3.2. Outbreaks Reported
At the time of this study outbreaks in the UK were not all routinely reported to national surveillance. Reporting was limited to laboratory-confirmed outbreaks (reported manually using a standard outbreak form) and to typing results, usually Salmonellas or E. coli O157, routinely sent from the reference laboratory to the national surveillance centre.
The highest monthly reports of enteric pathogens occurred during November and December 2004 and January 2005. The same seasonal affect was not as apparent in the two other years. The Winter peak was associated with a high incidence of norovirus. Although fewer in number, rotavirus infections were more likely to occur in the Spring, whereas in warmer months reports were more likely to be linked to foodborne transmission, notably Salmonella.
In 76 of the 279 suspected outbreaks (27.2%) no specimen was logged to the incident number (see Table 2
). When specimens were received the median number per outbreak was 4, with a range of 1 to 121. In this study the detection of two or more patients in a suspected outbreak with an indistinguishable organism was considered sufficient to signify microbiological cause. Using this definition a microbial cause was found in 96 (34.4%) of the outbreaks. In over one-third of incidents (107/279; 38.4%) either none or only one positive specimen was found and no cause could be confirmed. Norovirus was the most common organism identified (53/279; 19.0%) with elderly patients in either hospital or nursing/residential homes being particularly at risk. Rotavirus was identified in 13 (4.7%) outbreaks with the elderly or very young being the most likely targets. Salmonellas
were isolated in 15 outbreaks (5.4%), with 14 of these being linked to social events (wedding receptions or meals in hotels, restaurants or public houses). Three outbreaks were linked to E. coli
O157 infection. More than one agent was found in six outbreaks (possibly due to the simultaneous detection of a past infection) and, of these, four involved C. difficile
toxins and norovirus, one was C. difficile
and rotavirus and the remaining one was norovirus and adenovirus.. Astrovirus was found in eight suspected outbreaks but in each case only one positive patient was identified and given the outbreak definition no outbreak linked to this virus was confirmed.
Almost half of the outbreaks identified (137/279; 49.1% see Table 2
) occurred in residential/nursing homes. In 33 (24.1%) of these no specimens were received and in the others when specimens were submitted the median number per outbreak was 4, with a range of 1 to 17. With one exception (see 3.3 - outbreaks for which an outbreak control team was established) an outbreak control team was not set up for these incidents, however, a member of the HPU team did investigate, recommend control measures, as necessary, and produce a summary report. When a cause was established, the vast majority were due to norovirus, however, five (all in the Spring season) were associated with rotavirus. Fifty-one outbreaks were recorded in hospitals during the three year period. In 12 of these no specimens were submitted and in another 17 none or only one positive sample was identified. The median number of specimens received was 6 with a range from 2 to 121. Again norovirus was the most common cause (14 outbreaks), and C. difficile
toxin (alone or in combination) was recorded in 9 outbreaks and rotavirus in three. Of 31 outbreaks linked to social occasions, salmonellas were identified as the most likely cause. One outbreak was due to C. perfringen
s. The distribution of outbreaks for the other locations is given in Table 2
. Of the 60 incidents recorded, in 25 no specimens were received and a cause was identified in 15 (25%) with E. coli
O157 being found in three of them (see 3.3 - outbreaks for which an outbreak control team was established and Table 3
Of the 279 outbreaks only 11 were identified as probably foodborne. In three more the evidence pointed towards contaminated food but no specific item could be identified. One outbreak was linked to untreated water and another two were associated with contaminated kitchen equipment.
This study used a database as a tool for the retrospective analysis of outbreaks. The database enabled all records to be held in one place with the facility that new information could be easily added. Linking the ILOG number to the laboratory computer system also enabled microbiology test results to be included. One limitation is that the system depends upon one or more individuals voluntarily notifying the authorities in order to begin the process and, given that most cases of gastroenteritis are self-limited this probably accounted for the low rate of ascertainment. The information is also not always timely thereby increasing the difficulty of obtaining relevant epidemiological information and collecting specimens from those that were ill. Wider distribution of the database beyond the small group of specialists currently involved may enhance the value of the system.
We found that very few specimens were submitted to the laboratory in relation to the number of people experiencing symptoms. In point-source outbreaks late reporting could account for fewer specimens being submitted, as many cases would have recovered prior to notification, however, person to person spread played a major part in many of the outbreaks described here and therefore timing was not the only reason for few specimens being received. Whereas sampling of all ill cases in large outbreaks might overwhelm limited laboratory resources, the evidence here suggests that, even in larger outbreaks, sampling rates remain low. We found that in more than a quarter of the outbreaks no sample was logged. In order to increase specimen submission, both persons ill and the general practitioners treating them should consider two questions. First, are there others with a similar illness occurring at the same time, and secondly, have foods been eaten outside the home in the previous seven days e.g. from a restaurant or take-away premises or after attending a social function where food was served. An affirmative answer to one of these questions should prompt collection of a specimen and the recording of this highly relevant information on the request form. In this study we did not have information on incidents that were confined to one household, however, we recognize that outbreaks in the domestic setting are undoubtedly common and contribute to the overall picture of outbreaks.
Initially we were surprised that so few foodborne outbreaks were identified. However, outbreaks in long-term care were the most common and the vast majority of these appear to be spread person to person rather than linked to food preparation or consumption. We cannot rule out, however, that food did not play some part in those outbreaks for which no foodborne link could be established. Stopping outbreaks in care facilities is particularly difficult and early recognition and prompt action is needed. Recent guidelines [8
] emphasise key issues, notably a written policy (with access to advice at all times), isolation of infected residents, availability of additional care staff, good hand-washing practices, and early reporting to HPUs. It may well be that all these cannot be achieved within current resources and that, without substantial cost increases, these incidents remain very difficult to control effectively.
Using our outbreak definition of two or more laboratory-confirmed cases we identified a microbial cause in approximately one-third of the outbreaks reported. A recent one-year intensified study in The Netherlands [9
] identified a pathogen in 54% of outbreaks. Their outbreak definition was the occurrence of diarrhoea and/or vomiting in at least five cases with some common factor. As here noroviruses were the most common cause. The higher rate of detection of noroviruses in the Dutch study (55.2% compared to 19% here) may be partly linked to a more sensitive detection method (PCR performed in the national reference institute compared to routine EIA in a clinical laboratory) and also to their special effort to examine stool specimens from at least five ill patients in each outbreak. The frequent association between noroviruses and outbreaks is linked to their high infectivity and the short-lived immunity to them [10
]. Like others we found that Campylobacters
were commonly identified pathogens but none was reported as part of an outbreak. The reasons for this are unclear as there is plenty of evidence showing that Campylobacters
may be transmitted by a variety of foods, water and the environment, however, most infections, despite a relatively low infectious dose, remain sporadic or affect the members of one household only [11
]. C. difficile
was frequently isolated from cases of diarrhea, however, the high background levels, particularly in wards or homes for the elderly, can make it difficult to identify this organism as a cause of outbreaks [12
]. It is clear, however, that C. difficile
can cause serious outbreaks in debilitated patients [13
]. In this study about one-third of routine salmonella notifications were proven to be part of outbreaks. Undoubtedly some outbreaks were missed e.g. a significant number of isolates were from people returning from holidays abroad, from whom no further details were available, and therefore the proportion of Salmonella
infections matched to outbreaks is likely to be an underestimate.
This study supports the need for new surveillance systems to provide a more effective outbreak detection system [14
]. Syndromic surveillance systems are being used to try and capture data from changes in behaviour that might indicate the early stages of an outbreak [15
]. Unfortunately, because in most people, gastrointestinal infections are self-limiting and, symptoms disappear quickly, the majority do not change their behaviour in such a way that would trigger the surveillance system. However in some sections of the community, notably children, changes in practice are more likely to be seen.
Systems that utilize the speed and availability of the Internet are being introduced. An automated laboratory based system has been established in the Netherlands that is updated daily [16
]. In the Program for Monitoring Emerging Diseases (ProMED-mail) [17
], reports are screened, verified and then posted on the Internet. Rapid dissemination is also achieved via the Enter-net and Salm-gene databases [18
]. All these systems are intended for international monitoring and depend on laboratory confirmation as the source of information. The Internet, however, has been used for rapid reporting of local outbreaks. The RUsick2 Foodborne Disease Forum is a web-based system that encourages people with illness to contact a website and record information about their illness. The site is always open, information can be updated at any time, and is claimed to be user friendly. Operators look for common links between patients and then notify appropriate local health professionals Measures are in place to control bogus reports that may be either deliberate or suggested by reading other entries. A pilot study, admittedly actively supported by a publicity campaign, showed more than a four-fold increase in reporting [19
]. Recent government figures [20
] indicate that nearly two-thirds of adults in the UK access the Internet. Those without web access could enter data via a receptionist at a doctor’s surgery, however, in practice this is less likely to happen. Perhaps one limiting factor is the newness of this approach with people being either unwilling to share information over the Internet or being concerned about the lack of human intervention. People may be embarrassed by a request for the collection of faeces for laboratory testing and, unless symptoms are particularly severe, would rather wait until better rather than involve the public health agencies or their local doctor.