use "EMD_minimal_deidentified", clear 

*Table 1: Descriptive characteristics
preserve
keep if arm==1
tabstat age_exact, by(EMD) stat(min max mean sd p25 p50 p75)
tab EMD a04_sex, row
tab EMD c01_enrol, row
tab EMD c02_schtype, row
tab EMD sesquint, row
tab EMD pcgiver, row
tab EMD orphanhood, row
tab EMD vlcat, row
tabstat base25OHD, by(EMD) stat(min max mean sd p25 p50 p75)

*Table 2: Primary outcome 

*Unadjusted model
regress wk4825OHD i.EMD if wk4825OHD !=.

*Model adjusted for baseline 25(OH)D
regress  wk4825OHD i.EMD base25OHD if wk4825OHD !=.

*Table 2 secondary outcome 
regress adhpc48 i.EMD if wk4825OHD !=.

*Model adjusted for wk 0-24 pill count adherence 
regress adhpc48 i.EMD adhpc24 if wk4825OHD !=.

restore

*S1 Table: Descriptive characteristics comparing all participants following the week-24 re-randomization
tabstat age_exact, by(EMD) stat(min max mean sd p25 p50 p75)
tab EMD a04_sex, row
tab EMD c01_enrol, row
tab EMD c02_schtype, row
tab EMD sesquint, row
tab EMD pcgiver, row
tab EMD orphanhood, row
tab EMD vlcat, row
tabstat base25OHD, by(EMD) stat(min max mean sd p25 p50 p75)