1747. Confounding Contact: Modeling the Effect of Post-Randomization Patient Contact Rate Changes on Infection Control Interventions
Session: Oral Abstract Session: MRSA Prevention and Epidemiology
Saturday, October 29, 2016: 9:00 AM
Room: 288-290
Background: The randomized controlled trial is considered to be the gold standard of intervention research, with randomization ensuring there are no differences between study arms. Recently, multi-center trials of institution-level policies have come to the forefront of hospital epidemiology research. These trials can induce changes postrandomization that threaten the validity of their results.

Methods: We examine the results of a trial(1) on the effect of universal gowning and gloving on MRSA rates which reported healthcare workers (HCWs) visited patients 0.96 times/hour less in the treatment arm post-randomization. We use a mathematical model of a 12-bed ICU, calibrated to the study to simulate this change. This provides an estimate of how much of the study’s effect may be attributable to changes in contact. Two scenarios were considered – that the decline in contact results in less patient care activities, or that the reduced visits are the result of HCWs fitting more activities into a single visit. The differences between scenarios were analyzed using an ANOVA.

Results: The fitted model produced an average incidence of 28.68 cases/simulated year. Compared to the baseline, no-intervention scenario, the decrease in visits resulting in less contact resulted in a mean of 27.45 cases/year, a 4.26% decrease. If instead the decrease in the overall number of visits was the results of HCWs performing more tasks during a given visit, there was a mean of 29.6 cases/year, a 3.22% increase. All differences were statistically significant (p < 0.001).

Conclusion: The model predicts that the change in patient visitation post-randomization produced a meaningful change in MRSA incidence, biasing the study’s effect estimate. While statistically significant, these results are not enough to change the qualitative findings of the original study, though adjustment is warranted for secondary studies using the results. This study is an illustration of the utility of using mathematical models to extend and explore the results of randomized trials in hospital epidemiology.

1. Harris, A. D., et al. (2013). Universal glove and gown use and acquisition of antibiotic-resistant bacteria in the ICU: a randomized trial. Jama, 310(15), 1571–1580.

Eric Lofgren, MSPH, PhD, School for Global Animal Health, Washington State University, Pullman, WA

Disclosures:

E. Lofgren, None

Findings in the abstracts are embargoed until 12:01 a.m. CDT, Wednesday Oct. 26th with the exception of research findings presented at the IDWeek press conferences.