Evaluation of Patients’ Adverse Events Associated With Contact Isolation: Matched Cohort Study With Propensity Score

Abstract

Background: Contact isolation (ie, patient isolation with contact precautions) has been frequently used for preventing healthcare-associated infections caused by epidemiologically important pathogens (eg, vancomycin-resistant enterococcus [VRE]) via direct or indirect contact with patients. Based on ineffective components of routine contact isolations (eg, fewer healthcare provider visits), some studies have reported an association between the likelihood of adverse events and contact isolation. Objective: Given no strong evidence for this association due to most studies’ invalid study designs and systematic misclassification, we compared adverse events between a VRE isolation cohort and a matched comparison cohort, using a propensity score matching cohort study design. Methods: This study was conducted at a 1,337-bed, tertiary-care, university-affiliated, Korean hospital equipped with a full electronic medical record (EMR) system for all patient records. With institutional review board approval, all relevant EMR records were extracted for the study period 2015–2017. All contact isolation information of VRE patients were confirmed through EMR manual review by 1 trained research nurse. For propensity score matching, risk factors for adverse events (ie, decubitus ulcer, fall, and cardiopulmonary resuscitation [CPR]) were selected based on literature reviews: length of stay, age, gender, diabetes mellitus, hypertension, albumin, Charlson comorbidity index, Braden scale score, and Hendrich II fall risk. For each VRE case, the 1:1 matched case was selected through the nearest neighbor matching with calculated propensity scores. The retrospective observation period was from the cohort entry date (ie, contact isolation start date) to the cohort exit date (ie, discharge or discontinue of contact isolation). A time-to-event analysis with a Cox proportional hazard model was conducted using SAS version 9.4 software. Results: Among the 98,527 inpatients (323 VRE positive; 98,204 VRE negative), the VRE cohort (N = 141 of 216, 65% of total VRE patients admitted to general wards without adverse event history before contact isolation) and the matched comparison (no isolation) cohort (N = 141, 0.1%) showed no differences in characteristic comparisons (Table 1). The Cox proportional hazard model was not applicable for CPR because no CPR case was available in the matched comparison cohort. The hazard ratios for adverse events showed no statistically significant difference for both cohorts: decubitus ulcer (hazard ratio [HR], 1.049; 95% CI, 0.328–3.352; fall (HR, 0.418; 95% CI, 0.051–3.349) (Table 2). Conclusions: Based on the full EMR records for 3 years, our propensity-score–matched cohort study reported no association between the likelihoods of adverse events and contact isolation.

Funding: This work was supported by the Collaborative Research Program of Medical Science and Nursing Science from Seoul National University College of Medicine (Grant no. 800-20180001 & 810-20180001).

Disclosures: None.