70 Impact of Post-Discharge Events on Hospital Methicillin Resistant Staphylococcus aureus (MRSA) Rates

Friday, March 19, 2010: 10:30 AM
Centennial I-II (Hyatt Regency Atlanta)
Taliser R. Avery, BS , Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA
Susan S. Huang, MD, MPH , Division of Infectious Diseases and Health Policy Research Institute, University of California Irvine School of Medicine, Orange, CA

Background: Surveillance and reporting of hospital MRSA rates are increasingly mandated. However, such rates do not account for post-discharge MRSA detection even though they are considered healthcare-associated events by the CDC for 1 year following discharge.

Objective: We evaluated the impact of including post-discharge MRSA events in assessing hospital-specific and countywide MRSA rates.

Methods: We conducted a retrospective cohort study of patients ≥18yo admitted to all 28 hospitals serving adults in Orange County, CA from 2002-7 using a mandatory hospital discharge dataset including 25 diagnostic codes, with a present-on-admission (POA) indicators. MRSA carriers were identified by ICD-9 codes (041.11, 038.11, 482.41) with V09.0 resistance codes. We reported annual prevalence of admissions for MRSA carriers and assessed risk of incident MRSA admissions, i.e. first admission with MRSA carriage noted. Using each hospital's discharge data, we categorized incident cases as 1) hospital-onset MRSA (HO-MRSA, POA=N) or 2) community-onset MRSA (CO-MRSA, POA=Y).  Then, we recalculated HO-MRSA rates by assigning post-discharge incident CO-MRSA events to the most recent hospitalization within a year.  All HO-MRSA rates used at-risk denominators, which excluded admissions of ≤2days.  Incident HO-MRSA rates with and without post-discharge events were compared using paired t-tests.

Results: Countywide, there were 42,654 admissions among prevalent MRSA carriers over the 6y, increasing from 21/1,000 in 2002 to 41/1,000 in 2007 (p<.01). Total incident MRSA admissions were 13,760, increasing from 8/1,000 in 2002 to 11/1,000 in 2007 (p=NS). Of incident admissions, 9% were HO-MRSA and 91% CO-MRSA.

7,464 (60%) of incident CO-MRSA events occurred within 1 year of hospital discharge.  Median time from last hospitalization to post-discharge event was 5 wks (IQR = 2, 15).

Inclusion of post-discharge events in HO-MRSA rates, resulted in a 7-fold increase in overall HO-MRSA rates from 1.2 to 8.3 per 1,000 admissions (all yrs combined, p<.0001). Similarly, overall CO-MRSA rates decreased 60% from 9.3 to 3.7 per 1,000 (p<.001) (see figure for annual rates).  When hospitals were ranked into quartiles by HO-MRSA rates, inclusion of post-discharge events changed the quartile category for 75% (21/28) of hospitals.

Conclusions: Accounting for post-discharge events using datasets that allow patient tracking between hospitals can markedly change hospital and countywide rates of HO-MRSA and CO-MRSA.  Evidence that community-based MRSA events composed the bulk of incident MRSA carriage was reversed when community-onset post discharge events were reattributed to recent hospitalizations.  Post-discharge events increased HO-MRSA risks 7-fold and changed hospital rankings substantially.  The impact of including or excluding post-discharge events should be further assessed in this era of public reporting.