If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Epidemiology and Public Health, University of Maryland School of Medicine, Baltimore, MD, USAVeterans Affairs Maryland Healthcare System, Baltimore, MD, USACenter for Disease Dynamics, Economics and Policy, Washington, DC, USA
The 20-site Benefits of Universal Glove and Gown (BUGG) study found that wearing gloves and gowns for all patient contacts in the intensive care unit (ICU) reduced acquisition rates of meticillin-resistant Staphylococcus aureus (MRSA). The relative importance of gloves and gowns as a barrier, improved hand hygiene, and reduced healthcare worker (HCW)–patient contact rates is unknown.
Aim
To determine what proportion of the reduction in acquisition rates observed in the BUGG study was due to improved hand hygiene, reduced contact rates, and universal glove and gown use using agent-based simulation modelling.
Methods
An existing agent-based model to simulate MRSA transmission dynamics in an ICU was modified, and the model was calibrated using site-specific data. Model validation was completed using data collected in the BUGG study. A full 2k factorial design was conducted to quantify the relative benefits of improving each of the aforementioned factors with respect to MRSA acquisition rates.
Findings
Across 40 simulated replications for each factorial design point and intervention site, approximately 44% of the decrease in MRSA acquisition rates was due to universal glove and gown use, 38.1% of the decrease was due to improvement in hand hygiene compliance on exiting patient rooms, and 14.5% of the decrease was due to the reduction in HCW–patient contact rates.
Conclusion
Using mathematical modelling, the decrease in MRSA acquisition in the BUGG study was found to be due primarily to the barrier effects of gowns and gloves, followed by improved hand hygiene and lower HCW–patient contact rates.
A 20-site randomized controlled trial was undertaken to assess the benefits of universal glove and gown use [wearing gloves and gowns for all patient contacts in the intensive care unit (ICU)] in reducing the transmission of multi-drug-resistant organisms (MDROs) in acute care hospitals [
]. The Benefits of Universal Glove and Gown (BUGG) study focused on pre- and post-intervention acquisition rates for meticillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) in ICU settings.
The BUGG study found that universal glove and gown use had no effect on VRE acquisition rates; however, a large effect on MRSA acquisition rates was observed. In the intervention ICUs, acquisition rates decreased from 10.03 per 1000 patient-days in the baseline period to 6.00 per 1000 patient-days in the study period, whereas the control ICUs experienced a non-significant decrease in MRSA from 6.98 acquisitions per 1000 patient-days to 5.94 acquisitions per 1000 patient-days. Overall, there was a 40.2% relative reduction in MRSA acquisition rates in the intervention ICUs, compared with a 15.0% reduction in the control ICUs (P=0.046) [
However, the intervention ICUs also observed higher rates of hand hygiene compliance on exiting patient rooms (78.3% vs 62.9%), and lower contact rates between healthcare workers (HCWs) and patients (4.28 vs 5.24 entries/h), presumably due to the additional burden of donning gloves and gowns for each visit [
]. All three of these effects – the increased use of contact precautions, improved hand hygiene compliance on exiting patient rooms, and decreased patient visits by HCWs – could account for a decrease in MRSA acquisition, which has sparked a discussion within the healthcare epidemiology community about the relative importance of each of these factors [
Randomized controlled trials are expensive and time consuming to conduct; therefore, the authors' goal was to maximize the benefit of the BUGG study. The aim of the present study was to use agent-based mathematical modelling to mediate the results from the experimental BUGG study, leveraging the significant amount of data that was collected, and augmenting the conclusions drawn. Others have used this modelling approach to look at relative effects of infection control interventions in trials [
]. Specifically, the authors developed and calibrated an agent-based model of MRSA transmission using observational data collected during the BUGG study, and then used the model to determine what proportion of the reduction in acquisition rates was due to improved hand hygiene, reduced HCW–patient contact rates, and universal glove and gown use.
Methods
Many of the details of the BUGG study have been published previously [
], but the important points pertinent to the present study are as follows. The BUGG study was a cluster randomized trial conducted at 20 ICUs across the USA, and consisted of a three-month baseline period (beginning in September 2011) and a nine-month study period (beginning in January 2012). The intervention arm consisted of 10 ICUs where all HCWs were required to wear gloves and gowns for all patient contacts. The control arm consisted of 10 ICUs where the usual standard of care was followed, which required HCWs to follow US Centers for Disease Control and Prevention guidelines for contact precautions (i.e. gloves and gowns) when caring for patients known to have infection or colonization with antibiotic-resistant bacteria, such as VRE or MRSA. During the study period, contact precautions were enforced for all patients in the intervention arm and a mean of 10.5% of patients in the control arm. MRSA acquisition rates were calculated from nasal surveillance cultures obtained on ICU admission and discharge.
Hand hygiene compliance was measured via 30-min direct observation periods in a random sample of patient rooms. Site study staff observed HCWs covertly. Observations were made at various times of day for 2h/week over the entire study period. Hand hygiene was monitored on room entry and room exit. The recording form used was based on a form from the Institute for Healthcare Improvement [
]. The frequency of HCW visits was also recorded during these same 30-min observation periods.
An agent-based model was used to simulate MRSA transmission dynamics in an ICU. An existing model developed in NetLogo (v5.1.0) that has been used in previous studies [
Preventing the transmission of multidrug-resistant organisms (MDROs): modeling the relative importance of hand hygiene and environmental cleaning interventions.
] was modified for use in the present study. A conceptual diagram of the agent-based model is presented in Figure 1. In the model, patients are admitted to the ICU with parameterized probabilities of MRSA colonization (admission prevalence) and contact precautions status. On each visit to patients on contact precautions, HCWs probabilistically comply with wearing gloves and a gown, which affects the transmission probability from HCW to patient (and vice versa), noting that these transmission opportunities are only possible when one agent is colonized and the other is susceptible. If the HCW complies with wearing gloves and a gown, a gloves-and-gown effect is applied to the visit, which is a multiplicative factor that effectively reduces the transmission probability between the HCW and the patient. The transmission probability is unaffected for HCW visits to patients who are not on contact precautions. Hand hygiene opportunities are modelled on both entry to and exit from patient rooms, and successful removal of MRSA is dependent on satisfaction with both hand hygiene compliance and hand hygiene efficacy. Both hand hygiene compliance and gloves-and-gown compliance are specific to each type of HCW in the model (i.e. physicians, nurses and a generic ‘other’ class). It was assumed that each visit only involved one HCW interacting directly with the patient. Other HCWs could be present, but would not factor into transmission. It was also assumed that the ICU was fully occupied at all times; therefore, a new patient was admitted when another patient was discharged. Most model inputs for the simulation experiments were informed directly from the BUGG study. These parameters are summarized in Appendix A (see online supplementary material). In addition, the transmission probability and gloves-and-gown effect parameters were calibrated to be 0.05 and 0.6, respectively, using extensive experimentation that is described in detail elsewhere [
]. In the absence of data collected during the baseline period of the study, experiments were only run using the study period data when comparing simulated and observed MRSA acquisition rates.
Figure 1Conceptual patient flow diagram for agent-based model of meticillin-resistant Staphylococcus aureus (MRSA) transmission.
Using this calibrated agent-based model, a 2k factorial design was used to determine the independent impact of improving each of the three aforementioned factors with respect to reducing MRSA acquisition rates for each intervention site from the BUGG study. Specifically, simulation experiments were conducted across eight design intervention points (23) for which all combinations of baseline and improved levels of hand hygiene compliance on exiting patient rooms (factor 1), HCW–patient contact rates (factor 2), and the proportion of patients assigned to contact precautions status (factor 3) were simulated. The observed data collected during the study period were considered to represent the improved levels of these three factors, and therefore the baseline levels must be inferred. For hand hygiene compliance on exiting patient rooms and contact rates, baseline levels were assumed to be equal to the observed study period levels for each site minus the respective (and fixed) improvements of +15.4% and –0.96 contacts per patient per hour. For the proportion of patients assigned to contact precautions status, the authors began with the site-specific levels of this factor during the baseline period, and explored the impact of increasing this proportion to 100% (i.e. enforcing a universal glove and gown policy). Next, the main effect of each of these factors was calculated to estimate the direct impact on MRSA acquisition rates. The design of this experiment is summarized in Table I.
Forty simulation replications were conducted for each of the 20 participating hospitals during the study period. These experiments are summarized in Figure 2, which shows the distribution of model error (i.e. the difference between observed and simulated MRSA acquisition rates) for each site. On average, the model fits the observed data well; the mean model error across all sites is –0.51 acquisitions per 1000 patient-days, despite the fact that the model error was consistently large for a small number of sites (e.g. 6, 11, 15).
Figure 2Distribution summary of model errors between observed and simulated meticillin-resistant Staphylococcus aureus acquisition rates for each site. The mean model error across all sites is –0.51 acquisitions per 1000 patient-days.
Forty simulation replications were also conducted for each 2k factorial design point and intervention site. These results are summarized in Figure 3 for each intervention site, which shows fairly consistent trends with respect to each factor. The mean main effect is −4.05 acquisitions per 1000 patient-days [95% confidence interval (CI) −4.67 to −3.44] for hand hygiene compliance on exiting patient rooms, −1.54 acquisitions per 1000 patient-days (95% CI −1.91 to −1.17) for HCW contact rates, and –4.68 acquisitions per 1000 patient-days (95% CI −5.16 to −4.21) for universal glove and gown use. Overall, the improvement in all three factors led to a decrease in the MRSA acquisition rate from 13.18 acquisitions per 1000 patient-days (design point 1) to 2.54 acquisitions per 1000 patient-days (design point 8) when pooled across all sites. Approximately 44% of the decrease in MRSA acquisition was due to universal glove and gown use, 38.1% of the decrease was due to the improvement in hand hygiene compliance on exiting patient rooms, and 14.5% of the decrease was due to the reduction in HCW contact rates. The remaining 3.4% of this reduction was attributed to stochastic variability.
Figure 3Distribution summary (across 40 replications) of percentage reduction in meticillin-resistant Staphylococcus aureus (MRSA) acquisition rates for (a) hand hygiene compliance on exiting patient rooms, (b) contact rates, and (c) universal glove and gown use for each intervention site. The mean percentage reduction was 38.1% for hand hygiene compliance on exiting patient rooms, 14.5% for contact rates, and 44.0% for universal glove and gown use.
This study found that approximately 44.0% of the reduction in MRSA acquisition reported in the BUGG study was due to universal glove and gown use, 38.1% was due to improved hand hygiene on exiting patient rooms, and 14.5% was due to fewer HCW–patient contacts during HCW visits. This analysis allows for better understanding of the degree to which each component of universal glove and gown use led to a decrease in MRSA acquisition in the ICU setting. However, if universal glove and gown use led to improved hand hygiene and fewer HCW visits [
], and these factors all have an effect on acquisition of antibiotic-resistant bacteria, the effect of universal glove and gown use should also be considered as a whole or bundled approach of these factors. This is particularly true in light of the literature reporting that compliance with hand hygiene is difficult to achieve or sustain [
Potential positive and negative effects of contact precautions and fewer HCW visits merit discussion. The primary BUGG study and additional subanalyses have found that universal glove and gown use and contact precautions did not increase adverse events [
Benefits of Universal Glove and Gown Primary Investigators. The effect of universal glove and gown use on adverse events in intensive care unit patients.
]. The results of this randomized trial should hold more weight in the scientific literature than the often-quoted cohort study that suffered from confounding by indication [
The overall cost-effectiveness of a universal glove and gown policy in the ICU setting has yet to be determined and is not included in this model. The lack of consistency of the findings for MRSA and VRE has hindered enthusiasm and widespread adoption. Further work is underway to assess whether the intervention has a positive effect on antibiotic-resistant Gram-negative bacteria [
The modelling in this study has some limitations. The overall accuracy of the model was fairly high; however, there were some individual sites for which the simulated MRSA acquisition rates deviated significantly from those observed in the study. There are several reasons why these deviations may have occurred. First, the agent-based model is only an approximation of MRSA transmission dynamics in an ICU setting. The model does not include, for example, effects of environmental contamination, multi-use equipment and antibiotic usage that could also contribute to transmission. The authors have modelled these effects in other studies, but did not have access to data collected during the BUGG study to inform the site-specific model parameters [
Preventing the transmission of multidrug-resistant organisms (MDROs): modeling the relative importance of hand hygiene and environmental cleaning interventions.
]. Second, the model parameters were informed by data collected during the BUGG study. In some cases, the sample size for observed compliance rates for hand hygiene and glove and gown use was small, leading to highly uncertain estimates for the associated parameters. For example, there were limited observations of low physician compliance with hand hygiene and contact precautions at Site 6, which caused the simulation to overestimate transmission significantly for this particular site. Despite these limitations in model accuracy, the authors believe that there is still value in the relative contributions of each of the aforementioned three factors in reducing MRSA acquisition rates, and this model used actual clinical data for 20 hospitals unlike most models which are based on single-site data.
In conclusion, mathematical modelling showed that the decrease in MRSA acquisition in the BUGG study was primarily due to the barrier effect of universal glove and gown use, followed by improved hand hygiene compliance and fewer HCW visits. As it is likely that the mandatory universal gown and glove policy drove the higher hand hygiene compliance and reduced HCW–patient contacts, the BUGG bundle is an effective intervention in the MRSA prevention armamentarium. The authors believe that certain ICUs would benefit from the use of universal glove and gown policies for all patients.
Acknowledgements
The authors wish to thank M. Shardell, X. Liu, X. Li and Y. Zhou for their assistance with running simulations and analysing results during the various phases of this project.
Conflict of interest statement
None declared.
Funding sources
This research was supported by the Agency for Healthcare Research and Quality (Grant Nos. HHSA290200600015 and 1R18HS024045-01) and the National Institutes of Health (Grant No. 5K24AI079040-05).
Appendix A. Supplementary data
The following is the supplementary data related to this article:
Preventing the transmission of multidrug-resistant organisms (MDROs): modeling the relative importance of hand hygiene and environmental cleaning interventions.
Benefits of Universal Glove and Gown Primary Investigators. The effect of universal glove and gown use on adverse events in intensive care unit patients.
30142-1&id=gr1.jpg){kind=link}
30142-1&id=gr2.jpg){kind=link}
30142-1&id=gr3.jpg){kind=link}