Healthcare-associated infections (HAIs) cause morbidity, mortality, and excess costs.Reference Scott1, Reference Klevens, Edwards and Richards2 In 2011, there were an estimated 721,800 HAIs in US acute-care hospitals, and Clostridium difficile, Staphylococcus aureus, Enterococcus spp, and gram-negative bacilli are the most common HAI pathogens.Reference Magill, Edwards and Bamberg3 HAIs due to multidrug-resistant organisms (MDROs), such as methicillin-resistant S. aureus [MRSA], vancomycin-resistant Enterococcus (VRE), carbapenem-resistant Enterobacteriaceae (CRE), and other multidrug-resistant gram-negative bacteria, are of greatest concern.Reference Hidron, Edwards and Patel4 Intensive care units (ICUs) are epicenters of HAIs with MDROs as well as care for patients with multiple comorbidities and indwelling devices that can potentiate the risk for healthcare worker contamination and cross transmission of potential pathogens.Reference Brusselaers, Vogelaers and Blot5–Reference Bonten, Slaughter and Ambergen8
One putative path of hospital transmission of MDROs is by healthcare workers (HCWs) who contaminate their hands and/or clothes during patient contact.Reference Weinstein9 Previous studies have reported that gloves, gowns, and hands are frequently contaminated with MDROs after contact with patients and hospital environments during patient care.Reference Morgan, Rogawski and Thom10–Reference Ray, Hoyen, Taub, Eckstein and Donskey13 Although much is known about the importance of patient and environmental contamination as well as hand hygiene, the role of errors in donning and doffing of personal protective equipment (PPE) is incompletely understood. Studies under a simulated environment using bacteriophages and fluorescent markers have shown that errors in doffing often lead to contamination of unprotected sites on HCWs.Reference Tomas, Kundrapu and Thota14–Reference Kwon, Burnham and Reske17 On the other hand, failures in appropriate PPE use are common and types of failures and vary widely.Reference Krein, Mayer and Harrod18 The role of errors in donning and doffing PPE in MDRO transmission in real-world settings is unknown. We sought to assess the impact of PPE doffing errors on contamination of HCWs who were caring for patients on contact precautions for MDROs in adult ICUs.
Methods
Setting and enrollment
This study was conducted from September 2015 to February 2016 in 4 adult ICUs (104 total beds) at Rush University Medical Center, a 720-bed tertiary-care teaching hospital in Chicago, Illinois. The adult ICUs included medical, surgical, cardiac, and neurological ICUs. Patients in all ICUs received daily chlorohexidine gluconate baths as routine care. The Institutional Review Board of Rush University Medical Center reviewed and approved the study.
HCWs caring for patients on contact precautions (ie, HCWs were required to wear gloves and a gown during a patient interaction) due to colonization or infection by an MDRO were approached for enrollment. MDROs for which patients had been placed on contact precautions included MRSA, VRE, extended-spectrum β-lactamase (ESBL)–producing Klebsiella pneumoniae or Escherichia coli, and multidrug-resistant Pseudomonas aeruginosa (defined as resistance to 2 or more of antipseudomonal penicillins, cephalosporins, carbapenems or fluoroquinolones or as resistance to aminoglycosides). Microbiology and infection control data were used to identify eligible patients. HCWs eligible for enrollment included nurses, physicians, respiratory therapists, physical therapists, and patient care technicians. Because the focus of this study was HCW behavior, and to maximize power, each HCW was included in the study only once. Patients were eligible for enrollment twice.
Observations
A glove and gown donning and doffing checklist was developed based on the guidance document published by the Centers for Disease Control and Prevention (Supplemental Table 2 online19) and was used for HCW donning and doffing observations by 1 of 4 trained observers (K.O., Y.R., J.C., or K.J.P.). We monitored for proper order of doffing and whether components on the checklist were completed. Errors were coded differentially based on doffing method: (1) gowns and gloves removed together, (2) gowns and gloves removed separately with gloves first, and (3) gowns and gloves removed separately with gown first (not a CDC-recommended approach). HCWs were also observed while caring for patients. Patient care activities (eg, examining a patient, typing on the computer), HCW contact with sites (patient sites, environmental surfaces, and unprotected area of HCWs themselves), and duration of patient care were recorded. Prior to the start of the study, multiple pilot observation sessions were conducted by observers to ensure interobserver agreement of coding. In addition, simulated sessions were conducted every few weeks during the entire study period to ensure consistency in coding.
HCW and patient factors collected
We collected the following HCW data elements: job category; age; gender; body mass index (BMI); presence of jewelry, nail polish, lab coat, and hand abrasions/dermatitis; nail, sleeve, and hair length; and the number of other patients seen during the shift before study patient interactions. The following data elements for patients were collected: BMI; primary diagnosis; ambulatory status; hemodialysis; presence of endotracheal or tracheostomy tube, acute or chronic wound, urinary catheter, venous catheters, rectal tube, wound VAC, nasogastric tube, gastrostomy tube and device locations; and diarrhea or fecal incontinence.
Outcomes
We defined the primary outcome as MDRO acquisition by a HCW (ie, hands, clothes, PPE, or equipment) following a patient encounter if the positive site had no contamination when sampled before the encounter. Secondary outcomes were the proportion of HCWs with contamination following a patient encounter, regardless of contamination before the patient encounter. Separate binary variables (ie, for acquisition and contamination) for the following categories of sites were created: (1) hands (ie, right hand, left hand), (2) clothes (ie, ring, bracelet, watch, sleeves or forearm below elbow, ties, and clothes over abdomen), and equipment (ie, stethoscope, pager, personal cell phone, mobile phone for hospital use, and tablet device).
Microbiological methods
Swab samples for culture were obtained from HCWs before they donned their PPE (Supplemental Table 1 online). Gloves and gowns were cultured after donning.Reference Morgan, Rogawski and Thom10 After each patient–HCW interaction, gloves and gowns were cultured again before doffing. After doffing, HCW sites (Supplemental Table 1 online) were cultured again.
Within 2 hours before HCWs entered a patient room, cultures were obtained from multiple patient body sites, including anterior nares and throat (for MRSA), tracheostomy site, axillae, antecubital fossae, hands, inguinal area, perianal area, and ankles and/or area covered by sequential compression devices (Supplemental Table 1 online). Patients were swabbed at all body sites listed once per day. Similarly, cultures were obtained from 10×10-cm2 areas of high-touchReference Duckro, Blom, Lyle, Weinstein and Hayden12 environmental surfaces in the patient room within 2 hours before HCW–patient interactions (Supplemental Table 1).
Flocked nylon swabs (ESwab, Becton-Dickenson, Franklin Lakes, NJ) premoistened with Amies medium were used to obtain samples. The details of swabbing methods have been described previously.Reference Okamoto, Rhee and Schoeny20 To maximize the sensitivity of detecting the target organism, 8 environmental surfaces that were flat (ie, blood pressure cuff, call button, computer table/ keyboard/mouse, glove box, infusion pump, sink, and telephone) were also cultured using RODAC plates (25.6 cm2; D/E Neutralizing Agar or Trypticase Soy Agar with Lecithin and Polysorbate 80, Becton Dickinson, according to availability). An environmental site was considered culture positive if either a flocked nylon swab or a RODAC plate culture showed growth of the target organism associated with the specific patient interaction.
Broth enrichment was evaluated for each MDRO before the study began and was determined to improve sensitivity of detection of MRSA and VRE but not of gram-negative bacilli. For MRSA, swabs were inoculated into 5 mL tryptic soy broth + 6.5% NaCl, incubated 48 hours, then plated to MRSA Spectra agar (Remel, Lenexa, KS) and incubated 24 hours.Reference McAllister, Albrecht and Fosheim21 For VRE, swabs were inoculated into 5 mL bile esculin azide (BEA) broth (Remel), incubated 48 hours, then plated to BEA agar with 6 µg/mL vancomycin (Remel) and incubated 24 hours.Reference Ieven, Vercauteren, Descheemaeker, van Laer and Goossens22 For gram-negative bacilli, swabs were plated directly to MacConkey agar (Remel) and incubated 48 hours. Inoculated RODAC plates were incubated 24 hours according to the manufacturer’s instructions. Subsequently, all colonies growing on RODAC plates were inoculated into 3 mL sterile water. Then, 50 µL inoculum was plated to either MRSA Spectra agar (Remel) for MRSA, BEA agar with 6 µg/mL vancomycin (Remel) for VRE, or MacConkey agar for gram-negative bacilli and incubated for 24 hours at 35°C in ambient air. Representative colonies were identified to species level and tested for antimicrobial susceptibility by the MicroScan Walkaway System (Siemens, Tarrytown, NY). Detection was determined qualitatively as growth or no growth.
Statistical analysis
Continuous variables were compared using the Student t test or the Mann-Whitney U test, and categorical variables were compared using the χ2 test or the Fisher exact test, as appropriate. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Statistical significance was defined at p < .05 (2-tailed). All statistical analyses were performed using SPSS version 23 software (IBM Corp, Armonk, NY) and SAS version 9.4 software (SAS Institute, Cary, NC).
Using methods specified by Fleiss, κ coefficients for 4 trained observers, simultaneously coding in 7 simulated sessions, were calculated for each item on the donning and doffing checklists.Reference Fleiss23 The average coefficient was κ = .84 and all coefficients were greater than κ = .67. These values represent good to very good agreement according to guidelines suggested by Altman.Reference Altman24
Results
Basic characteristics
In total 125 HCWs were approached; all participated in the study (Table 1). Of these, 83 (66.4%) were nurses; 24 (19.2%) were physicians: 5 attending physicians, 6 fellow physicians, and 13 resident physicians. The HCWs spent a median of 7 minutes in patient rooms (interquartile range [IQR], 3–10). Furthermore, 64 HCWs (51.2%) had received formal training for PPE for Ebola virus in the previous year. Also, 113 HCWs (90.4%) had received formal training (excluding Ebola virus training) for PPE donning and doffing within the previous 5 years. The reasons for contact precautions were MRSA (n = 95, 76%), VRE (n = 16, 12.8%), ESBL-producing E. coli or K. pneumoniae (n = 7, 5.6%), and MDR P. aeruginosa (n = 7, 5.6%). The study included 125 encounters with 79 participating patients: 45 (36.0%) in the medical ICU, 34 (27.2%) in the surgical ICU, 24 (19.2%) in the neurological ICU, and 22 (17.6%) in the cardiac ICU. The timing of the last patient room cleaning was available for 118 encounters. Cleaning was done a median of 10.5 hours (range, 0–23) prior to study encounters.
Table 1. Characteristics of 125 Healthcare Workers who Participated in the Study
Contamination of HCWs, patients, and environmental surfaces
In total, 6,089 samples from 5,093 sites were obtained: 1,296 (25.4%) from HCWs before a patient encounter, 1,232 (24.2%) from HCWs after a patient encounter, 1,783 (35.0%) from environmental surfaces, and 782 (15.4%) from patient sites. The rates of MDRO contamination were 1.2% (n = 16) for HCW sites before a patient encounter, 6.7% (n = 82) for HCW sites following a patient encounter, 26.3% (n = 206) for patient sites, and 19.5% (n = 348) for environmental surfaces.
Of 125 HCWs, 10 (8%) had at least 1 site contaminated with the target organism before a patient encounter (Table 2). Following an encounter, 43 HCWs (34.4%) acquired the target organism (Table 2), with 38 (30.4%) acquiring the target MDRO on gloves and/or gowns: 26 HCWs on gloves alone, 1 HCW on the gown alone, and 11 HCWs on both gloves and gowns. Notably, 4 HCWs (3.2%) had acquisition on hands after patient encounters and before hand hygiene was performed. In addition, 4 HCWs (3.2%) acquired the target MDRO on their clothes (3 on sleeves and/or forearms and 1 on a ring), and 5 HCWs (4.0%) acquired the target MDRO on their equipment (3 stethoscope and 2 mobile phones dedicated to in-hospital use).
Table 2. Healthcare Worker (n = 125) Contamination and Acquisition* with the Target Multidrug Resistant Organism Before and After a Patient Encounter
a Items not brought into the room (e.g., stethoscope) as well as a sign-out paper were not cultured after a patient-HCW encounter.
b One of two HCWs contaminated before patient interaction did not have contamination after patient interaction.
* Note. MDRO acquisition by a HCW was defined as contamination (i.e., hands, clothes, PPE, or equipment) following a patient encounter if the positive site had no contamination when sampled before the encounter. Contamination indicates any growth of the target MDRO following a patient encounter, regardless of contamination before the patient encounter.
Patient samples were obtained before 111 encounters (Table 3). Overall, 89 encounters (79.5%) yielded at least 1 positive culture from a patient body site. The positivity rates varied by site, ranging from 9% at inguinal areas to 80% at tracheotomy sites.
Table 3. Patient and Environmental Sites Colonized or Contaminated with the Target Multidrug Resistant Organism Before 125 Patient Encounters
Note. Patients were not cultured before 14 patient encounters because patients were cultured only once daily.
Environmental cultures were obtained before all 125 patient encounters (Table 3). Environmental sites were positive for the target MDRO in 86 encounters (68.8%). Overall, more contamination was found at sites close to patients (blood pressure cuff, 37.6%; call button, 37.6%; bed line, 34.4%; bed rail, 33.9%) than at sites distant from patients (door handle, 8.1%; toilet handle, 4.0%; glove box, 4.0%).
MDRO contamination of HCW PPE was more common in settings of higher patient and environmental contamination, especially when patients had diarrhea and when areas close to patients were contaminated (Table 4). Similarly, HCW MDRO acquisition on other sites (ie, ring, sleeves, clothes over abdomen, stethoscope, and mobile phones) was associated with heavier patient colonization and environmental contamination.
Table 4. Healthcare Worker Contamination or Acquisition by Multidrug-Resistant Organisms and Associated Factors*
HCW; healthcare worker. PPE; personal protective equipment. SD, standard deviation. CI, confidence interval. N/A, not applicable.
* Note. MDRO acquisition by a HCW was defined as contamination of hands, clothes, PPE, or equipment following a patient encounter if the positive site was not contaminated with the target MDRO when sampled before the encounter. Contamination indicates any growth following a patient encounter, regardless of contamination with the target MDRO before the patient encounter.
a Urine bag, stool bag, bed linen, and blood pressure cuff.
b Could not estimate because no HCW acquisition occurred without environmental site contamination near the patient.
Donning and doffing of gloves and gown
The donning and doffing methods of 123 HCWs (98.4%) were recorded. Of those, 69 HCWs (56.1%) doffed gloves and gown together, whereas 29 HCWs (23.6%) doffed gown first and 25 (20.3%) doffed gloves first.
Overall, 105 (84.8%), 16 (12.9%), and 47 (37.6%) HCWs made multiple errors in donning gown, donning gloves, and doffing respectively. Examples of errors included touching the inside the gown or glove with a gloved hand, touching the outside of the gown or glove with bare hands, and not unfastening the gown at the neck. The rates of making multiple errors varied depending on the doffing method: doffing gown and gloves together (12 of 69, 17.4%), gown first (15 of 29, 51.7%), gloves first (18 of 25, 72.0%) (P < .001).
HCWs who made multiple doffing errors were more likely to have contaminated clothes following a patient encounter (OR, 13.23; 95% CI, 1.31–133.26; P = .026) (Table 4 and Supplementary Table 3). In addition, there was a higher risk of hand contamination when gloves were removed before gowns during PPE doffing (OR, 5.56; 95% CI, 1.07–28.80); P = .052).
Discussion
In this prospective observational study of 125 HCWs in adult ICUs, 34.4% of HCWs acquired the MDRO that was colonizing or infecting a patient during a patient encounter. More than one-third of HCWs made multiple errors in doffing PPE, which may have resulted in contamination of HCW clothes. Furthermore, 3.2% of HCWs hands were contaminated after a patient encounter, often in the context of an HCW removing gloves before a gown during PPE doffing.
Previous studies reported that 15%–62% of HCWs had contamination with clinically important bacterial pathogens (MRSA, VRE, MDR Pseudomonas, or Acinetobacter baumannii) following care of an ICU patient who was colonized or infected with one of these pathogens.Reference Morgan, Rogawski and Thom10, Reference Hayden, Blom, Lyle, Moore and Weinstein11, Reference Thom, Rock and Jackson25–Reference Jackson, Harris and Magder29 The rate of HCW contamination in our study is consistent with these results. The relation of HCW contamination to PPE doffing errors reinforces the importance of adherence to contact precautions and of appropriate PPE donning and doffing practices for preventing the spread of potential pathogens in ICU settings. Others have demonstrated violations in doffing protocols and contamination of HCW gown, gloves, and clothes/skin in simulated environments with the use of fluorescent materials and/or bacteriophages.Reference Tomas, Kundrapu and Thota14–Reference Kwon, Burnham and Reske17 Our findings extend these observations to the real world of a busy clinical setting where it is likely that HCWs may deviate from PPE donning and doffing protocols, even if they have received formal training.Reference Krein, Mayer and Harrod18, Reference Doll, Feldman and Hartigan30 Our findings demonstrate that in a real-world setting, errors in PPE doffing result in HCW contamination, which creates the potential for cross-transmission of pathogens to other patients. Greater training of medical students and residents might improve the PPE donning and doffing practice in HCWs in the long run.Reference John, Tomas, Hari, Wilson and Donskey31
We found that 3.2% of HCWs had hand contamination with an MDRO after PPE removal. This finding confirms a previous study in which 4 of 84 HCWs who wore gloves and cared for VRE-colonized patients had VRE on their hands after glove removal.Reference Hayden, Blom, Lyle, Moore and Weinstein11 We extend this work by identifying that hand contamination following PPE removal was associated with removal of gloves before removal of gowns. We speculate that removal of gloves first leads to removal of the gown with bare hands, which would increase the chance of HCW hand contamination if a contaminated part of the gown were touched during gown doffing. Alternatively, hands could have been contaminated possibly due to leaks in gloves.Reference Tenorio, Badri and Sahgal32, Reference Olsen, Lynch, Coyle, Cummings, Bokete and Stamm33 Removing gloves first in PPE doffing is an approach outlined in the CDC doffing protocol; a prior study noted 43% of HCWs removed gloves first when doffing PPE,Reference Zellmer, Van Hoof and Safdar34 and we observed that nearly 20% of HCWs removed gloves before gowns when doffing. Although further work is needed to determine whether HCW self-contamination risk with gloves-first PPE doffing is generalizable, removing gloves first strategy might need to be reconsidered.
Our study has several limitations. First, observations were performed in a clinical setting, and the complex nature and rapidity of PPE donning and doffing could have led to inaccurate coding. However, research personnel underwent multiple training sessions before and during the study to ensure accuracy and consistency of coding of observations. Second, HCW practices could have been modified by the presence of study personnel (ie, the Hawthorne effect). Although this possibility cannot be discounted, we purposefully explained to HCWs that we were observing their entire regular patient care activities, not specifically mentioning or emphasizing the donning and doffing of PPE. Third, some contamination may not have been detected. However, we used both swabs and RODAC plates to improve the sensitivity of detection of environmental contamination,Reference Okamoto, Rhee and Schoeny20 and we used sensitive selective culture methods for all MDROs, including broth enrichment to improve the detection of MRSA and VRE. The objective of this study was to evaluate clinically important bacteria in a real-world setting; therefore, we adapted previously published methodologyReference Morgan, Rogawski and Thom10 to ensure the comparability of the results. Fourth, we did not measure culture results quantitatively, nor did we measure the intensity of patient care activities performed by HCWs. Thus, we cannot comment on whether those factors affected the frequency of HCW contamination.Reference Boyce, Potter-Bynoe, Chenevert and King35 Fifth, swabs used to culture patient sites were cultured without neutralizer for chlorohexidine, which might have caused “false negative” result (ie, no growth due to residual chlorohexidine), despite the diluting effect of the broth. Lastly, we did not use molecular methods to verify that MDRO strains isolated from HCWs, patients, and environment during patient interactions were the same. However, whole-genome sequencing is underway to better elucidate transmission dynamics for HCW contamination in adult ICUs.
In conclusion, the risk of hand contamination was higher when gloves were removed before gowns during PPE doffing. In addition, HCWs who made multiple PPE doffing errors were more likely to have contaminated clothes following a patient interaction. When caring for patients on contact precautions for an MDRO, HCWs appear to have differential risk for hand contamination based on their method of doffing PPE. An intervention as simple as reinforcing the preferred order of doffing may reduce HCW contamination with MDROs.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2019.33
Author ORCIDs
Koh Okamoto, 0000-0002-4072-9096
Acknowledgments
We thank Efrain Salazar and George P. Warpinski for patient enrollment and sample collection, and Melanie Carr, Patricia Mangan, and Grace McIlroy for microbiological analysis. We also thank all patients and healthcare workers in the 4 intensive care units at Rush University Medical Center who kindly participated in this study.
Financial support
This work was supported in part by the CDC Prevention Epicenters Program under Epicenters for the Prevention of Healthcare-Associated Infections, Antimicrobial Resistance and Adverse Events—A Multicenter Program Expansion (grant no. U54 RFA-CK-11-0010501SUPP15; principal investigator, R.A.W.).
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
