Does Hospital Infection Reporting Affect Actual Infection Rates, Reported Rates, or Both? A Case Study of Pennsylvania

Abstract:     
We contribute to two literatures. The first involves reputation markets and whether public reporting of quality measures (here, hospital infection rates) leads to higher quality. The second involves performance measurement, and the impact of performance reporting rules on the reporting (here, reported infection rates, controlling for actual infection rates). We find evidence of both effects for hospital reporting of central-line associated bloodstream infection (CLABSI) in Pennsylvania (PA).

Health-care associated infections (HAIs) kill about 100,000 people annually; many are preventable. In response, 20 states now require hospitals to publicly report their infection rates and national reporting of CLABSI is planned. We provide initial evidence on how HAI reporting affects hospital behavior. CLABSI rates in PA and control states, estimated from inpatient data (“inpatient rates”), vary similarly during the pre-reporting period (1998-2003). During the 2004-2008 reporting period, Pennsylvania rates drop by 14%, versus a 9% increase in control states. This estimate is hopefully reliable because hospitals do not publicly report inpatient rates and have no direct incentive to manipulate them. The drop is strongest for hospitals in the top third of reported rates.

PA publicly reported CLABSI rates fall by 40% over 2005-2007; over twice the 17% drop in inpatient rates. This difference suggests time-inconsistent public reporting and confirms the importance of using non-reported measures (here, our inpatient rate) to measure the effect of performance reporting on quality.

Number of Pages in PDF File: 53

Keywords: CLABSI, health-care reporting, infection rates, performance measurement

JEL Classification: I18, K32

working papers series