Impact of environmental hygiene interventions on healthcare-associated infections and patient colonization: a systematic review

Healthcare-associated infections (HAI) are one of the gravest threats to patient safety worldwide. The importance of the hospital environment has recently been revalued in infection prevention and control. Though the literature is evolving rapidly, many institutions still do not consider healthcare environmental hygiene (HEH) very important for patient safety. The evidence for interventions in the healthcare environment on patient colonization and HAI with multidrug-resistant microorganisms (MDROs) or other epidemiologically relevant pathogens was reviewed. We performed a systematic review according to the PRISMA guidelines using the PubMed and Web of Science databases. All original studies were eligible if published before December 31, 2019, and if the effect of an HEH intervention on HAI or patient colonization was measured. Studies were not eligible if they were conducted in vitro, did not include patient colonization or HAI as an outcome, were bundled with hand hygiene interventions, included a complete structural rebuild of the healthcare facility or were implemented during an outbreak. The primary outcome was the comparison of the intervention on patient colonization or HAI compared to baseline or control. Interventions were categorized by mechanical, chemical, human factors, or bundles. Study quality was assessed using a specifically-designed tool that considered study design, sample size, control, confounders, and issues with reporting. The effect of HEH interventions on environmental bioburden was studied as a secondary outcome. After deduplication, 952 records were scrutinized, of which 44 were included for full text assessment. A total of 26 articles were included in the review and analyzed. Most studies demonstrated a reduction of patient colonization or HAI, and all that analyzed bioburden demonstrated a reduction following the HEH intervention. Studies tested mechanical interventions (n = 8), chemical interventions (n = 7), human factors interventions (n = 3), and bundled interventions (n = 8). The majority of studies (21/26, 81%) analyzed either S. aureus, C. difficile, and/or vancomycin-resistant enterococci. Most studies (23/26, 88%) reported a decrease of MDRO-colonization or HAI for at least one of the tested organisms, while 58% reported a significant decrease of MDRO-colonization or HAI for all tested microorganisms. Forty-two percent were of good quality according to the scoring system. The majority (21/26, 81%) of study interventions were recommended for application by the authors. Studies were often not powered adequately to measure statistically significant reductions. Improving HEH helps keep patients safe. Most studies demonstrated that interventions in the hospital environment were related with lower HAI and/or patient colonization. Most of the studies were not of high quality; additional adequately-powered, high-quality studies are needed. Systematic registration number: CRD42020204909


Background
Clean healthcare facilities look appealing, offer a sense of security and increase patient satisfaction [1]. Although visually clean facilities have become the standard of healthcare settings in high-income countries, cleanliness not only plays a role in quality of care, but in its safety. The microbiological aspect of cleanliness, healthcare environmental hygiene (HEH), has remained a neglected field, with little investment beyond what is considered the norm. Few high-quality studies link interventions in HEH to a reduction in either patient colonization with epidemiologically relevant pathogens or healthcare-associated infections (HAI). Though there are many reasons for this, one is the lack of literature critically evaluating the role of HEH in patient safety.
HAI are acquired during hospital stay [2] and cause more deaths worldwide than malaria, tuberculosis, and AIDS combined, and the burden of the six main types of HAI is higher than the total burden of the 32 major communicable diseases [3,4]. These infections also increase morbidity, prolong hospital stay, and are a major financial burden to healthcare systems [5,6]. The total annual global cost for five of the most common types of HAI is estimated at $8.3-$11.5 billion [7]. Despite their ubiquity, still much is unknown about how to prevent HAI, and no single hospital or healthcare facility in the world can claim to be unaffected.
While HAIs are usually the result of an infection with the patient's own flora, this flora can change due to colonization with hospital pathogens through HCWs' hands or from the hospital environment. Definitively knowing whether an HAI came from the patient's environment or from another source is difficult. Though it is known that some bacteria are more often transmitted through the patient environment than others, it is comparatively rare that extensive investigations are performed at the time of diagnosis. Usually such investigations are reserved for unusual infections or outbreak situations, in hospitals with sufficient resources to undertake them.
Over the past 25 years, best practice interventions such as hand hygiene in patient care have reduced the number of HAIs [8,9]. Poor hand hygiene has been recognized as being one of the main drivers of HAIs among patients [9]. Even if such practices can reduce HAIs by up to 50%, there is still a remaining proportion that needs to be addressed and where HEH may play a role [10]. A prerequisite for addressing some of these challenges is to review the literature to evaluate whether HEH interventions have a direct effect on HAI and thus, on patient safety.
HEH is essential for all types of healthcare facilities, from hospitals and long-term care facilities to home care environments. Environmental hygiene builds on both technical and human components, and it includes all aspects of the healthcare environment that are not the patient or the HCWs themselves. The technical component includes cleaning and disinfection of surfaces, water management, air control, waste management, laundry, and sterilization and device reprocessing. The human component includes best practice implementation, staff management, and environmental services departments' structural organization [11]. This component includes the evaluation of the cost and value of HEH interventions and programs, the training and monitoring of staff, their career development and workflow organization. Both of these components carry major implications for the wellbeing of patients, HCWs, the community and the larger natural environment.
Beyond the biological plausibility that the healthcare environment has a direct effect on patient safety, a number of reports over the last decades increasingly highlighted the potential impact of environmental hygiene on health [12,13]. Most common healthcare-associated pathogens are known to survive on surfaces for hours or days, some for weeks and a few for over a year [14,15]. It has been shown that hygiene failures correlate strongly with HAI in an ICU setting [16]. There is an increase of 150-500% in the chance of acquiring a pathogen if the prior room occupant was colonized with it [17].
This paper reviews the evidence-base for the ability of interventions in the hospital environment to reduce patient colonization with multidrug-resistant microorganisms (MDROs) and other epidemiologically relevant pathogens, and to prevent HAI. This exercise is difficult for a number of reasons. First, high-quality randomized  11:38 controlled trials in HEH are sparse. Secondly, the bulk of studies are retrospective or prospective before-andafter studies with limited methodological quality. Third, there is heterogeneity of the field about "clean environment" and how environmental hygiene is defined. Finally, HEH interventions are often combined with other infection prevention and control (IPC) interventions such as hand hygiene or a reorganization of patient care. These confounding factors can cause difficulty when determining whether outcomes are a direct effect of an HEH intervention.

Methods
We performed the systematic review protocol according to the PRISMA checklist [18], in both the PubMed and Web of Science databases. The full search strategies are available in the Additional file 1. The primary outcome is a comparison of the measure of patient colonization or HAI compared to baseline/control. HAI was defined according to the WHO definition [2]. The secondary outcome was environmental bioburden as defined as either cultured environmental samples or adenosine tri-phosphate (ATP) sampling. Although ATP sampling is technically a proxy measure of bioburden, it correlates closely with microbiological sampling in the literature [19]. Other proxy measures for bioburden such as the use of florescent dye were not included. Though the use of fluorescent techniques can show a measurable improvement in cleaning procedures, they do not necessarily demonstrated an impact on bioburden, depending on what is being used to remove the fluorescent dye. Therefore, studies that used improved cleaning practices or fluorescent marking as a proxy measure of bioburden were marked as "NA".
All original studies were eligible if they were published before December 31, 2019, and if they measured the effect of an HEH intervention on HAI or patient colonization. Studies with an English abstract were eligible when published in English, French, German, or Spanish and only included if they were original research.
Studies were not eligible if they were conducted in vitro, did not include patient colonization or HAI as an outcome, were bundled with hand hygiene interventions, or were implemented during an outbreak. Outbreaks were excluded because outbreak management broadens the intervention, and it would not be possible to adjust for that effect. Complete structural rebuilds were excluded, because interventions such as renovating a building or replacing a plumbing system are not feasible HEH interventions in most contexts. There is also evidence that such interventions result in reduction of the studied pathogen for a limited time, after which the environment can become recolonized [20].
Interventions of interest were either mechanical, chemical, or they applied a human factors design. The standardized extraction forms included type of intervention, study title, authors, year of publication, study design, type of intervention(s), intervention(s), sample size or sample size proxy, control, microorganisms studied, outcome, whether the method is recommended for application by the authors, quality score and grade, reduction in bioburden, and comments.
Interventions were stratified into chemical, mechanical, human factors, and bundles of combining two or more of the aforementioned categories. Titles, abstracts and the full text of all potentially eligible studies were screened independently by at least two reviewers. Inclusions and exclusions were recorded following the PRISMA guidelines, and reasons for exclusion were detailed. Data were extracted by two authors. Any disagreement was resolved through discussion with a third author. Any missing data was requested from original study authors by email. Ethical approval was not required for this review.
As a wide variety of procedures and methodologies were identified, a descriptive analysis with a narrative synthesis was performed. Due to this heterogeneity, additional sub-group analyses by type of intervention, type of microorganism, and study quality were performed.
The study designs were divided into the following categories: randomized controlled trials (RCTs), quasiexperimental studies (prospective and retrospective), and before-and-after studies (prospective and retrospective). Sample sizes were categorized by ranges from less than 10 to more than 100′000 patients/patient-days/room cleanings. Presence of a study control was adjusted to include proxies for a control. The main confounding factors that were analyzed included hand hygiene compliance, antibiotic use, and the seasonality of certain HAI.
Available tools for analyzing study quality were assessed, and selected using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for conducting observational studies which had been previously used for such a review [21,22]. The STROBE checklist was, however, difficult to apply to some HEH interventions, in particular when a study had no control, its primary outcome was laboratory-based or based on bioburden measurements. We therefore also constructed a specifically-designed quality scoring system which included what the reviewers deemed the most important elements in the studies. Obviously, this scoring system is only meant to compare this specific list of studies and is not applicable in other contexts. After discussion in a working group, the following five elements were included in the quality assessment: study design, sample size, control, confounders, and issues with reporting. Among issues with reporting, conflict of interest (COI) was defined as minor if less than half of the authors disclosed a COI, such as having worked for industry as a consultant in the same field, and major if more than half of authors were funded by industry for the study. Table 1 summarizes the quality scoring scale used in the review. Studies were graded from 0 to 20 points. "High quality" studies referred to studies that received an A or B grade according to the quality scale (Table 1). Some studies that ranked lower on the quality scale were well-performed, but simply not designed or powered to determine significant changes in patient colonization or HAI.
Over half (15/26, 58%) of the studies demonstrated a significant decrease in patient colonization or HAI following the chosen intervention for all microorganisms tested [24, 26, 29, 31, 33, 35-38, 40, 41, 43-46]. In one study, the reduction was not significant for all patient groups [26]. If additional interventions that demonstrated a reduction in all microorganisms tested were included, whether significant or not, this increased to 69% [23,28,32]. If the additional interventions that demonstrated a reduction in at least one of the microorganisms tested (significant or not) were included, this increased to 88% [25,27,34,47,48].
Sub-group analyses were conducted for the most frequently implemented interventions (Table 3): ultraviolet-C light (UVC), hydrogen peroxide (both liquid and gaseous), and human factors. UVC interventions were implemented in six studies [23,24,27,29,30,48]. Of these, one study was bundled [48]. The interventions were recommended for application by the authors in four (67%) of the studies [24,27,29,48]. Reductions in colonization/HAI were significant in those same four studies, though not for all microorganisms tested [27,48].
Five studies assessed the implementation of gaseous hydrogen peroxide [31,35,36,45,46]; two were bundled interventions [45,46]. The interventions were recommended for application by authors in all studies, and all reductions were statistically significant. Three studies assessed liquid hydrogen peroxide [32,33,46]. The interventions were recommended in all studies, and the reductions in colonization/HAI were statistically significant in two studies [33,46].
Human factors studies encompassed all interventions that included training and education, monitoring and feedback, and promotion of institutional safety climate. Nine studies assessed the implementation of human factors [38][39][40][41][42][44][45][46][47]; six were bundled interventions [41,42,[44][45][46][47]. The interventions were recommended by the authors in 78% (7/9) of the studies [38,40,41,[44][45][46][47], though one only recommended it for VRE [47]. Reductions in colonization/HAI were significant in those same studies. One study performed a cost analysis. The installation of high efficiency particulate air (HEPA) filters was found to decrease the cost per patient; it is to note that these findings were significant in both $ and €, but did not reach the threshold for significance in Turkish Lira [26]. Another article suggested that gaseous hydrogen peroxide decontamination was cost-effective for C. difficile, based on the estimated minimum cost of nosocomial C. difficile infection per year [36].
Seven studies assessed the effect of interventions on Gram negative bacteria [25,29,30,34,41,43,48]. Three studies observed the impact of an intervention on A. baumannii (including carbapenem-resistant and multidrug-resistant strains) [34,41,48], and three on Pseudomonas (two on P. aeruginosa and one on Pseudonomas spp.) [25,30,43]. Klebsiella, extended spectrum beta-lactamase Enterobacteriaceae, S. maltophilia, Proteus sp. and coliform bacilli were each analyzed by only one study [25,30,43]. Fiftyseven percent of interventions (4/7) were recommended for application by the authors, each of which demonstrated a significant decrease in HAI/colonization [29,41,43,48]. One older study [28] evaluated the role of negative air pressure rooms to prevent Varicella zoster and Herpes zoster infection. Although statistical significance was not calculated, there were no new cases after the intervention and the method was recommended by the authors [28]. Another study demonstrated the effect of air control to prevent invasive fungal infections during construction and showed an effect among oncology-haematology patients [26].  Analysis by quality (Table 4) The quality scoring system (Table 1) considered study design, sample size, whether there was a control, how the study adjusted for confounding factors, and issues in reporting. Table 4 shows the detailed quality scoring system results for the 26 studies. Forty-two percent of the studies (11/26) were considered to be of high-quality (grade A or B, Table 4). All studies that were of quality "A" and 1 study of quality "B" were RCTs [32,39,47,48]. 27% of high-quality study interventions (3/11) were not recommended for application by the authors [23,25,39]. The interventions in 64% (7/11) of these studies significantly reduced colonization/HAI [33-35, 38, 44, 47, 48]. In 43% (3/7) of these studies, the reduction was only significant for specific bacteria [34,44,47]. Fifty-eight percent of the studies (15/26) were of lower quality (grade of C or D, Table 4). Eighty-six percent of these (13/15) significantly reduced colonization/HAI [24, 26-29, 31, 36, 37, 40, 41, 43, 45, 46]. In one of these studies, the reduction was only significant for specific bacteria [27]. A further analysis was conducted which included only the higher quality studies that used a true control, and the most commonly studied microorganisms (S. aureus, C. difficile, and VRE), in order to assess whether there was a significant reduction per pairing of each microorganism and intervention (Table 5). This resulted in 15 of pairings from five studies [32,39,44,47,48]. The distribution included five interventions for each S. aureus, C. difficile, and VRE. Eighty-seven percent of the pairings (13/15) demonstrated a reduction in colonization or HAI [32,44,47,48], but only 27% of them (4/15) demonstrated a significant reduction in patient colonization or HAI [44,47,48]. Studies were too heterogenous to perform any kind of metanalysis, and in those high quality studies, no two interventions on the same microorganism were comparable. Future studies in the field should aim to calculate sample sizes and be adequately powered to be able to demonstrate such reductions.

Interpretation
This systematic review demonstrated that interventions in environmental hygiene were often associated with a reduction in HAI in a seemingly causal way. Over half of studies demonstrated a significant decrease in colonization or HAI for all of the microorganisms tested. These results are indicative of the importance of environmental hygiene in patient safety. There were major issues with both the heterogeneity of the interventions and the settings, as well with the quality in a number of the studies, hence the sub analyses. There are relatively few high quality studies in HEH compared to other fields, and even the use of RCTs in the field is exceedingly rare [11]. One high-quality study [49] in particular would have been useful for the review, but was excluded due to a hand hygiene intervention. Often, the primary study outcome evaluated environmental bioburden. Though HAI or patient colonization was a secondary outcome obtained from hospital data, these studies were not necessarily designed and powered to analyze this outcome. The measurable impact of HEH is likely to be more apparent if future studies are sufficiently powered.
Most of the studies that did not show a statistically significant reduction in HAI or patient colonization nonetheless recommended their interventions for application because they did greatly reduce environmental bioburden [28,32,38]. Though eight studies had controls [25,32,37,39,42,44,47,48], many had before-and-after study designs [23, 24, 26-31, 34-36, 40, 41, 43, 45, 46], and thus did not implement appropriate controls. Two used similar institutions as "proxy" controls [33,38]. Often, studies used the baseline rate of colonization or HAI before the intervention was implemented, and attempted to account for some confounding factors such as hand hygiene, antimicrobial use, and seasonality of the diseases of interest. In retrospect, it may have been more useful to only analyze more recent studies, because the two that were published before 2000 [25,28] (in 1971 and 1985, respectively) were exploring different research questions and microorganisms.
The success of the interventions also depended on which microorganisms were studied, and how successfully or not specific pathogens spread through the healthcare environment. For example, VRE, known to spread through the environment, was sometimes more successfully reduced than pathogens known to frequently spread through hands from patient to patient. One study [26] testing air filters gave further support to the fact that not all microorganisms are able to be transmitted by air, unlike what some manufacturers claim.
Considering the subset analysis targeted on specific pathogens, it is important to note that not all studies were designed to demonstrate the efficacy of a particular intervention on colonization/HAI, as this was not always the primary outcome. Some interventions were recommended by the authors for application because they demonstrated a significant reduction in some pathogens but not in others. Though these outcomes were often coupled with a significant decrease in environmental bioburden, some studies were not sufficiently powered to demonstrate that the reduction was statistically significant.
Overall, the selected studies were very heterogenous; both in terms of the types of interventions and their quality. The review attempts to address some of these limitations by performing subset analyses. However, the Studies were selected if they had a quality rating of "A" or "B" (Table 4), used a control and if they studied the three most commonly-examined microorganisms Significance of individual experiments on commonly studied microorganisms per method of intervention; systematic review