Using the SHEA/HICPAC Metrics to Measure MRSA in Hospitals: Factors and Practical Implications

Background:  The Society for Healthcare Epidemiology of America (SHEA) and CDC’s Healthcare Infection Control Practices Advisory Committee (HICPAC) published recommendations on how hospitals should measure multidrug-resistant organisms.

Objective: To test the SHEA/HICPAC metrics in a real-world setting.

Methods:  Using electronically obtained data, hospital-onset incidence of methicillin-resistant Staphylococcus aureus (MRSA) was determined using the SHEA/HICPAC definitions for 3 hospitals (>1100 beds) within a healthcare network. Rates were calculated for 1 year using patient days (PD) as the denominator, including clinical cultures only, and using 3 calendar days to define hospital onset. Additionally, we tested the impact of: 1) including surveillance cultures; 2) using time periods other than 3 calendar days to determine hospital-onset incidence; 3) excluding non “at risk” patients from the denominator; and 4) using different time periods of historical microbiologic data when excluding patients with a prior history of MRSA. Negative binomial regression models were used to test for differences between different rate definitions.

Results:  For all 3 hospitals combined, HO-MRSA was 0.49/1000 PD. When surveillance cultures were included, the rate was 0.56/1000 PD (p=ns); notably however, MRSA active surveillance is only performed upon admission in 2 adult ICUs and transfers into 2 neonatal ICUs in the hospitals. Using 72 hours to define hospital-onset was no different than using 3 calendar days (0.44 vs 0.49, p=ns); however, using 48 hours or 2 calendar days led to significantly higher HO-MRSA rates (0.53 and 0.61 respectively; p<0.05 compared to 0.49). Excluding non “at risk” patients from the denominator when calculating hospital-onset incidence led to a denominator of 191,626 patient days as compared to 381,685, and using this denominator led to a nearly doubling of the HO-MRSA rate (0.98 vs 0.49, p<0.05). When excluding patients with a prior history of MRSA, rates were similar whether we looked at 1, 2, or 3 years worth of historical microbiologic data (0.5, 0.5, and 0.49 respectively; p=ns).

Conclusions: The SHEA/HICPAC metrics can lead to significantly different estimates of HO-MRSA depending on how data are utilized. Including surveillance cultures in the estimate may lead to a significantly different number if active surveillance is performed widely throughout the institution. Three calendar days is a reasonable approximation of 72 hours to define the time period of hospital onset and may be easier for some hospitals to utilize. Excluding patients with a prior history of MRSA can be cumbersome based on available data formats, and we found that looking back for only 1 year of historical data was sufficient. However, our experience within a large academic healthcare network may not be generalizable to differing hospital settings and patient populations.