Hand-hygiene compliance by hospital staff and incidence of health-care-associated infections, Finland

Abstract Objective To determine changes in hand-hygiene compliance after the introduction of direct observation of hand-hygiene practice for doctors and nurses, and evaluate the relationship between the changes and the incidence of health-care-associated infections. Methods We conducted an internal audit survey in a tertiary-care hospital in Finland from 2013 to 2018. Infection-control link nurses observed hand-hygiene practices based on the World Health Organization’s strategy for hand hygiene. We calculated hand-hygiene compliance as the number of observations where necessary hand-hygiene was practised divided by the total number of observations where hand hygiene was needed. We determined the incidence of health-care-associated infections using a semi-automated electronic incidence surveillance programme. We calculated the Pearson correlation coefficient (r) to evaluate the relationship between the incidence of health-care-associated infections and compliance with hand hygiene. Findings The link nurses made 52 115 hand-hygiene observations between 2013 and 2018. Annual hand-hygiene compliance increased significantly from 76.4% (2762/3617) in 2013 to 88.5% (9034/10 211) in 2018 (P < 0.0001). Over the same time, the number of health-care-associated infections decreased from 2012 to 1831, and their incidence per 1000 patient-days fell from 14.0 to 11.7 (P < 0.0001). We found a weak but statistically significant negative correlation between the monthly incidence of health-care-associated infections and hand-hygiene compliance (r = −0.48; P < 0.001). Conclusion The compliance of doctors and nurses with hand-hygiene practices improved with direct observation and feedback, and this change was associated with a decrease in the incidence of health-care-associated infections. Further studies are needed to evaluate the contribution of hand hygiene to reducing health-care-associated infections.


Introduction
Evidence shows that improving hand-hygiene practices reduces the frequency of health-care-associated infections in hospitals. 1,2 Studies have examined the association between hand-hygiene and health-care-associated infections: some were conducted over relatively short periods, 3,4 some focused on specific infections, such as central-line-associated infections or Staphylococcus aureus bacteraemia, [5][6][7][8][9][10][11] and some relied on mathematical models to predict the incidence of such infections. 3,9 Several studies have demonstrated improvements in handhygiene compliance after implementation of interventions to promote hand hygiene, 3,[9][10][11] but which individual measures are most effective or how to maintain long-term improvements is still unknown. 12 Despite promising new electronic systems for automatic monitoring of hand-hygiene compliance, these systems have notable limitations in real-world settings; in particular, they are expensive and require special methods, such as wireless technology. [13][14][15] These systems also typically only provide data on hand-hygiene compliance when entering or leaving patient rooms. In addition, very little evidence exists on the effectiveness of hand-hygiene improvements at reducing the incidence of health-care-associated infectionss. 12,15,16 We conducted a 6-year hospital-wide internal audit survey based on the World Health Organization's (WHO) model for hand hygiene 1 to evaluate hand-hygiene practices by direct observation by trained observers and provide immediate feedback. This method is considered the gold standard for monitoring hand-hygiene performance. 1,17 At the same time, we recorded the incidence of health-care-associated infections across the hospital using a semi-automated electronic incidence surveillance programme. We hypothesized that changes in hand-hygiene compliance observed during our survey would be reflected in the incidence of health-careassociated infections.

Research
Hand-hygiene compliance and health-care-associated infections, Finland Helena Ojanperä et al.

Observation of hand-hygiene
In January 2013, infection control link nurses began regular direct observation of hand-hygiene compliance at inpatient and outpatient locations to improve hand-hygiene compliance in health-care workers at the hospital. 19 All doctors and nurses treating patients at the hospital were the study population. The link nurses informed doctors and nurses when they were making observations of hand-hygiene compliance and explained that the observations were part of the hospital's quality evaluation process. In addition, the link nurses explained to patients that the observations were only on staff behaviour with the aim of improving professional practices.
The link nurses recorded information on the following variables during each observation: (i) duration of hand rubbing (in seconds); (ii) the observed moment according the WHO strategy (before touching a patient, before a clean or aseptic procedure, after touching a patient, after risk of exposure to a bodily fluid and after touching patient surroundings); 1 (iii) profession of the person observed (doctor or nurse); and (iv) the ward where the observation was made. The target number of observations was at least 10 observations per ward a month. This observation and reporting exercise requires about 4-6 working hours for each ward per threeweek cycle.
Between 2013 and 2016, the link nurses recorded hand-hygiene observations on paper (using stopwatches to measure timings) and transferred the data to an Excel spreadsheet (Microsoft, Redmond, United States of America). To reduce the time required to perform direct observations, the hospital's infection control unit, OYS TestLab, which offers a systematic approach to enable and support the development of health and medical technology products at Oulu University Hospital, and FCG Flowmedik Oy (Helsinki, Finland) developed a mobile device (the web-based eRub-tool) that facilitates the coding of observations. Therefore, since 2017, the link nurses have made their observations using this device. The reports on the total number of hand-hygiene observations and the type of hand-hygiene moment are made available immediately after the observation on the hospital's intranet. Data gathered between 2013 and 2016 were also transferred to the eRub-tool.
The implementation costs of this study were negligible because the hospital had established the link nurse programme several years before this survey began. Infection-control link nurses spent one working day every 3 weeks observing hand hygiene and checking all registered antibiotic prescriptions following patient discharge. Since our hospital was a pilot hospital where the web-based eRub-tool was developed, the use of this system was free of charge for our hospital.

Use of handrubs
We obtained data on the yearly use of alcohol-based handrub in litres per 1000 patient-days from the hospital's financial records. We determined the number of 1000 patient-days per month or per year by summing the corresponding number of overnight stays made by patients at the hospital and dividing the total by 1000.

Hand-hygiene compliance
We calculated hand-hygiene compliance as the number of observations where the necessary hand hygiene was practised divided by the total number of observations where hand hygiene was needed. We also extracted the duration of hand rubbing as recorded in the eRub database. We calculated compliance with the hand-rubbing time on a monthly basis over the 6-year study period. The length of hand rubbing recommended by WHO is 20-30 s. 1 Depending on the size of hands we recommended one or two measures of handrub from the dispenser; so the approximate volume of handrub was 1.6 mL or 3.2 mL).
At the individual level, the link nurses gave verbal feedback on handhygiene performance immediately after the hand-hygiene observation. At the group level, feedback was given during regular ward meetings to the personnel on the wards by the link nurses and infection control nurses. At the organization level, the hand-hygiene compliance results were available to all staff of the hospital who have access to the hospital's intranet.

Health-care-associated infections
We determined the incidence of healthcare-associated infections by analysing the hospital's records. The hospital uses a semi-automated electronic incidence surveillance programme that is linked to all the electronic databases of the hospital. 20 When an antibiotic is added to the patient's prescription, the programme automatically opens an enquiry form that doctors have to complete. Doctors are required to indicate whether the antibiotic was prescribed to treat a healthcare-associated infection acquired in the hospital or a community-acquired infection. In each ward, two infectioncontrol link nurses checked all of the registered antibiotic initiations during the hospital stay after patient discharge. In our hospital, health-care-associated infections are classified according to a modified version of the criteria proposed by the United States Centers for Diseases Control and Prevention. 21 We calculated the incidence of health-careassociated infections per 1000 patientdays on a monthly or annual basis over the study period.

Analyses
We did not calculate a sample size before the survey. Our goal was to obtain at least 50 000 hand-hygiene observations, which we considered sufficient for comparison of monthly hand-hygiene Hand-hygiene compliance and health-care-associated infections, Finland Helena Ojanperä et al.
compliance figures with the incidence of health-care-associated infections. We used SAS, version 9.4 (SAS Institute, Cary, USA) for all analyses. We calculated the monthly change in incidence of health-care-associated infections (from 1 May 2013 to 31 December 2018) using a Poisson regression model. We give the results of the Poisson regression analysis as incidence rate ratio. We calculated the Pearson correlation coefficient (r) to evaluate the relationship between the incidence of health-care-associated infections and compliance with hand hygiene. For each year of the study, we calculated the mean monthly frequency of hand-hygiene compliance and 95% confidence interval (CI). We calculated the median and 25th and 75th centiles of rubbing times.

Ethical considerations
The Hospital District Medical Director and Chief Nursing Officer of Oulu University Hospital approved this audit survey (registration number 246/2018). In Finland, the Medical Research Act (no. 488/1999) states that approval from the local ethics committee is not required for register-based surveys that do not process identifiable information. We consulted the secretary of the Regional Ethics Committee of the Hospital District who confirmed that our study was conducted in accordance with all applicable research regulations and norms of Finland.

Results
Between May 2013 and December 2018, the link nurses made 52 115 observations where hand hygiene was needed (Table 1). In the past 3 years of the survey (2016-2018), these nurses made more than 10 000 observations each year.
T h e a n n u a l h a n d -h y g i e n e compliance increased from 76.4% (2762/3617) to 88.5% (9034/10 211; P < 0·0001; Table 1) and the monthly hand-hygiene compliance increased from 77.5% (328/423) in May 2013 to 94.4% (456/483) in December 2018 ( Table 2 and Fig. 1). The median hand-rubbing time over the 6 years was 21 s, with the 25th and 75th centile times being 13 s and 30 s, respectively. The consumption of alcohol-based handrub was 57 L per 1000 patientdays in 2013 and 74 L per 1000 patient-days in 2018.
The number of health-care-associated infections decreased from 2012 infections in 2013 to 1831 in 2018 and the incidence of health-care-associated infections decreased from 14.2 (95% CI: 13.4-14.9) to 11.7 (95% CI: 11.1-12.2) per 1000 patient-days (incidence rate ratio for monthly change = 0.999, P < 0.0001; Table 1 and Fig. 1). As shown in Fig. 2, we found a weak but statistically significant negative correlation between the monthly incidence of health-care-associated infections and hand-hygiene compliance (r = −0.48; P < 0.001).

Discussion
Our results show that during the 6-year period of regular hand-hygiene observations and immediate feedback, the frequency of annual hand-hygiene compliance of doctors and nurses improved from 76.4% in 2013 to 88.5% in 2018 (P < 0.0001). At the same time the hospital-wide incidence of healthcare-associated infections decreased significantly.
Two previous studies have examined the association between hand-hygiene compliance and all hospital-wide health-care-associated infections. The first study showed that over a 4-year period when hand-hygiene improved from 47.6% (1349/2834) to 66.2% (1701/2569), the prevalence of healthcare-associated infections decreased from 16.9% to 9.9%. 2 The second study examined hand-hygiene compliance after implementation of an infectioncontrol programme and health-careassociated infections in both general wards and an intensive care unit. 4 Over a 16-month follow-up period, hand-hygiene compliance increased from 41.0% (2235/5454) to 50.5% (3246/6428). Over the same time, the incidence of health-care-associated infections in the general ward was unchanged, but the number of severe health-careassociated infections in the intensive care unit decreased. 4 In our study, when monthly handhygiene compliance was more than 80.0% over 2 years, the incidence of health-care-associated infections started to decrease (Fig. 2). In earlier studies where health-care-associated infections decreased, hand-hygiene compliance was at least 66.2% at the end of the study (number of opportunities = 2569 during the last observation period, December 1997) 2 and the duration of the study period was at least 17 months. 2,5,6,9,22 In a study with exceptionally high initial hand-hygiene compliance of 82.6%, compliance increased to 95.9% while the rate of health-care-associated infections fell by 6.0% during the 17-month study period. 22 Further reductions in the incidence of health-care-associated infections at our hospital could therefore be possible if the annual hand-hygiene

Research
Hand-hygiene compliance and health-care-associated infections, Finland Helena Ojanperä et al.
compliance rate at the hospital were raised above 90.0%. We think that the increased compliance with hand hygiene is the most plausible explanation for the reduced incidence in health-care-associated infections in our study because the infection control practices applied at the hospital had remained largely unchanged in the years before the study. Although we cannot completely exclude the possibility that other factors may have contributed to the decline in health-care-associated infection rates, no other hospital-wide infection prevention strategies were introduced during the study period. However, the negative correlation between health-care-associated infections and hand-hygiene compliance was relatively weak (r 2 = 0.23), suggesting that only 23% of the observed variation in the incidence of health-care-associated infections was related to changes in hand-hygiene practices. Therefore, some unidentified confounding factors could have contributed to the decrease in health-care-associated infections.
Previous reports have emphasized the need for multimodal approaches to achieve and sustain permanent improvements in hand-hygiene compliance. 1 We believe that this is probably true when starting a hand-hygiene programme in a hospital with a low level of hand-hygiene compliance, as was demonstrated in an earlier hospital-wide compliance project. 2 However, our results indicate that direct observation and immediate feedback on hand-hygiene procedures can induce a sustained increase in hand-hygiene compliance even when the annual compliance rate is relatively high to begin with (76.4% in our case). This relatively high initial compliance rate was a result of years of intensive efforts to improve hand-hygiene practices. Before this survey, four of the five components of the WHO's multimodal hand-hygiene guideline 1,23 had already been implemented at the hospital. However, this policy by itself did not reduce the incidence of health-care-associated infections to the levels observed after introducing regular observation and feedback. The important role of performance feedback in promoting and sustaining good hand-hygiene behaviour in hospital health-care workers has also been highlighted in previous reports. 24 Direct observation makes it possible to assess compliance rates for all of the WHO hand-hygiene moments, and is seen as the gold standard for monitoring hand-hygiene compliance. However, direct observation is time-and resource-intensive, and can only be done in a small proportion of hand-hygiene opportunities. 14 Furthermore, the risk of bias due to the Hawthorne effect can occur as health workers may improve their practices when under observation. 25,26 Regardless of any possible Hawthorne effect, to improve your behaviour when you are observed, the infection-control link nurses made regular hand-hygiene observations in their own wards in a similar way over several years and gave feedback to their co-workers. We think that the Hawthorne effect might have had a positive influence on our results; it has earlier been shown that the Hawthorne effect can be used to encourage hand-hygiene compliance. In an ideal situation, the Hawthorne effect would be sustained with continuous observations improving hand-hygiene compliance and decreasing numbers of health-careassociated infections. 27 Our study has several limitations. First, it is a non-randomized real-world internal audit survey conducted in a single university hospital in Finland. Further studies in other types of hospitals and countries will be needed to test the generalizability of these findings. Second, the link nurses only made hand-hygiene observations during the day shifts on week days, so the findings may not apply to night and weekend shifts. Third, our analysis is based only on hand-rubbing time, without the evaluation of rubbing technique. The observed median rubbing time of 21 s is acceptable because WHO recommends rubbing for 20-30 s. 1 However, some of the staff observed did not spend enough time on rubbing because the 25th centile rubbing time was only 13 s. In addition, in the last year of the study (2018), no handrubbing was done at all in 11.5% (1177/10 211) of the occasions when hand hygiene was needed. Fourth, our survey did not assess whether the hand-hygiene technique was properly performed or if gloves were used. Fifth, we cannot rule out possible confounding factors (e.g. seasonal influenza, norovirus outbreaks in the community or holiday periods of health-care workers) that may have affected handhygiene compliance or the incidence of health-care-associated infections. Sixth, because we were interested only in hospital-level changes, we do not know which health-care-associated infections (e.g. surgical site infections or hospitalacquired pneumonia) decreased more after increased hand-hygiene compliance. We also did not compare hand-hygiene compliance between doctors and nurses or in different hospital wards.
A strength of this survey is the fact that observations of hand-hygiene compliance and health-care-associated infection surveillance were done regularly on a monthly basis over several years in a real-life setting. Health-care-associated infections were followed using a semiautomated electronic surveillance system. 20 Implementation of electronic surveillance is considered a feasible way to identify health-care-associated infections. 28 20 We were not able to obtain the number of discharged patients as in this ear-lier study, so could report incidence of health-care-associated infections per 1000 patients-days. These health-careassociated infections were recorded using the same semi-automated electronic surveillance system used during the survey. The introduction of the eRub-tool made easier to observe hand-hygiene and saved the link nurses time. Similarly, positive outcomes have been achieved using mobile tools for hand-hygiene observation in other studies. 5,29 Further studies are needed to evaluate the contribution of hand-hygiene techniques to reducing the incidence of health-care-associated infections and to determine which health-care-associated infections are most effectively prevented by improving hand-hygiene compliance.

Resumen
Cumplimiento de la higiene de las manos entre el personal de los hospitales y la incidencia de las infecciones relacionadas con la atención médica, Finlandia Objetivo Determinar los cambios del cumplimiento de la higiene de las manos tras las observaciones directas en las prácticas de higiene de las manos que realizan los médicos y las enfermeras, además de evaluar la relación entre los cambios y la incidencia de las infecciones relacionadas con la atención médica. Métodos Se realizó una encuesta de auditoría interna en un hospital de atención terciaria en Finlandia de 2013 a 2018. Las enfermeras de control de infecciones observaron las prácticas de higiene de las manos según la estrategia de la Organización Mundial de la Salud para la higiene de las manos. Se estimó que el cumplimiento de la higiene de las manos equivalía a dividir el número de las observaciones en las que era necesario practicar la higiene de las manos por el número total de observaciones en las que se requería la higiene de las manos. Se determinó la incidencia de las infecciones relacionadas con la atención médica mediante un programa semiautomatizado y electrónico de vigilancia de la incidencia. Se calculó la correlación de Pearson (r) para evaluar la relación entre la incidencia de las infecciones relacionadas con la atención médica y el cumplimiento de la higiene de las manos.