forked from joshuawlambert/NutritionNHANES
-
Notifications
You must be signed in to change notification settings - Fork 0
/
DiabetesReport_counts.Rmd
182 lines (138 loc) · 5.59 KB
/
DiabetesReport_counts.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
171
172
173
174
175
176
177
178
179
180
181
---
title: "Diabetes MH and PH Report -- Analyzing as Counts"
author: "Joshua Lambert, PhD"
date: "11/6/2019"
output: pdf_document
---
## Loading the data
```{r,echo=FALSE, message=FALSE,}
m3 <- read.csv("~/CON_PROJECTS/NHANES_Alc/m3_counts.csv") #file created by datapackages.R
library(survey)
library(tableone)
library(knitr)
library(nhanesA)
library(plyr)
library(rFSA)
library(sjPlot)
library(sjmisc)
#dim(m3)
m3<-m3[which(m3$RIDAGEYR>=18),] #remove those less than 18.
#dim(m3)
adjust4b<-c('RIDAGEYR','gender','race','edu','ses','BMXBMI',
'SDMVPSU','SDMVSTRA','WTINT2YR')
nutr<-unique(c(grep(pattern = "lbx",x = tolower(colnames(m3))),grep(pattern = "lbd",x = tolower(colnames(m3)))))
candidate_vars<-c(adjust4b,colnames(m3)[nutr],"diab","mh","ph")
dat<-m3[,unique(candidate_vars)]
#dim(dat)
#factoring mh and ph as binary
#dat$mh<-factor(dat$mh)
#dat$ph<-factor(dat$ph)
#dichotomize countinuous variables 1=above median, 0=below median
medians<-lapply(X = dat[,c(10:54)],FUN = function(x){median(x,na.rm = TRUE)})
numNA<-lapply(X = dat[,c(10:54)],FUN = function(x){sum(is.na(x))})
mat<-cbind(t(t(medians)),t(t(numNA)))
kable(mat[-grep(rownames(mat),pattern = "SI"),],col.names = c("Median","NumNA"))
vita<-ifelse(dat$LBXVIA<60,yes = "Low-Average",no = "High")
dat[,c(10:54)]<-lapply(X = dat[,c(10:54)],FUN = function(x){factor(dicho(x,dich.by ="median"))})
dat$RIDAGEYR<-as.numeric(dat$RIDAGEYR)
dat$BMXBMI<-as.numeric(dat$BMXBMI)
dat$mh<-as.numeric(dat$mh)
dat$ph<-as.numeric(dat$ph)
vec<-grep(colnames(dat),pattern = "SI")
#vec<-which(apply(X = dat,MARGIN = 2,FUN = function(x){sum(is.na(x))})>8588) #columns with more than 50% data missing
dat<-dat[,-vec]
#leaves 17176 observations and 34 columns
dat$ph2<-ifelse(dat$ph>1,yes = 1,no = 0)
dat$mh2<-ifelse(dat$mh>1, yes=1, no=0)
```
\clearpage
\pagebreak
```{r, echo=FALSE}
kableone(CreateTableOne(vars=colnames(dat[,-c(7:10)]),data = dat[,-c(7:10)]),strata = c("diab"))
```
\clearpage
\pagebreak
```{r, echo=FALSE}
#ph
kableone(CreateTableOne(vars=colnames(dat[,-c(7:10,32,34,35,36)]),data = dat[,-c(7:10)],strata = c("ph2","diab")))
```
\clearpage
\pagebreak
```{r, echo=FALSE}
#mh
kableone(CreateTableOne(vars=colnames(dat[,-c(7:10,32,33,35,36)]),data = dat[,-c(7:10)],strata = c("mh2","diab")))
```
\clearpage
\pagebreak
# Mental Health
Analysis based on as low as 7110 subjects after removing all missingness.
```{r, echo=FALSE, include=FALSE}
##########################
#### Mental Health #######
##########################
nhanesdesign<-svydesign(id=~SDMVPSU,
strata=~SDMVSTRA,
weights=~WTINT2YR,
nest=TRUE,
data=dat)
null_model<-svyglm(formula = "mh~RIDAGEYR+gender+race+edu+ses+BMXBMI+ph+diab",design=nhanesdesign,family=quasipoisson,data=dat)
#race and BMI not significant will remove from null model and won't use in fixvar
null_model<-svyglm(formula = "mh~RIDAGEYR+gender+edu+ses+ph+diab",design=nhanesdesign,family=quasipoisson,data=dat)
dim(dat[complete.cases(dat[,-c(7:10)]),])
###########
#2-Way Int#
###########
fsafit_mh_2<-FSA(formula = "mh~1",fitfunc = svyglm,family=quasipoisson,data=dat[,-c(7:10,35,36)],design=nhanesdesign,m = 2,
numrs = 30,cores = 40,interactions = TRUE,criterion = int.p.val,minmax = 'min',
,fixvar = c(adjust4b[c(1:2,4:5)],"ph")
)
tab1<-data.frame(print(fsafit_mh_2))
```
```{r, echo=FALSE}
kable(tab1, caption = "Mental Health Results for Two-Way Interactions")
dat$via<-as.factor(vita)
nhanesdesign<-svydesign(id=~SDMVPSU,
strata=~SDMVSTRA,
weights=~WTINT2YR,
nest=TRUE,
data=dat)
plot_model(svyglm(formula = "mh~RIDAGEYR+gender+edu+ses+ph+LBXVIA*diab",
design=nhanesdesign,family=quasipoisson,data=dat), type = "int",
title = "LBXVIA*diab")
summary(svyglm(formula = "mh~RIDAGEYR+gender+edu+ses+ph+LBXVIA*diab",
design=nhanesdesign,family=quasipoisson,data=dat))
f1<-svyglm(formula = "mh~RIDAGEYR+gender+edu+ses+ph+LBXVIA*diab",
design=nhanesdesign,family=quasipoisson,data=dat)
```
\clearpage
\pagebreak
# Physical Health
```{r, echo=FALSE, include=FALSE}
##########################
#### Physical Health #######
##########################
nhanesdesign<-svydesign(id=~SDMVPSU,
strata=~SDMVSTRA,
weights=~WTINT2YR,
nest=TRUE,
data=dat)
null_model<-svyglm(formula = "ph~RIDAGEYR+gender+race+edu+ses+BMXBMI+mh",design=nhanesdesign,family=quasipoisson,data=dat)
#race not significant will remove from null model and won't use in fixvar
null_model<-svyglm(formula = "ph~RIDAGEYR+gender+edu+ses+BMXBMI+mh+diab",design=nhanesdesign,family=quasipoisson,data=dat)
###########
#2-Way Int#
###########
fsafit_ph_2<-FSA(formula = "ph~1",fitfunc = svyglm,family=quasipoisson,data=dat[,-c(7:10,35,36)],design=nhanesdesign,m = 2,
numrs = 30,cores = 40,interactions = TRUE,criterion = int.p.val,minmax = 'min',
fixvar = c(adjust4b[c(1:2,4:6)],"mh")
)
tab1<-data.frame(print(fsafit_ph_2))
```
```{r, echo=FALSE}
kable(tab1, caption = "Physical Health Results for Two-Way Interactions")
plot_model(svyglm(formula = "ph~RIDAGEYR+gender+edu+ses+BMXBMI+mh+LBXVIE*diab",
design=nhanesdesign,family=quasipoisson,data=dat), type = "int",
title = "LBXVIE*diab")
f2<-svyglm(formula = "ph~RIDAGEYR+gender+edu+ses+BMXBMI+mh+LBXVIE*diab",
design=nhanesdesign,family=quasipoisson,data=dat)
```