Skip to main content
. 2015 Mar 12;81(7):2635–2650. doi: 10.1128/AEM.03793-14

TABLE 6.

Final mixed-effect negative binomial hurdle model of risk factors associated with the count of generic Escherichia coli (log10 CFU/g) on spinach (with the random effects farm and date)a

Variable (comparison level) Reference or unit Coeffb SEc OR or RRd 95% CIe P value
Binary part (binomial with logit link)
    Intercept −5.97 1.58 NAf NA <.001
    Hygiene-field status (level 1)g 0 −2.88 0.86 0.06 0.01–0.30 0.001
    pd29h mm 1.24 0.38 3.5 1.7–7.2 0.001
    Manure_application (yes) No 3.96 1.37 52.2 3.6–761.4 0.004
    State (Texas) Colorado 4.68 1.52 108.1 5.5–2128.7 0.002
P_count part (zero-truncated negative binomial with log link)
    Intercept −29.08 12.02 NA NA 0.016
    pd29i mm 0.38 0.11 1.5 1.2–1.8 <.001
    txd9j °C 2.50 1.04 12.2 1.6–94.2 0.017
    txd92 °C −0.0523 0.02 NA NA 0.021
a

Variance component values (standard deviation) were 3.53 (1.88) for date and 0.10 (0.32) for farm in the binary part and 1.62e−7 (4.00e−4) for date and 0.77e−7 (2.77e−4) for farm in the P_count part; in the final P_count part, the random effects were dropped because their effect was negligible. For the intercept-only model, variance component values were 6.11 (2.47) for date and 0.94 (0.97) for farm in the binary part and 0.16 (0.40) for date and 0.11 (0.33) for farm in the P_count model.

b

Coeff, value of the regression coefficient.

c

SE, standard error.

d

OR, odds ratio; RR, relative risk. In the column, ORs are shown for the binary part and RRs are shown for the P_count part.

e

CI, confidence interval.

f

NA, not applicable.

g

The estimated OR (95% CI) applies to all factors within the composite variable “hygiene-field status” group. Here, “level 1” indicates the presence of toilet training and use of toilets and washing stations but absence of field grazing and hay production before planting of the spinach during the current growing season.

h

pd29, mean amount of rain between the day of sample collection (SC) and day 29 prior to SC.

i

Estimate for the variable pd29 when txd9 is 0.

j

Mean of the maximum daily temperatures between the day of SC and day 9 prior to SC; here the estimate is for the variable txd9 when pd29 is 0.