use "D:\S3_Dataset.DTA.dta", clear

//Descriptive statistics
drop prevalent_TB
tabstat Patient_Age, statistics( mean sd median min max iqr ) columns(statistics)
tabulate Age
tabulate Gender
tabulate Stage
tabstat BASE_CD4, statistics( mean sd median min max iqr ) columns(statistics)
tabulate Base_CD4
gen prevalent_TB=0
replace prevalent_TB=1 if (TB_date-Patient_ARTstartDateWestern)>-240 & (TB_date-Patient_ARTstartDateWestern)<3000
tabulate prevalent_TB
tabulate Adherence_2
tabulate Last_status
tabulate Age Active, chi2 row
tabulate Gender Active, chi2 row 
tabulate Stage Active, chi2 row
tabulate Base_CD4 Active, chi2 row
//Immunological failure at first evaluation
tabulate Outcome
//Immunological failure at first evaluation excluding not evaluated
proportion Outcome if Outcome<2
//Immunological failure by type of criteria 
tab First_failure_type if Outcome==1
//Immunologic failure at last evaluation
tab failure_final
//Secondline for those with first immunological failure
tabulate Secondline if Outcome==1
//tab failure_confirmed
//Transition after last immunological evaluation
tab failure_confirmed if Outcome==1
//Secondline for those with confirmed immunological failure
tabulate Secondline if failure_confirmed==4
//Supporting information Table 1: IMMUNOLOGIC FAILURE STATUS AFTER FOLLOW-UP IMMUNOLOGIC EVALUATION BY IMMUNOLOGIC FAILURE DIGNOSTIC CRITERIA 
tab First_failure_type failure_confirmed if Outcome==1 & failure_confirmed!=2, row chi2
//Supporting information Table 2: IMMUNOLOGIC FAILURE STATUS AFTER FOLLOW-UP IMMUNOLOGIC EVALUATION BY BASELINE CHRACTERISTICS 
tab Age failure_confirmed if Outcome==1 & failure_confirmed!=2, row chi2
tab Gender failure_confirmed if Outcome==1 & failure_confirmed!=2, row chi2
tab Stage failure_confirmed if Outcome==1 & failure_confirmed!=2, row chi2
tab Base_CD4 failure_confirmed if Outcome==1 & failure_confirmed!=2, row chi2
//Setting up for survival analysis
drop if Outcome==2
stset time, id(Card) failure(Outcome==1) scale(1)
//Descriptive survival analysis
stdescribe
by Outcome, sort : stdescribe
//Life table
ltable time Outcome, survival
//Splitting data by tuberculosis status
//generate tbtime=time if TB==0
drop Tuberculosis
generate tbtime=(TB_date-Patient_ARTstartDateWestern)/365 if TB==1 & (TB_date-Patient_ARTstartDateWestern)>=180 & (TB_date-Patient_ARTstartDateWestern)<10000
//replace tbtime=time if TB==1 & (TB_date-Patient_ARTstartDateWestern)<=180
stsplit Tuberculosis, at(0) after(_t=tbtime)
replace Tuberculosis = Tuberculosis + 1
gen tb=1 if tbtime>0 & tbtime<10
replace Outcome=0 if tb==1 & Outcome==.
replace Tuberculosis=0 if Tuberculosis==1 & tbtime==.
//Splitting data by treatment interruption status
generate losttime=(first_lost-Patient_ARTstartDateWestern)/365
stsplit off_trt, at(0) after(_t=losttime)
replace off_trt = off_trt + 1
gen lost=1 if losttime>0 & losttime<10
replace Outcome=0 if lost==1 & Outcome==.
replace off_trt=0 if off_trt==1 & losttime==.
//Regression analysis
drop if Outcome==2
stset time, id(Card) failure(Outcome==1) scale(1)
stcox  Patient_Age, strata(site) nohr
stcox  Stage, strata(site) nohr
stcox  Gender, strata(site) nohr
stcox  Tuberculosis, strata(site) nohr
stcox  BASE_CD4, strata(site) nohr
stcox  off_trt, strata(site) nohr
stcox  BASE_CD4 Stage, strata(site) nohr
estimate store test1
stcox  BASE_CD4, strata(site) nohr
lrtest . test1
stcox  Stage, strata(site) nohr
lrtest . test1
stcox BASE_CD4 Stage Gender, strata(site) nohr
lrtest . test1
stcox BASE_CD4 Stage Patient_Age, strata(site) nohr
lrtest . test1
stcox BASE_CD4 Stage Tuberculosis, strata(site) nohr
lrtest . test1
stcox BASE_CD4 Stage off_trt, strata(site) nohr
lrtest . test1
//Interaction
stcox BASE_CD4 ib(0).Stage c.BASE_CD4#b(0).Stage, strata(site)
lrtest . test1
//Preliminary model
stcox BASE_CD4 Stage, strata(site) nohr