Table 2. Results from multi-level multivariable and exact a logistic regression models showing associations between the occurrence of Salmonella in raccoon fecal and soil samples with respect to raccoon sex for raccoon samples, year, season, rainfall and interaction effects in Ontario, Canada.
| Multivariable models for Salmonella according to sample type | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Predictor | Sub-Category | Raccoon feces b | Soil b, c | Manure Pit b, d | ||||||
| (n = 1092) | (n = 1609) | (n = 69) | ||||||||
| Odds Ratio | 95% CI | P | Odds Ratio | 95% CI | P | Odds Ratio | 95% CI | P | ||
| Sex | Female | REF | — | — | — | — | — | — | ||
| Male | 2.06 | 1.28–3.32 | 0.003 | — | — | — | — | — | — | |
| Season | May to July | REF | REF | REF | ||||||
| Aug. to Nov. | 5.27 | 2.18–12.77 | < 0.001 | 2.74 | 1.66–4.54 | < 0.001 | 3.86 | 1.00–17.68 | 0.044 | |
| Year e | 2012 (2011 REF) | — | — | — | 1.26 | 0.78–2.04 | 0.335 | 0.42 | 0.09–1.75 | 0.210 |
| 2013 (2011 REF) | — | — | — | 0.45 | 0.24–0.83 | 0.011 | 0.06 | 0.006–0.42 | < 0.001 | |
| Rainfall | Sum prior 30 days | 1.00 | 0.99–1.00 | 0.155 | — | — | — | — | — | — |
| Interactions f | Sex*Season | 0.50 | 0.26–0.94 | 0.031 | — | — | — | — | — | — |
| Rainfall*Season | 0.99 | 0.98–1.00 | 0.014 | — | — | — | — | — | — | |
| Variance [VPC] | Site-level | 0.12 [2.8] | 0.02–0.58 | — | — | — | — | — | — | |
| Animal-level | 0.79 [18.7] | 0.32–1.93 | — | — | — | — | — | — | ||
| Sample-level | [78.5] | — | — | — | — | — | — | — | ||
a Exact logistic regression was used to model manure pit samples.
b Random effects included site and animal. Significant differences are in bold, n = total number of samples, the dash indicates “Not Applicable”, REF = referent group, and CI = confidence interval.
c Random effect for site was not included in the model because it did not improve model fit based on AIC and BIC; it explained only a small amount of the variance (0.023), and its removal had little to no impact on the coefficients in the model.
d Interactions were not tested for manure pit samples and random effects were not included in the model.
e Wald’s χ2 test for year was P = 0.004 for soil samples. Results for 2013 versus 2012 for soil samples and manure pit samples were (OR = 0.39; 95% CI, 0.21–0.72; p = 0.003) and (OR = 0.16; 95% CI, 0.01–0.97; p = 0.029), respectively.
f To interpret season, sex, rainfall, and their interaction effects, refer to contrasts in Table 3.