Sequential and simultaneous exposure–response modeling 85
the joint model (Section 4.2.3). However, the SE s here are lower: 0.157 (rather
than 0.196) for ec50 and 0.056 (rather than 0.079) for emax, as the variation
in the fitted dose–exposur e model has not been counted for.
Next we implement the IPPSE approach in SAS, in a similar way as in
NONMEM. To this end, one needs to predict the individual parameters (here
cl and v) for each patient. The SEs of the predicted values are generated in
the output dataset. In the proc NLMIXED call for fitting the ER model, the
calculation o f c oncentrations is repeated with the predicted parameter values
plus random effects b1 and b2. Note that they have a different meaning than
in the model for the dose-PK relationship, ...