Mean and individual variability of pharmacokinetic parameters
| Effect | Δ211 | Parametric Model | Semiparametric Model | ||||
|---|---|---|---|---|---|---|---|
| keo | kmo | Cm50 | keo | kmo | Cm50 | ||
| Heart rate | −15.9 | 2.1 | 0.032 | 9.4 (6.6, 12.0) | 2.6 (1.7, 8.6) | 0.048 (0.028, 0.089) | 8.9 |
| Systolic pressure | −7.6 | – | 0.0097 (0, 0.043) | 6.7 (0.01, 12.4) | – | 0.0065 | 5.8 |
| Diastolic pressure | −1.7 | – | 0.070 (0.014, 0.19) | 9.9 (5.1, 19.4) | – | 0.066 | 9.0 |
| Mean pressure | −7.4 | – | 0.016 (0, 0.054) | 8.3 (0.03, 12.6) | – | 0.0082 | 6.5 |
| Energy expenditure | −65.7 | 1.6 | 0.00000094 | 0.000025 | 3.8 (2.0, 24.9) | 0.19 (0.13, 0.27) | 9.4 |
| Epinephrine | −1.3 | – | 0.0020 | 1.8 | – | 0.018 (0, 0.087) | 12.5 |
| Free fatty acids | −11.6 | 0.013 | 0.0000002 | 488 | 0.010 (0.006, 0.016) | – | – |
The values in brackets are the 95% confidence intervals obtained with a likelihood ratio profile. No lower bound for the confidence interval could be determined for some values of k mo, which indicates that the likelihood ratio profile in the corresponding direction is very flat and, accordingly, any (positive)k mo value below the upper bound cannot be excluded based on the data available. In these cases, a lower bound of 0 is indicated in the table. Note that C m 50 in the parametric model is the extrapolated one assuming that tolerance can suppress the effect completely, whereas in the semiparametric model, it is the observed one, i.e., the concentration that gives 50% of maximal observed tolerance. Because theC m 50 value in the nonparametric model is calculated from the tolerance model obtained and is not a parameter of the model itself, it is not feasible to run a likelihood ratio profile on this parameter, and correspondingly, no confidence intervals are available. Parameters in bold correspond to the model selected as the best model according to the Akaike criteria.