Excess Length of Stay due to Methicillin-Resistant Staphylococcus aureus and Vancomycin-Resistant Enterococci Infections at ICUs across the United States
Background: Understanding the impact of hospital-acquired infections on excess length of stay (LOS) can help to identify variation in practice and opportunities to improve care. However, time-dependent bias, which occurs when the timing of infections is not appropriately taken into account, can inflate the estimates of excess LOS due to HAIs. We used a multistate model to examine the excess ICU LOS after methicillin-resistant Staphylococcus aureus (MRSA) and vancoymycin-resistant enterococci (VRE) positive clinical cultures.
Methods: We conducted a retrospective study using clinical culture results from 13,278 patients admitted to 18 ICUs in the United States from April 2005 – August 2006. Our multistate model (Figure 1) incorporated an intermediate event, and two endpoints: death and discharge. Because death and discharge are competing events, we modeled them separately to improve our interpretation. We obtained median LOS estimates, interquartile range [IQR] and estimates of excess LOS using the multistate model with 95% Confidence Intervals (CIs) using 1000 bootstrap samples. Additionally, we examined variation in and associations of excess ICU LOS estimates across the ICUs due to positive MRSA and VRE cultures.
Results: Of patients admitted to ICU, 13.2% died before discharge or ICU transfer. Median ICU LOS [IQR] for those without a positive culture and those with positive MRSA and VRE cultures were 4 days [2, 7], 9 days [4, 20] and 9 days [4, 21], respectively. Multistate model estimates of excess ICU LOS [95% CI] were 3.7 days [2.3, 5.1] and 5.3 days [3.6, 7.5] for MRSA and VRE positive clinical cultures, respectively. Estimates of excess ICU LOS were positively correlated (Figure 2), ranging from -2.0 and 32.0 days and -1.7 and 42.0 days for MRSA and VRE positive clinical cultures, respectively.
Conclusion: Overall, excess ICU LOS was longer after positive VRE cultures compared to positive MRSA cultures. Estimated excess ICU LOS were similar to the differences in the median ICU LOS for those with MRSA and VRE positive cultures. There was a positive correlation in excess ICU LOS related to MRSA and VRE infections, suggesting possible ICU-level factors contributing to excess LOS. Future analyses that adjust for confounders are warranted.
R. E. Nelson, Roche: Consultant and Grant Investigator, Consulting fee, Research grant and Research support
M. Jones, None
V. Stevens, None
W. C. Huskins, None
M. Samore, None