Patient Cohorting In Controlling Outbreaks Health And Social Care Essay

This has been a study to compare two forms of patient cohorting; ward closures and bay closures, used to control norovirus outbreaks in acute trust hospitals, and to determine which one of the two is comparatively more effective in minimizing the impact and duration of norovirus outbreaks. The main findings from this study have shown that hospitals adopting bay closures for the control of norovirus outbreaks had significantly longer duration of ward infections (p < 0.018) and all the outbreaks too lasted for significantly longer periods (p < 0.032) than in hospitals adopting implementing ward closures as part of their infection control policies. Additionally it was observed that patients admitted to hospitals adopting bay closures for the control of norovirus outbreaks were one and a half times more likely to get infected with the virus during an outbreak, which subsequently, led to more blocked hospital beds and thereby more bed days being lost.

Diarrhoea and vomiting are major symptoms experienced in norovirus gastroenteritis (Chadwick et al. 2000). This is also reflected in this study, with 80% and 50% of all norovirus cases suffering from diarrhoea and vomiting, respectively. This finding is similar to what has been described in other hospital outbreaks documented in the literature (Mattner et al. 2006, Vardy, Love and Dignon 2007, Tsang et al. 2008, Tseng et al. 2010), where diarrhoea was seen to be the major clinical symptom experienced followed by vomiting. Additionally, more than 50% of cases of norovirus gastroenteritis from both cohorts had vomiting, a characteristic symptom of norovirus infection, and this complied with Kaplan’s diagnostic criteria. Illness durations of 12-60 hours (Kaplan et al. 1982, Kapikian et al. 1972) have been reported in epidemiological studies on norovirus and this duration of infection has become one of the criterion used for the clinical diagnosis of norovirus gastroenteritis. Other community studies have shown the average duration of symptoms being less than two days (Rockx et al. 2002). However in this study it was found that more than 50% of cases in both study cohorts had symptoms lasting more than 3 days (72 hours) while an average duration of symptoms in cases in both cohorts was calculated to be 3.4 days (81 hours). This is similar to what had been seen in other hospital outbreaks of norovirus gastroenteritis in which symptoms lasted an average of 72 hours (Lopman et al. 2004, Graham et al. 1994) to 3.75 days (90 hours) (Lopman et al. 2004b). This is in contrast to the 12-60 hours proposed by Kapikian et al. (1972) and used in the Kaplan criteria (Kaplan et al. 1982). One explanation for this could be that the studies by Kapikian et al. (1972) and Kaplan et al. (1982) were carried out on healthy adult volunteers and not in cases in hospital based outbreaks. Similarly study by Rockx et al. (2002) was a community based analysis and thus study participants would have had a stronger immunity when compared to people admitted in hospitals. Compared to adult volunteers in the population, patients in hospitals have lowered immunity due to their coexistent illness and are usually in the older age group; two characteristics that have shown to be linked to prolonged symptoms of norovirus gastroenteritis (Turcois et al. 2006, Lopman et al. 2004, Lopman et al. 2004b), with symptoms sometimes lasting for as long as two weeks. Such prolonged symptoms in both cohorts, with some cases having symptoms as long 31 days (Bay cohort) and 25 days (ward cohort). However, since this study was based on retrospective data hence, it was not possible to determine whether the patients that had symptoms lasting more than expected, were having relapses of infection or just rebound of symptoms following rehydration. Even though highly specific, the Kaplan diagnostic criteria has been said to have a sensitivity of about 60% (Turcois et al. 2006), from this study it can be seen that its applicability in the diagnosis of cases based on the duration of their symptoms may be questionable.

The greater number of norovirus outbreaks reported from ward cohort could be explained by the shorter duration of infection that occurred in hospitals in the cohort. Norovirus outbreaks lasted for significantly shorter periods in the ward cohort (19.9 days versus 31.5 days, p< 0.032) with the shortest being 4 days as opposed to 11 days in the bay cohort. This meant that the hospitals in the ward cohort, would have resumed normal functioning much earlier than those in a bay cohort. Ward cohort hospitals would have experienced relatively longer norovirus-free periods, before the virus would inevitably be reintroduced into the hospital by a hospital staff, a visitor or a new admission thereby initiating a new wave of infection which would be reported as a new outbreak. Such new waves of new infections may have also occurred in the bay cohort hospitals, but outbreaks in this cohort were significantly longer, the new outbreaks may have been counted as part of an on-going outbreak.

When both cohorts were compared with one another, wards in the bay cohort had significantly longer duration infection than those in the ward cohort (p< 0.018). This average outbreak duration was also much longer than what has been documented in other hospital norovirus outbreak studies (Tseng et al. 2010, Harris et al. 2010). The finding of longer duration of ward infections and closure in the bay cohort in this study is however contrary to observations made by Orr et al. (2010). He studied norovirus outbreaks in a NHS hospital over two consecutive years and noted that if norovirus was successfully contained to a closed bay, the length of closure averaged 3.9 days, in contrast to 8.3 days when a ward closures were used. However in his study, Orr et al highlighted that an enlarged infection control team was available for support during the outbreaks in which bay closures were practiced. This may have been the reason for units reopening when bay closures were used.

Outbreak policies from hospitals in the bay cohort revealed that even though bay closures were used to cohort nurse norovirus cases, when two or more bays in a ward were affected, then cohort nursing became impracticable and ward closures were instigated. Outbreak records from the various outbreaks that were reported from this cohort revealed that all the 107 wards included in this study had to be eventually closed at a certain point because infection had spread from an index bay to other bays in the ward.

One possible explanation for the longer durations of ward infections observed in the bay cohort could be the average lapse between diagnosis of the first case in the ward and closure of the ward in which he was admitted. When compared to hospitals in the ward cohort, this time lapse was found to be an average of 1.36 days greater than what occurred in the ward cohort (3.5 days in bay cohort versus 2.14 days in ward cohort). In an outbreak study, Friesema et al. (2009), showed that if wards are closed within three days of onset of symptoms in the first patient, its reopening time can be shortened by as much as three days. The average delay of 3.5 days in ward closures that occurred in bay cohort resulted in a similar outcome, with bay cohort wards being re-opened an average of three days later than those in the ward cohort (11.03 days versus 7.47 days). Additionally, in contrast to the hospital wards in the ward cohort, bay cohort wards still had patients admitted in to their open bays. This meant there was a higher turnover of patients in these wards, a feature that has been related to prolonged ward infections and duration of closure (Billgren et al. 2002).

When compared to norovirus outbreaks in non healthcare settings such as ships, schools and camps (Widdowson et al. 2005, Neri et al. 2008, Fretz et al. 2005b), the average duration of norovirus outbreaks observed in both cohorts lasted much longer. This could have two possible explanations. Firstly, unlike hospitals, non-health settings are self censoring, having the ability to completely stop the entry of more susceptible individuals. Usually one group leaves and cleaning occurs before another group arrives. In this way exposure to susceptible individuals is kept to a bare minimum. In acute hospitals this is not possible, as admission of patients cannot stop. There was a constant turnover of susceptible people in the hospitals in both cohorts despite an ongoing norovirus outbreak and this could be the reason why they may have experienced longer outbreaks. A similar reason was identified as the cause of the prolonged outbreak in an acute hospital in England (Cooper et al. 2011). Secondly, the longer outbreaks observed in both cohorts could have been due to the delayed recognition by nursing staff of index cases of gastroenteritis. Patients in a hospital setting are not as healthy as the normal and healthy people and thus, recognizing outbreaks of viral gastroenteritis can be difficult because of the higher frequency of incontinence and other causes of gastroenteritis. Thus, a case of diarrhoea that would have been rapidly detected in a healthy population may not be identified as easily in a hospital. Nursing staff would need to be extra vigilant and a high index of suspicion to identify a case of norovirus induced diarrhoea. Outbreak reports from hospitals of both cohorts revealed that in seven (58%) of the outbreaks in the bay cohort and four (19%) in the ward cohort, there was a delay in nursing staff recognising index cases at the beginning of these outbreak. This resulted in delayed notification to infection control teams in the hospitals resulting in a considerable longer time lapse before infection control measures could be implemented, and would have given ample time for the virus to spread form the index patient to other patients in a bay and ward. Such delays have been linked to prolongation of hospital outbreaks of norovirus (Tseng et al. 2010, Lynn et al. 2004, Marx et al. 1999).

However, when compared with one another, average duration of norovirus outbreaks were significantly longer in the bay cohort hospitals (p<0.032). The delay in ward closures may have been a major factor contributing to the longer outbreaks in this cohort, with evidence from different studies highlighting that delays greater than three days may prolong outbreak durations in a hospital, from a few days (Ward et al. 2000, Navarro et al 2002) to as long as week (Lopman et al. 2004). In view of the short duration of illness and infectiousness of a norovirus case, the fact that in all reported outbreaks in both cohorts, new cases of norovirus infection continued to occur across hospital wards long after the first case indicates that the outbreaks had a propagated rather than point source pattern. With bay cohorts hospitals having longer sustained outbreaks and greater number of wards and patients affected per outbreak indicates that the virus was able to spread more efficiently between wards in these hospitals. This may have been due to bay cohort outbreaks being caused by a new and more virulent strain of norovirus to which the patients were previously unexposed and non-immune. However, since both cohorts were located in same the geographical region, a similar pattern should have also occurred in the ward cohort hospital outbreaks, since they would also have been exposed to the same strains of norovirus affecting the bay cohort hospitals. A more plausible explanation for the longer and larger outbreaks in the bay cohort hospitals lies on its biological characteristics and the different means by which it can spread.

Firstly, in outbreak situations norovirus can achieve an estimated Reproduction Rate (RR) of over 14 (Heijne et al. 2009), which implies that a primary case can infect 14 secondary cases. This reproduction rate is much higher than that of other viruses such as certain strains of pandemic influenza (RR of 2-3) (Mills, Robin and Lipsitch 2004) and the polio virus which is designated as a highly contagious enteric virus has a RR of only 5-7 (CDC 2009). Furthermore Gotz et al. (2001) observed that with this RR, norovirus can generate much more secondary cases with peak infectivity occurring 2.6 days after an index case became symptomatic. Additionally, Lee et al. (2010) demonstrated that early closure of a ward was most economical as this was when the transmission efficacy of norovirus was at its maximum (>90%). Thus late closure of a ward would have lost its purpose of preventing spread of infection, since the virus would have already infected many people. In view of this high RR and peak duration of infectivity of 2.6 days, despite closing only an infected bay at first, the average delay of 3.5 days in closing the ward in bay cohort hospitals, meant that maximum spread of the virus had would have already occurred to a new ward and infected more people. A similar delay in closing this new ward would have given the same result, initiating a vicious cycle. In effect, fewer people would have benefited from the delay ward closures which would have explained the greater number of cases that occurred in hospitals in the bay cohort. Furthermore, the continued admission of patients into open bays of norovirus infected wards in the bay cohort also meant that more patients were exposed to the virus at a potential increased risk of being infected. This was supported by the calculated logistic regression in this study which showed that patients in the bay cohort were one half times more likely to get infected with norovirus during an outbreak. This increased risk of infection was reflected in the higher average number of cases that occurred and the greater number of wards that were affected per outbreak in the bay cohort.

Secondly, the low infectious dose of the virus and its ability to be spread by aerosalization is a factor to consider. Contamination of open bays by suspended aerosolized viral particles and subsequent infection of other patients and health care workers cannot be ignored. The initial closure of only an infected bay rather than a whole ward only restricts movement in and out of the closed bay, while the rest of the ward (open bays) would experience normal flow of visitors, hospital staff and continued admission and discharge of patients. Even though patients in a closed bay would have been cohort nursed as recommended in the guidelines by Chadwick et al. (2000), the aerosalization of the virus in vomit as described by Caul (1994) could have spread the virus to other bays in the ward. This method of spread was linked to outbreak propagation in a stadium (Lee et al. 2007), restaurant (Marks et al. 2000) and a hospital emergency room (Sawyer et al. 1988). Ward activities such as drug rounds, ward cleaning and even use of a commodes would have occurred in the open bays. All these have been shown increase the concentration of particulate pathogens in the air by as much as 400%, thereby aiding their spread within a ward and contaminating the environment (Roberts et al 2008). Norovirus viral particles could have spread by similar mechanisms and settled on and contaminate environmental surfaces or could have been ingested by patients and/or health care workers in the open bays. Airborne transmission via aerosolized droplets in vomitus within a defined space and contact with contaminated environmental surfaces in confined spaces can prolong the duration of a norovirus outbreak (Marks et al. 2003, Siegel et al. 2007). Similarly, open bays having asymptomatic cases or patients in the incubation period could have contributed norovirus contamination of their environment. Norovirus excretion has been documented in asymptomatic cases (Graham et al. 1994) and also in people who are still in the incubation period of infection (Goller et al. 2004). Even though the average age of norovirus cases in both cohorts was not known due to limitations of the data available for this study, it is well recognised that the average age of patients in acute trust hospitals has been increasing (NAO 2004) and epidemiological studies have shown that elderly patients excrete norovirus just as well in formed stools as in liquid/loose stools (Goller et al. 2004). All the hospitals included in this study used the Bristol Stool Chart (Appendix IV) to monitor patient diarrhoeal stools and patients were considered symptomatic and infectious only if they had diarrhoea of grade VI and VII (Appendix IV). Since stool specimens of asymptomatic patients are usually not subjected to microbiological investigations, patients in open bays could have been wrongly classified as non-infectious simply based simply on their stool consistency, even though they might have been excreting the virus and contaminating environmental surfaces and fomites. Environmental persistence of norovirus on ward surfaces even after they had been cleaned has been reported (Morter et al 2010), which lead to spread of the norovirus across wards in a hospital through the hands of healthcare workers and patients. Since open bays are not subjected to the similar enteric precautions and increased cleaning as closed bays, ward surfaces had a higher chances of the virus persisting on environmental surfaces in an open bay. In contrast to the ward cohort hospitals, partially open wards in the bay cohort hospitals still had normal movement of patients and healthcare workers through open bays, increasing the risk of fomite contamination and person to person transmission to other wards in the hospital. With norovirus infection having an incubation period as short as 12 hours (Chadwick et al. 2000), the average delay of 32 hours (1.36 days) in ward closures that occurred in the bay cohort can be significant enough time for a contaminated surface in the open bay of a ward to infect the hands of a healthcare worker or ward staff and be transmitted to other hospital units. The role of health care workers in spreading norovirus via person to person transmission has been found to play a crucial role in propagation of outbreaks in hospital (Gellert et al. 1990, Margalit-Calderon et al. 2005, Lopman et al. 2004b). However, since the data for this study were principally obtained from records of retrospective outbreaks, and information on staff flow pattern during outbreaks could not be determined, therefore the degree of contribution of healthcare workers in spreading the virus across wards in both cohorts could not be quantified in this study.

Thirdly, air currents have been shown to augment the airborne transmission of viruses such as the influenza virus (Wong et al. 2010) and a similar theory for norovirus was described by Caul (1995). Air currents generated by movement of people within a partially open norovirus infected ward have been shown to cause infection in people walking in hospital corridors past the ward (Sawyer et al. 1988). Air currents generated by the throughput of people moving in and out of open bays may have assisted in the spread of aerosolized norovirus. In one of the ward cohort hospitals, air filters were installed in a few wards of one of the hospitals in the ward cohort in the later part of 2009, and it was noticed that there was a concurrent reduction in secondary cases in infected wards and ward re-opening times were reduced to as low as three days. Similarly in one of the bay cohort hospitals, during an outbreak that lasted 56 days, it was noticed that new cases stopped occurring late into the outbreak only after air purifiers were installed in a few wards. However, since the air purifiers were used late in the outbreak, the reduction in attack rate of the virus observed could have been due to fewer susceptible patients left in the wards. More controlled comparative studies would be required to determine, whether the reduction in lengths of outbreaks experienced in such wards was entirely due to the air filters trapping aerosolized norovirus particles or other confounding variables such as better enteric precautions and isolation procedures.

Additionally this study revealed that hospitals in the bay cohort had significantly more beds blocked and bed days lost per outbreak than hospitals in the ward cohort with an average of more than a hundred extra beds blocked per outbreak. Reports prepared infection control teams highlighted that due to the excessive number of blocked beds the hospital experienced during most of their outbreaks, it became difficult to adhere to infection control guidelines as these outbreaks progressed and spread through the hospitals. New cases of norovirus gastroenteritis were occurring so rapidly and across so many wards that it exceeded the available resources for isolation and cohorting. This problem was more evident in six of the twelve outbreaks in this cohort. These particular outbreaks had the following in common; they lasted more than thirty days, resulted in more than a hundred patients with gastroenteritis and the most number of ward infections and beds blocked during each outbreak. A similar problem has been reported to occur in a norovirus outbreak in a Swiss hospital (Khanna et al. 2003) which lasted thirty one days causing 63 cases who could not be isolated/ cohorted due to many blocked beds and also lack of adequate number of side rooms. Other authors have acknowledged the difficulty faced adhering to published infection control guidelines can become difficult (Chadwick et al. 2000, Garner 1996, Angelillo et al. 2001, Colville 2011), especially when outbreaks of norovirus gastroenteritis spiral out of control, identifying staff shortages, lack of adequate isolation facilities and non compliance with enteric precautions as possible reasons. The greater number of wards affected per outbreak in the bay cohort would have implied that less hospital staff may have been available for reallocation of resources during shifts. This is because staff in infected wards cannot be moved to other wards (Chadwick et al. 2000). Additionally even though there was no data on hospital staff illnesses in this study bay cohort outbreak reports revealed that a problem of staff shortages during the outbreak, necessitating available staff to work overtime and also hiring agency and bank staff for cleaning purposes. Shortage of staffing during norovirus outbreaks have been associated with reduced compliance to good infection control practices (Khanna et al. 2003). It is possible that such staff shortages experienced could have led to a cycle that perpetuated outbreaks affecting the bay cohort However due to the nature of data available for this study it was not possible to quantify the actual contribution of the staff shortages and hiring of temporary agency staff to infection spread in the bay cohort. Additionally in two of the three bay cohort hospitals, as norovirus outbreaks became prolonged, both suspected and confirmed cases of norovirus had to be kept in make shift bays in the Medical Admissions Unit (MAU) sometimes for as long as 48 hours because most of the wards had closed and beds were blocked. Delayed transfer of norovirus cases from an admissions unit was shown to facilitate spread of the virus through aerosol transmission (Sawyer et al. 1988). In order to create beds previously closed wards had to be reopened temporarily to enable the infection control team to move patients across wards and create bed spaces for patients in the MAU awaiting admission. A similar movement of symptomatic patients between wards was shown to increase the spread of norovirus in a hospital and lead to an increased number of cases and prolongation of norovirus outbreaks in hospitals (Lopman et al. 2004)

From the findings in this study, it can be seen how using bay instead of ward closures to control norovirus outbreaks could create a hospital environment that promotes and facilitates the spread of norovirus between wards. Furthermore, it was noticed that all the 107 bay cohort wards that were included in this study, at a particular point in time during an outbreak, they had to be completely closed in order for the respective infection control teams to be able to control spread of the virus. In a nutshell, this means that the final outcome was a ward closure, which was what the ward cohort hospitals were practicing in the first place.

IMPLICATIONS FOR PRACTICE

The practice of bay and ward closures to control norovirus infections are at two ends of a spectrum with both having their own problems and limitations. The inconvenient occurrence of most norovirus outbreaks occurring in the winter months, when bed occupancy is at its highest and the turnover of patients is slower and the increasing pressure to admit patients, closing a hospital ward instead of a bay can severely disrupt hospitals activities and have an impact on the business continuity of the hospital. This problem can be more severe if several wards are closed.

If a ward is closed due to a case of norovirus in one of its bays, the discharge of patients in other bays will be severely disrupted and potential beds that can be used for continued admissions would be inevitably blocked. Patients will have to stay longer than necessary on a ward and this adds additional costs to the hospital. A closed ward will also delay investigative procedures that may be needed for some patients and choice and justification of preventing movement of asymptomatic patients from a closed ward to other departments for routine investigations can be very challenging. Additionally, the inability to admit patients into a closed ward can affect the admission targets of a hospital. Such targets are based on performance indicators which are set by the NHS. Failure to meet these indicators can have contractual consequences and lead to severe financial implications for the hospital. When bay closures are used it gives the opportunity to maximize the available bed capacity and maintain a continuous flow of patients and business continuity of a hospital. This has been one of the reasons why some acute trusts in the NHS have used bay instead of ward closures (Orr et al. 2010, Wyeth et al. 2010). However, the advantage of achieving these targets and continued admission that bay closures make possible, is only felt in the early stages of an outbreak.

The longer duration of ward infections and subsequent ward closures that results from practicing bay closures can once again increase the length of stay of patients in a hospital. The increased length of stay of patients can be attributed to the higher proportion of patients staying longer due to infection and also the need to delay discharge of patients to other care settings so as to prevent the outbreak of norovirus gastroenteritis in these settings. Patients admitted in a ward for longer than necessary increases the risks of patients being exposed to and contracting nosocomial infections. Cancellation of elective procedures is another effect of prolonged outbreaks. Patients have to wait longer than necessary before receiving treatment, affecting the quality of care they are entitled to receive. In a study by Cooper et al. (2010), on a hospital outbreak of norovirus where bay closures were used, revealed that patient satisfaction was affected due o increased length of stay and also due to cancelled elective procedures. In the same study, the outbreak there was a significant rise in the number of cancelled elective procedures which was necessary in order to release beds. Having a greater number of norovirus cases arising due to bay closures imply that bed availability is reduced, which may create difficulties in accommodating emergency admissions and simultaneously increase the time patients would spend in the Accident and Emergency (A&E) before they can be admitted. At the same time prolonged waiting times in the A&E have in itself been shown to disrupt the normal availability of hospital beds, with an increased risk of patients being admitted to a ward outside their speciality. This was observed to have occurred in the current study. Such patients are called ‘outliers’ (Colville 2011) and they can be exposed to a number of potential hazards. Firstly, they may be exposed to nosocomial infections. Longer outbreaks and greater number of norovirus cases have been shown to increase nursing staff needs (Perry 2003), which can be increased with simultaneous staff shortages that may occur from staff absences due to illness. In the midst of such potential staff shortages occurring, ‘outliers’ can stretch the limits of care, with health and specialist teams required to cover a greater number of wards, which in itself increases the risk of spread of infection. Thirdly, relevant nursing skills may not be available in the wards these ‘outliers’ are accommodated which can seriously affect the quality of care received by a patient.

With exception of newly built hospitals built in the United Kingdom which have 90-100% of their beds as single-bed rooms (McDonald 2010), existing structures are expected to have minimum requirement of only 20% of beds as single rooms(Hurst et al. 2008). This lack of single occupancy rooms in already existing hospitals in the NHS can create difficulties in isolating norovirus cases during outbreaks. Thus closure of an entire ward can significantly disrupt hospital activities, whereas having multiple wards closed at the same time can have a larger impact. Inability to achieve these performance indicators can have contractual consequences for acute trust hospitals and result in substantial financial consequences.

Even though findings from this study have shown that ward closures are more effective that bay closures in shortening the duration and extent of a norovirus outbreak, before recommending it to be adopted across all acute trust hospitals across England its feasibility must be considered. Ward closures can have their own limitations and financial implications. Hospital/ward design and resources available to a hospital are all factors to consider before an acute trust hospital can adopt using ward closures.

First of all, hospital design is a major factor determining whether a ward closure can be implemented. Ward closures are applicable in nightingale and bay ward designs, but not so in hospitals having the ‘race-track’ ward designs. In the ‘race-track’ wards design, access to some wards is only possible through other wards (Hurst et al. 2008), therefore if a norovirus infected ward is closed, access to unaffected wards would be affected. In this scenario the only possible option would be bay closures. Cooper et al. (2010) documented a similar problem during a norovirus outbreak in an acute trust hospital which was designed with race-tract wards. In addition to ward designs in a hospital, the type of ward in which norovirus infection occurs has to be taken into account. Specialist wards cannot be closed since they have to remain open for patients requiring specialist care.

Finally, the ward closures can place excessive demands on the availability of staff a hospital may have. Since staff working in a hospital ward that has been closed due to norovirus, are not allowed to move between hospital units during a shift (Chadwick et al. 2000) this can stretch the availability of staff across the hospital and maintain safe staffing levels.

Risk management has to be considered when infection control teams have to chose between bay and ward closures. The risk of having patients being admitted into an open bay and acquiring norovirus infection is balanced against the risk of not receiving medical treatment because a ward is closed. Additionally, choices for patient safety need are a priority. The longer an outbreak persists, the greater the risk of norovirus exposure to patients. On the contrary, the need to continue admitting emergency patients can be an equally demanding patient safety priority. Considering the balancing both these demands for patient safety can be a challenge for infection control teams when choosing between bay and ward closures.

. LIMITATIONS OF FINDINGS

Practicing only bay or ward closures are not sufficient enough in achieving norovirus outbreak control within a hospital setting. They have to be applied in combination with other infection control measures which are equally important components of norovirus infection control guidelines. Such infection control measures include; environmental decontamination, adherence to excluding sick staff from work for 48 hours, compliance of health care workers with hand hygiene. Each of these have individually or in combination could have influenced the results obtained in this study. The degree to which hospital employees working in norovirus infected wards in both cohorts, adhered to these other control measures could not be quantified in this study.

Outbreak reports from hospitals in both cohorts revealed that hospital workers continued to work despite being symptomatic with diarrhoea and vomiting during a norovirus outbreak and did not report illness to the infection control team. This problem was more evident in the early stages of an outbreak. This problem of health care workers working while being symptomatic has been identified as a risk factor for norovirus outbreak prolongation and spread of the virus between wards of a hospital (Khanna et al. 2003), increasing the secondary attack rate of the infection. Ideally symptomatic staff is expected to be excluded from work till 48 hours after cessation of symptoms (Chadwick et al. 2000). Since data for this study was retrospective in nature, it was not possible to quantify, how much this lapse in infection control may have contributed and influenced the results obtained. Similarly, it was not possible to determine how adequately health care workers adhered to hand hygiene when moving across hospital wards in between shifts. This too could have influenced the outbreak trend.

Secondly, this study showed that hospitals in the bay cohort had outbreaks that were more expensive than those in the ward cohort. This estimate was however based primarily on the costs acquired due to blocked beds and bed days lost during an outbreak. Additional costs that may have been occurred due to factor such as; staff absences due to illness, extra cost of cleaning wards, cancelled elective procedures, and hiring of agency or bank staff could not be included. Each of these has been considered to be significant contributors to the economic costs of a norovirus outbreak (Lopman et al 2003, Hansen et al 2004). Even though, a study by Zingg et al. (2005) showed that loss of revenue during norovirus outbreaks was three times higher than losses of productivity costs due to staff absences and cancelled procedures, the latter can become very significant if outbreaks lasted longer and affected numerous hospital wards. Prospective controlled studies have to carried out, taking in to account all these additional costs, before the actual economic impact either method of closure can be quantified and interpreted.