In multiple smaller studies, looking at one or two organisms, different activities have been associated with a greater likelihood of healthcare worker (HCW) clothing contamination including contact with wound dressing, artificial airways, side rails, linens, infusion pumps, catheters or drain, and direct patient contact including performing a physical examination or spending a longer duration in a room (4, 10, 11). Studies have not assessed common risk factors for contamination with the most common MDR bacteria and have been limited by clustering of patients or repeated measurements of the same HCW. A. baumannii may be more likely than other MDR bacteria to contaminate HCW clothing or the environment, although it has not been directly compared to methicillin-resistant Staphylococcus aureus (MRSA) or vancomycin-resistant Enterococci (4). Understanding factors that lead to contamination of HCW clothing, and thus increase potential for transmission, may help lead to interventions to prevent transmission of MDR bacteria. To our knowledge, no study has assessed the importance of environmental contamination leading to contamination of HCW clothing and thus the potential causal role of the environment inpatient-to-patient transmission of MDR bacteria.

To evaluate the differential rate of contamination by MDR A. baumannii compared with other MDR bacteria as well as investigating the importance of environmental contamination in the transfer of MDR bacteria to HCW clothing, we studied a cohort of ICU-based HCWs performing routine patient care.

We cultured HCWs during routine clinical care activities with patients that were on contact precautions for MDR bacteria. Patients were identified as MDR bacteria colonized or cocolonized on the basis of active surveillance or clinical cultures obtained as part of routine care; MDR bacteria status led to these patients being on contact precautions. Patients selected for study were part of a convenience sample. Patients had to have been identified as having MDR bacteria within the prior year. Active surveillance is performed at the time of admission for MRSA in all units with nares culture, for VRE in the medical ICU and surgical ICU with perirectal cultures, and in all units for MDR A. baumannii when a patient is admitted from another facility (consisting of culturing a patient’s sputum, groin, perirectal area, and wound, if present). We do not perform active surveillance for MDR Pseudomonas aeruginosa. In our hospital, contact precautions are used for all patients with MRSA and VRE as well as MDR A. baumannii susceptible to two or less antimicrobial classes and MDR P. aeruginosa one or less antimicrobial class not including tigecycline or polymyxins. (In this study, all isolates of either species were imipenem-resistant.)

Sequential HCWs (registered nurses, patient care technicians, respiratory therapists, occupational/physical therapists, and physicians) were approached for participation in the study before engaging in routine, nonemergent clinical care activities for these patients. HCW cultures were obtained as described previously (4, 11). Briefly, sterile rayon-tipped applicators were moistened with Amies transport media (Becton Dickinson 220093 CultureSwab, liquid Amies, Franklin Lakes, NJ). Hand and glove samples were obtained with a standardized process by swabbing the dorsum of each finger three times and the palm of each hand two times with a twirling motion of the swab with a single swab for both hands or both gloves. Gowns were sampled by swabbing each forearm twice and then swabbing a “W” along the beltline, all with a single swab and performed with a twirling motion. Hands were first sampled before donning gowns and gloves. Data from observations in which HCW hands were colonized before room entry with the same organism colonizing the patient were excluded from the analysis.

For risk factor collection, before the start of the study, we formed a “work group” of investigators to identify domains of interest for risk factor collection. Risk factors were grouped into domains related to HCW activities, HCW types, type of unit, and patient-specific factors (that could be noted by observation alone). HCW activities were observed and documented including the duration and nature of HCW–patient contact. On completion of activities and before exiting the patient room, HCW gloves and gown were swabbed by investigators. The bare hands of the HCWs were immediately sampled a second time before hand hygiene (4, 11).

After enrollment of five HCWs on the same day, using a standardized protocol, environmental sampling was performed on nine sites in the patient room: sink, bed rails, bedside table, vital sign monitor, supply cart, door handle, intravenous pump, ventilator, and floor. At each site, an area of approximately 10 cm² was sampled using a sterile cotton swab previously moistened with phosphate-buffered saline (2).

For MDR Gram-negative bacteria, swabs were vortexed in 5 mL of brain heart infusion broth with 6 μg/mL of imipenem (imipenem was added to enhance detection of MDR A. baumannii and MDR P. aeruginosa, because all isolates were carbapenem-resistant). For patients known to be colonized with MRSA and VRE, the swabs were placed in brain heart infusion with 2.5% NaCl. Broths were placed at 35°C overnight. For MDR Gram-negatives, 50 μL was plated to MacConkey agar with/without 6 μg/mL of imipenem. For MRSA, 50 μL was plated to ChromAgar MRSA and for VRE, 50 μL was plated to Enterococcus agar. Plates were incubated 24–48 hrs at 35°C. Nonlactose fermenting colonies were subcultured to trypticase soy agar with 5% sheep’s blood. Oxidase negative isolates were identified as A. baumannii by analytical profile index 20NE test strip or Vitek II (Biomerieux, Durham, NC). Oxidase-positive isolates were subcultured onto Triple Sugar Iron and Pseudomonas (P-agar) slants. Triple Sugar Iron slants were incubated overnight at 35°C; P-agar slants were placed at 42°C overnight. P. aeruginosa isolates were identified as K/K on Triple Sugar Iron slants and were pyocyanin-producing on P-agar. Colonies of MRSA and VRE were identified using standard laboratory procedures. Susceptibilities were performed following Clinical Laboratory and Standards Institute guidelines.

Pulsed field gel electrophoresis was performed on isolates following protocols from the Centers for Disease Control and Prevention’s pulsenet with modifications. Briefly, DNA was digested with restriction enzymes including SmaI, ApaI, and SpeI as appropriate according to manufacturer recommendations (New England Biolabs, Beverly, MA [1]). DNA was separated in 1% agarose on a contour-clamped homogeneous-field machine (CHEF-DR II; Bio-Rad, Richmond, CA). Electrophoresis was performed. After electrophoresis, gels were stained with ethidium bromide and photographed under ultraviolet illumination. Photographic images of gels were saved digitally with the Geldoc EQ (BioRad Laboratories, Hercules, CA) and saved as .tiff files for gel analysis with Gel Compare (Applied Math, Kortrijk, Belgium). The resulting band patterns were compared with the Dice coefficient 100 using the unweighted pair group method to define pulsed field type clusters using the criteria established by Tenover et al (12). Isolates selected for pulsed field gel electrophoresis were those in which patient clinical isolates from the same organism were available to compare with the environmental isolate and HCW isolate after patient contact. Identical isolates had no band differences, closely related isolates had three or less band difference, partially related had six or less band difference, and unrelated had less than seven band difference as per Tenover et al (12).

Overall, 120 of 585 (20.5%) HCW–patient interactions resulted in contamination of HCWs’ gloves or gowns. The frequency of detection of VRE, MDR A. baumannii, MRSA, and MDR P. aeruginosa on HCW gloves, gowns, or hands after patient care are reported in Table 1. Before room entry, 8.2% (20 of 245) of HCW room entries had at least one MDR bacteria on their hands. The frequency of specific organisms is presented in Table 1.

Forty-nine observation periods occurred with individual environmental sampling after each period of observation. Of all rooms sampled, 35 of 49 (71%) had positive environmental cultures for an MDR organism known to be colonizing the patient. A total of 77.8% (28 of 36) of rooms housing patients with MDR A. baumannii had environmental cultures for MDR A. baumannii. A total of 35.1% (13 of 37) of rooms housing patients with VRE had environmental cultures for VRE. A total of 28.1% (nine of 32) of rooms housing patients with MRSA had environmental cultures for MRSA. A total of 22.2% (four of 18) of rooms housing patients with MDR P. aeruginosa had environmental cultures for MDR P. aeruginosa.

We were unable to determine how many individual HCWs were contaminated with MDR bacteria, because we did not collect HCW names (130 HCWs indicated it was their first time participating in the study—accounting for 318 interactions with individual HCWs). Pulsed field gel electrophoresis was performed on environmental isolates for which an additional clinical isolate was available from a patient. (Table 2). Clinical isolates for MRSA, MDR P. aeruginosa, and MDR A. baumannii are routinely saved by the clinical laboratory. There were 22 cases (four MRSA, one MDR P. aeruginosa, and 17 MDR A. baumannii) selected for pulsed field gel electrophoresis, which compared 1) patient clinical isolate; 2) environmental isolate; and 3) HCW isolate after patient contact. Of the 22 cases, 18 (82%) of HCWs had a strain related to the patient and 20 HCWs had a strain related to the environment.

Risk factors for MDR A. baumannii contamination of HCW gown or gloves after patient care were evaluated with bivariable analysis and then multivariable modeling adjusting for multiple measurements with generalized estimating equations (Table 3). The final model showed that positive environmental culture OR 4.2; 95% CI 2.7–6.5), duration in room for 5 mins (OR 2.0; 95% CI 1.2–3.4), performing a physical examination (OR 1.7; 95% CI 1.1–2.8), and contact with the ventilator (OR 1.8; 95% CI, 1.1–2.8) were associated with contamination of HCW clothing by MDR bacteria (Table 4).

HCWs frequently become contaminated with MDR bacteria. Using a highly sensitive method, Hayden and colleagues (13) found that 62% of HCWs who entered the room of patients known to be colonized with VRE contaminated their gloves or hands during care. In experiments of transmission in which investigators touched prespecified areas on a patient or patient room and then imprinted their gloved hands to culture medium, MRSA was found to be transferred to gloves after approximately 30% to 60% of contacts (13, 14). Using the same method used in the current study with fewer patients and statistical analysis that did not account for clustering, we have examined frequency of contamination with MRSA and VRE (11) and MDR A. baumannii or MDR P. aeruginosa (4). We found a similar frequency of transmission of MDR bacteria in these studies. The impact of contamination of HCW clothing is not precisely understood. Presumably, HCW contamination with MDR bacteria is a key step in transmission to other patients. A study by Duckro et al (15) demonstrated that contaminated HCW hands were capable of transmitting VRE to inanimate surfaces. The frequency that HCW contamination results in transmission to patients is unknown. In our study, even after appropriate use of gloves and gown while caring for a patient on contact precautions, 1.7% to 4.2% of HCWs had a MDR bacterium on their hands without hand hygiene. This emphasizes the importance of hand hygiene after removal of gloves and gown (9).

Overall, 8% of HCW room entries occurred with a HCW who had MDR bacteria on their hands. The source of contamination of HCW hands may have been prior patient contact or contact with the environment. HCW hand contamination represents a significant potential for transmission and supports Centers for Disease Control and Prevention recommendations to perform hand hygiene before room entry (9). Alternatively, universal use of gown and gloves for all patient contact in the ICU may best address this problem (16).

MDR A. baumannii contaminated HCWs and patient rooms more frequently than other MDR bacteria. MDR A. baumannii was present in 78% of rooms housing a patient colonized with MDR A. baumannii. Similarly, HCWs entering the room of patients had MDR A. baumannii on their hands approximately 4% of the time, twice as often as other bacteria.

The current study supports the hypothesis that MDR A. baumannii is transmitted more frequently than other MDR bacteria. This increased propensity to contaminate hands, clothing, and the environment may, in part, explain the recent emergence of MDR A. baumannii (8). Future research is needed to explore characteristics of MDR A. baumannii that lead to this result.

A greater risk of contamination was observed when HCWs performed a physical examination or remained in the room longer, which are similar to risks of contamination described previously (4, 10, 11). Contact with a ventilator was associated with a greater risk of contamination, most likely because the respiratory tract is often heavily colonized with MDR bacteria and contact with respiratory equipment may pose a particular risk (4, 10, 11). The strongest predictor of HCW contamination was a patient room being found to be contaminated with MDR bacteria. HCWs entering a room that had environmental contamination with MDR bacteria were over four times as likely to become contaminated with that organism. Although studies have found an association between environmental cleaning and decreased risk of hospital-associated infections (17, 18), the mechanism of and interventions for environmental contamination and transmission of MDR bacteria are not well defined. Our study supports the idea that contamination of a patient’s room increases the likelihood of transmission (19–22).

The reason environmental contamination poses such a significant risk for HCW contamination is unclear. In our study, model building included a variable describing number of times a HCW contacted the environment. Frequency of HCW contact with the environment was not significant in the final model, whereas environmental contamination was. This dichotomy is perplexing but an important finding. It may be that patients who contaminate the environment with MDR bacteria are also more likely to contaminate HCW gowns and gloves through direct patient contact. We did not culture patients for MDR bacteria and cannot address the possibility that patients who contaminated the environment had a greater burden of organism. This should be addressed in future studies. Alternatively, contamination of the environment may be the critical step in contamination of HCWs. The latter option would indicate that transmission of MDR bacteria could be significantly decreased with more aggressive daily environmental cleaning. We also did not collect data relative to most recent environmental cleaning, although this would have been done daily, typically within 2–4 hrs of data collection. At a minimum, contamination of the environment appears to be a marker for a patient room at greater likelihood of HCW contamination.

The best method to control the spread of MDR bacteria in the ICU is unknown. Recent studies have focused on detecting and isolating patients for a specific organism such as MRSA (23, 24). Detecting and isolating patients has been studied alone (23) or combined with other interventions, including better hand hygiene, compliance with isolation, and preventive methods for specific healthcare-associated infections (24). Large multicenter studies such as these are necessary to evaluate infection prevention initiatives. Evidence from ours and other studies (13) suggest that environmental cleaning may be appropriate to study in a multicenter intervention trial.

Limitations of this study include that it was a single-center study with a modest number of patient rooms sampled. Because HCW participation was anonymous, we could not adjust for repeated measurements from the same HCW, although we adjusted for multiple observations on the same patient using generalized estimating equations. Although we knew 318 interactions occurred with a HCW on their first experience with the study, we did not know how many times other HCWs may have been observed. Although active surveillance for MDR bacteria is different in different ICUs, this potential bias is unlikely to have had an effect on reported rates of contamination because frequency of transmission did not differ by individual ICU. Although we had a significant detection rate of MDR A. baumannii, our sampling technique may have been insensitive compared with others (25). We did not record time of most recent environmental cleaning and did not assess effectiveness of cleaning. We were unable to assess which or how many patients’ anatomic sites were originally positive as a measure of organism burden, which has been associated with transmission of MRSA (10).