Introduction
Nosocomial infections, also called healthcare-associated infections (HCAIs), are reported to account for approximately 7% of all infections in developed countries and 10% in developing countries [Reference Danasekaran, Mani and Annadurai1, Reference Haque, Sartelli, McKimm and Abu Bakar2]. Recent evidence suggests that nosocomial infections affect nearly 15% of all hospitalised patients [Reference Raoofi, Pashazadeh Kan, Rafiei, Hosseinipalangi, Noorani Mejareh, Khani, Abdollahi, Seyghalani Talab, Sanaei, Zarabi, Dolati, Ahmadi, Raoofi, Sarhadi, Masoumi, sadat Hosseini, Vali, Gholamali, Asadi, Ahmadi, Ahmadi, Beiramy Chomalu, Asadollahi, Rajabi, Gharagozloo, Nejatifar, Soheylirad, Jalali, Aghajani, Navidriahy, Deylami, Nasiri, Zareei, Golmohammadi, Shabani, Torabi, Shabaninejad, Nemati, Amerzadeh, Aryankhesal and Ghashghaee3] and are associated with prolonged hospital stay, significant disability, and economic burden. Studies conducted in high-income settings in the USA and Europe showed that the incidence density of such infections is around 13–20 episodes per 1000 patient days [Reference Khan, Baig and Mehboob4] and is associated with a high financial burden [Reference Gidey, Gidey, Hailu, Gebreamlak and Niriayo5]. This burden is expected to be much higher with the increasing emergence of multi-drug-resistant organisms (MDROs) [Reference Gidey, Gidey, Hailu, Gebreamlak and Niriayo5]. Recent studies have shown that >70% of the bacteria implicated in HCAI are usually resistant to one or more of the antimicrobials used for the initial treatment of patients [Reference Marschall, Agniel, Fraser, Doherty and Warren6], and the attributable cost increase in treating resistant organisms ranges from 4000 to 4500 USD per infection per patient [Reference Jia, Li, Hou, Ma, Yang, Wu, Liu, Wen, Yang, Luo, Xing, Zhang, Wu, Ding, Liu, Lin, Li, Chen and Li7, Reference Johnston, Thorpe, Jacob and Murphy8].
Multi-drug-resistant bacteria can survive in the hospital environment for long periods [Reference Simmons and Larson9], and all surfaces, porous or non-porous, in patient’s rooms are highly susceptible to contamination [Reference Leas, Sullivan, Han, Pegues, Kaczmarek and Umscheid10]. Consequently, effective infection prevention programmes have environmental hygiene as an integral component. A wide range of chemical disinfectants are commonly used in healthcare settings and include surface disinfectants such as quaternary ammonium compounds, sodium hypochlorite, peracetic acid, and liquid hydrogen peroxide [Reference Leas, Sullivan, Han, Pegues, Kaczmarek and Umscheid10]. No-touch technologies in addition to conventional cleaning measures are commonly used in hospital settings and include exposure to ultraviolet light or hydrogen peroxide vapour or mist [Reference Otter, Yezli, Perl, Barbut, French and Walker11–Reference Passaretti, Otter, Reich, Myers, Shepard, Ross, Carroll, Lipsett and Perl13]. Ultraviolet light sources are broadly categorised as UV-C devices and those that utilise pulsed xenon–UV light (PX-UVL). The former consist of mercury bulbs that emit continuous radiation of wavelength ranging from 200 to 270 nm [Reference Weber, Anderson and Rutala14, Reference Nerandzic, Thota, Sankar C., Jencson, Cadnum, Ray, Salata, Watkins and Donskey15], while a PX-UVL system is characterised by short high-intensity bursts of radiation of UV wavelengths (100–280 nm) and visible (380–700 nm) spectra [Reference Weber, Anderson and Rutala14, Reference Nerandzic, Thota, Sankar C., Jencson, Cadnum, Ray, Salata, Watkins and Donskey15].
A meta-analysis by Dong et al. [Reference Dong, Zhou, Liu and Zhao16] has shown that PX-UV may be useful in reducing the incidence rate of infections with Clostridium difficile (CD) and methicillin-resistant Staphylococcus aureus (MRSA) but was not effective in reducing the rates of vancomycin-resistant enterococcal (VRE) infection. Likewise, Marra et al. [Reference Marra, Schweizer and Edmond17] pooled data on both types of UV technologies and found a statistically significant decrease in both CDI and VRE infection rates, but rates of MRSA and Gram-negative MDROs were unaffected. However, the latter analysis included only two studies utilising UV-C.
Recently, several studies have examined the effect of UV-C disinfection system on the rates of MDRO infections in hospitals. The current review aimed at synthesising the findings of all studies assessing the impact of UV-C room disinfection on reducing HCAI infection rates. In particular, the outcomes were related to the effect of UV-C disinfection on the risk of CDI, VRE, and Gram-negative multi-drug-resistant pathogens. The decision to focus only on UV-C disinfection systems (and exclude PX-UVL studies) was driven by the specific research question and objectives of the meta-analysis. The primary focus of our analysis was on disinfection systems that utilise continuous-wave UV-C light. PX-UVL systems differ from the latter in terms of the light source, technology, and application protocols. Nevertheless, the methods aim to achieve disinfection by damaging the genetic material of microorganisms. We considered that these differences made it challenging to directly compare the outcomes of PX-UVL evaluations with conventional UV-C systems.
Methods
Selection of studies
The review protocol was registered at PROSPERO (registration number CRD42023405885). Given its nature, oversight by an institutional board was not required before registration. The search strategy is presented in Supplementary Table S1. Three databases, that is PubMed, EMBASE, and Scopus, were screened for English language studies published up to 15 February 2023, in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, McDonald, McGuinness, Stewart, Thomas, Tricco, Welch, Whiting and Moher18]. The inclusion criteria were as follows: studies that had assessed the efficacy of an ultraviolet C (UV-C) disinfectant system for reduction in the incidence of MDRO infections; were conducted in all healthcare settings; of all sample sizes and types of patient population; and randomised controlled trials (RCTs), quasi-experimental studies, and before–after interventions. The exclusion criteria were as follows: those assessing the effect of PX-UVL; evaluations of the efficacy of UV-C in combination with other infection control measures; and reviews, case reports, and case series. After the removal of duplicates, the titles and abstracts of studies were independently screened by two of the study investigators for potential inclusion. The full texts of the remaining studies were then read, and final decisions for inclusion were made. A senior author was consulted in case of any discrepancies.
Data extraction and statistical analysis
Data were extracted using a pretested electronic template that consisted of variables related to study identifiers (author’s name, year of publication, study design, and country of study), type of health facility where the study was conducted, duration, and findings of relevance. Statistical analysis was conducted using STATA 16 software (Texas, USA). The pooled effect sizes were reported as an incidence rate ratio (IRR) with 95% confidence intervals (CI). We decided, a priori, to use a random-effects model for all analyses to account for potential variability such as characteristics of the hospitals studied, study design, and method of data collection. It was assumed that these differences would have led to substantial heterogeneity in the reported findings. A revised Cochrane risk-of-bias tool for randomised trials (RoB 2) was used [Reference Sterne, Savovic, Page, Elbers, Blencowe, Boutron, Cates, Cheng, Corbett, Eldridge, Emberson, Hernán, Hopewell, Hróbjartsson, Junqueira, Jüni, Kirkham, Lasserson, Li, McAleenan, Reeves, Shepperd, Shrier, Stewart, Tilling, White, Whiting and Higgins19]. For ‘before–after’ studies, the quality assessment tool developed by the National Heart Lung and Blood Institute [20] was used, and Egger’s test and funnel plots were used for the assessment of publication bias [Reference Egger, Smith, Schneider and Minder21]. P < 0.05 was statistically significant. A subgroup analysis was conducted based on the study design (i.e. RCT/quasi-experimental and before–after design), for the risk of CDI as an outcome.
Results
A systematic search across three databases identified 1380 studies. After the removal of the 455 duplicates, 925 unique studies remained. Screening based on their title and abstract led to the further exclusion of 900 studies. The full texts of the resulting 25 studies were screened, and an additional 16 were excluded (Figure 1) leaving nine studies [Reference Rock, Hsu, Curless, Carroll, Ross Howard, Carson, Cummings, Anderson, Milstone and Maragakis22–Reference Schaffzin, Wilhite, Li, Finney, Ankrum and Moore30] for this meta-analysis (summarised in Table 1). All studies were conducted in the USA and five were of a ‘before–after’ design, two were classed as quasi-experimental, and two were of a cluster-randomised crossover design. The included studies examined mainly CD, VRE, Gram-negative rod infection, and MRSA. Two studies specifically studied Gram-negative MDROs including Klebsiella, Acinetobacter, Pseudomonas, and E. coli. Schaffzin et al. [Reference Schaffzin, Wilhite, Li, Finney, Ankrum and Moore30] defined a Gram-negative as an isolate that was non-susceptible (intermediate or resistant) to at least one agent in at least three of nine antibiotic classes (anti-pseudomonal penicillin, third- or fourth-generation cephalosporin, carbapenem, fluoroquinolone, aminoglycoside, penicillin and beta-lactamase inhibitor, monobactam, polymyxin, and folate inhibitor). Five of the studies were conducted in academic medical facility or tertiary care hospitals, two in community hospitals, and one each in Veteran’s Health Administration (VHA) hospitals, and in a tertiary or community or VHA setting. The quality assessments for the cluster-randomised and quasi-experimental studies and for ‘before–after’ design studies are presented in Supplementary Figure S1 and Supplementary Table S2, respectively. It was concluded that most studies had a moderate risk of bias.
Figure 1. Selection process of studies included in the review.
Table 1. Characteristics of the studies included in the meta-analysis
Pooled analysis indicated no statistically significant reduction in CD infection rates with the use of ultraviolet C disinfection systems (IRR: 0.90, 95% CI; 0.62–1.32, I2 = 71.9%, N = 7) (Figure 2). Subgroup analysis based on the study design also showed no effect of this system on risk of CD infection for both RCT/quasi-experimental studies (IRR: 1.04, 95% CI; 0.72–1.50, I2 = 43.2%, N = 3) and ‘before–after’ design (IRR: 0.80, 95% CI; 0.40–1.59, I2 = 81.4%, N = 4) (Supplementary Figure S2). No publication bias was shown by Egger’s test (P = 0.728) and funnel plots (Supplementary Figure S3). Similarly, no significant reduction in VRE infection was observed (IRR 0.72, 95% CI; 0.38–1.37, I2 = 65.4%, N = 3) (Figure 3), and no evidence of publication bias was found (Supplementary Figure S4). UV-C systems appeared to reduce the risk of Gram-negative rod infection (IRR 0.82, 95% CI; 0.68–0.99, I2 = 0.0%, N = 2), but the number of studies reporting this outcome was small (Figure 4). No evidence of publication bias was found (Supplementary Figure S5).
Figure 2. Forest plot of IRRs of Clostridioides difficile (CD) infection for UV-C versus control.
Figure 3. Forest plot of IRRs of vancomycin-resistant enterococcal infection (VRE) for UV-C versus control.
Figure 4. Forest plot of IRRs of Gram-negative rod infection for UV-C versus control.
Discussion
Our study did not find any evidence of the benefit of using UV-C-based disinfectant systems in healthcare facilities to reduce the incidence of nosocomial infections, particularly MDROs such as CD and VRE. Nevertheless, some studies showed that UV-C (wavelength 200–270 nm) is effective in reducing the risk of Gram-negative rod infections by inducing DNA and RNA damage, through dimerisation of pyrimidine molecules, thereby reducing the replication of microorganisms [Reference Dai, Vrahas, Murray and Hamblin31, Reference Chang, Ossoff, Lobe, Dorfman, Dumais, Qualls and Johnson32]. UV-C at the wavelengths of 250–270 nm appears to be the most efficient due to its maximal absorption by microbial nucleic acids [Reference Yin, Dai, Avci, Jorge, de Melo, Vecchio, Huang, Gupta and Hamblin33]. However, one of the disadvantages of UV-C is that its penetration is affected by the presence of organic matter [Reference Otaki, Okuda, Tajima, Iwasaki, Kinoshita and Ohgaki34]. Additionally, there is an issue of costs and the requirement for the training of the personnel. We found no advantages to the use of UV-C in healthcare settings as an adjunct to conventional infection prevention modalities to reduce the incidence of MDRO.
The aspect of cost-effectiveness and staff training should also be considered further. Two previous studies that have conducted cost-effectiveness evaluation of ultraviolet disinfection systems after terminal cleaning [Reference Pegues, Han, Gilmar, McDonnell and Gaynes28, Reference Vianna, Dale, Simmons, Stibich and Licitra35] have shown the cost to be around 200,000 to 300,000 USD per year. However, on average, cases of CDI and VRE can lead to a cost of c. 14,000 USD/case [Reference Dubberke and Olsen36, Reference Cosgrove37]. Nursing professionals are essential for implementing disinfection protocols in healthcare facilities, as they are directly involved in the cleaning and disinfection of shared medical monitoring devices. It is therefore critical that any advancement in technology be known to them. In this regard, the current review further emphasises the importance of ‘no-touch’ disinfection systems in health facilities.
With the assumption that, in the included studies, the dosing and duration of exposure were appropriate, there may be some possible explanations for the lack of effectiveness of UV-C disinfectant systems. One reason could be the use of disinfectant as part of the standard protocol offered in both study groups. High compliance with disinfectants such as bleach or standard quaternary ammonium compounds may have led to relatively few residual spores for the UV-C device to eliminate. Second, UV-C disinfection relies on direct line-of-sight exposure to effectively kill microorganisms, and inadequately exposed surfaces to UV-C light could result in incomplete disinfection. Recent findings suggest that the role of the environment in the transmission of C. difficile may not be as significant as previously believed. Eyre et al. [Reference Eyre, Cule, Wilson, Griffiths, Vaughan, O’Connor, Ip, Golubchik, Batty, Finney, Wyllie, Didelot, Piazza, Bowden, Dingle, Harding, Crook, Wilcox, Peto and Walker38] examined 1250 isolates from cases of symptomatic CDI over four years using whole-genome sequencing. Surprisingly, possible environmental contamination was found to be responsible for linking only 2% of patients with genetically related C. difficile isolates. Furthermore, differences in study design, protocols, equipment, or the specific UV-C system most likely contribute to variations in effectiveness, as well as room size and layout, UV-C system placement, and operator training, which can influence the overall performance and outcomes of studies. On the contrary, Steele et al. [Reference Steele, Hurtado, Rychlik, Bonebrake, Bovee, O’Donnell, Perryman and Kociolek26] showed that UV-C irradiation reduced the risk of CDI in a paediatric haematology–oncology unit and suggested that this might depend on the pre-intervention CDI burden and that such high-risk units could have maximum benefit from use of such systems.
There are some limitations to our study. First, all reviewed studies were conducted in the USA, that is a high-income setting, and therefore, the generalisability of the findings is limited. Second, most of the included studies had a ‘before–after’ design, compared current data with a historical control group, and therefore did not consider changes in hospital practices that occur with time. Third, the included studies varied in terms of intervention sites and standard methods of disinfection used. For instance, some studies were conducted in cancer and solid organ transplant in-patient units where patients would have lower immunity and thus be at an increased risk of HCAI. Finally, our approach of limiting the study scope to English language studies could have introduced a degree of bias through missing relevant studies, as well as geographic, cultural, and publication bias.
Conclusion
The use of UV-C had no measurable impact on the incidence of CD and VRE infections but might be of some advantage in reducing the risk of Gram-negative rod infection, albeit with low confidence. Further studies to support or refute the outcome of this meta-analysis are needed.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0950268823001371.
Data availability statement
The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.
Author contribution
Conceptualization: Q.W., Y.S., Q.W.; Funding acquisition: Q.W.; Validation: Q.W., J.L., Y.S., Q.W.; Visualization: Q.W., Y.S., Q.W.; Data curation: J.L., Y.S., Q.W.; Formal analysis: J.L.; Methodology: J.L., Y.S., Q.W.; Project administration: J.L., Q.W.; Writing – original draft: J.L., Y.S., Q.W.; Writing – review & editing: J.L., Q.W.; Investigation: Q.W.; Resources: Q.W.
Competing interest
The authors declare none.
Open access
