Table G.15.
Summary of dose–response analysis of incidence of vomiting in cats reported by Hughes et al. (1999) using BMDS software
| Models | Number of parameters | Minus Log‐likelihood | p‐value | AIC | BMC10 | BMCL10 | Comment |
|---|---|---|---|---|---|---|---|
| (mg/kg feed) | |||||||
| Full model | 6 | 6.93 | na | 25.86 | na | na | |
| Null (reduced) model | 1 | 10.64 | < 0.001 | 22.64 | na | na | |
| Probit | 2 | 9.25 | 0.33 | 22.49 | 4.21 | 2.36 | Not selected |
| Logistic | 2 | 9.24 | 0.33 | 22.49 | 4.54 | 2.50 | Not selected |
| LogProbit | Failed | ||||||
| LogLogistic | Failed | ||||||
| Multistage | Failed | ||||||
| Multistage Cancer | 2 | 9.19 | 0.34 | 22.38 | 2.80 | 1.05 | Selected |
| Quantal‐Linear | 1 | 9.24 | 0.47 | 20.47 | 2.03 | 1.04 | AIC minimum |
| Weibull | Failed | ||||||
| Gamma | Failed | ||||||
DON: deoxynivalenol.
The models selected to identify the lowest BMCL10 following EFSA guidance (EFSA Scientific Committee, 2017) are identified in the column ‘Comments’ and include the model with the lowest AIC.
BMC10: benchmark concentration response of 10%; BMCL10: 95% lower confidence limit for the benchmark concentration response (BMR) of 10%; na: not available, including cases where the BMC/Ls were not calculated, the fit was incomplete, or serious or conflicting comments were noted (e.g. parameters reaching boundary).