Together with Campylobacter spp.
, the non-typhoid serovars of Salmonella enterica
(hereafter referred to as “salmonella”) are the most commonly diagnosed bacterial cause of foodborne infections in Europe [1
] and other industrialized countries, e.g. the USA [2
], Canada [3
], Australia [4
]. Symptoms range from mild, self-limiting diarrhoea to systemic infection with fatal outcome. Acute salmonella infection may be complicated by serious sequelae, such as reactive arthritis [5
]. In the USA, non-typhoid salmonella are estimated to be the leading cause of hospitalization and deaths attributable to consumption of contaminated food, causing 35% of such hospitalizations and 28% of such deaths [6
Published data on the incidence of salmonella infections are generally based on notifications of stool culture-confirmed cases [1
]. These cases constitute only a small fraction of all cases occurring in the community. A sequence of events must occur so that a sick person in the community gets registered as a case in a surveillance system: the person must consult a health care provider, he/she must be asked to submit a stool sample, he/she must comply with this, the stool sample must be sent to and arrive at a laboratory in satisfactory condition, it must be tested for salmonella, the test must be positive, and the positive test result must be reported. All these factors may differ considerably among countries or states, due to differences in patients’ health seeking behaviour and accessibility of health services, in clinical practices regarding stool examination, in diagnostic practices and test sensitivity in clinical laboratories. These factors are influenced by cultural, infrastructural and economic aspects and determine the degree of “underdiagnosis”. Finally, a diagnosed case will go unnoticed by the surveillance system if not reported (“underreporting”). We use the term “underascertainment” for the joint effect of underdiagnosis and underreporting. Because of the varying extent of underascertainment direct comparison of reported incidence rates from different countries or states is potentially misleading.
Little is known about the true community incidence of salmonella infections in Europe. Several studies have aimed to estimate so-called multipliers, i.e. the number of cases occurring in the community per one reported case. Community surveys of acute gastrointestinal infection (AGI) prevalence by telephone interviews were done in several countries, e.g. Norway [7
], Ireland [8
], Malta [9
], Denmark [10
], France [11
], and Poland [12
]. However, due to their retrospective design these studies mostly lack aetiological diagnoses and are prone to recall bias. Prospective community cohort studies of AGI including microbiological diagnostics were undertaken in England in 1993–1996 [13
], in the Netherlands in 1998–1999 [14
], and in the United Kingdom (UK) in 2008–2009 [16
]. The estimated community incidence of salmonella-associated AGI was similar in the Netherlands and England (3.3 and 2.2 per 1,000 person-years, respectively) in the 1990s, but the degree of underascertainment was markedly higher in the Netherlands than in England (multipliers of 14.3 and 3.2, respectively). The recent UK study revealed a lower population incidence (0.6 per 1,000 person-years), but the higher multiplier of 4.7 indicates increasing underascertainment compared to the situation in England in 1993–1996. Due to their very high cost and demanding logistics such cohort studies cannot be easily replicated.
In the USA, community incidence of salmonella-associated AGI was estimated by combining surveillance data with information on health seeking behaviour and diagnostic practices from laboratory and population surveys in FoodNet areas [17
], yielding an incidence estimate of 5.2 per 1,000 (multiplier ~39) for the period 1996–1999 [18
]. For the period 2000–2008, the incidence estimate was 3.4 per 1,000 (90% credible interval [CI] 2.2-5.6) and the multiplier 29 (90% CI 18–48) [6
]. Similar such “multiplier studies” in Canada and Australia showed similar results. For Canada, the estimated incidence was 2.5-6.9 per 1,000 (multiplier 13–37) in 2000–2001 [3
]. For Australia, it was 2.6 (95% CI 1.5-6.2) per 1,000 (multiplier 7 [95% CI 4–16]) in 2005 [4
]. However, as discussed in [4
], such estimates are mostly based on extrapolations from limited data and/or expert assumptions about the various steps leading to underascertainment, resulting in large uncertainties of the estimates.
As a basis for decision making in public health and for the assessment of the health and economic burden of salmonellosis, more reliable estimates of the true community incidence of salmonella infections (and other foodborne pathogens) are highly desirable. We therefore strived to develop an alternative method of estimating the community incidence of salmonella infections, which should be affordable and independent of ascertainment artefacts, expert opinion and accuracy of interviewees’ recall. To that end, we estimated the community incidence of salmonella infections from measurements of salmonella-specific antibodies in cross-sectional sero-surveys of the general population.
We present here the results of a pilot study in eight European Union member states. We compare our so-called ”sero-incidence” estimates [19
] with the incidence of salmonella cases reported through the countries’ respective surveillance systems and with published incidence estimates derived from infection risk in returning Swedish travellers [20
], representing an alternative surveillance approach insensitive to differences in case ascertainment among countries.