Effectiveness of Ultraviolet-C Disinfection on Hospital-Onset Gram-Negative Rod Bloodstream Infection: A Nationwide Stepped-Wedge Time-Series Analysis
Michihiko Goto, Shinya Hasegawa, Erin Balkenende, G Clore, Nasia Safdar, Eli N. Perencevich, Suzanne Bradley, Daniel J. Morgan, Kalpana Gupta, Christopher J. Hostler, Charlesnika T. Evans, Matthew Bidwell Goetz, Heather Schacht Reisinger, Nasia Safdar, Gio J Baracco Lira, Aaron DeVries, Bryan Harris, Marvin J. Bittner, Christopher D. Pfeiffer, Michael Rubin, Jose Cadena-Zuluaga, Katie J. Suda
Abstract
BACKGROUND: The effectiveness of enhanced terminal room cleaning with ultraviolet C (UV-C) disinfection in reducing gram-negative rod (GNR) infections has not been well evaluated. We assessed the association of implementation of UV-C disinfection systems with incidence rates of hospital-onset (HO) GNR bloodstream infection (BSI). METHODS: We obtained information regarding UV-C use and the timing of implementation through a survey of all Veterans Health Administration (VHA) hospitals providing inpatient acute care. Episodes of HO-GNR BSI were identified between January 2010 and December 2018. Bed days of care (BDOC) was used as the denominator. Over-dispersed Poisson regression models were fitted with hospital-specific random intercept, UV-C disinfection use for each month, baseline trend, and seasonality as explanatory variables. Hospitals without UV-C use were also included to the analysis as a nonequivalent concurrent control group. RESULTS: Among 128 VHA hospitals, 120 provided complete survey responses with 40 reporting implementations of UV-C systems. We identified 13 383 episodes of HO-GNR BSI and 24 141 378 BDOC. UV-C use was associated with a lower incidence rate of HO-GNR BSI (incidence rate ratio: 0.813; 95% confidence interval: .656-.969; P = .009). There was wide variability in the effect size of UV-C disinfection use among hospitals. CONCLUSIONS: In this large quasi-experimental analysis within the VHA System, enhanced terminal room cleaning with UV-C disinfection was associated with an approximately 19% lower incidence of HO-GNR BSI, with wide variability in effectiveness among hospitals. Further studies are needed to identify the optimal implementation strategy to maximize the effectiveness of UV-C disinfection technology.