Table 3.

Main steps in population model buildinga

1One-compartment model, CL = θ1CLCR 1,476ςɛ 2= 16.8
2Two-compartment model, CL = θ1CLCR 1,325ςɛ 2 = 0.441; much better than step 1
3Infusion duration implemented as a random variable1,325Not better than step 2
4 C̅L̅ = θ1CLCR + θ2 1,290Better than step 2
5 C̅L̅ = θ1 × 6 × [θ3 − (age/100)]/(bw/S CR) + θ4, withi = 1 for males and 2 for females1,281Better than step 4c
6Similar to step 5, but θ1 = θ2 1,285Sex is a significant covariatec
7 to 10Typical values of either CL, V 1, CLD, orVt are allowed to differ according to o.d. or b.i.d. regimen1,282 to 1,287No influence of dosing regimen
11 to 14Similar to steps 7 to 10 but with separated η’s according to o.d. or b.i.d. regimen1,282 to 1,289No influence of dosing regimen
15 ̅V̅ ̅ ̅t̅ ̅ ̅ = θ7 + [(θ8 − θ7) × time]/(θ9 + time)1,281θ8 tends to θ7; no influence of time on Vt
16Similar to step 15, but forV 1 1,281No influence of length of therapy
17 ̅V̅ ̅1̅ ̅ = θ5 × (bw/65) + θ6 1,280Not significantc
18 ̅V̅ ̅ ̅t̅ ̅ ̅ = θ7(bw/65) + θ8 1,281Not significantc
19Similar to step 5, but with FOCE η-ɛ interaction method1,248ςɛ 2 = 0.189
20Similar to step 19, but C =Ĉ + ɛ1 Ĉb + ɛ2 with var (ɛ2) fixed to 0.25 1,258Not better than step 19
• a After step 2 the two-compartment model was always used. After step 6 the clearance model described in step 5 was always used.

• b OBFV, objective function value.

• c Additional criteria (see Materials and Methods section) were also considered for the decision.