IOV and covariate effects on logit TVF
Dear NMusers,
In the past I have used the logit transform to obtain IIV on F in NONMEM and
used the following equations to get F1 and the SD on F1:
TVF = THETA(.)/(1-THETA(.))
LTVF= LOG(TVF) + ETA(.)
F1 = EXP(LTVF)/(1+EXP(LTVF))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
s.d.F1=SQRT(OMEGA)*THETA(.)*(1-THETA(.))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
However, recently I had to implement IOV and covariate effects on TVF. I came
up with these equations but I am not sure if the SD equations below are valid.
Furthermore, can I give a broader meaning or definition to the SD other than
the SD for a median BMI of 25.
OC1 = 0
OC2 = 0
OC3 = 0
IF (OCC.EQ.1) OC1 = 1
IF (OCC.EQ.2) OC2 = 1
IF (OCC.EQ.3) OC3 = 1
QCC1=OC1*ETA(x)
QCC2=OC2*ETA(y)
QCC3=OC3*ETA(z)
IOV=QCC1+QCC2+QCC3
TVF = THETA(x)*((BMI/25)**(THETA(y)))
IF (TVF.GE.1) TVF=.99
TPH = LOG(TVF/(1-TVF))
F1 = EXP(TPH+ETA(.)+IOV)/(1+EXP(TPH+ETA(.)+IOV))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
s.d.F1.IIV=SQRT(OMEGA.IIV)*(TVF.for.BMIof25)*(1-(TVF.for.BMIof25))
s.d.F1.IOV=SQRT(OMEGA.IOV)*(TVF.for.BMIof25)*(1-(TVF.for.BMIof25))
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Kindly advice...MNS