generated from opensafely/research-template
/
MICE_data_prep1.R
170 lines (103 loc) · 5.98 KB
/
MICE_data_prep1.R
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
## Specify libraries
library(pacman)
library(tidyverse)
library(Hmisc)
library(here)
library(arrow)
library(purrr)
library(broom)
library(data.table)
library(forcats)
library(rstatix)
library(janitor)
library(lubridate)
library(skimr)
library(ggplot2)
library(mice)
BMI_trajectories <- read_feather (here::here ("output/data", "BMI_trajectories_final_demog.feather"))
## Remove infinity time change values
## Due to way BMI is extracted 141 patients with a value recorded on 1st March 2018 were counted in two time windows
## This created a time difference of 0 and therefore an infinity value with BMI change/time
## create a flag to identify when a time difference between BMI measures is recorded as '0' Then filter these out.
BMI_trajectories <- BMI_trajectories %>%
dplyr::mutate(timechange1_check = time_change1)
BMI_trajectories <- BMI_trajectories %>%
dplyr::mutate(time_change_error = timechange1_check)
BMI_trajectories$time_change_error[BMI_trajectories$time_change_error == 0] <- 1
BMI_trajectories$time_change_error[BMI_trajectories$time_change_error != 0] <- 0
BMI_trajectories %>%
tabyl(time_change_error)
########### IMPORTANT
## need to add this code to other trajectory analyses or NA will be filtered out
## Actually just want to filter out the infinity values
BMI_trajectories <- BMI_trajectories %>%
replace_na(list(time_change_error = 0))
BMI_trajectories <- BMI_trajectories %>%
dplyr::filter(time_change_error == 0)
## order the age-groups for ordered plots
BMI_trajectories$age_group_2 <- factor(BMI_trajectories$age_group_2, # Reordering group factor levels
levels = c("18-29", "30-39", "40-49", "50-59", "60-69", "70-79", "80+"))
BMI_trajectories$age_group <- factor(BMI_trajectories$age_group, # Reordering group factor levels
levels = c("18-39", "40-65", "65-80", "80+"))
BMI_trajectories$smoking_status <- factor(BMI_trajectories$smoking_status,
levels = c('N',"S", "E", "M"))
## selected the variables for analysis
## remove redundant columns
BMI_trajectories <- BMI_trajectories %>%
dplyr::select(-c(type1_diabetes,
type2_diabetes,
unknown_diabetes,
sbp_date_measured,
dbp_date_measured,
diabetes_type,
year,
had_bmi,
age_group,
ethnic_no_miss,
cholesterol_test,
insulin_meds,
oad_meds,
sbp,
dbp,
ends_with("_date"))) %>%
dplyr::rename(precovid_change = yearly_bmi_change1) %>%
dplyr::rename(postcovid_change = yearly_bmi_change2)
colnames(BMI_trajectories)
## filter out extreme BMI Change values. Likely to be error entries. Limits cover >95.5% of population
BMI_trajectories <- BMI_trajectories %>%
dplyr::mutate(post_covid_missing = postcovid_change) %>%
dplyr::mutate(post_covid_missing = replace_na(post_covid_missing, "missing"))
BMI_trajectories <- BMI_trajectories %>%
dplyr::filter(precovid_change>-6 & precovid_change<6) %>%
dplyr::filter((postcovid_change>-6 & postcovid_change<6) | post_covid_missing == "missing" )
## develop a variable for change in BMI_trajectory
BMI_trajectories <- BMI_trajectories %>%
ungroup()
BMI_trajectories <- BMI_trajectories %>%
dplyr::mutate(trajectory_change = postcovid_change - precovid_change)
## Categorise BMIs at different stages
BMI_trajectories <- BMI_trajectories %>%
dplyr:: mutate (base_bmi_category = base_bmi) %>%
dplyr::mutate(precovid_bmi_category = precovid_bmi) %>%
dplyr::mutate(postcovid_bmi_category = postcovid_bmi)
## categorise patients based on BMI at base, precovid and postcovid.
## done in base R for increased efficiency
BMI_trajectories$base_bmi_category[BMI_trajectories$base_bmi_category < 18.5] <- "underweight"
BMI_trajectories$base_bmi_category[BMI_trajectories$base_bmi_category >= 18.5 & BMI_trajectories$base_category <25] <- "healthy"
BMI_trajectories$base_bmi_category[BMI_trajectories$base_bmi_category >= 25 & BMI_trajectories$base_category <30] <- "overweight"
BMI_trajectories$base_bmi_category[BMI_trajectories$base_bmi_category >= 30 & BMI_trajectories$base_category <99] <- "obese"
BMI_trajectories$precovid_bmi_category[BMI_trajectories$precovid_bmi_category < 18.5] <- "underweight"
BMI_trajectories$precovid_bmi_category[BMI_trajectories$precovid_bmi_category >= 18.5 & BMI_trajectories$precovid_bmi_category <25] <- "healthy"
BMI_trajectories$precovid_bmi_category[BMI_trajectories$precovid_bmi_category >= 25 & BMI_trajectories$precovid_bmi_category <30] <- "overweight"
BMI_trajectories$precovid_bmi_category[BMI_trajectories$precovid_bmi_category >= 30 & BMI_trajectories$precovid_bmi_category <99] <- "obese"
BMI_trajectories$postcovid_bmi_category[BMI_trajectories$postcovid_bmi_category < 18.5] <- "underweight"
BMI_trajectories$postcovid_bmi_category[BMI_trajectories$postcovid_bmi_category >= 18.5 & BMI_trajectories$postcovid_bmi_category <25] <- "healthy"
BMI_trajectories$postcovid_bmi_category[BMI_trajectories$postcovid_bmi_category >= 25 & BMI_trajectories$postcovid_bmi_category <30] <- "overweight"
BMI_trajectories$postcovid_bmi_category[BMI_trajectories$postcovid_bmi_category >= 30 & BMI_trajectories$postcovid_bmi_category <99] <- "obese"
## Select only variable that could influence the imputation and used in models: ## can calculate postcovid_change from trajectory change and precvid change
BMI_trajectories <- BMI_trajectories %>%
dplyr::select(-c("bmi_change1", "bmi_change2", "time_change1", "time_change2",
"period1_missing", "period2_missing", "complete_bmi_data", "timechange1_check", "time_change_error","post_covid_missing", "postcovid_change"))
## Save
BMI_trajectories
write.csv (BMI_trajectories, here::here ("output/data", "imputation_data_long.csv"))