Further considerations on model-independent bioavailability estimation.

Assumptions attendant to model-independent bioavailability estimation were reexamined. Particular attention was given to the situation where an intravenous reference is not available and nonrenal clearance is assumed to be constant between treatments. Under these circumstances, the previously proposed approximation was compared with other bioavailability estimators. On the basis of error analysis, a procedure was devised to yield optimal relative bioavailability estimates.