作者: Vanessa W. Stevens , Karim Khader , Richard E. Nelson , Makoto Jones , Michael A. Rubin
DOI: 10.1017/ICE.2015.132
关键词:
摘要: BACKGROUND Standard estimates of the impact Clostridium difficile infections (CDI) on inpatient lengths stay (LOS) may overstate care costs attributable to CDI. In this study, we used multistate modeling (MSM) CDI timing reduce bias in excess LOS. METHODS A retrospective cohort study all hospitalizations at any 120 acute facilities within US Department Veterans Affairs (VA) between 2005 and 2012 was conducted. We estimated LOS using an MSM address time-dependent bias. Bootstrapping generate 95% confidence intervals (CI). These were compared unadjusted differences mean for with without RESULTS During period, there 3.96 million 43,540 CDIs. comparison means suggested 14.0 days (19.4 vs 5.4 days). contrast, only 2.27 (95% CI, 2.14–2.40). The mild-to-moderate 0.75 0.59–0.89), severe CDI, it 4.11 3.90–4.32). Substantial variation across Veteran Integrated Services Networks (VISN) observed. CONCLUSIONS significantly contributes LOS, but magnitude its is smaller when methods are that account time-varying nature infection. greatest occurred among patients Significant geographic variability a useful tool obtaining more accurate Infect. Control Hosp. Epidemiol. 2015;36(9):1024–1030