Risk factors for transmission of multidrug-resistant organisms and acquisition of healthcare-associated infections

Abstract

Background: Healthcare-associated infections (HAI) increase hospitalized patients' morbidity and mortality. In order to prevent the spread of HAI among patients in the healthcare setting, a better understanding of transmission dynamics and how patient-level factors influence acquisition is needed. Objectives: To estimate the association between patients' bacterial burden of vancomycin-resistant Enterococcus (VRE) and transmission to others and to build risk adjustment models taking into account patient case-mix for a more accurate comparison of rates HAIs between hospitals. Methods: Using a prospective cohort design, quantitative cultures (perianal, skin, and stool samples) were obtained from VRE-colonized ICU patients at the University of Maryland Medical Center. The association between patient bacterial burden and transmission to healthcare workers (HCW) gloves or gowns was estimated using generalized estimating equations. HCWs were observed during patient care to identify risk factors for transmission. Retrospective cohorts of surgical and ICU patients in 28 US hospitals were assembled to build risk adjustment models using hospital discharge codes for surgical site infections and central line-associated bloodstream infections using mixed models. Results: There were 71 transmission events among 479 HCW-patient interactions. VRE transmission was associated with VRE burden on the perianal swab (OR: 1.37 [95% CI 1.19, 1.57]); skin swabs (OR: 2.14 [95% CI: 1.51, 3.02)]; and in the stool (OR: 1.95 [95% CI: 1.39, 2.72]). Independent risk factors for transmission of VRE to HCWs in a multivariable model were touching the patients' skin (OR: 2.18 [95% CI: 1.15, 4.13]) and transferring the patient in/out of bed (OR: 2.66 [95% CI: 1.15, 6.43]). There were 573 (1.3%) surgical site infections among the 45,394 surgical patients and the risk adjustment model yielded a C-statistic was 0.73 (95% CI, 0.71-0.75). Of the 85,849 ICU patients, 162 (0.2%) developed a central line-associated bloodstream infection and the risk adjustment model yielded a C-statistic of 0.64 (95% CI, 0.60-0.69). Conclusions: Our first study shows that ICU patients with higher bacterial burden are more likely to transfer VRE to HCWs. Our second study demonstrates the importance of including comorbidities in risk adjustment models. These findings have implications for infection control interventions and HAI rate comparisons.