Table 7.

Coefficients, P values, and ORs in logistic regression models describing associations between STEC virulence factors and origin of STEC isolates (human with severe disease or uncomplicated diarrhea and healthy individuals versus bovine)a

Gene or charac-teristicModel 1Model 2Model 3
βPORβPORβPOR
ehxA
espP−1.560.00940.21−1.430.0350.14−1.600.02820.20
stx1−1.780.00060.17−1.940.00060.24−2.200.00460.11
stx2−1.880.04350.15
HEp-20.950.00472.591.220.00453.39
  • a Model 1, logistic regression model obtained by backward-elimination procedure and describing associations between STEC virulence factors and human isolates in comparison to bovine isolates; model 2, logistic regression model obtained by backward-elimination procedure and describing associations between virulence factors and isolates from severe disease in comparison to bovine isolates; model 3, logistic regression model obtained by backward-elimination procedure and describing the association between virulence factors and isolates from patients with uncomplicated diarrhea and healthy individuals in comparison to bovine isolates. The backward-elimination procedure had a threshold of 5% significance. HEp-2, level of adherence on HEp-2 cell cultures measured in proportion to a positive control; β, coefficient; −, variables not significant at the 5% level and not included in the model. The results indicate, for instance, that when all the other factors are kept constant, anespP-positive isolate is five times less likely (equivalent to 0.20 times more likely) to originate from a healthy human or a human with simple diarrhea than from cattle (model 3). Similarly, for each increase of 100% in the adherence level (in comparison to the positive control), an isolate is 2.59 times more likely to originate from humans than from cattle (model 1).