/**** ***** ***** ***** ***** ***** ***** ***** ***** *****
[Author]
Toshiharu Mitsuhashi@Okayama University Hospital
[manuscript title]
Effects of Two-week e-Learning on eHealth Literacy: ///
A Randomized Controlled Trial of Japanese Internet Users
***** ***** ***** ********** ***** ***** ***** ***** ****/

version 15

set more off
capture log close
log using result_of_analysis, replace
use Dataset.dta, clear

***** ***** ***** ***** ***** ***** ***** *****
***** ***** Multiple Imutation (MI) ***** *****
***** ***** ***** ***** ***** ***** ***** *****

// MI general Setting
mi set mlong
mi misstable patterns, freq
mi register imputed diff_eHEALS
mi register imputed diff_HEL
mi register imputed diff_obj
local mi_pr sex area child marital job edu srh int_se age access
mi register regular `mi_pr'
// Execute MI(predictive mean matching;PMM)
mi impute chained (pmm, knn(5)) diff_eHEALS diff_HEL diff_obj= `mi_pr' ,add(30) rseed(43982)
// Save as a different file name
save Dataset_MI.dta, replace

***** ***** ***** ***** ***** ***** ***** ***** ***** *****
***** ***** Main Analysis (Intention To Treat)  ***** *****
***** ***** ***** ***** ***** ***** ***** ***** ***** *****

use Dataset.dta, replace
/* (Table.1) Baseline characteristics of the sample */
// Categorical variable
tab sex     intv,col
tab edu     intv,col
tab child   intv,col
tab hincome intv,col m
tab marital intv,col
tab job     intv,col
tab srh     intv,col
tab int_se  intv,col
// Continuous variable
sum age
bysort intv:sum age

/* (main text - Material & Method) Cronbach’s alpha */
// Cronbach's alpha of eHEALS
alpha pre_eHEALS1-pre_eHEALS8
alpha post_eHEALS1-post_eHEALS8
// Cronbach's alpha of HEL
alpha pre_HEL1-pre_HEL5
alpha post_HEL1-post_HEL5

/* (main text - Discussion) e-learning */
tab access if intv            // 27 subjects didn't learn at all.
gen complete=(percent==100)
tab complete if intv & access // 10 subjects learned only part of content.

/* (Table.2) Outcomes' Score (Continuous) */
// (Score at baseline)
bysort intv: sum pre_eHEALS  if fas
bysort intv: sum pre_HEL     if fas
// (Score at follow-up)
bysort intv: sum post_eHEALS if fas
bysort intv: sum post_HEL    if fas
// (Score change = Score at follow-up minus Score at baseline)
bysort intv: sum diff_eHEALS if fas
bysort intv: sum diff_HEL    if fas
// Liner regression analysis
regress diff_eHEALS intv if fas
regress diff_HEL    intv if fas
// Cohen's d
esize twosample diff_eHEALS if fas, by(intv)
esize twosample diff_HEL    if fas, by(intv)

/* (Table.3) Outcome's Score (Nomial) */
tab    pre_obj  intv if fas, col         // (At baseline)
tab    post_obj intv if fas, col         // (At follow-up)
tab    diff_obj intv if fas, col         // (Change=follow-up minus baseline)
mlogit diff_obj intv if fas, base(0) rrr // Multinomial logistic regression analysis

/* (Table.4) Estimation using MI dataset */
use Dataset_MI.dta, clear
mi estimate     : regress diff_eHEALS intv
mi estimate     : regress diff_HEL    intv
mi estimate, rrr: mlogit  diff_obj    intv, base(0)

***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ******
***** ***** Ancillary analysis (Per Protocol analysis) ***** *****
***** ***** ***** ***** ***** ***** ***** ***** ***** ***** ******

use Dataset.dta, clear
/* (Table.S1) Outcomes' Score (Continuous)*/
// (Score at baseline)
bysort intv: sum pre_eHEALS  if pps
bysort intv: sum pre_HEL     if pps
// (Score at follow-up)
bysort intv: sum post_eHEALS if pps
bysort intv: sum post_HEL    if pps
// (Score change = Score at follow-up minus Score at baseline)
bysort intv: sum diff_eHEALS if pps
bysort intv: sum diff_HEL    if pps
// Liner regression analysis
regress diff_eHEALS intv if pps
regress diff_HEL    intv if pps
// Cohen's d
esize twosample diff_eHEALS if pps, by(intv)
esize twosample diff_HEL    if pps, by(intv)

/* (Table.S2) Outcome's Score (Nomial) */
tab    pre_obj  intv if pps, col         // (At baseline)
tab    post_obj intv if pps, col         // (At follow-up)
tab    diff_obj intv if pps, col         // (Change=follow-up minus baseline)
mlogit diff_obj intv if pps, base(0) rrr // Multinomial logistic regression analysis

log close
set more on
exit