library(tidyverse)
## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.2.1 ✔ readr 2.2.0
## ✔ forcats 1.0.1 ✔ stringr 1.6.0
## ✔ ggplot2 4.0.3 ✔ tibble 3.3.1
## ✔ lubridate 1.9.5 ✔ tidyr 1.3.2
## ✔ purrr 1.2.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library(readr)
library(dplyr)
library (lubridate)
library (ggplot2)
library (rmarkdown)
library(openxlsx)
library(writexl)
library(rstatix)
##
## Attaching package: 'rstatix'
##
## The following object is masked from 'package:stats':
##
## filter
library(eeptools)
## Registered S3 method overwritten by 'lme4':
## method from
## na.action.merMod car
library(ggpubr)
library(readxl)
library(BlandAltmanLeh)
library(openxlsx)
PSFI_df_malnutrition <- read_xlsx("PSFI_final_malnutrition.xlsx")
Create malnutrition categories and rules
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
filter(case_control == 1)
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
mutate(
wfhz_outlier = wfhz < -5 | wfhz > 5,
baz_outlier = baz < -5 | baz > 5,
wfhz_clean = ifelse(wfhz_outlier, NA, wfhz),
baz_clean = ifelse(baz_outlier, NA, baz)
)
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
mutate(
use_wfh = age_years < 5,
use_muac = age_months >= 3 & age_years < 5,
use_baz = age_years >= 2,
has_valid_measure =
(use_wfh & !is.na(wfhz)) |
(use_muac & !is.na(muac)) |
(use_baz & !is.na(baz)),
mam_muac = use_muac & !is.na(muac) & muac >= 11.5 & muac < 12.5,
sam_muac = use_muac & !is.na(muac) & muac < 11.5,
mam_wfh = use_wfh & !is.na(wfhz) & wfhz >= -3 & wfhz < -2,
sam_wfh = use_wfh & (
(!is.na(wfhz) & wfhz < -3) |
(!is.na(ht) & ht < 45)
),
overweight_wfh = use_wfh & !is.na(wfhz) & wfhz > 2 & wfhz <= 3,
obesity_wfh = use_wfh & !is.na(wfhz) & wfhz > 3,
mam_baz = use_baz & !is.na(baz) & baz >= -3 & baz < -2,
sam_baz = use_baz & !is.na(baz) & baz < -3,
overweight_baz_2_5 = use_baz & age_years <= 5 & !is.na(baz) & baz > 2 & baz <= 3,
obesity_baz_2_5 = use_baz & age_years <= 5 & !is.na(baz) & baz > 3,
overweight_baz_gt5 = use_baz & age_years > 5 & !is.na(baz) & baz > 1 & baz <= 2,
obesity_baz_gt5 = use_baz & age_years > 5 & !is.na(baz) & baz > 2,
overweight_baz = overweight_baz_2_5 | overweight_baz_gt5,
obesity_baz = obesity_baz_2_5 | obesity_baz_gt5,
mam = mam_wfh | mam_baz | mam_muac,
sam = sam_wfh | sam_baz | sam_muac,
overweight = overweight_wfh | overweight_baz,
obesity = obesity_wfh | obesity_baz,
malnutrition_who5 = case_when(
sam ~ "SAM",
mam ~ "MAM",
obesity ~ "Obesity",
overweight ~ "Overweight",
has_valid_measure ~ "Normal",
TRUE ~ NA_character_
)
)
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
mutate(
measure_used = case_when(
sam_muac ~ "MUAC",
sam_wfh ~ "WFH",
sam_baz ~ "BMI-for-age",
mam_muac ~ "MUAC",
mam_wfh ~ "WFH",
mam_baz ~ "BMI-for-age",
obesity_wfh | overweight_wfh ~ "WFH",
obesity_baz | overweight_baz ~ "BMI-for-age",
use_wfh & !is.na(wfhz) ~ "WFH",
use_baz & !is.na(baz) ~ "BMI-for-age",
use_muac & !is.na(muac) ~ "MUAC",
has_valid_measure ~ "Normal",
TRUE ~ NA_character_
)
)
measure_used_table <- PSFI_df_malnutrition %>%
count(measure_used) %>%
mutate(
percent = round(
100 * n / sum(n),
1
)
)
measure_used_table
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
filter(
case_control == 1
)
overview_table_long <- tibble(
category = c(
"SAM",
"MAM",
"Overweight",
"Obesity",
"Concurrent SAM + Overweight",
"Concurrent MAM + Overweight",
"Normal",
"Total"
),
n = c(
sum(PSFI_df_malnutrition$sam, na.rm = TRUE),
sum(
PSFI_df_malnutrition$mam &
!PSFI_df_malnutrition$sam,
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$overweight &
!PSFI_df_malnutrition$obesity &
!PSFI_df_malnutrition$sam &
!PSFI_df_malnutrition$mam,
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$obesity &
!PSFI_df_malnutrition$sam &
!PSFI_df_malnutrition$mam,
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$sam &
PSFI_df_malnutrition$overweight,
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$mam &
PSFI_df_malnutrition$overweight &
!PSFI_df_malnutrition$sam,
na.rm = TRUE
),
sum(
!PSFI_df_malnutrition$sam &
!PSFI_df_malnutrition$mam &
!PSFI_df_malnutrition$overweight &
!PSFI_df_malnutrition$obesity,
na.rm = TRUE
),
nrow(PSFI_df_malnutrition)
)
) %>%
mutate(
percent = round(100 * n / nrow(PSFI_df_malnutrition), 1)
)
overview_table_long
measure_table <- tibble(
measure = c(
"WFH used",
"MUAC used",
"BMI-for-age used",
"WFH + MUAC overlap",
"WFH + BMI overlap",
"MUAC + BMI overlap",
"All three measures"
),
n = c(
sum(
PSFI_df_malnutrition$use_wfh &
!is.na(PSFI_df_malnutrition$wfhz),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_muac &
!is.na(PSFI_df_malnutrition$muac),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_baz &
!is.na(PSFI_df_malnutrition$baz),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_wfh &
!is.na(PSFI_df_malnutrition$wfhz) &
PSFI_df_malnutrition$use_muac &
!is.na(PSFI_df_malnutrition$muac),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_wfh &
!is.na(PSFI_df_malnutrition$wfhz) &
PSFI_df_malnutrition$use_baz &
!is.na(PSFI_df_malnutrition$baz),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_muac &
!is.na(PSFI_df_malnutrition$muac) &
PSFI_df_malnutrition$use_baz &
!is.na(PSFI_df_malnutrition$baz),
na.rm = TRUE
),
sum(
PSFI_df_malnutrition$use_wfh &
!is.na(PSFI_df_malnutrition$wfhz) &
PSFI_df_malnutrition$use_muac &
!is.na(PSFI_df_malnutrition$muac) &
PSFI_df_malnutrition$use_baz &
!is.na(PSFI_df_malnutrition$baz),
na.rm = TRUE
)
)
) %>%
mutate(
percent = round(
100 * n / nrow(PSFI_df_malnutrition),
1
)
)
measure_table
WHO5
PSFI_df_malnutrition <- PSFI_df_malnutrition
bar_x <- PSFI_df_malnutrition %>%
mutate(
malnutrition_who5 = factor(
malnutrition_who5,
levels = c(
"SAM",
"MAM",
"Normal",
"Overweight",
"Obesity"
)
)
) %>%
count(malnutrition_who5, .drop = FALSE)
print(bar_x)
## # A tibble: 6 × 2
## malnutrition_who5 n
## <fct> <int>
## 1 SAM 154
## 2 MAM 107
## 3 Normal 378
## 4 Overweight 55
## 5 Obesity 60
## 6 <NA> 1
barplot(
bar_x$n,
names.arg = bar_x$malnutrition_who5,
col = rgb(1, 0.8, 0.8, 1),
border = "white",
ylab = "Count",
las = 2,
cex.names = 0.8
)
malnutrition_who5_table <- PSFI_df_malnutrition %>%
mutate(
malnutrition_who5 = factor(
malnutrition_who5,
levels = c(
"SAM",
"MAM",
"Normal",
"Overweight",
"Obesity"
)
)
) %>%
count(malnutrition_who5, .drop = FALSE) %>%
mutate(
percent = round(100 * n / sum(n), 1)
)
malnutrition_who5_table
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
mutate(
# WHO5-style hierarchical category
# 0 = Normal
# 1 = MAM
# 2 = SAM
# 3 = Overweight
# 4 = Obesity
malnutrition_who5 = case_when(
sam ~ 2L,
mam ~ 1L,
obesity ~ 4L,
overweight ~ 3L,
has_valid_measure ~ 0L,
TRUE ~ NA_integer_
),
# Optional labels
malnutrition_who5_label = factor(
case_when(
malnutrition_who5== 0 ~ "Normal",
malnutrition_who5 == 1 ~ "MAM",
malnutrition_who5 == 2 ~ "SAM",
malnutrition_who5 == 3 ~ "Overweight",
malnutrition_who5 == 4 ~ "Obesity"
),
levels = c(
"SAM",
"MAM",
"Normal",
"Overweight",
"Obesity"
)
)
)
WHO3
PSFI_df_malnutrition <- PSFI_df_malnutrition %>%
mutate(
malnutrition_who3 = case_when(
is.na(malnutrition_who5) ~ NA_integer_,
malnutrition_who5 == 1 ~ 1L, # Malnutrition
malnutrition_who5 == 2 ~ 1L, # Malnutrition
malnutrition_who5 == 3 ~ 2L, # Overweight
malnutrition_who5 == 4 ~ 2L, # Overweight
TRUE ~ 0L # Normal
),
malnutrition_who3_label = factor(
case_when(
malnutrition_who3 == 0 ~ "Normal",
malnutrition_who3 == 1 ~ "Malnutrition",
malnutrition_who3 == 2 ~ "Overweight"
),
levels = c(
"Malnutrition",
"Normal",
"Overweight"
)
)
)
PSFI_df_malnutrition <- PSFI_df_malnutrition
bar_x <- PSFI_df_malnutrition %>%
count(malnutrition_who3_label, .drop = FALSE)
print(bar_x)
## # A tibble: 4 × 2
## malnutrition_who3_label n
## <fct> <int>
## 1 Malnutrition 261
## 2 Normal 378
## 3 Overweight 115
## 4 <NA> 1
barplot(
bar_x$n,
names.arg = bar_x$malnutrition_who3_label,
col = rgb(1, 0.8, 0.8, 1),
border = "white",
ylab = "Count",
las = 2,
cex.names = 0.9
)
# =========================================
# WHO3 TABLE
# =========================================
malnutrition_cat_table_who3 <- PSFI_df_malnutrition %>%
mutate(
malnutrition_who3_label = factor(
malnutrition_who3_label,
levels = c(
"Malnutrition",
"Normal",
"Overweight"
)
)
) %>%
count(malnutrition_who3_label, .drop = FALSE) %>%
mutate(
percent = round(
100 * n / sum(n),
1
)
)
malnutrition_cat_table_who3
PSFI_df_malnutrition %>%
summarise(
missing_who5 = sum(is.na(malnutrition_who5)),
missing_who3 = sum(is.na(malnutrition_who3))
)
write.xlsx(PSFI_df_malnutrition, file = "PSFI_final_malnutrition2.xlsx")
Post analysis
discordant_cases <- PSFI_df_malnutrition %>%
filter(
# malnut_mod_sev = 1 but WHO5 says Normal/Overweight/Obesity
(malnut_mod_sev == 1 &
malnutrition_who5 %in% c(0, 3, 4)) |
# malnut_mod_sev = 0 but WHO5 says MAM/SAM
(malnut_mod_sev == 0 &
malnutrition_who5 %in% c(1, 2))
) %>%
select(
record_id,
malnut_mod_sev,
malnutrition_who3_label,
malnutrition_who5_label,
measure_used,
age_months,
muac,
baz,
bmiz,
wfhz,
whz,
sam,
ht,
wt,
ss_skin___5,
)
discordant_cases
discordant_summary <- discordant_cases %>%
count(
malnut_mod_sev,
malnutrition_who5_label
)
discordant_summary
malnutrition_categories_table <- PSFI_df_malnutrition %>%
select(
subject_id,
record_id,
malnut_mod_sev,
malnutrition_who3,
malnutrition_who5,
malnutrition_who3_label,
malnutrition_who5_label
)
malnutrition_categories_table
write.xlsx(
list(
malnutrition_categories_table = malnutrition_categories_table,
discordant_cases = discordant_cases
),
file = "malnutrition_categories_v2.xlsx",
overwrite = TRUE
)