Modelling interventions and contact networks to reduce the spread of carbapenem-resistant organisms between individuals in the ICU

Background: Contact precautions are widely used to prevent the transmission of carbapenem-resistant organisms (CROs) in hospital wards. However, evidence for their effectiveness in natural hospital environments is limited.

Objective: To determine which contact precautions, healthcare worker (HCW)-patient interactions, and patient and ward characteristics are associated with greater risk of CRO infection or colonization.

Design, setting and participants: CRO clinical and surveillance cultures from two high-acuity wards were assessed through probabilistic modelling to characterize a susceptible patient's risk of CRO infection or colonization during a ward stay. User- and time-stamped electronic health records were used to build HCW-mediated contact networks between patients. Probabilistic models were adjusted for patient (e.g. antibiotic administration) and ward (e.g. hand hygiene compliance, environmental cleaning) characteristics. The effects of risk factors were assessed by adjusted odds ratio (aOR) and 95% Bayesian credible intervals (CrI).

Exposures: The degree of interaction with CRO-positive patients, stratified by whether CRO-positive patients were on contact precautions.

Main outcomes and measures: The prevalence of CROs and number of new carriers (i.e. incident CRO aquisition).

Results: Among 2193 ward visits, 126 (5.8%) patients became colonized or infected with CROs. Susceptible patients had 4.8 daily interactions with CRO-positive individuals on contact precautions (vs 1.9 interactions with those not on contact precautions). The use of contact precautions for CRO-positive patients was associated with a reduced rate (7.4 vs 93.5 per 1000 patient-days at risk) and odds (aOR 0.03, 95% CrI 0.01-0.17) of CRO acquisition among susceptible patients, resulting in an estimated absolute risk reduction of 9.0% (95% CrI 7.6-9.2%). Also, carbapenem administration to susceptible patients was associated with increased odds of CRO acquisition (aOR 2.38, 95% CrI 1.70-3.29).

Conclusions and relevance: In this population-based cohort study, the use of contact precautions for patients colonized or infected with CROs was associated with lower risk of CRO acquisition among susceptible patients, even after adjusting for antibiotic exposure. Further studies that include organism genotyping are needed to confirm these findings.