Study design and participants

We performed a prospective cohort study to determine which HCW types and patient care activities are risk factors for VRE transmission to HCWs’ gloves or gowns, a surrogate for transmission to other patients in the ICU. Between January 1, 2017, and November 15, 2017, 100 VRE-colonized patients from the medical and surgical ICUs were enrolled at the University of Maryland Medical Center. The medical ICU is a 29-bed unit providing medical care to adult patients with acute or life-threatening conditions, while the surgical ICU is a 24-bed unit which cares for adult surgical patients. These ICUs screen for VRE on admission, discharge, and once weekly as part of the VRE infection prevention active surveillance program. On each study day, e-mail alerts associated with the hospital microbiology laboratory notified research staff of recent (within 72 hours) VRE positive cultures. Patients with positive rectal surveillance cultures and 5 HCWs per patient were eligible for enrollment in the study. The Institutional Review Board of the University of Maryland, Baltimore, granted approval for the study and waived consent of participants.

Data collection

All HCW–patient activities were recorded by research staff on a standardized data collection form. All HCW activities were categorized into 2 domains: interactions with the patient (patient domain) or the patient’s environment (environmental domain). The patient domain included direct contact with the patient (eg, bathing/hygiene, wound dressing, or physical examination) or contact with the objects in direct contact with a patient (eg, bed rail, bedding, catheter or drain, artificial airway, vital signs, giving oral meds, IV tubing or IV meds, or rectal tube or bag). The environmental domain included contact with items in the ICU room environment (eg, the sink, bedside table, supply cart, lift, curtain, trash, computer medical equipment, and room furniture; see online appendix for the complete list). These patient care activities and interactions were chosen based on prior literature showing that these interactions were associated with increased transmission of several types of multidrug-resistant organisms.Reference Hayden, Blom, Lyle, Moore and Weinstein ^{9 ,} Reference Furuno, Perencevich and Johnson ^{11 –} Reference Pineles, Morgan and Lydecker ¹³

Following patient care, but prior to doffing, the gloves and gown of each HCW were sampled for the presence of VRE with BBL dual Culturettes (BBL, Becton Dickinson, Sparks, MD). With a twirling motion, the swab was rubbed along each finger and the palms of both gloved hands. All HCW gowns were sampled twice along both forearms and then in a “W” pattern along the beltline using a twirling motion.

Patient demographics including age, sex, and race were abstracted form the electronic medical record. The Elixhauser Index, a validated comorbidity score used for hospitalized patients,Reference Elixhauser, Steiner, Harris and Coffey ^{14 ,} Reference Quan, Sundararajan and Halfon ¹⁵ was calculated using diagnostic codes from the International Classification of Diseases, Tenth Revision. Study researchers conferred with the nursing staff to obtain clinical characteristics including the presence or absence of invasive devices, including an artificial airway (endotracheal or tracheostomy tube), Foley catheter, central venous catheter (central line), chest tube, surgical drain, rectal tube, nasogastric tube, and whether the patient had diarrhea or wounds. To quantify VRE burden, we cultured each patient’s perianal area by gently rubbing ESwabs (Copan Diagnostics, Murrieta, CA) on the skin immediately around the anus.

Laboratory procedures

All HCW gown and glove swabs were placed into tryptic soy broth with 6.5% NaCl and were incubated for 24 hours at 35±2°C. After incubation, 50 µL from each broth tube was inoculated onto a Bile Esculin Azide Agar with 6 µg/mL vancomycin plate (Remel, Lenexa, KS) and was incubated aerobically at 35±2°C for 48 hours. The isolates were frozen in tryptic soy broth with 15% glycerol and stored at −80°C. A 1-mL aliquot of each patient perianal sample was serially diluted using Butterfield buffer. The agar plate was inoculated with 100 µL of each serial dilution and distributed evenly onto each agar plate using a cell spreader. Next, 100 µL of the original sample was inoculated into tryptic soy broth with 6.5% NaCl. The plates and broth tubes were incubated for 48 hours aerobically at 35±2°C, after which the number of bacterial colonies was counted. If there was no growth on the inoculated plates, 100 µL from the previously inoculated tryptic soy broth with 6.5% NaCl tubes was inoculated onto a plate. If there was growth on the agar after 48 hours at 35±2°C, a count of 1 colony-forming unit (CFU) was given. The colonies were subcultured and identified to the species level using VITEK II (BioMérieux, Marcy-l’Étoile, France). Antimicrobial susceptibility testing was conducted according to the Clinical Laboratory Standards Institute guidelines using the Kirby-Bauer disk diffusion method.

Statistical analysis

Frequencies, proportions, means, and standard deviations were calculated to describe patient demographics and clinical characteristics. Patient bacterial burden (x + 1) was log transformed and expressed in log₁₀ CFU per milliliter (CFU/mL). We estimated the following associations with HCW glove or gown contamination: (1) HCW type (ie, physician or nurse practitioner, nurse, patient care technician, physical or occupational therapist, respiratory therapist, environmental service worker, or other) and (2) specific patient-care activities or interactions. Risk factors significant at α≤0.10 in the previous analyses were considered candidate predictors for the multivariable model. All models were built using logistic regression models fit by generalized estimating equations with an exchangeable correlation matrix to consider within-patient correlation and were conducted in a stepwise fashion, where the model with the lowest QIC was chosen as the final multivariable model. Potential confounders were selected a priori for all models and included patient age, race, Elixhauser Index, invasive devices, diarrhea, bacterial burden, and duration of time the HCW spent in the room.

All analyses were conducted using SAS version 9.4 software (SAS Institute, Cary, NC).