** Extrapolate the biomarker data to year 2014


include ../../../fem_env.do

use $outdata/hrs_analytic_recoded.dta

keep if wave >=8
 
keep hhidpn wave sysbp died iwstat wtresp age_yrs male black hispan smoken bmi hearte stroke diabe hibpe cancre rxchol

#d ;

merge 1:1 hhidpn wave using "$dua_rand_hrs/bio_hrs_recoded_impfinal.dta", keepusing(hhidpn wave biowgtr
                                                                                hdl hdl_imp 
                                                                                tchol tchol_imp
                                                                                a1c a1c_imp 
                                                                                sysbp_imp 
                                                                                crp_imp
                                                                                ) keep (master match) nogen

	;
#d cr




replace sysbp = . if sysbp == -333 

replace sysbp_imp = sysbp if wave == 12

xtset hhidpn wave

foreach v in hdl tchol a1c crp {

replace `v'_imp = `v'_imp[_n-1] - `v'_imp[_n-2] + `v'_imp[_n-1]  if `v'_imp == . & !died & wave == 12

}

sum hdl_imp
sum tchol_imp
sum a1c_imp
sum sysbp_imp
sum crp_imp


save $outdata/bio_temp.dta, replace

** Cap the values after the extrapolation

replace hdl_imp = 12 if hdl_imp < 12
replace hdl_imp = 146 if hdl_imp >= 146 & hdl_imp != .

replace tchol_imp = 61 if tchol_imp < 61
replace tchol_imp = 641 if tchol_imp >= 641 & tchol_imp != .

replace a1c_imp = 3 if a1c_imp < 3
replace a1c_imp = 17.6 if a1c_imp >= 17.6 & a1c_imp != .

replace sysbp_imp = 37 if sysbp_imp < 37
replace sysbp_imp = 227 if sysbp_imp >= 227 & sysbp_imp != .

gen logcrp_imp = log(crp_imp)
 


tempfile bio_temp
save `bio_temp'


	********************************************************
* Impute ldl data by NHANES, using imputed HDL and tchol
********************************************************
gen logbmi = log(bmi)
gen age_l60 = age_yrs<=60
gen age_6070 = (age_yrs>60 & age_yrs<=70) 
gen age_7080 = (age_yrs>70 & age_yrs<=80)  
gen age80p = (age_yrs >80) 
gen male_black = male*black
gen male_hispan = male*hispan

drop hdl tchol

rename hdl_imp hdl
rename tchol_imp tchol

count if hdl == . & !died
count if tchol == . & !died


gen tchol_hdl  = tchol*hdl
gen tchol_hearte = tchol*hearte
gen tchol_storke = tchol*stroke
gen tchol_diabe = tchol*diabe
gen tchol_hibpe = tchol*hibpe
gen tchol_cancre = tchol*cancre

gen hdl_hearte = hdl*hearte
gen hdl_storke = hdl*stroke
gen hdl_diabe = hdl*diabe
gen hdl_hibpe = hdl*hibpe
gen hdl_cancre = hdl*cancre

gen rxchol_stroke = rxchol*stroke
gen rxchol_hearte = rxchol*hearte
gen rxchol_diabe  = rxchol*diabe
gen rxchol_cancre = rxchol*cancre



count if tchol == . | hdl == . | tchol_hdl == . | age_l60 == . | age_6070 == . | age_7080 == .| age80p == .| male == . | black == . | hispan == . | smoken == . | logbmi == . | hearte == . | stroke == . | diabe == . | hibpe == . | cancre == . | rxchol == . 

est use "$local_path/Estimates/nhanes_biomarkers/nhanes_ldl.ster"

predict ldl_imp, xb

sum ldl_imp [aw=wtresp], detail

est use "$local_path/Estimates/nhanes_biomarkers/nhanes_ldl_minimal.ster"

predict ldl_min, xb

replace ldl_imp = ldl_min if ldl_imp == .

replace ldl_imp = 10.6 if ldl_imp <= 10.6

keep hhidpn wave ldl_imp

tempfile ldl_imp
save `ldl_imp'

	
use `bio_temp'

merge 1:1 hhidpn wave using `ldl_imp', keep (master match) nogen

save $outdata/bio_temp2.dta, replace

***************************************************************************************************
* Available imputed biomarker - ldl hdl tchol a1c sysbp
* Generate the biomarker_re as placing the population mean for the missing biomarkers
* Generate missing flag variables
* Generate l4 for all the variables
***************************************************************************************************


* Make a panel
xtset hhidpn wave


gen ldl = ldl_imp 
replace ldl = . if (hdl== . | tchol == .)

* Gen 2 year lagged variable for imputed hdl tchol ldl a1c sysbp
gen l2ldl_imp = L.ldl_imp
gen l2hdl_imp = L.hdl_imp
gen l2tchol_imp = L.tchol_imp
gen l2a1c_imp = L.a1c_imp
gen l2sysbp_imp = L.sysbp_imp
gen l2crp_imp = L.crp_imp
gen l2logcrp_imp = L.logcrp_imp


gen l2a1c = L.a1c
gen l2hdl = L.hdl
gen l2tchol = L.tchol
gen l2sysbp = L.sysbp
gen l2ldl = L.ldl
gen l2crp = L.crp



** Create biomarker_re, biomarker_miss

local bio hdl ldl tchol a1c sysbp crp logcrp

foreach v of local bio {
gen `v'_miss = (`v' == .)
label var `v'_miss "Whether `v' value is missing"

egen `v'_mean = mean(`v'), by(wave)
gen `v'_re = `v'

replace `v'_re = `v'_mean if `v' == .
label var `v'_re "Recoded `v' by placing the mean for missing values"

drop `v'_mean


gen l2`v'_miss = (l2`v'== .)
label var l2`v'_miss "Whether l2`v' value is missing"

}

save $outdata/bio_hrs_recoded_impfinalv2.dta, replace


*erase "$outdata/bio_temp.dta"
*erase "$outdata/bio_temp2.dta"