Infectious intestinal disease (IID) causes substantial disease burden in both high- and low-income countries. In a population-based study conducted in 2008–9 (IID2), it was estimated that a quarter of the UK population experiences IID annually [1]; norovirus and Campylobacter, the most common viral and bacterial pathogens, account for 3 million and 0.5 million annual cases respectively. The sheer volume of IID cases results in considerable, potentially preventable economic burden, both through healthcare use and lost productivity. In a study of the health and economic burden of IID (IID1), conducted in the 1990s, the societal cost, including both costs to the health service and to patients, was estimated to be nearly £750 million annually (1994–5 prices), with the greatest shares coming from Campylobacter (£69.6 million), E. coli (£69.3 million, including non-Shiga toxin-producing strains), Salmonella (£46.4 million), norovirus (£24.4 million) and rotavirus (£18.2 million) [2]. These costs are a gross underestimate for norovirus, as current molecular diagnostics are far superior to electron microscopy used in IID1 [3][4]. Overall rates of norovirus disease in IID2, based on polymerase chain reaction (PCR), were 3.7 times higher than those in IID1 [1,5]. Furthermore, changes to the UK health service mean that these earlier estimates may no longer reflect accurately the cost of IID to the health service, or the distribution of cost between the health service and patients; since IID1 was conducted, telephone information and advice lines and out-of-hours services have been introduced, while data from IID2 indicate that IID-related general practice (GP) consultations halved between 1995 and 2009 [1]. Finally, while these costs include only the costs of acute illness, several pathogens also have longer-term sequelae, such as Campylobacter-associated Guillain-Barré syndrome [6]. We therefore estimated the societal cost due to three IID pathogens of major public health significance in the UK: Campylobacter, norovirus and rotavirus. These pathogens were selected because of their considerable burden, the fact that their incidence has not decreased in the past two decades, and because of their policy relevance. Campylobacter and norovirus are both high priority pathogens in the UK Food Standards Agency’s foodborne disease strategy [7], while the recent introduction of routine rotavirus immunisation, and increasing prospects for introduction of norovirus vaccines make these two pathogens very relevant to control policy.

We used data from IID2 to estimate the incidence and health service usage due to these three pathogens in the United Kingdom (comprising England, Wales, Scotland and Northern Ireland) in 2008–9, when IID2 was conducted. The UK population in 2009 was 61.8 million. Health services measured included use of telephone health and advice services; in-person, telephone, and out-of-hours GP consultations; Accident & Emergency department visits; and hospital admissions. Due to uncertainty in estimating hospitalisations, we used three different approaches to estimate the number of hospital admissions for each pathogen. We used NHS reference costs and data from previous studies to estimate costs of illness to patients and the health service. For Campylobacter, we additionally estimated the cost of Guillain-Barré syndrome-related hospitalisation, using published GBS incidence data [6].

We expressed costs in monetary terms, because in a universal, provider-pays healthcare system such as the UK National Health Service (NHS), this facilitates comparison of relative costs borne by the health service and patients. A monetary metric also provides baseline data to evaluate the cost-effectiveness of future interventions which, for zoonotic pathogens such as Campylobacter, involve costs to other sectors, including agriculture and food production.

The total costs to patients were £45.3 million for Campylobacter, £69.3 million for norovirus and £14.1 million for rotavirus. Point estimates of healthcare costs, including primary care and hospitalisations, ranged between £4.2–£5.1 million for Campylobacter, £11.9 –£17.7 million for norovirus and £5.1–£10.9 million for rotavirus, depending on the method for estimating hospitalisation costs.

We estimate that the economic cost of Campylobacter, norovirus and rotavirus disease in the UK exceeds £150 million annually; 80% of this cost is borne by patients, through lost employment and out-of-pocket expenses resulting from illness. The annual cost of norovirus disease exceeds £80 million. Sporadic norovirus disease accounts for more healthcare consultations than either Campylobacter or rotavirus, while previous work indicates that norovirus outbreaks in healthcare settings cost the NHS over £100 million annually (2002–3 prices) [17].

Our work makes several improvements on previous estimates. Previous UK rotavirus estimates included only hospitalisation costs [18,19], total healthcare costs [11], or societal costs from patients <5 years seeking healthcare [20]. Only IID1 included costs to patients not seeking healthcare, but that study lacked data to estimate hospitalisation costs [2]. Our estimates comprise costs to the NHS and to patients, including those not seeking healthcare. Further, our study provides the first reliable estimates of the societal cost of non-outbreak related norovirus, using empirical incidence data and modern diagnostics. Our estimates are three-fold higher higher than previously reported [2].

Wherever possible, we used pathogen-specific incidence and healthcare usage data, except for telephone information and advice consultations, for which data were sparse. We also accounted for uncertainty in incidence estimates. For hospitalisations, we used three different estimation methods for each pathogen. Hospitalisation estimates from outbreak data were higher than cohort-based estimates from IID1 and IID2; it is likely that there were insufficient cases in IID1 and IID2 to reliably estimate hospitalisation. The pathogen-specific epidemiologies should also be considered. For rotavirus, hospitalisation estimates from outbreak data were comparable with those from Harris et al. among children <5 years [11], in whom most hospitalisations occur. For norovirus, however, estimates from outbreak data were considerably higher than other data sources. Most reported norovirus outbreaks occur in institutional settings and involve vulnerable patients with higher hospitalisation risk. Estimates from longitudinal studies (scenario 2) are likely to be more reliable, and these were similar to those of Haustein among adults and the elderly [10]. This also implies that the hospitalisation burden from norovirus falls primarily on adults and the elderly rather than children. For rotavirus, the converse is true, and the total cost was more sensitive to the choice of hospitalisation scenario, because healthcare-associated costs comprise a greater fraction of total costs compared with Campylobacter and norovirus.

We estimated the cost of Campylobacter-associated GBS hospitalisation at £11,000 per case. This is an underestimate, as it excludes long-term disability costs. Around 15% of GBS patients experience persistent disability at one year [21], around a third have restricted mobility three to six years after onset, and around 40% have to change jobs and alter leisure activities [22]. In addition, NHS reference costs capture average costs for patients undergoing similar procedures, but GBS patients may require more intensive treatment. We addressed this by accounting for longer average duration of GBS hospitalisation and case mix (emergency, elective and day case); previous work indicates that GBS hospitalisation costs correlate strongly with hospitalisation duration [23].

Our study has several limitations. First, we considered costs borne by the NHS and by patients, but not wider societal costs from mortality, illness in caregivers, and costs of lost productivity to employers and the wider economy. Data on these additional costs are currently lacking. Rotavirus mortality is believed to be rare, and has not been considered in previous UK analyses [11]. Harris et al. estimated 85 annual norovirus-associated deaths in UK elderly. We are not aware of similar estimates for Campylobacter. Scandinavian studies have described excess mortality among Campylobacter patients up to one year post-infection, but these estimates are difficult to generalise to the UK, because of differences in reporting systems and distribution of population risk factors. Further, attributing deaths to gastrointestinal pathogens is problematic, because deaths occur primarily in vulnerable patients with pre-existing conditions. Thus, in some cases an IID pathogen may be causally related to mortality, in others infection will be coincidental, while in others it may precipitate deaths that would eventually have occurred even in the absence of infection.

We did not explicitly take account of hospitalisation duration in our cost estimates. The NHS reference cost for a hospitalisation due to “infectious or non-infectious gastroenteritis” is based on the average duration of hospital stay for a patient with this code, which is 1 day. Hospitalisation data suggest that average length of hospitalisation is longer for Campylobacter compared with rotavirus (6 days vs 2 days) [8]. However, inferring differences in hospitalisation duration from these data is problematic. Average hospitalisation duration for Campylobacter might also be influenced by sequelae such as Guillain-Barré syndrome. Secondly, average hospitalisation duration among recorded admissions for norovirus is 14 days [8]. However, hospitalised patients with diagnosed rotavirus and particularly norovirus infections represent only a fraction of all hospitalisations from these pathogens—previous work indicates that a large fraction of these are recorded as unspecified viral infections (see our references 10 and 11), which have an average hospitalisation of 1 day. This suggests that viral infections are more likely to be diagnosed in patients with more severe illness or with underlying conditions, and adjusting for length of stay using these data risks grossly overestimating hospitalisation costs. For this reason, we have decided to follow the methodology used in previous economic costing studies for rotavirus in the UK, which have not adjusted further for length of hospitalisation [11]. For Campylobacter, it is possible that our conservative approach underestimates hospitalisation costs. However, hospitalisation due to Campylobacter is uncommon; even assuming that hospitalisation costs increased by 5-fold, the result would be to double the health service costs. Qualitatively, our results would not change dramatically; the cost to patients would still dominate the overall economic burden.

Our analysis excludes costs from long-term disability. While we included costs of Campylobacter-related GBS hospitalisation, data on GBS rehabilitation costs were unavailable. Campylobacter can cause other sequelae, including reactive arthritis in approximately 1% cases [24], and irritable bowel syndrome in approximately 6% of patients [25]. Again, data on costs associated with these conditions were unavailable.

We estimated costs across all age groups, although health service costs could vary by age group. However, NHS reference costs are not stratified by age group. Further, they do not include information on variability in healthcare costs. As with other previous studies [11], we could not account for uncertainty in treatment costs.

We used the best available data sources, but some of our estimates are subject to considerable uncertainty due to sparse data, particularly for both hospitalisation costs and costs to patients. For rotavirus, we assumed that the cost to patients not seeking healthcare is zero; IID1 included economic data from a small number of rotavirus patients, all of whom sought healthcare. Patient costs due to rotavirus are therefore underestimated.

Overall societal costs of Campylobacter disease in our study (£50 million) were lower than in IID1 (£70 million) [2], although campylobacteriosis incidence has remained stable [1,5]. This decrease in cost results from reductions in GP usage; IID-related GP consultation rates halved between 1995 and 2009 [1]. This has also reduced the cost per case (from £164 in IID1 to £85 here). Our estimates, however, should be considered in the context of expenditure to control Campylobacter. These costs fall under other economic sectors, including agriculture, food production and retail, and are poorly characterised. Studies considering the cost of such expenditures relative to the cost of illness averted are important to evaluate the cost-effectiveness of Campylobacter interventions, which will be useful for guiding control policies in addition to the wider socio-political and public health considerations. For norovirus, the improved ascertainment of cases in IID2, most of whom do not seek healthcare, has resulted in much higher estimates of overall cost, while the cost per case is 30% lower than previously reported in IID1. For rotavirus, overall costs are somewhat higher (£25 million vs £18 million) due to increases in incidence since IID1, although the cost per case is about 45% lower. These costs will continue to change over time, following the recent introduction of rotavirus vaccination into the UK immunisation programme. Our results therefore provide a baseline with which to evaluate the cost effectiveness of rotavirus immunisation in future.