Comparisons of health care-associated infections identification using two mechanisms for public reporting

Background: Many states have or are in process of legislating hospitals to report health care-associated infections (HAI). The purpose of this article is to compare two methods currently in use by different states: 1) selected infections due to medical care Patient Safety Indicator (PSI-7) and 2) Centers for Disease and Prevention Control (CDC) protocols for central line-associated bloodstream infections (CLA-BSI). Methods: Data came from a multihospital study. Site coordinators provided lists of elderly Medicare patients admitted in an enrolled intensive care unit in 2002 cross referenced with patient specific data on CLA-BSI following CDC protocols. PSI-7 was identified using Medicare data and the Agency for Healthcare Research and Quality PSI software version 2.1. Results: The full sample comprised records from 14,637 patients from 41 intensive care units in 24 hospitals. Patients were excluded if they did not meet the PSI-7 denominator criteria. In a sample of 9,948 patients, both methods identified infections in 89 (0.89%) patients. The methods had little concordance with only 8 patients identified using both methods. Conclusions: Inconsistencies that we identified in this study are concerning given the fact that reports of HAI generated by different methods vary widely. Mandatory reporting mechanisms should be standardized and their accuracy confirmed.