Data_cleaning.Rmd

---
title: "prevalent_data_cleaning_1015"
author: "Mingzhou_Fu"
date: "10/15/2019"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r Basic setups, message=FALSE, warning=FALSE}
rm(list = ls())
library(tidyverse)
library(writexl)
library(haven)   # read in SAS file
library(compareGroups)

raw_data_path = '/Users/Mingzhou/Desktop/AD_Grant/HRS_raw/'
work_data_path = '/Users/Mingzhou/Desktop/AD_Grant/Cholesterol/data/'
output_path = '/Users/Mingzhou/Desktop/AD_Grant/Cholesterol/output/prevalent/'

# Original dataset
raw_dataset = 'hrs_democoggen_wide20190813.Rda'
load(paste0(raw_data_path, raw_dataset)) # -- N = 43,232
# Biomarkers
load(paste0(raw_data_path, "biomk06bl_r.rda"))
load(paste0(raw_data_path, "biomk08bl_r.rda"))
load(paste0(raw_data_path, "biomk10bl_r.rda"))
load(paste0(raw_data_path, "biomk12bl_r.rda"))
# Medication variables
load(paste0(raw_data_path, "h06n_r.rda"))
load(paste0(raw_data_path, "h08n_r.rda"))
load(paste0(raw_data_path, "h10n_r.rda"))
load(paste0(raw_data_path, "h12n_r.rda"))
# Physical activity variables
load(paste0(raw_data_path, "h06c_r.rda"))
load(paste0(raw_data_path, "h08c_r.rda"))
load(paste0(raw_data_path, "h10c_r.rda"))
load(paste0(raw_data_path, "h12c_r.rda"))
# Cognition plus
load(paste0(raw_data_path, "cogimp9216a_r.rda"))
# Kunkle PGS
load(paste0(raw_data_path, "european_kunkle_pgss.rda"))
load(paste0(raw_data_path, "african_kunkle_pgss.rda"))
```

# Combine all data together
```{r Join all the data, message=FALSE, warning=FALSE}
med06 = 
  med06 %>% 
  dplyr::select(HHID, PN, KN360)
med08 = 
  med08 %>% 
  dplyr::select(HHID, PN, LN360)
med10 = 
  med10 %>% 
  dplyr::select(HHID, PN, MN360)
med12 = 
  med12 %>% 
  dplyr::select(HHID, PN, NN360)

# Join all the waves together, restrict to 2010, 2012, 2014 -- N = 24,916
pre_full_data = 
  hrs %>% 
  left_join(biomk06, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(biomk08, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(biomk10, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(biomk12, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(med06, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(med08, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(med10, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(med12, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(h06c_r, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  mutate(KC223 = replace(KC223, KC223 == 8 | KC223 == 9, NA)) %>% 
  mutate(KC224 = replace(KC224, KC224 == 8 | KC224 == 9, NA)) %>% 
  mutate(KC225 = replace(KC225, KC225 == 8 | KC225 == 9, NA)) %>% 
  mutate(vigorous_06 = case_when(
    KC223 == 7 ~ 5,
    KC223 == 1 ~ 4,
    KC223 == 2 ~ 3,
    KC223 == 3 ~ 2,
    KC223 == 4 ~ 1
  )) %>% 
  mutate(moderate_06 = case_when(
    KC224 == 7 ~ 5,
    KC224 == 1 ~ 4,
    KC224 == 2 ~ 3,
    KC224 == 3 ~ 2,
    KC224 == 4 ~ 1
  )) %>% 
  mutate(mild_06 = case_when(
    KC225 == 7 ~ 5,
    KC225 == 1 ~ 4,
    KC225 == 2 ~ 3,
    KC225 == 3 ~ 2,
    KC225 == 4 ~ 1
  )) %>% 
  mutate(exercise_06 = (vigorous_06 + moderate_06 + mild_06)/3) %>% 
  left_join(h08c_r, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  mutate(LC223 = replace(LC223, LC223 == 8 | LC223 == 9, NA)) %>% 
  mutate(LC224 = replace(LC224, LC224 == 8 | LC224 == 9, NA)) %>% 
  mutate(LC225 = replace(LC225, LC225 == 8 | LC225 == 9, NA)) %>% 
  mutate(vigorous_08 = case_when(
    LC223 == 7 ~ 5,
    LC223 == 1 ~ 4,
    LC223 == 2 ~ 3,
    LC223 == 3 ~ 2,
    LC223 == 4 ~ 1
  )) %>% 
  mutate(moderate_08 = case_when(
    LC224 == 7 ~ 5,
    LC224 == 1 ~ 4,
    LC224 == 2 ~ 3,
    LC224 == 3 ~ 2,
    LC224 == 4 ~ 1
  )) %>% 
  mutate(mild_08 = case_when(
    LC225 == 7 ~ 5,
    LC225 == 1 ~ 4,
    LC225 == 2 ~ 3,
    LC225 == 3 ~ 2,
    LC225 == 4 ~ 1
  )) %>% 
  mutate(exercise_08 = (vigorous_08 + moderate_08 + mild_08)/3) %>% 
  left_join(h10c_r, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  mutate(MC223 = replace(MC223, MC223 == 8 | MC223 == 9, NA)) %>% 
  mutate(MC224 = replace(MC224, MC224 == 8 | MC224 == 9, NA)) %>% 
  mutate(MC225 = replace(MC225, MC225 == 8 | MC225 == 9, NA)) %>% 
  mutate(vigorous_10 = case_when(
    MC223 == 7 ~ 5,
    MC223 == 1 ~ 4,
    MC223 == 2 ~ 3,
    MC223 == 3 ~ 2,
    MC223 == 4 ~ 1
  )) %>% 
  mutate(moderate_10 = case_when(
    MC224 == 7 ~ 5,
    MC224 == 1 ~ 4,
    MC224 == 2 ~ 3,
    MC224 == 3 ~ 2,
    MC224 == 4 ~ 1
  )) %>% 
  mutate(mild_10 = case_when(
    MC225 == 7 ~ 5,
    MC225 == 1 ~ 4,
    MC225 == 2 ~ 3,
    MC225 == 3 ~ 2,
    MC225 == 4 ~ 1
  )) %>% 
  mutate(exercise_10 = (vigorous_10 + moderate_10 + mild_10)/3) %>% 
  left_join(h12c_r, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  mutate(NC223 = replace(NC223, NC223 == 8 | NC223 == 9, NA)) %>% 
  mutate(NC224 = replace(NC224, NC224 == 8 | NC224 == 9, NA)) %>% 
  mutate(NC225 = replace(NC225, NC225 == 8 | NC225 == 9, NA)) %>% 
  mutate(vigorous_12 = case_when(
    NC223 == 7 ~ 5,
    NC223 == 1 ~ 4,
    NC223 == 2 ~ 3,
    NC223 == 3 ~ 2,
    NC223 == 4 ~ 1
  )) %>% 
  mutate(moderate_12 = case_when(
    NC224 == 7 ~ 5,
    NC224 == 1 ~ 4,
    NC224 == 2 ~ 3,
    NC224 == 3 ~ 2,
    NC224 == 4 ~ 1
  )) %>% 
  mutate(mild_12 = case_when(
    NC225 == 7 ~ 5,
    NC225 == 1 ~ 4,
    NC225 == 2 ~ 3,
    NC225 == 3 ~ 2,
    NC225 == 4 ~ 1
  )) %>% 
  mutate(exercise_12 = (vigorous_12 + moderate_12 + mild_12)/3) %>% 
  left_join(kunkle_europ, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  left_join(kunkle_afric, by = c('HHID' = 'HHID', 'PN' = 'PN'))
# N = 43,232

# Select potential useful variables
pre_full_data = 
  pre_full_data %>% 
  mutate(PGS_AD_IGAP = case_when(
    !is.na(AA_PGS3_AD_IGAP13) ~ AA_PGS3_AD_IGAP13,
    !is.na(EA_PGS3_AD_IGAP13) ~ EA_PGS3_AD_IGAP13
  )) %>% 
  mutate(PGS_AD_Kunkle = case_when(
    !is.na(AAPH13_KunkleNAPOE_PT_1) ~ AAPH13_KunkleNAPOE_PT_1,
    !is.na(EAPH13_KunkleNAPOE_PT_1) ~ EAPH13_KunkleNAPOE_PT_1
  )) %>% 
  mutate(gen_ancestry = case_when(
    !is.na(AA_PGS3_HDL_GLGC13) ~ "African",
    !is.na(EA_PGS3_HDL_GLGC13) ~ "European")) %>% 
  # combine all PGS regardless of ancestry
  mutate(PGS_HDL = case_when(
    !is.na(AA_PGS3_HDL_GLGC13) ~ AA_PGS3_HDL_GLGC13,
    !is.na(EA_PGS3_HDL_GLGC13) ~ EA_PGS3_HDL_GLGC13
  )) %>% 
  mutate(PGS_TC = case_when(
    !is.na(AA_PGS3_TC_GLGC13) ~ AA_PGS3_TC_GLGC13,
    !is.na(EA_PGS3_TC_GLGC13) ~ EA_PGS3_TC_GLGC13
  )) %>% 
  mutate(PC1_5A = case_when(
    gen_ancestry == "European" ~ eaPC1_5A,
    gen_ancestry == "African" ~ AAPC1_5A
  )) %>% 
  mutate(PC1_5B = case_when(
    gen_ancestry == "European" ~ eaPC1_5B,
    gen_ancestry == "African" ~ AAPC1_5B
  )) %>% 
  mutate(PC1_5C = case_when(
    gen_ancestry == "European" ~ eaPC1_5C,
    gen_ancestry == "African" ~ AAPC1_5C
  )) %>% 
  mutate(PC1_5D = case_when(
    gen_ancestry == "European" ~ eaPC1_5D,
    gen_ancestry == "African" ~ AAPC1_5D
  )) %>% 
  mutate(PC1_5E = case_when(
    gen_ancestry == "European" ~ eaPC1_5E,
    gen_ancestry == "African" ~ AAPC1_5E
  )) %>% 
   # complete dbs: no missing in baseline DBS HDL/TC
  mutate(dbs_2006 = case_when(
    (!is.na(KHDL_ADJ) | !is.na(KTC_ADJ)) ~ 1,
    TRUE ~ 0
  )) %>% 
  mutate(dbs_2008 = case_when(
    (!is.na(LHDL_ADJ) | !is.na(LTC_ADJ)) ~ 1,
    TRUE ~ 0
  )) %>% 
  mutate(dbs_2010 = case_when(
    (!is.na(MHDL_ADJ) | !is.na(MTC_ADJ)) ~ 1,
    TRUE ~ 0
  )) %>% 
  mutate(dbs_2012 = case_when(
    (!is.na(NHDL_ADJ) | !is.na(NTC_ADJ)) ~ 1,
    TRUE ~ 0
  )) %>% 
  mutate(PGS_GENCOG = case_when(
    !is.na(AA_PGS3_GENCOG_CHRG15) ~ AA_PGS3_GENCOG_CHRG15,
    !is.na(EA_PGS3_GENCOG_CHRG15) ~ EA_PGS3_GENCOG_CHRG15
  )) %>% 
  dplyr::select(HHID, PN, STUDY, AD04, AD06, AD08, AD10, AD12, AD14, STROKE06, STROKE08, STROKE10, STROKE12, 
                PGS_HDL, PGS_TC, PGS_AD_IGAP, PGS_AD_Kunkle, PGS_GENCOG, 
                KHDL_ADJ, LHDL_ADJ, MHDL_ADJ, NHDL_ADJ, KTC_ADJ, LTC_ADJ, MTC_ADJ, NTC_ADJ, 
                gen_ancestry, BIRTHYR, GENDER, SCHLYRS, BEALE2003_06, BEALE2003_08, BEALE2003_10, BEALE2003_12, APOE2010_bin, 
                PC1_5A, PC1_5B, PC1_5C, PC1_5D, PC1_5E, dbs_2006, dbs_2008, dbs_2010, dbs_2012, 
                # HTN 
                R8HIBPE, R9HIBPE, R10HIBPE, R11HIBPE, 
                # Diabetes
                R8DIABE, R9DIABE, R10DIABE, R11DIABE, 
                # BMI
                R8BMI, R9BMI, R10BMI, R11BMI, 
                # Smoking status
                smoke06, smoke08, smoke10, smoke12, 
                # Alcohol
                R8DRINK, R9DRINK, R10DRINK, R11DRINK, 
                # medication
                KN360, LN360, MN360, NN360, 
                # Physical activity
                exercise_06, exercise_08, exercise_10, exercise_12,
                # Proxy level
                KA009, LA009, MA009, NA009)
# N = 43,232
```

# Wave selection
```{r Wave selection, message=FALSE, warning=FALSE}
# Filter observations have at least one DBS data -- N = 18,700
dbs_full = 
  pre_full_data %>% 
  dplyr::filter(dbs_2006 == 1 | dbs_2008 == 1 | dbs_2010 == 1 | dbs_2012 == 1) 
```

# Recode some of the variables
```{r}
full_sample = 
  dbs_full %>% 
  mutate(wave_adj = case_when(
    dbs_2006 == 1 ~ '2006',
    dbs_2008 == 1 ~ '2008',
    dbs_2010 == 1 ~ '2010', 
    dbs_2012 == 1 ~ '2012'
  )) %>% 
  mutate(AD = case_when(
    wave_adj == '2006' ~ AD06,
    wave_adj == '2008' ~ AD08,
    wave_adj == '2010' ~ AD10,
    wave_adj == '2012' ~ AD12
  )) %>% 
  mutate(AD_former = case_when(
    wave_adj == '2006' ~ AD04,
    wave_adj == '2008' ~ AD06,
    wave_adj == '2010' ~ AD08,
    wave_adj == '2012' ~ AD10
  )) %>% 
  mutate(HDL = case_when(
    wave_adj == '2006' ~ KHDL_ADJ,
    wave_adj == '2008' ~ LHDL_ADJ,
    wave_adj == '2010' ~ MHDL_ADJ,
    wave_adj == '2012' ~ NHDL_ADJ
  )) %>% 
  mutate(TC = case_when(
    wave_adj == '2006' ~ KTC_ADJ,
    wave_adj == '2008' ~ LTC_ADJ,
    wave_adj == '2010' ~ MTC_ADJ,
    wave_adj == '2012' ~ NTC_ADJ
  )) %>% 
  mutate(age = case_when(
    wave_adj == '2006' ~ 2006 - BIRTHYR,
    wave_adj == '2008' ~ 2008 - BIRTHYR,
    wave_adj == '2010' ~ 2010 - BIRTHYR,
    wave_adj == '2012' ~ 2012 - BIRTHYR
  )) %>% 
  mutate(sex = GENDER - 1) %>% 
  mutate(living = case_when(
    wave_adj == '2006' ~ BEALE2003_06,
    wave_adj == '2008' ~ BEALE2003_08,
    wave_adj == '2010' ~ BEALE2003_10,
    wave_adj == '2012' ~ BEALE2003_12
  )) %>% 
  mutate(education = SCHLYRS) %>% 
  mutate(stroke = case_when(
    wave_adj == '2006' ~ STROKE06,
    wave_adj == '2008' ~ STROKE08,
    wave_adj == '2010' ~ STROKE10,
    wave_adj == '2012' ~ STROKE12
  )) %>% 
  mutate(med = case_when(
    wave_adj == '2006' ~ KN360,
    wave_adj == '2008' ~ LN360,
    wave_adj == '2010' ~ MN360,
    wave_adj == '2012' ~ NN360
  )) %>% 
  mutate(med_lipid_cat = case_when(
    med == 1 ~ "Yes",
    med == 5 ~ "No"
  )) %>% 
  mutate(HTN = case_when(
    wave_adj == '2006' ~ R8HIBPE,
    wave_adj == '2008' ~ R9HIBPE,
    wave_adj == '2010' ~ R10HIBPE,
    wave_adj == '2012' ~ R11HIBPE
  )) %>% 
  mutate(diabetes = case_when(
    wave_adj == '2006' ~ R8DIABE,
    wave_adj == '2008' ~ R9DIABE,
    wave_adj == '2010' ~ R10DIABE,
    wave_adj == '2012' ~ R11DIABE
  )) %>% 
  mutate(BMI = case_when(
    wave_adj == '2006' ~ R8BMI,
    wave_adj == '2008' ~ R9BMI,
    wave_adj == '2010' ~ R10BMI,
    wave_adj == '2012' ~ R11BMI
  )) %>% 
  mutate(smoke = case_when(
    wave_adj == '2006' ~ smoke06,
    wave_adj == '2008' ~ smoke08,
    wave_adj == '2010' ~ smoke10,
    wave_adj == '2012' ~ smoke12
  )) %>% 
  mutate(drink = case_when(
    wave_adj == '2006' ~ R8DRINK,
    wave_adj == '2008' ~ R9DRINK,
    wave_adj == '2010' ~ R10DRINK,
    wave_adj == '2012' ~ R11DRINK
  )) %>% 
  mutate(proxy = case_when(
    wave_adj == '2006' ~ KA009,
    wave_adj == '2008' ~ LA009,
    wave_adj == '2010' ~ MA009,
    wave_adj == '2012' ~ NA009
  )) %>% 
  mutate(exercise = case_when(
    wave_adj == '2006' ~ exercise_06,
    wave_adj == '2008' ~ exercise_08,
    wave_adj == '2010' ~ exercise_10,
    wave_adj == '2012' ~ exercise_12
  )) %>% 
  dplyr::select(HHID, PN, STUDY, AD, AD_former, HDL, TC, PGS_HDL, PGS_TC, PGS_AD_IGAP, PGS_AD_Kunkle, PGS_GENCOG, 
                gen_ancestry, age, sex, living, education, stroke, wave_adj, APOE2010_bin, med_lipid_cat,
                PC1_5A, PC1_5B, PC1_5C, PC1_5D, PC1_5E, HTN, diabetes, BMI, smoke, drink, proxy, exercise)

# Make categorical variables
full_sample = 
  full_sample %>% 
  mutate(AD_cat = case_when(
    AD == 3 ~ "Normal",
    AD == 2 ~ "CIND",
    AD == 1 ~ "Dementia"
  )) %>% 
  mutate(AD = case_when(
    AD_cat == "Normal" ~ 0,
    AD_cat == "CIND" ~ 1,
    AD_cat == "Dementia" ~ 2
  )) %>% 
  mutate(AD_cat_former = case_when(
    AD_former == 3 ~ "Normal",
    AD_former == 2 ~ "CIND",
    AD_former == 1 ~ "Dementia"
  )) %>% 
  mutate(AD_former = case_when(
    AD_cat_former == "Normal" ~ 0,
    AD_cat_former == "CIND" ~ 1,
    AD_cat_former == "Dementia" ~ 2
  )) %>% 
  mutate(sex_cat = case_when(
    sex == 0 ~ "Male",
    sex == 1 ~ "Female"
  )) %>% 
  mutate(living_cat = case_when(
    living == 1 ~ "Urban",
    living == 2 ~ "Suburban",
    living == 3 ~ "Exurban"
  )) %>% 
  mutate(stroke_cat = case_when(
    stroke == 1 ~ "Yes",
    stroke == 0 ~ "No"
  )) %>% 
  mutate(APOE2010_cat = case_when(
    APOE2010_bin == 0 ~ "No",
    APOE2010_bin == 1 ~ "Yes"
  )) %>% 
  mutate(HTN_cat = case_when(
    HTN == 0 ~ "No",
    HTN == 1 ~ "Yes"
  )) %>% 
  mutate(diabetes_cat = case_when(
    diabetes == 0 ~ "No",
    diabetes == 1 ~ "Yes"
  )) %>% 
  mutate(smoking_cat = case_when(
    smoke == 0 ~ "Never",
    smoke == 1 ~ "Former",
    smoke == 2 ~ "Current"
  )) %>% 
  mutate(drinking_cat = case_when(
    drink == 0 ~ "Never",
    drink == 1 ~ "Ever"
  )) %>% 
  mutate(med_lipid = case_when(
    med_lipid_cat == "Yes" ~ 1,
    med_lipid_cat == "No" ~ 0
  )) %>% 
  mutate(proxy_cat = case_when(
    proxy == 1 ~ 'Self',
    proxy == 2 | proxy == 3 | proxy == 4 ~ 'Proxy'
  ))
```

# Insert the domain cognition data
```{r Insert domain, message=FALSE, warning=FALSE}
domain_cog = 
  cognition_plus %>% 
  dplyr::select(HHID, PN, 
                R8IMRC, R9IMRC, R10IMRC, R11IMRC,    # immediate recall
                R8DLRC, R9DLRC, R10DLRC, R11DLRC,    # delayed recall
                R8SER7, R9SER7, R10SER7, R11SER7,    # serial's 7
                R8BWC20, R9BWC20, R10BWC20, R11BWC20,   # backward counting from 20
                R8VOCAB, R9VOCAB, R10VOCAB, R11VOCAB,    # vocabulary
                R8TR20, R9TR20, R10TR20, R11TR20,    # total recall
                R8MSTOT, R9MSTOT, R10MSTOT, R11MSTOT,   # Total mental status
                R8COGTOT, R9COGTOT, R10COGTOT, R11COGTOT   # Total cognition
                )

domain_full = 
  full_sample %>% 
  left_join(domain_cog, by = c('HHID' = 'HHID', 'PN' = 'PN')) %>% 
  mutate(im_recall = case_when(
    wave_adj == "2012" ~ R11IMRC,
    wave_adj == "2010" ~ R10IMRC,
    wave_adj == "2008" ~ R9IMRC,
    wave_adj == "2006" ~ R8IMRC
  )) %>% 
  mutate(dl_recall = case_when(
    wave_adj == "2012" ~ R11DLRC,
    wave_adj == "2010" ~ R10DLRC,
    wave_adj == "2008" ~ R9DLRC,
    wave_adj == "2006" ~ R8DLRC
  )) %>% 
  mutate(total_recall = case_when(
    wave_adj == "2012" ~ R11TR20,
    wave_adj == "2010" ~ R10TR20,
    wave_adj == "2008" ~ R9TR20,
    wave_adj == "2006" ~ R8TR20
  )) %>% 
  mutate(serial7 = case_when(
    wave_adj == "2012" ~ R11SER7,
    wave_adj == "2010" ~ R10SER7,
    wave_adj == "2008" ~ R9SER7,
    wave_adj == "2006" ~ R8SER7
  )) %>% 
  mutate(backwc = case_when(
    wave_adj == "2012" ~ R11BWC20,
    wave_adj == "2010" ~ R10BWC20,
    wave_adj == "2008" ~ R9BWC20,
    wave_adj == "2006" ~ R8BWC20
  )) %>% 
  mutate(vocab = case_when(
    wave_adj == "2012" ~ R11VOCAB,
    wave_adj == "2010" ~ R10VOCAB,
    wave_adj == "2008" ~ R9VOCAB,
    wave_adj == "2006" ~ R8VOCAB
  )) %>% 
  mutate(ms_total = case_when(
    wave_adj == "2012" ~ R11MSTOT,
    wave_adj == "2010" ~ R10MSTOT,
    wave_adj == "2008" ~ R9MSTOT,
    wave_adj == "2006" ~ R8MSTOT
  )) %>% 
  mutate(cog_total = case_when(
    wave_adj == "2012" ~ R11COGTOT,
    wave_adj == "2010" ~ R10COGTOT,
    wave_adj == "2008" ~ R9COGTOT,
    wave_adj == "2006" ~ R8COGTOT
  )) %>% 
  dplyr::select(-c(R8IMRC, R9IMRC, R10IMRC, R11IMRC, 
                R8DLRC, R9DLRC, R10DLRC, R11DLRC, 
                R8TR20, R9TR20, R10TR20, R11TR20, 
                R8SER7, R9SER7, R10SER7, R11SER7, 
                R8BWC20, R9BWC20, R10BWC20, R11BWC20, 
                R8VOCAB, R9VOCAB, R10VOCAB, R11VOCAB, 
                R8MSTOT, R9MSTOT, R10MSTOT, R11MSTOT, 
                R8COGTOT, R9COGTOT, R10COGTOT, R11COGTOT))

domain_full = 
  domain_full %>% 
  mutate(HDL_clinic = case_when(
    HDL < 40 & sex_cat == 'Male' ~ 1,
    HDL < 50 & sex_cat == 'Female' ~ 1,
    !is.na(HDL) ~ 0
  )) %>% 
  mutate(TC_clinic = case_when(
    TC >= 240 ~ 2,
    TC >= 200 ~ 1,
    !is.na(TC) ~ 0
  )) %>% 
  mutate(HDL_clinic_cat = case_when(
    HDL_clinic == 1 ~ 'Risk',
    HDL_clinic == 0 ~ 'Under'
  )) %>% 
  mutate(TC_clinic_cat = case_when(
    TC_clinic == 2 ~ 'High',
    TC_clinic == 1 ~ 'Boarderline',
    TC_clinic == 0 ~ 'Good'
  )) %>% 
  mutate(TC_HDL_ratio = TC/HDL)

save(domain_full, file = paste0(work_data_path, "prevalent_full.rda"))
```


# Sample selection
```{r Sample selection steps, message=FALSE, warning=FALSE}
# Get rid of those who don't have genetic data -- N = 14,064
genetic_full = 
  domain_full %>% 
  dplyr::filter(!is.na(gen_ancestry))

# Exclude those w/ missing values in age, sex, education, living, cohort, PC1_5A -- N = 11,362
# Updated -- also excluded those w/ missing values in health behaviors: stroke, BMI, HTN, diabetes, smoking, drink status -- N = 11,173
complete_data = 
  genetic_full %>% 
  dplyr::filter(!is.na(age) & !is.na(sex) & !is.na(education) & !is.na(wave_adj) & !is.na(PC1_5A) & !is.na(med_lipid_cat)) %>% 
  dplyr::filter(!is.na(stroke) & !is.na(BMI) & !is.na(HTN) & !is.na(diabetes) & !is.na(smoking_cat) & !is.na(drink))

# Restrict to people age <=90 to avoid survival bias -- N = 10,911
age_restrict = 
  complete_data %>% 
  dplyr::filter(age <= 90 & age >= 50)

# No reversed cognition -- N = 29
reverse_sample_AD =
  age_restrict %>% 
  filter((AD_cat_former == "Dementia") & AD_cat == "Normal") 
final_sample = 
  age_restrict %>% 
  anti_join(reverse_sample_AD, by = c('HHID' = 'HHID', 'PN' = 'PN'))
# N = 10,882

final_sample = 
  final_sample %>% 
  mutate(TC_clinic_bin = case_when(
    TC_clinic_cat == 'Good' | TC_clinic_cat == 'Boarderline' ~ 0,
    TC_clinic_cat == 'High' ~ 1
  )) %>% 
  mutate(TC_clinic_bin_cat = case_when(
    TC_clinic_bin == 0 ~ "Good",
    TC_clinic_bin == 1 ~ "High"
  ))
# N = 10,882
# Exclude those in the AHEAD and CODA
# study_restrict = 
  # age_restrict %>% 
  # filter(STUDY != 11 & STUDY != 21)
```

# Separate by Ancestry
```{r Separate by ancestry, message=FALSE, warning=FALSE}
europ =
  final_sample %>% 
  dplyr::filter(gen_ancestry == 'European')
# N = 8,781
afric = 
  final_sample %>% 
  dplyr::filter(gen_ancestry == 'African')
# N = 2,101

save(final_sample, file = paste0(work_data_path, "prevalent_eligible.rda"))
save(europ, file = paste0(work_data_path, "prevalent_europ.rda"))
save(afric, file = paste0(work_data_path, "prevalent_afric.rda"))
```