The study population was defined as all human cases of laboratory-confirmed Campylobacter
infection with onset from April 2003 through March 2006, reported by residents in 4 local authorities in northwestern England, as previously described (7
). All case-patients were asked detailed questions about their illnesses and possible exposures.
Water samples were collected at least each fortnight from October 2003 through December 2005 as 2-L grab samples from sampling points on 2 rivers associated with the study area (River Mersey and River Wyre). Water samples were transported to the Food and Environmental Microbiology Laboratory, Royal Preston Hospital. Campylobacter
species were isolated by the addition of 10 mL of the water sample to 90 mL of warmed Campylobacter
enrichment broth (product CM0983, Oxoid Ltd, Basingstoke, UK) and incubated at 37°C for 24 hours, followed by incubation at 42°C for 24 hours. The enrichment broths were subcultured onto Campylobacter
blood-free selective agar (charcoal cefoperazone deoxycholate agar product CM0739, Oxoid Ltd) at 37°C for 48 hours microaerobically, by using a microaerobic gas generating kit (product CN0025, Oxoid, Ltd). Campylobacter
colonies were identified by morphologic features and confirmed by microaerobic and aerobic growth on blood agar. The colonies were then placed in Amies transport and sent to the laboratory Health Protection Agency regional laboratory in Manchester, UK, for DNA extraction and characterization. C. jejuni
isolates were identified to species and typed by multilocus sequence typing as previously described (7
Case–case methodology was used to compare exposures between sequence types of Campylobacter
). Statistical analysis was performed by using STATA version 9.2 (StataCorp, College Station, TX, USA). Two-way tabulations and Fisher exact test were used to estimate the direction and size of association of individual variables with sequence types. Logistic regression was performed to estimate the significance of these findings and to examine them in a multivariate model. Data were collected from 2 distinct locations, and, thus, area of residence was controlled for in analysis as a stratified variable. Likelihood ratio tests were used to assess the significance of including or excluding variables from multivariate analysis.
Among the typed isolates of C. jejuni (n = 1,104), ST-45 (n = 49) was the third most prevalent sequence type reported in the study (after ST-257 and ST-21) and highly restricted to cases reported between late April (week 17) and early August (week 32) (). When week of diagnosis was categorized into 2 seasons (early summer [weeks 17–28] and all other weeks), logistic regression of season by sequence type confirmed that, when compared with all other typed cases of C. jejuni, cases of ST-45 were more likely to be reported in early summer than during the rest of the year (odds ratio 2.79, confidence interval 1.56–4.99, p = 0.001; 1,008 observations), and this relationship was not seen with other prevalent sequence types.
Stacked area charts showing the number of human case reports of the 4 most prevalent sequence types (STs) during the study, by 4-week intervals.
Seventy-four (37%) of 198 river water samples were positive for C. jejuni, and among these, 28 sequence types () were identified, 11 of which were also reported among human cases in the study population. The most prevalent sequence type identified was ST-45, and the seasonality of isolation appears to closely correlate with the seasonality of human disease onset caused by this type ().
Isolates of Campylobacter jejuni recovered from river water samples in 2004 and 2005, showing clonal complex and ST*†
Figure 2 Numbers and sequence types (STs) of isolates of Campylobacter jejuni from surface water samples compared with numbers of human cases of ST-45 (line) for 2004 and 2005, by 4-week interval. Only the 4 most prevalent human types also identified in water (more ...)
Univariate logistic regression analysis identified that, in addition to illness during early summer, the statistically strongest associations (p<0.1) with ST-45 were being <5 years of age and living in the more rural part of the study area. These 3 factors remained significantly and independently more associated with ST-45 infection than with other sequence types when combined in a multivariate model (tested by using likelihood ratio tests) (). When this model was used, subsequent estimation of the association of food and exposure variables showed that consumption of chicken at least once was less associated and consumption of home-delivered milk, and going fishing were more associated with ST-45 than with other types. This finding was sustained in a multivariate model (also including the demographic variables previously mentioned) (). However, information on exposure variables was not available for all 1,008 cases of C. jejuni, which resulted in a loss of statistical power.
Two-step multivariate logistic regression analysis of epidemiologic variables associated with cases of ST-45, in comparison with all other sequence types of Campylobacter jejuni *