-
Notifications
You must be signed in to change notification settings - Fork 0
/
AppendixB.Rmd
171 lines (145 loc) · 6.39 KB
/
AppendixB.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
---
title: "AppendixB - EDA"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(ggplot2)
library(ggpubr)
library(gridExtra)
library(grid)
library(knitr)
```
### EDA of NHANES data
```{r}
# Load merged dataset for the years 2007-2008 through 2017-2018
load('./Data/masterDF.Rda')
df <- df %>% mutate(RIAGENDR = as.factor(RIAGENDR),
RIDRETH3 = as.factor(RIDRETH3),
DMDEDUC2 = as.factor(DMDEDUC2),
INDFMIN2 = as.factor(INDFMIN2),
DRQSDIET = as.factor(DRQSDIET),
DPQ020 = as.factor(DPQ020),
DIQ010 = as.factor(DIQ010)) %>%
mutate(Year = ifelse(origin.x == "DEMO_E.XPT", 2008,
ifelse(origin.x == "DEMO_F.XPT", 2010,
ifelse(origin.x == "DEMO_G.XPT", 2012,
ifelse(origin.x == "DEMO_H.XPT", 2014,
ifelse(origin.x == "DEMO_I.XPT", 2016, 2018)))))) %>%
filter(RIDAGEYR >=20)
# Add categorical variable for UWCB; UWCB=NA if no information is available for any of
# the criteria
df$UWCB <- NA
for (i in 1:nrow(df)){
if (is.na(df$DPQ050[i]) & is.na(df$WHD080E[i]) &
is.na(df$WHD080J[i]) & is.na(df$WHD080K[i])){
df$UWCB[i] = NA
}
else if((!is.na(df$DPQ050[i]) & df$DPQ050[i] %in% c(2,3)) | (!is.na(df$WHD080E[i]) & df$WHD080E[i] == 14) |
(!is.na(df$WHD080J[i]) & df$WHD080J[i] == 32) | (!is.na(df$WHD080K[i]) & df$WHD080K[i] == 33)){
df$UWCB[i] = 1
} else{
df$UWCB[i] = 0
}
}
table(df$UWCB)
sum(is.na(df$UWCB))
# remove observations with NA outcome
df <- df %>% drop_na(UWCB)
# Recode outcome as factor
df <- df %>% mutate(UWCB = as.factor(UWCB))
df2 <- df %>% dplyr::select(UWCB, RIDAGEYR, RIAGENDR, DMDEDUC2, INDFMIN2, DRQSDIET,
DPQ020, DR1TKCAL, BMXBMI, RIDRETH3, DMDHHSIZ,
DR1TPROT,DR1TCARB,DR1TSUGR, DR1TTFAT, DR1TCAFF, DR1TALCO, DSDCOUNT,
DIQ010,SLD012) %>% drop_na()
df2 <- df2 %>% mutate(UCWBlab = ifelse(UWCB==1, "UWCB", "No UWCB"))
```
```{r, fig.width= 10, warning=FALSE}
#Special Diet
options(scipen = 100)
df %>% summarize(`weight loss diet` = mean(df$DRQSDT1, na.rm=TRUE)/1,
`low fat diet` = mean(df$DRQSDT2, na.rm=TRUE)/2,
`low salt diet` = mean(df$DRQSDT3, na.rm=TRUE)/3,
`sugar free diet` = mean(df$DRQSDT4, na.rm=TRUE)/4,
`low fiber diet` = mean(df$DRQSDT5, na.rm=TRUE)/5,
`high fiber diet` = mean(df$DRQSDT6, na.rm=TRUE)/6,
`diabetic diet` = mean(df$DRQSDT7, na.rm=TRUE)/7,
`weight gain diet` = mean(df$DRQSDT8, na.rm=TRUE)/8,
`low carbs diet` = mean(df$DRQSDT9, na.rm=TRUE)/9,
`high protein diet` = mean(df$DRQSDT10, na.rm=TRUE)/10,
`other special diet` = mean(df$DRQSDT91, na.rm=TRUE)/91) %>%
gather("Diet", "Prevalence")
#Nutrition Intake
protein <- df %>%
ggplot(aes(y = DR1TPROT)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Protein intake (gm)")
carbs <- df %>%
ggplot(aes(y = DR1TCARB)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Carbohydrate intake (gm)")
sugars <- df %>%
ggplot(aes(y = DR1TSUGR)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Sugars intake (gm)")
fiber <- df %>%
ggplot(aes(y = DR1TFIBE)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Dietary fiber intake (gm)")
fat <- df %>%
ggplot(aes(y = DR1TTFAT)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Fat intake (gm)")
caffeine <- df %>%
ggplot(aes(y = DR1TCAFF)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Caffeine intake (mg)")
alcohol <- df %>%
ggplot(aes(y = DR1TALCO)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Alcohol intake (gm)")
water <- df %>%
ggplot(aes(y = DR1_320Z)) +
geom_boxplot(fill="grey") +
theme_minimal() + ylab("Water intake (gm)")
grid.arrange(protein, carbs, sugars, fiber, fat, caffeine, alcohol, water, ncol = 4)
```
### Assessing missingness for variables of interest
```{r}
# Assess missingness for covariates of interest
features = c("RIDAGEYR", "RIAGENDR", "DMDEDUC2", "INDFMIN2", "DRQSDIET",
"DPQ020", "DR1TKCAL", "BMXBMI", "RIDRETH3",
"DMDHHSIZ","DR1TPROT","DR1TCARB","DR1TSUGR",
"DR1TTFAT", "DR1TCAFF", "DR1TALCO", "DSDCOUNT", "DIQ010","SLD012")
tab <- as.data.frame(matrix(nrow=length(features), ncol=2))
tab$V1 <- c("Age", "Gender", "Education", "Income", "Special diet", "Depression",
"Energy intake (kcal)", "BMI",
"Ethnicity", "Household size", "Protein intake",
"Carbohydrate intake", "Sugar intake", "Fat intake",
"Caffeine intake", "Alcohol intake", "Dietary supplements",
"Diabetes", "Sleep")
colnames(tab) <- cbind("Variable", "Missingness [%]")
for (x in 1:length(features)){
tab$`Missingness [%]`[x] <- sum(is.na(df[features[x]]))/nrow(df)
}
tab <- tab %>% arrange(desc(`Missingness [%]`))
knitr::kable(tab)
```
### EDA plots
```{r}
#education,fewer participants with college graduate or above had UWCBs
df2 %>% ggplot(aes(x = UCWBlab,fill = DMDEDUC2)) +
geom_bar(position = "fill") +
labs(x="",y = "Proportion") +
scale_fill_discrete(name="Highest education level",
breaks=c("1", "2","3","4","5","77","99"),
labels=c("<9th grade","9-11th grade","high school graduate","college/AA degree","college graduate or above","Refused","Don't know"))
#total family income, fewer participants with high income had UWCBs
df2 %>% ggplot(aes(x = UCWBlab,fill = INDFMIN2)) +
geom_bar(position = "fill") +
labs(x="",y = "Proportion") +
scale_fill_discrete(name="Total family income",
breaks=c("1", "2","3","4","5","6","7","8","9","10","12","13","14","15","77","99"),
labels=c("0-4999","5000-9999","10000-14999","15000-19999","20000-24999","25000-34999","35000-44999","45000-54999","55000-64999","65000-74999", ">20000","<20000","75000-99999",">100000", "Refused","Don't know"))
```