Birds of a Feather? Hospital-level predictors of community-associated versus hospital-associated methicillin-resistant Staphylococcus aureus (CA-MRSA vs. HA-MRSA) colonization among intensive care unit (ICU) patients

Background: In surveys of ICUs in 26 hospitals, we noted clustering of CA- vs HA-MRSA patient colonization. Although patient-level predictors of CA-MRSA and HA-MRSA colonization among ICU patients have been studied, hospital characteristics are unknown.

Objective: To identify hospital characteristics associated with CA-MRSA and HA-MRSA colonization among ICU patients in a region.

Methods: We recruited Chicago hospitals with ≥10 ICU beds for 2 single-day point prevalence MRSA colonization surveys separated by six months, 7/08 - 7/09. All ICU patients were cultured (adult and pediatric: nose and groin; neonatal: nose and umbilicus). Cultures were processed centrally for methicillin resistance and pulsed-field gel electrophoresis genotype (CA-MRSA defined as USA300/400/1000/1100; HA-MRSA, all others).We used a 2007 public database to obtain hospital-level variables: hospital bed size; ICU bed size, yearly admissions, occupancy rate, and average daily census; hospital bed to infection preventionist (IP) ratio; Medicare and Medicaid payer proportion. We surveyed hospitals regarding laboratory methods (on- vs. off-site processing; polymerase chain reaction [PCR] vs culture-based MRSA screening). Continuous variables were dichotomized by the median; univariate analyses were performed with prevalence ratios (PR) of CA-MRSA and HA-MRSA as separate outcomes. Chi-square test was used to assess significance.

Results: All 26 eligible hospitals participated; 1716 patients were cultured. Hospital-level ICU variables denoting larger and busier wards (greater than median ICU size, yearly admissions, occupancy rate, and average daily census) were generally associated with decreased risk of CA-MRSA and HA-MRSA (Table). High Medicare payer mix was associated with a 2-fold increased prevalence of CA-MRSA (PR 2.21, 95% confidence interval [CI] 1.32 – 3.72, P <0.01) and HA-MRSA (PR 2.12, CI 1.42 - 3.19, P <0.01), while high Medicaid payer mix was associated only with CA-MRSA (PR 1.98, CI 1.19 – 3.30, P <0.01). Hospitals with off-site MRSA laboratory processing had a greater risk of CA-MRSA (PR 2.74, CI 1.29 - 5.82, P <0.01) and HA-MRSA (PR 3.32, CI 1.97 - 5.60, P <0.01).

Conclusions: Hospitals with smaller, less busy ICUs and socioeconomically disadvantaged patient populations (based on payer mix) had higher MRSA colonization prevalence, particularly CA-MRSA. Some of the associations may be markers for unmeasured differences among hospitals, such as variations in population catchment, readmission rates, or infection control resources. For public reporting of hospital MRSA rates, stratification along some of these hospital-level covariates may lead to more meaningful inter-hospital comparisons.