Methods

We conducted a before-and-after, multimodal study over a 6-month period in five hospital units in Tuscany to improve hand hygiene. The study included direct observation of clinicians, identification of physician and nurse champions, and an intense 2-week educational intervention. This intervention was developed in collaboration with healthcare staff, including directors of the hospital units involved. We retained the same six external observers (nursing students) who observed and recorded the data during the before period of our previous study for this study as well, as these individuals were uniquely positioned to notice the circumstances around non-compliance with hand hygiene for the healthcare workers being studied. Kampf’s work,16 as we explain in the discussion, also informed the development of the intervention package.

The intervention included several components. First, we presented information regarding current hand hygiene practice on the units to both physician and nursing leaders to reinforce their vigorous support for making a change. Second, we educated nurses and doctors about infection and hand hygiene through seminars and individualised educational sessions, focusing on practice before first contact with patients. In general, a doctor taught doctors, while a nurse taught nurses. The education emphasised the use of alcohol-based hand rub as a suitable substitute for soap and water in most instances. Third, we asked doctor and nurse champions on the units to wear green buttons that read “Chiedimi se mi sono lavato le mani” (“Ask me if I have washed my hands”). Our point was to make the issue of hand hygiene salient. These same champions were also asked to model proper hand hygiene behaviour. Fourth, we supplied personal use bottles (100 ml) of alcohol-based hand rub to all doctors and nurses to be kept in their pockets. The nurse manager of each unit kept hand rub supplies restocked and was also responsible for ensuring that additional hand rub was available in every place where doctors and nurses had patient contact. Finally, we created a “sense of urgency”17 to prevent delay in adoption by asking the doctor and nurse managers on each unit to implement a plan to improve hand hygiene over weeks rather than months.

The same six students from the nursing school of the University of Florence who observed doctor–patient and nurse–patient interactions in the previous study did so for this one. As in the previous study, the first author (SS), who was on sabbatical in Tuscany, also performed observations. We considered any hand hygiene effort, regardless of duration or technique, a “pass.” The original training included a 2-h session on how to monitor hand hygiene adherence and background information on healthcare-associated infection and hand hygiene. In addition, background information on the study objective was provided, together with materials from WHO (used in its hand hygiene campaign), including a modified observation form. The form used in the study included the name of the observer, start and end time of the observation period, whether each observed subject was a physician or nurse, and what type of hand hygiene was practiced before patient contact (soap and water, hand rub, or neither). Observers were given strict instructions not to interfere with the clinical care being delivered. If they were asked to leave for medical reasons or reasons related to confidentiality, they complied immediately. Healthcare workers participating in our study were aware that they were being observed to monitor infection control practices; workers were generally aware that hand hygiene practices were a focus of the observations.

The before period of observation occurred from January to March 2008. We implemented the intervention over the first 2 weeks of April 2008 (to coincide with the date of the Italian elections to ensure the date was well known to all healthcare workers). The after period of observation took place in April, May and June (observations that occurred during the implementation were considered to be in the after period).

Two individuals who had access to the paper observation forms entered data into a Microsoft Access database designed specifically for this purpose. Study staff assessed the data for entry errors, resolving discrepancies. Study staff performed data analyses in SAS V.9.1 and Computer Programs for Epidemiologic Analysis (PEPI, USD, Stone Mountain, Georgia, USA). The first set of analyses grouped all units and both clinician types together for a baseline comparison. Then, staff performed sub-analyses for all doctors and all nurses separately, followed by unit-specific analyses, first for each unit as a whole, then for each clinician type separately. When we report a percentage (eg, in table 1), the numerator consists of the number of instances of hand hygiene (one per clinician-patient interaction), and the denominator the total number of clinician-patient interactions. We used Pearson χ² tests to assess associations for binomial data, with α set at 0.05 (two-tailed).

The first author also conducted in-person, unstructured interviews with either the doctor or the nurse leader from each unit (or their representatives). He also spoke with the nursing student observers as well as several front-line doctors and nurses from several of the units to better understand the variability of hand hygiene improvement. He interviewed a total of 17 individuals and took notes during these discussions; comments from interviewees, however, were not tape recorded.

Ethics committees from the two participating hospitals reviewed and approved this study.

Results

During the 6 months of the observation period, we observed a total of 3987 clinician-patient interactions, 1526 doctor-patient observations and 2461 nurse–patient interactions. The seven external observers performed a total of 173 observation sessions, with the mean observation time being 36 min. An average observation period recorded 23 interactions. On average, each observer participated in 25 observation sessions.

The results of before-and-after comparisons are shown in table 1, where all differences are reported in terms of absolute percentages. Overall healthcare worker hand hygiene increased from 31.5% to 47.4% (p<0.001); adherence among nurses increased from 33.7% to 47.9% (p<0.001), whereas among doctors it increased from 27.5% to 46.6% (p<0.001). Improvement was statistically significant in three of five units; in the remaining units, the change was in the expected direction. However, units differed considerably in the magnitude of their improvement, from a low of 3.5% (p = 0.27) to a high of 30.6% (p<0.001).

Although both doctors and nurses improved, the improvement was greater for doctors. In the before period, doctors were less likely than nurses to clean their hands.15 In the after period, given the improvement by doctors, there was no longer a statistically significant difference between doctors and nurses (p = 0.56). The absolute percentage increase in hand hygiene was 15.9% overall, 14.2% for nurses and 19.1% for doctors. The relative percentage increase was 50.5% overall, 42.1% for nurses and 69.5% for doctors.

The use of alcohol-based hand rub as compared with washing with soap and water or doing nothing increased overall from 8.4% to 17.0% (p<0.001) in absolute terms. For nurses, the increase was from 6.5% to 11.4% (p<0.001), and for doctors, from 11.7% to 25.4% (p<0.001). If we limit the data just to the interactions where one of the two modalities of hand hygiene was practiced, the use of hand rub, as compared with soap and water, increased from 26.7% to 35.8% (p<0.001). Whereas the increase for nurses (19.1% to 23.8%) was not statistically significant (p = 0.08), the increase for doctors (42.6% to 54.6%) was significant (p = 0.007).

The correlation between the compliance for nurses and doctors in the same unit in the after period was 0.61. The correlation between improvement for nurses and doctors in the same unit (before vs after) was 0.66.

The interviewees perceived that the ability to identify a motivated champion (either nurse or doctor) was an important determinant for the intervention’s success. In addition, interviewees perceived that the extent to which the nurse or physician leader of each unit was an effective and engaged leader also generally predicted whether or not the intervention was successful. Finally, the interviewees related the extent to which the physician leader appeared to be motivated by the preintervention results (unit-specific and overall preintervention results were shared with each unit’s physician director) to the ultimate result after the intervention.

Discussion

Our multicentre interventional study in Florence, Italy, included observations from almost 4000 encounters between clinicians and patients. Before the intervention, the overall rates of hand hygiene adherence among nurses were comparable to those found when the Tuscan Regional Health Authority began a hand hygiene campaign 3 years ago.3 Focusing on overall hand hygiene rates, however, might be misleading given the substantial unit-level variability that appeared to exist. After the multimodal intervention, overall healthcare worker adherence with hand hygiene before patient contact increased by almost 16% in absolute terms but, again, with substantial variability. Despite the overall improvement, however, we found that healthcare workers in our two hospitals still did not use hand hygiene more than half the time before touching the patient. It is possible, however, that a more sustained intervention or one that was reinforced at intervals may have led to a more dramatic improvement.

Semmelweis recognised the importance of hand hygiene in preventing nosocomial infections more than 100 years ago,18 and improving hand hygiene practice appears to reduce healthcare-associated infection and the transmission of infectious pathogens between staff and patients.18 19 20 21 22 23 It is surprising, therefore, that despite this accumulated evidence, adherence to hand hygiene recommendations is so low, often <50%.9 24 25 26

Promoting hand hygiene is a complex issue. Many simple low-cost interventions based on ergonomic principles (eg, enhancing the visibility and reducing the difficulty of access to sinks)27 have been used to attempt to improve hand hygiene. Kampf16 has suggested six rules to improve compliance in hand hygiene: (1) select an alcohol-based hand rub that has good skin tolerance and is acceptable to healthcare workers; (2) ensure that this hand rub is easily available; (3) implement educational interventions promoting hand hygiene; (4) create a budget that covers all costs relevant to preventable nosocomial infection; (5) ensure that senior staff set a good example for junior staff; and (6) ensure an appropriate patient–staff ratio.

We generally incorporated these rules into the multimodal intervention used. We also incorporated suggestions from the observers in the before period of the study. Specifically, observers encouraged us to promote the wide availability of alcohol-based hand rub, including delivering it to all the healthcare workers in the form of a personal bottle that could be placed in the pocket. We realised that relying on hospital-based maintenance workers to affix hand hygiene dispensers on the walls of the hospital corridors and/or patient rooms would not be possible given the timeframe of our study.

As we discussed in our previous study,15 although alcohol-based hand rubs existed in the units, they were not often used. Nurses, in particular, used them minimally. Informally, we were given various explanations for this, including the unclear benefit of alcohol-based hand rub compared with soap and water, inadequate availability of alcohol-based hand rub in the facility (sometimes because the dispenser was not filled when it became empty), allergic reactions to the hand rub and the fear that alcohol-based hand rub was carcinogenic.15 Although an educational component of the intervention was obviously critical to enhance healthcare workers’ knowledge, we suspected that a single intervention based on short teaching sessions was unlikely to be successful for even a short time. This result led us to develop a multimodal intervention.

We also focused on the availability of the alcohol-based hand rub. Limited accessibility is well demonstrated as one of the risk factors associated with poor compliance; the most successful strategies aim to improve accessibility to hand hygiene agents.28 Because easy access to a hand hygiene agent is strongly associated with physician adherence,29 we considered various changes to enhance availability, including giving the responsibility for restocking to the nurse managers.

Finally, we attempted to convey what Kotter17 refers to as “a sense of urgency.” He notes that company executives commonly fail to motivate employees because they do not adequately emphasise the unpleasant facts and serious consequences related to business as usual. Executives also tend to underestimate the difficulty of motivating people to change. As Kotter points out, one way to enhance the likelihood of success is to bring in outsiders who communicate a “big picture” rationale for the changes that are needed. In fact, the presence of an American researcher (SS) served as the kind of “outsider” that Kotter recommends. Furthermore, we processed data from the before period soon after it was collected so that we could provide early feedback on the “unpleasant facts” regarding hand hygiene.

There is a considerable body of research on multimodal/multifaceted strategies; however, an exhaustive review of this literature is beyond the scope of this article. In brief, multimodal strategies have been assessed in both infectious and non-infectious problems with evidence of some benefit.30 31 32 33 34 Of particular relevance to our work is both Pronovost’s study of vascular catheter “bundles,” which included hand hygiene as part of the intervention,35 and Pittet’s multimodal hand hygiene study.9 We used such an approach as well, basing it largely on the WHO hand hygiene campaign.13

Pittet’s study, like our own, had an educational component, visual reminders (posters), individual bottles of hand rub and carefully built institutional commitment. Unlike our study, Pittet’s occurred over a period of 3 years. Compliance in Pittet’s study improved from 48% to 66% (an 18% absolute increase) during the study period. This result is similar to the 15.9% improvement we found. Interestingly, Pittet also found that nursing staff compliance improved but physician compliance did not. In contrast, we found that improvement occurred both for nurses (14.2%) and for doctors (19.1%). Furthermore, as these percentages show, doctors improved more than nurses. Pittet found that doctors (who did not improve overall) switched from soap and water to hand rub. We found as well that doctors increased their use of hand rub. Interestingly, in our study, nurse compliance improvements resulted mostly from increased use of soap and water.

The apparent effect of our intervention in the cardiology unit needs some comment. From the data we can surmise that there was no unit-specific effect at play that discouraged compliance, as nurses improved markedly. Although we cannot definitely explain why the hand hygiene adherence rate among the cardiologists in this particular hospital was so low (and remained so after the intervention), our interviews were instructive. We believe that physician performance was related to a lack of a committed physician champion on this unit. There seemed to be relatively little normative pressure exerted on the cardiologists to motivate them to practice good hand hygiene.

Our findings should be interpreted in the context of certain limitations. First, we focused on only one of the “five moments” of hand hygiene (which represent the five key times during patient care when appropriate hand hygiene is important in avoiding the spread of infection).36 However, we believe that this moment is a critically important one from the patient’s perspective. In addition, healthcare workers are likely to be more diligent in washing their hands after patient contact as a mechanism of self-protection against harmful organisms. Second, we used direct observation to collect data. Direct observation carries the risk of changing the observed subject’s behaviour, at least in the short term. Specifically, it has been shown that the awareness of being observed is strongly associated with adherence.37 However, we used this same approach in both the before and after parts of this study. Therefore, the differences we observed are less likely to be an artifact of the observation itself. Despite this, it is still possible that hand hygiene, having been made more salient, led clinicians to be more attentive to the observers than they were in the before period. In addition, a Hawthorne effect was possible when the intervention was introduced. However, if such an effect drove the results, it would be difficult to explain one of our key findings: variability across units. Third, we assessed five units in one city, and thus our findings may not be widely generalisable. And although we included a variety of patient care units in our study—surgical, medical, and emergency units—we could not include units representing all possible specialties. Our findings may not apply to hand hygiene practices in other healthcare facilities or for the types of patient care units we did not evaluate. Finally, we could not correlate rates of hand hygiene adherence to those of hospital-acquired infection because neither hospital reliably conducts surveillance for common hospital-acquired infections on the evaluated units.

Limitations notwithstanding, we believe our five-unit Italian study indicates the promise of a multimodal intervention in a real-world healthcare setting for at least temporary change in compliance behaviour regarding hand hygiene. Importantly, we also identified some reasons underlying the variability observed. What we do not know, however, is whether this improvement will be sustained or how to ensure that it will be. To this end, we plan to conduct a follow-up evaluation of hand hygiene practices in the same five units approximately 1 year after the intervention occurred. We anticipate that each unit will develop additional approaches to improve hand hygiene among their healthcare workers.