Clinical Manifestation of Norovirus Gastroenteritis in Health Care Settings

Abstract

Noroviruses are generally believed to cause relatively mild gastroenteritis of short duration in otherwise healthy adults. However, outbreaks in health care settings are common and affect vulnerable populations. During 2002–2003, a total of 4 major hospitals, 11 community hospitals, and 135 nursing homes in the county of Avon, England, were prospectively monitored for outbreaks of gastroenteritis. For 482 hospital staff, 166 nursing home staff, and 266 nursing home residents, the median duration of norovirus gastroenteritis was 2 days, with 75% achieving complete recovery within 3 days. The median duration of norovirus gastroenteritis for 730 hospital patients was 3 days (75% of the patients achieved complete recovery within 5 days), which was significantly longer than that for all other groups (P < .001). Therefore, infection in hospitalized persons may be more severe than that in other groups in the community at large. This increased duration of acute illness should be considered when implementing measures to prevent transmission in hospital settings.

In both developing and developed countries, infections of the gastrointestinal tract are common [1]. These infections are important causes of child mortality in developing countries, whereas in developed countries, their impact is typically felt in terms of economic loss and morbidity. Community-based studies have shown that, in developed countries, 20%–25% of individuals have an episode of gastroenteritis annually [2, 3]. Noroviruses (previously known as “Norwalk-like viruses” and “small, round-structured viruses”) have been shown to be the most common etiological agent in cohort analyses involving Dutch [2] and English [3] communities. It has been estimated that 650,000 cases of norovirus gastroenteritis occur in England and Wales annually [4].

Noroviruses are small (diameter, 35–40 nm), single-stranded RNA viruses that belong to the Caliciviridae family [5]. Because of their low infectious dose and environmental stability, they are transmitted by a number of routes [6]. Spread by means of food and water consumption and person-to-person contact can occur by the fecal-oral route, and airborne transmission can occur when an affected individual vomits. Vomitus spread is most commonly recognized in semiclosed communities, such as nursing homes, hospitals, and cruise ships [7].

The syndrome resulting from norovirus infection has historically been described as mild and self-limiting. These descriptions originate from studies performed in the 1970s that involved human volunteers [8, 9], as well as from numerous outbreak investigations [10–18]. Such reports were the basis for the criteria used by Kaplan et al. [19] to discern outbreaks with a viral etiology. These criteria included a short incubation period (24-60 h), a short infection duration (12–60 h), and a high frequency of vomiting (⩾50% of cases) [19]. However, these analyses were based on infection in otherwise healthy adults and, therefore, may not be representative of the range of symptoms in the community at large. The recent analysis by Rockx et al. [20] of a community-based cohort of patients with norovirus infection revealed a median duration of infection of 6 and 3 days in infants <1 year of age and children 1-4 years of age, respectively.

It is well known that persons in health care settings, such as hospital staff and patients and nursing home staff and residents, are at high risk for outbreaks of infectious diseases [21–25]. More than one-half of the 5257 reports collected during broad-based surveillance of gastroenteritis outbreaks in England and Wales occurred in residential homes or hospitals [26]. However, detailed descriptions of outbreaks of norovirus infection in these communities of (principally) older individuals are lacking. Here, we analyze case series' from a prospective study of outbreaks of gastroenteritis in hospitals and nursing homes in the county of Avon, England.

Methods

Population, follow-up, and clinical definitions. The study design is briefly summarized below; a detailed description is published elsewhere (BA Lopman et al., unpublished data). Cases were ascertained in an active prospective study of gastroenteritis outbreaks in hospitals and nursing homes. Four major acute care hospitals and 11 community hospitals that operate in the county of Avon, England, were monitored in the gastroenteritis surveillance network from April 2002 through March 2003. Combined, there were 2900 beds allocated for acute care in these hospitals. Nursing homes were defined as institutions that provide inpatient care for people whose infirmity, illness, or injury requires nursing care on a regular basis. In England and Wales, such care can only be provided by a qualified nurse or under the direct supervision of a nurse, and institutions must be registered with the local health authority. All such nursing homes (n = 152) registered in Avon were invited to join the surveillance system. A total of 135 nursing homes (89%) agreed to participate and completed the full 1-year follow-up period. Including beds in these nursing homes, ∼4500 beds were covered by the surveillance system. Ethical approval for this work was obtained from the South West Multi-Centre Research Ethics Committee (Bristol, United Kingdom).

The surveillance system was designed to detect outbreaks of gastroenteritis, and, therefore, a 2-tiered definition of cases and outbreaks was employed. Cases were defined as illness in hospital patients, nursing home residents, and health care staff that was associated with (1) ⩾2 episodes of vomiting, (2) ⩾3 episodes of diarrhea, or (3) both (1) and (2) during a 24-h period. Persons with illness that was believed to meet this definition were excluded from the study if the symptoms were due to incontinence or ingestion of laxative drugs. An outbreak was defined as the occurrence of ⩾2 cases in a hospital functional care unit (i.e., a ward) with dates of onset within 7 days of each other.

Outbreak investigation and diagnostic evaluation. Staff who were managing outbreaks were recommended to obtain specimens from the first 10 patients with an outbreak-associated case for virological analysis and from the first 3 patients for bacterial analysis. Recovery of a such a large number of specimens was suggested because of the low sensitivity of viral diagnostic tests [27]. Explicit instruction on how to obtain and send samples for analysis was based on standard operating procedure published by the Health Protection Agency (London) [28, 29]. Specimens were tested for viral pathogens at the regional public health laboratory using an in-house norovirus ELISA, followed by RT-PCR and then by electron microscopy [30, 31].

Case data were recorded on dedicated forms on a daily basis, ensuring a high level of accuracy. As an outbreak progressed, forms were completed by infection-control nurses in hospitals and by environmental-health officers in nursing homes. Information, including date of onset, first symptom-free day, and presence of vomiting and/or diarrhea, was collected. The duration of illness was defined as the number of days (inclusive) between the first and final dates of symptoms. Therefore, whole days, instead of hours, were counted. Individuals were considered to be “symptom free” when they had no episodes of vomiting or diarrhea during a 1-day period.

Data handling and statistical analysis. Forms were returned by post soon after an outbreak ended. Data were entered and stored in an Access database (Microsoft). Analyses were performed using Excel (Microsoft) and Stata, version 8.0 (Stata) [32]. The χ² test was used to compare proportions. Student's t test was used to compare mean values of normally distributed data, and the (nonparametric) Wilcoxon rank sum test was used to compare skewed data.

Results

During the follow-up period, there were 271 events that met the outbreak definition (33 occurred in nursing homes, and 238 occurred in hospital units). In these outbreaks, case-defined illness occurred in 4378 individuals, who were distributed as follows: 2154 were hospital patients, 1360 were hospital care staff, 505 were nursing home residents, and 358 were nursing home staff. A total of ⩾2 fecal specimens were available for analysis in 122 (51%) of the hospital outbreaks and in 19 (56%) of the nursing home outbreaks. Among these, norovirus was the confirmed etiological agent (based on detection in ⩾2 specimens) in 61 hospital outbreaks (50%) and 14 nursing home outbreaks (74%). Below, the case series' from these norovirus-confirmed outbreaks are described and compared with cases from outbreaks caused by other pathogens and from outbreaks in which no causative organisms were identified.

Age and Sex Distribution of Persons with Case-Defined Illness

The median age of hospital patients with case-defined illness (81 years; range, 0–98 years) was similar to that of affected nursing home residents (87.5 years; range, 3–101 years). The median age of nursing home staff with case-defined illness was 39 years (range, 16–64 years). Age data were not available for hospital staff affected in outbreaks, but all such persons were assumed to be 15–64 years old in the analyses presented below. Rates were highest in pediatric and geriatric hospital populations, as illustrated in figure 1.

Female sex was over-represented in all at-risk groups: 58% of hospital patients, 87% of hospital staff, 74% of nursing home residents, and 93% of nursing home staff were female. Hospital patients were the only group in which the age and sex distribution of the entire population (i.e., those with and those without case-defined illness) was known, thus allowing for the calculation of age-specific rates. Among children <5 years of age and those aged 5–15 years, rates of illness were somewhat higher in male subjects, although the difference was not statistically significant (P = .061 and P = .30, respectively, by the χ² test). However, in populations of older individuals, female sex was associated with significantly higher rates of illness among those aged 75–84 and ⩾85 years (P < .0001 for both age groups, by the χ² test).

Symptomatology

General characteristics. Information on gastroenteritis symptoms was available for 4192 (96%) of 4378 cases. Overall, 80% of persons with case-defined illness experienced diarrhea, and 61% experienced vomiting. Vomiting—the characteristic symptom of norovirus infection—was experienced by 67% of individuals with case-defined illness who were affected in norovirus-confirmed outbreaks. Vomiting was less common among hospital patients than among persons in the other groups (56% vs. 69%; P < .0001, by the χ² test), whereas diarrhea was more common (85% vs. 68%; P < .0001, by the χ² test). No patients with cases associated with norovirus outbreaks were reported to have bloody diarrhea.

Table 1 shows that all cases in outbreaks that were not confirmed to be caused by norovirus (i.e., those in which only 1 specimen was positive for norovirus), outbreaks in which specimens were negative for all pathogens, and outbreaks in which no specimens were analyzed had a similar clinical picture, based on the frequency of vomiting and diarrhea, to those in outbreaks that were confirmed to be caused by norovirus. Likewise, cases in outbreaks caused by rotavirus and Clostridium difficile did not have clinical pictures that were markedly different from those in outbreaks associated with norovirus.

Duration of symptoms, illness, and recovery. Information on duration of symptoms was available for 3217 (73%) of 4378 cases. Overall, the median duration of illness was 2 days, with a range of 1 to 32 days (75th percentile, 4 days). In norovirus-confirmed outbreaks, a similar overall pattern was seen (median duration, 2 days; range, 1–21 days; 75th percentile, 4 days). Hospital staff, nursing home staff, and nursing home residents all experienced symptoms for a median duration of 2 days (75th percentile, 3 days). However, the median duration of symptoms among hospital patients was 3 days (75th percentile, 5 days) (table 2).

The mean duration of illness among hospital patients with case-defined illness was 3.75 days, although this figure is right-skewed. In norovirus-confirmed outbreaks, the duration of illness was significantly longer in hospital patients than in hospital staff (P < .001, by the Wilcoxon rank sum test), nursing home residents (P < .001, by the Wilcoxon rank sum test), and nursing home staff (P < .001, by the Wilcoxon rank sum test). Figure 2B demonstrates that this pattern was also observed for persons with microbiologically confirmed case-defined illness, and table 2 shows that this observation was consistent across all age groups.

Recovery was slowest in the oldest age group (individuals ⩾85 years of age), with 40% of hospital patients still symptomatic after 4 days. We also investigated the time to recovery stratified according to hospital ward, but no distinct patterns were observed (data not shown).

Discussion

This analysis of a prospective study of a large number of outbreaks of gastroenteritis in health care settings has revealed that these infections may be more severe in hospital patients than in other affected groups. In outbreaks in which norovirus was the confirmed etiological agent, affected patients had a median duration of illness of 3 days, and, in the oldest age group (individuals ⩾85 years of age), 40% were still symptomatic after 4 days. Rates of infection in these health care settings paralleled those in the community, in that the highest incidence of infection occurred in pediatric and geriatric populations—although it is important to note that patients, nursing home residents, and health care staff in all age groups were affected.

Results of our study are from outbreaks that were microbiologically confirmed, which is essential for analysis in this type of study. Kaplan et al. [19] developed their criteria in 1982, when virological confirmation was much more difficult to achieve than that based on modern molecular analysis and ELISA. Perhaps more importantly, Kaplan et al. [19] did not consider hospital-based outbreaks in their classic study. Our study suggests that the criteria of a mean or median duration of illness of 12–60 h reported by Kaplan et al. [19] is incorrect for hospital patient populations. In the present study, they had mean and median durations of illness of 80 and 72 h, respectively.

Also, we used a broad clinical outbreak definition; cases were considered to be part of the same outbreak if they occurred within 7 days of each other. Diagnostic tests were thus important in the confirmation of outbreaks. However, practically speaking, it was typically quite clear when outbreaks were occurring, because of the rapid onset of multiple cases.

The strength of our study is derived from the fact that cases were ascertained using an active system of surveillance that was employed prospectively. Infection-control nurses (in hospitals) and environmental-health officers (in nursing homes) personally observed all cases and were responsible for reporting events. Because the same protocol was employed in hospitals and nursing homes, 4 “at-risk” groups were available for comparison: hospital staff and patients and nursing home staff and residents.

Information on the underlying conditions of hospital patients (i.e., their reason for hospital admission) was not available. Thus, the longer illness duration in hospital patients may have been due to underlying conditions, but it remains unknown exactly what medical conditions are associated with slower recovery from norovirus infection. Although we had information on the duration of illness in general, such information was not available for each individual symptom. Earlier reports have indicated that vomiting is limited to the first day of the illness, whereas diarrheal episodes occur throughout the course of the illness [20, 33]. Also, serial fecal specimens were not available for investigating the duration of viral shedding in these infections. Furthermore, it is possible that persons with outbreak-associated cases were symptomatic for reasons other than those for infections caused by the pathogen detected in the outbreak. Because diarrhea among institutionalized older persons is very common [24, 34], episodes in hospital patients or nursing home residents that resulted from long-term incontinence or ingestion of laxative drugs may have been incorrectly enumerated as outbreak-associated cases, even though the case definition attempted to exclude these noninfectious syndromes. However, the observation that recovery in microbiologically confirmed cases was similar to that in cases clinically associated with an outbreak suggests that this was not a major bias (figures 2A and 2B).

Is it possible that many of the cases in hospital patients that were determined to be part of outbreaks were in fact cases of severe community-acquired infection that had resulted in hospitalization? We think this is highly unlikely. First of all, hospitalization due to norovirus gastroenteritis is rare. Officially, only 143 such hospitalizations were reported nationally in England in 2002–2003 [35]. Although this official figure is likely an underestimate, there have been no direct studies of hospitalization rates associated with community-acquired norovirus infection. Adak et al. [4] have estimated that 1 of 1666 cases (∼410 annually) result in hospitalization. The period 2002–2003 was clearly an epidemic year for noroviruses. A novel genogroup II.4 norovirus variant emerged, and many European health care systems were affected by nosocomial outbreaks [36]. However, even a 5-fold increase (over the 1995 figures) in the number of cases of community-acquired norovirus infection would result in an estimated 33 hospitalizations in the study hospitals (based on 1 hospitalization per 1666 cases; the study hospitals serve 1.6% of the English population) and, therefore, would be unlikely to bias these results, which are based on 730 patient cases. Furthermore, any patients with norovirus infection who were hospitalized would have been isolated and, therefore, would unlikely have been associated with outbreaks.

Most descriptions in the medical literature of norovirus gastroenteritis have come from volunteer studies involving healthy adults or from investigation of individual outbreaks. These observations have resulted in norovirus gastroenteritis being referred to as a mild and self-limiting disease, in that the duration of symptoms lasts 12–60 h and resolves without therapy [19]. The course of illness observed in hospital staff, nursing home staff, and nursing home residents in the present study follows these descriptions closely. The median duration of illness was 2 days in these groups. However, hospital patients who were affected in outbreaks had a mean duration of illness of 3.75 days and a median duration of symptoms of 3 days (75th percentile, 5 days). Other recent studies and outbreak reports have demonstrated that the historical description may be inadequate for some groups. Rockx et al. [20] found that the median duration of norovirus illness was 4 days in the community at large, and, in children <1 year of age, the median duration of illness was 6 days. In military camps, outbreaks are common and have also been associated with severe infection [37, 38]. An outbreak among British military personnel involved in the war in Afghanistan resulted in soldiers being evacuated to England and Germany after complaining of headache, neck stiffness, photophobia, obtundation, and gastrointestinal symptoms [39].

However, it should be noted that other cohort studies and reports of large outbreaks have not necessarily confirmed these findings. For example, in a prospective Finnish follow-up study, the median duration of norovirus infection in children <2 years of age was 2 days [33], and, in a retrospective cohort study of a large outbreak in 30 day care centers in Sweden, the median duration of illness was only 15 h for children [40]. And, although there are numerous detailed published reports of hospital outbreaks of norovirus infection, they typically describe gastroenteritis lasting ⩽2 days [41–43]. However, it should be noted that the present analysis goes beyond those reports, in that these data are prospective and have been collected from >200 ward outbreaks.

Further work should determine what underlying factors are associated with prolonged illness in hospitalized populations. In addition, these data should inform epidemic models of transmission of norovirus in hospitals. The finding that infected hospitalized persons are acutely ill for a mean of nearly 4 days will be important for understanding the transmission and impact of control measures implemented in outbreak situations. These data demonstrate that hospitalized populations are vulnerable to more-severe illness resulting from infection with norovirus and, therefore, underscore the ethical importance of preventing hospital-acquired infection.

Acknowledgments

We thank the following members of the Enhanced Surveillance for Gastroenteritis Outbreaks study team: Helen Tucker, Lauren Tew, Dawn Hill, Maja Rollings, Samantha Matthews, Tracey Halladay, Christine Perry, Liz Bowden, Joanna Davies, Stephanie Carroll, Denise Myers, Dawn Hill, Christine Perry, and Sarah King (infection-control nurses); Erwin Brown, John Leeming, Eleri Davies, and Kim Jacobsen (microbiologists); Nicky Lambourne, Dianne Lloyd, Dave Thomas, James Barrow, Marc Hollier, John Jefferies, John Buckingham, and Andrew Tanner (environmental health officers); and numerous nursing home staff. We also thank Richard Slack, Keith Neal, David Carrington, and Owen Caul, for the helpful discussions, and Celia Penman, Kirsty Alexander, Maggie Roebuck, Katie Christou, and Meera Sirvanesan, for their support.

Financial support. Health Protection Agency and European Commission.

Conflict of interest. All authors: No conflict.

References