rm(list=ls())

d<-read.csv("new_faces_demo_07may2013.csv")

site<-d$SITE
year.dx<-d$FIRST_INFECTION_YEAR
year.enrol<-d$YEAR_OF_ENROLLMENT
aids.enrol<-with(d, ifelse(is.na(ADE_PRIOR_FIRST_VISIT),"Unknown",
                           ifelse(ADE_PRIOR_FIRST_VISIT==1,"AIDS","not AIDS")))
male<-ifelse(d$SEX=="Male",1,0)
age.enrol<-d$AGE_AT_FIRST_VISIT
age.dx<-d$AGE_AT_FIRST_INFECTION
msm<-ifelse(d$RISK_FACTOR_1=="MSM",1,0)
mode<-ifelse(d$RISK_FACTOR_1=="Hetero","Heterosexual",
             ifelse(msm==1,"MSM",
                    ifelse(d$RISK_FACTOR_1=="Unknown","Unknown","Other")))
cd4.enrol<-with(d,ifelse(AGE_AT_BASELINE_CD4C-AGE_AT_FIRST_VISIT< -180/365.25 | 
                           AGE_AT_BASELINE_CD4C-AGE_AT_FIRST_VISIT>180/365.25,NA,
                         BASELINE_CD4_COUNT))

d.vandy<-data.frame(site,male,age.enrol,age.dx,
                    year.enrol,year.dx,msm,mode,
                    aids.enrol,cd4.enrol)

save(d.vandy,file="vandy-anna.Rda")