Hospital water systems are recognized as sources of healthcare-associated infections. Waterborne outbreaks are a threat to high-risk patients and the risk factors have been described in a previous systematic review,Reference Kanamori, Weber and Rutala1 but this association has never been quantitatively estimated. We systematically reviewed the scientific literature to determine the types of exposure that represent a risk to patients on high-risk units and to quantitatively estimate their risk.
Methods
We systematically reviewed the scientific literature with the primary objective of identifying risk factors for the acquisition of waterborne HAI or colonization in high-risk units. Our data sources were PubMed, Web of Science, and Embase for the period extending from 1966 to July 10, 2018. Water supplies and sources included pipes and plumbing fixtures, with a focus on taps and sinks. Three independent reviewers screened and selected the clinical studies, and 2 authors extracted data; disagreements were resolved by discussion. If consensus was not reached, a third reviewer was consulted. The review exclusively considered randomized controlled trials (RCTs), non-RCTs and analytical observational studies. We excluded articles related to Legionella pneumophila and Staphylococcus aureus.
We used the Newcastle-Ottawa Scale (NOS) and the Grading of Recommendations Assessment Development and Evaluation (GRADE) system to assess the overall quality of evidence. An NOS score ≥7 indicates a higher study quality; scores range from 0 to 10 for both case-control and cohort studies. Two reviewers assessed risk of bias in all reviewed studies. We grouped studies depending on populations (ie, adult, pediatric, and neonatal populations) and study type.
Clinical and methodological heterogeneity was evident across studies, so we used a random effects model. We obtained summary estimates using a form of inverse variance-weighted averages of logarithmic odds ratios (ORs). We evaluated the heterogeneity among studies using the Q test (Cochran’s χ2 test) and the Higgins I 2 statistic (range, 0–100%). I 2 values were interpreted as low (<25%), moderate (25%–50%), and high heterogeneity (50%–75%). We did not assess potential publication bias due to the small number of studies included. We used Review Manager software (Revman 5.3; Cochrane Collaboration, Oxford, UK) to draw forest plots and perform statistical analyses.
Results
We identified 301 potentially relevant studies and selected 10 eligible studies for review (Supplemental Fig. 1 online). The characteristics of the studies and NOS scores are displayed in Table 1.
Table 1. Baseline characteristic of study participants of case-control and cohort studies examining waterborne healthcare associated colonization and/or infections
Note. NICU, neonatal intensive care unit; SICU, surgical intensive care unit; MICU, medical intensive care unit; NOS, Newcastle-Ottawa scale.
Neonatal population
Three case-control studies reported risk factors related to the water system as the reservoir and acquisition of waterborne-HAI or colonization in neonates.Reference Von Dolinger Brito, Matos, Abdalla, Filho and Pinto2–Reference Kinsey Bicking, Koirala and Solomon4 Waterborne HAI and colonization were linked to contaminated tap water and the hospital wastewater system. Gram-negative bacilli (Pseudomonas aeruginosa and Serratia marcescens) were the most frequently identified pathogens.
The risk factors reported in the reviewed studies included care in a room with no point-of-use filter installed on the sink faucet during the 7 days before positive culture (OR, 37.55; 95% confidence interval [CI], 7.16–196.93)Reference Kinsey Bicking, Koirala and Solomon4, invasive ventilation (OR, 5.79; 95% CI, 1.29–30.62) after adjustment for gestational age,Reference Kinsey Bicking, Koirala and Solomon4 and exposure to humidity (OR, 1.87; 95% CI, 0.47–7.91) (Supplemental Fig. 2 online).Reference Kinsey Bicking, Koirala and Solomon4
Two case-control studies addressed the association between central venous catheter (CVC) exposure and waterborne HAI or colonization.Reference Maltezou, Tryfinopoulou and Katerelos3, Reference Kinsey Bicking, Koirala and Solomon4 The summary OR was 8.02 (95% CI, 2.19–29.31), with low heterogeneity (I 2 = 0%). CVC exposure had an attributable fraction for waterborne HAI or colonization acquisition of 35.5% (95% CI, 31.8%– 49.4%). The summary OR for exposure to an umbilical catheter was 2.41 (95% CI, 0.20–28.33).Reference Maltezou, Tryfinopoulou and Katerelos3, Reference Kinsey Bicking, Koirala and Solomon4 Two case-control studies assessed the association between total parenteral nutrition (TPN) and waterborne HAIsReference Maltezou, Tryfinopoulou and Katerelos3, Reference Kinsey Bicking, Koirala and Solomon4; the summary OR was 11.52 (95% CI, 0.55–240.94) (Fig. 1).
Fig. 1. Forest plot of types of exposure related to waterborne-HAIs in Neonatal population admitted to NICU
a. Central vascular catheter exposure
b. Umbilical catheter exposure
c. Total parenteral nutrition exposure
Adult population admitted to ICU
a. Age (years)
b. Length of stay overall (days)
CI: confidence interval, IV: interval variable
Adult population
We identified 5 cohort studiesReference Berthelot, Grattard and Mahul5–Reference Zhou, Hu, Gao, Bao, Chen and Li9 and 2 case-control studiesReference Dandalides, Rutala and Sarubbi10, Reference Salm, Deja and Gastmeier11 that reported plausible evidence of risk factors related to waterborne HAI or colonization in adults. Waterborne HAI and colonization were linked to contaminated tap water and the hospital wastewater system (ie, soiled utility sink and sinks in patient rooms). Gram-negative bacilli (eg, Pseudomonas aeruginosa, Serratia marcescens, Klebsiella pneumoniae, Proteus mirabilis, and Enterobacter spp) were the most frequently detected pathogens.
Environmental exposures associated with acquisition were contaminated tap water in patient rooms (hazard ratio [HR], 1.76; 95% CI, 1.09–2.84)Reference Reuter, Sigge, Wiedeck and Trautmann6 and stay in a room with a contaminated sink (OR, 11.2; 95% CI, 1.92–65.68).Reference Salm, Deja and Gastmeier11 HemofiltrationReference Salm, Deja and Gastmeier11 (OR, 22.9; 95% CI, 1.63–293.91) and cumulative duration of mechanical ventilation for ≥10 days (HR, 2.56; 95% CI, 1.46–4.50)Reference Venier, Leroywe and Slekovec8 were also reported as a risk factors for acquisition (Supplemental Fig. 3 online).
A retrospective case-control studyReference Salm, Deja and Gastmeier11 examined the association between immunosuppression and the risk for colonization or infection by the outbreak strain (OR, 7.87; 95% CI, 0.942–65.73), and only 1 prospective study reported duration of antibiotic treatment associated with a decrease of risk of P. aeruginosa (OR, 0.78; 95% CI, 0.69–0.87).Reference Berthelot, Grattard and Mahul5
Two case-control studies contributed to a meta-analysis. The length of stay before colonization was also identified as a risk factor (standardized mean difference [SMD], 0.65; 95% CI, 0.26–1.05; P = .01) for case-control studies (Fig. 1).
We could not calculate summaries of other exposures mentioned in previous articlesReference Kanamori, Weber and Rutala1 because they were not analytical observational studies in this regard.
Quality of evidence
The quality of evidence as defined by GRADE rating system for every exposure was low, depending on the risk factor (Supplemental Fig. 4 online).
Discussion
The principal contaminated reservoirs of water system reported and included in this systematic review were the distal ends (ie, tap water) and sink drains.
The absence of protective barriers installed on the distal ends of a plumbing installation could play an important role for microorganism dispersion and colonization or infection by a waterborne microorganism.
We highlight the relevance of these aqueous reservoirs not only for the identification of several enterobacteria and nonfermentive gram-negative bacilliReference Roux, Aubier, Cochard, Quentin and van der Mee-Marquet12 but also because most outbreaks due to contaminated sinks are related to multidrug-resistant organisms.Reference Berthelot, Grattard and Mahul5 Therefore, barriers that reduce contamination of the areas around the sink by splashes from the drain should be considered to prevent patient or device contamination.
Although exposure to invasive procedures was associated with acquisition, these life support systems could act not only as risk factors but also as markers to identify more vulnerable neonates. Longer exposure to the ICU environment, especially if sources of water contamination are present, facilitates the acquisition of waterborne pathogens. Staying in a room where the sink or the tap water are contaminated was the most relevant environmental exposure. These findings are consistent with previous studies reporting that the distance between the sink and the patient bed was very often between 1 and 2 m, so close that it allows water splashing around the sink to reach patients.Reference Roux, Aubier, Cochard, Quentin and van der Mee-Marquet12
Quantifying patient risk factors for waterborne-associated HAI or colonization in high-risk units could help clinicians estimate patient risk for acquisition and evaluate who needs enhanced prevention measures while admitted to these units. Exposures related to waterborne infection or colonization should give rise to the implementation of measures to strengthen key strategies for prevention of invasive device–associated HAIs. Improvement of hand hygiene, avoiding indwelling invasive device insertion, early removal of and/or alternatives to indwelling invasive devices should be considered.
The general limitations of our study include the small number of studies included, their clinical and methodological heterogeneity, and some potential publication bias that may not have been detected.
In conclusion, CVC exposure for neonatal patients and length of stay for the adult population are risk factors for waterborne HAI or colonization in high-risk units.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2019.326
Acknowledgments
None.
Financial support
No financial support was provided relevant to this article.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
