Design and setting
This observational cross-sectional study was conducted on 15 wards of the Jewish General Hospital, Montreal, Canada, a 637-bed, tertiary medical centre. In 2013, 166 cases of healthcare-associated CDI occurred in our institution (incidence rate, 8.3/10,000 patient-days), due to a great extent to the high prevalence (55 %) of the epidemic NAP1/057 strain. Global hand hygiene compliance rate in 2013 was 47 % (1721/3635 opportunities). The hospital has 4 main types of accommodation for patient hospitalization: private rooms and 2-, 3- and 4-bedded rooms. Intensive care units (ICU) and maternity and psychiatry wards were excluded from the study as their configurations are very different from regular care units. On average, there was 1 sink per 10.7 beds available for HCWs on the study wards. The Jewish General Hospital Institutional Review Board approved the study as a quality improvement project.
Survey of sinks
We determined the location of sinks available to HCWs to hand wash after contact with patients with CDI. Institutional policy requires HCWs to wear gloves and perform hand washing at the closest available sink after contact with these patients or their immediate surroundings (also called the “patient zone”) [2]. In accordance with national recommendations, the use of patient personal sinks (i.e., those present at the bedside for their personal hygiene) is not permitted as they are likely to be contaminated with spores [8]. We used the infection control electronic database to identify the rooms that accommodated ≥1 patient with CDI between April 6, 2012 and March 31, 2013. An auditor then visited the premises and identified the nearest available sink outside of the patient zone and recorded the following information: location (e.g., hallway, clean utility, nursing station, medication room, etc.); distance from the patient zone; whether the sink was directly visible from the patient’s room; and if it was not directly visible upon exiting the patient room, the number of 90° turns required to reach the sink. Distances were measured with a measuring wheel in meters. As newly diagnosed patients are usually transferred to a single patient room following CDI diagnosis, all these data were recorded twice for every patient: 1) on the day of diagnosis of CDI, and 2) 24 h after diagnosis. This allowed to determine whether patient transfer to a private room had any impact on the proximity of sinks.
Handwashing compliance audits
From September to December 2013, an audit of HCWs’ hand hygiene compliance after CDI patient care was performed. This survey was conducted as part of our routine audits mandated by Accreditation Canada and performed by trained observers through direct unobtrusive observation according to the World Health Organization (WHO) recommendations [2]. Only opportunities after contact with CDI patients or surfaces within the patient zone (WHO, Moments 4 and 5; Canada, Moment 4) were audited [2]. Glove use, proper timing of glove removal (i.e., removal after leaving the patient zone and before touching any surface outside of the patient zone) and the type of hand cleansing method (rubbing or washing) were recorded. To decrease the Hawthorne effect, audits were limited to 30 min sessions and performed by a HCW who did not work in the areas where the audits were conducted [2].
Analysis
Standard descriptive statistics were used to calculate the proportion of room types, location of sinks, median distances, and the number of 90° turns between patient rooms and sinks. Medians and interquartile range (IQR) were used for non-normally distributed data. Geospatial characteristics at t = 0 h and t = 24 h were compared. Medians were compared using the Mann–Whitney test for unpaired variables and the Wilcoxon rank-sum test for paired samples. Categorical data were compared using χ
2 and Fisher’s exact test.
Predictors of hand washing compliance were analyzed by univariate and multivariate logistic regression. To adjust for potential confounders, all variables found to be associated with hand washing (P < 0.05) by univariate analysis were considered for inclusion in a multivariate model [9]. Five potential variables were identified: “glove use”, “direct visualisation of the sink”, “number of 90° turns”, “proper glove removal” and “distance to the sink”. The variable “direct visualization of the sink” was removed as it was strongly correlated with the variable “number of 90° turns”. All data related to a single respondent were excluded when any of the variables included in the model had missing values. Given that HCWs who did not wear gloves during patient care had a missing value for the variable “proper glove removal’, the variable “glove use” would have been uniformly positive in a multivariate model; it was thus excluded from the final model. Hence, we built a forced-entry model with the variables “proper glove removal”, “distance to sink”, and “number of 90° turns”. The magnitude of the association between outcomes and explanatory variables was measured by odds ratios (OR) and corresponding 95 % confidence intervals (CI). The number of 90° turns was treated as an ordinal variable for the univariate analysis to facilitate reporting, but as a continuous variable for the multivariate analysis. All tests were two-tailed and a P-value <0.05 was defined as statistically significant. Statistical analyses were performed with R (version 3.02; R Foundation for Statistical Computing; Vienna, Austria) and PASW Statistics version 18.0 (SPSS Inc. Chicago, IL).