Patients and methods

In all, 286 children with positive stool cultures were reported to the regional health authorities. A few days after the outbreak the regional health authorities were asked to deliver detailed information to the parents of all affected children about the possibility of developing ReA days to weeks after the onset of diarrhoea. The same information was handed out to all paediatricians and general practitioners in the region. Medical help and advice in cases of suspected musculoskeletal symptoms were offered to the affected families.

Since no case of arthritis was reported directly to us during the following 6 weeks, we contacted all general practitioners and paediatricians in the region by telephone to identify patients with ReA. Ten weeks after the outbreak a questionnaire on the severity of enteritis, joint symptoms, joint distribution, enthesitis, and extra-articular symptoms suggestive of ReA was mailed to all parents of affected children through the regional health authorities.

Four months after the outbreak all paediatricians and general practitioners in the region were contacted again by one of us (MR) to identify further cases of possible ReA which had not been reported to us before but which had, in the meantime, sought medical help from the local doctor.

Results

Discussion

In this large cohort of 286 children affected by salmonella food poisoning only six children (2%) developed oligoarticular arthralgia of short duration. Not a single case of frank arthritis was reported. Various efforts were undertaken to deliver information to and obtain information from all parents and doctors at several time points, making it unlikely that clinically relevant cases were missed. Likewise, according to the regional health authorities and local doctors, visits of parents and children to doctors outside the region can be excluded.

The reason for the low incidence of ReA in this cohort of children is not clear but may be related to a lower susceptibility to ReA in young children in general, or to the bacterial strain (S enteritides lysotype 8/7) which may be less likely to cause arthritis. In Germany about 20 different strains ofS enteritides are isolated each year from stool cultures, 15 of which occur in frequencies higher than 0.2% of all outbreaks. The most common lysotype is S enteritides lysotype 4/6 which accounts for 70–80% of all salmonella infections, followed by lysotype 8/7 which accounts for 5% of all infections, and the lysotypes 6/6 and 21/1 which each account for 3–5% of all infections. Thus, the strain S enteritides lysotype 8/7 isolated in this outbreak is a common strain in Germany. These strains are not known to differ in their clinical manifestations. Furthermore, there are no data on differences between the strains in their ability to cause arthritis (B Gericke, Reference Centre for Salmonella Infections, Wernigerode, Germany; personal communication).

Underreporting of cases or methodological problems are unlikely explanations, given the extent of the efforts made to identify patients with ReA in this cohort. A particularly low frequency of HLA-B27 in the region where the outbreak took place can also be excluded since the area of the outbreak is located just 60 km outside Berlin which has an HLA-B27 prevalence of 9.3%.5

From the literature it appears that, in general, children are less susceptible to ReA (or joint symptoms) after enteric infections than adults (table 1) and joint symptoms are often milder in children than in adults.1-4 6-12 It is noteworthy that in all the salmonella outbreaks which affected both adults and children, the incidence of ReA was always lower in children than in adults.1 ,3 ,4 Although we could not compare children and adults in this outbreak simply because no adults were affected, our data confirm earlier reports on the generally low incidence of ReA in children. The reduced susceptibility to ReA of children compared with adults cannot be accounted for by genetic factors. Methodological difficulties in assessing ReA in children or in recognising very mild forms of ReA which were otherwise unnoticed, may contribute to the difference between children and adults. However, if the difference is found to exist, it may well relate to differences in the immune systems of children and adults—for example, viral infections such as the Epstein Barr virus cause no or only mild symptoms in children but usually have a much more severe clinical picture and often a prolonged course in adults. The immune system of children may be less educated or more “naive” and may, in general, also be less likely to mount an autoimmune response than that of adults. Alternatively, children might mount a more effective antimicrobial immune response resulting, in general, in an early, fast, and efficient elimination of bacteria.13