLABSI rates dropped by 14 per cent, versus a 9 per cent increase in control states. The overall drop comes primarily from hospitals in the highest third of reported rates. Reported CLABSI rates fell much faster, by 40 per cent, from 2005 to 2007. This difference suggests time-inconsistent reporting. Practical implications: Much more research is needed before we can have confidence that public reporting affects HAI rates (and for which HAIs), or know how to design an effective reporting scheme. HAI reporting cannot yet be considered to be "evidence based." National reporting mandates will foreclose the state experiments needed to address these questions. Originality/value: What little we know about impact of public reporting on HAI rates comes in significant part from the case study of Pennsylvania described in this article.