R Code: This is the coding used to compute the statistics using R.

The actual data cannot be posted by our agreement with GSK [in order to use their secure data portal for access]. So what's below is the basic form for the CSV files imported into R for the analysis, followed by the code for the calculations.

The Continuous Variables:

The .csv file format
PID,SITE,TRXNAME,WK1,WK2,WK3,WK4,WK5,WK6,WK7,WK8,WK8LOCF
329.001.00061,X001,IMIPRAMINE,0,0,-6,-9,-7,NA,NA,NA,-7
329.001.00066,X001,IMIPRAMINE,5,-10,NA,NA,NA,NA,NA,NA,-10
etc

And as it appears in R
              PID SITE    TRXNAME WK1 WK2 WK3 WK4 WK5 WK6 WK7 WK8 WK8LOCF
1   329.001.00061 X001 IMIPRAMINE   0   0  -6  -9  -7  NA  NA  NA      -7
2   329.001.00066 X001 IMIPRAMINE   5 -10  NA  NA  NA  NA  NA  NA     -10
etc.

The code
>  # ------------------------------------------------
>  # OMNIBUS WK8 OC
>  hamd.lm<-lm(WK8~TRXNAME*SITE,hamd)
>  anova(hamd.lm)
results 
[note: if the interaction term is not significant(<0.10), it is dropped from the model]
>  # ------------------------------------------------
>  hamd.lm<-lm(WK8~TRXNAME+SITE,hamd)
>  anova(hamd.lm)
results        
>  # ------------------------------------------------
>  lsmeans(hamd.lm,~TRXNAME)
results
>  # ------------------------------------------------
>  # OMNIBUS WK8 LOCF
>  hamd.lm<-lm(WK8LOCF~TRXNAME*SITE,hamd)
>  anova(hamd.lm) 
results 
[note: if the interaction term is not significant(<0.10), it is dropped from the model]   
>  # ------------------------------------------------
>  hamd.lm<-lm(WK8LOCF~TRXNAME+SITE,hamd)
>  anova(hamd.lm)  
results      
>  # ------------------------------------------------
>  lsmeans(hamd.lm,~TRXNAME) 
results
>  # ------------------------------------------------

The Categorical Variables:

The .csv file format
TRXNAME,SITE,WK1YES,WK1NO,WK2YES,WK2NO,WK3YES,WK3NO,WK4YES,WK4NO,WK5YES,WK5NO,WK6YES,WK6NOWK7YESWK7NOWK8YESWK8NOLOCF8YESLOCF8NO
IMIPRAMINE,X001,0,5,2,3,2,1,2,0,2,0,1,0,1,0,1,0,3,2
IMIPRAMINE,X002,0,11,3,7,6,4,4,4,5,3,3,3,5,1,5,1,6,5
etc

And as it appears in R
      TRXNAME SITE WK1YES WK1NO WK2YES WK2NO WK3YES WK3NO WK4YES WK4NO WK5YES WK5NO WK6YES WK6NO WK7YES WK7NO WK8YES WK8NO LOCF8YES LOCF8NO
1  IMIPRAMINE X001      0     5      2     3      2     1      2     0      2     0      1     0      1     0      1     0        3       2
2  IMIPRAMINE X002      0    11      3     7      6     4      4     4      5     3      3     3      5     1      5     1        6       5
etc

The code

# ------------------------------------------------
> # WEEK 8   HAM-D OMNIBUS  OC
> #
> hamd.glm <- glm(cbind(WK8YES, WK8NO) ~ TRXNAME * SITE, family=binomial, data=hamd)
> anova(hamd.glm, test="Chisq")
results 
[note: if the interaction term is not significant(<0.10), it is dropped from the model] 
#-------------------------------------------------
> # WEEK 8   HAM-D OMNIBUS  OC
> #
> hamd.glm <- glm(cbind(WK8YES, WK8NO) ~ TRXNAME + SITE, family=binomial, data=hamd)
> anova(hamd.glm, test="Chisq")
results
# ------------------------------------------------
> # WEEK 8   HAM-D OMNIBUS  LOCF
> #
> hamd.glm <- glm(cbind(LOCF8YES, LOCF8NO) ~ TRXNAME * SITE, family=binomial, data=hamd)
> anova(hamd.glm, test="Chisq")
results 
[note: if the interaction term is not significant(<0.10), it is dropped from the model] 
# ------------------------------------------------
> # WEEK 8   HAM-D OMNIBUS  LOCF
> #
> hamd.glm <- glm(cbind(LOCF8YES, LOCF8NO) ~ TRXNAME + SITE, family=binomial, data=hamd)
> anova(hamd.glm, test="Chisq")
results