-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathHIV Paper R Script.R
416 lines (357 loc) · 15.1 KB
/
HIV Paper R Script.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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
library(haven) #To read Stata dataset into R
library(descr)
# SECTION 1: LOADING AND CLEANING DATASET
wom<- read_dta("C://Dataset//ZAIR71FL.DTA") #Import women's dataset
hiv<- read_dta("C://Dataset//ZAAR71FL.DTA") #Import HIV dataset
#Select relevant variables from the main dataset
sub<- subset(wom, select = c(v001,v002,v003,v005,v012,v013,v025,v133,v149,v745a,v170,v169a,v190,v201,v501,v502,v531,v836,v834a,v714,v731,v739,
v833a,v766b,v850a,v850b,d104,d106,d107,d108,v746,v473b, d005))
#Create similar "ID" variables in the women's dataset and HIV dataset for merging
sub$ID <- as.numeric(paste(sub$v001, sub$v002, sub$v003, sep = ""))
hiv$ID <- as.numeric(paste(hiv$hivclust, hiv$hivnumb, hiv$hivline, sep = ""))
#Merge both datasets by women's ID
mer<-merge(sub,hiv,by="ID")
#Retain people who have ever been in a union (currently or previously had a partner or husband)
mer<- subset(mer, mer$v502!=0 )
#Create HIV assay result variable
mer$hiv<-NA
mer$hiv[mer$hiv03==0]<-0
mer$hiv[mer$hiv03==1]<-1
freq(mer$hiv)
mer<- subset(mer, mer$hiv!="NA" )
#Age
mer$age<- mer$v012
mer$ageCAT<-NA
mer$ageCAT[mer$age>17 & mer$age<=29]<-1
mer$ageCAT[mer$age>=30 & mer$age<=39]<-2
mer$ageCAT[mer$age>=40 & mer$age<=49]<-3
mer$ageCAT <- factor(mer$ageCAT,
levels = c(1,2,3),
labels = c("18-29", "30-39", "40-49"))
freq(mer$age)
#Residence
mer$rsd<-NA
mer$rsd[mer$v025==1]<-1
mer$rsd[mer$v025==2]<-2
mer$rsd<- factor(mer$rsd,
levels = c(1,2),
labels=c("Urban", "Rural"))
freq(mer$rsd)
#Education
mer$edu<- mer$v133 #education in years
mer$edu1<-NA
mer$edu1[mer$v149==0]<-4
mer$edu1[mer$v149==1]<-4
mer$edu1[mer$v149==2]<-3
mer$edu1[mer$v149==3]<-3
mer$edu1[mer$v149==4]<-2
mer$edu1[mer$v149==5]<-1
mer$edu1<- factor(mer$edu1,
levels = c(1,2,3,4),
labels=c("Higher","Completed Secondary","Completed Primary","Less than primary"))
freq(mer$edu1)
mer$edu2<-NA
mer$edu2[mer$v149==0]<-2
mer$edu2[mer$v149==1]<-2
mer$edu2[mer$v149==2]<-2
mer$edu2[mer$v149==3]<-2
mer$edu2[mer$v149==4]<-1
mer$edu2[mer$v149==5]<-1
mer$edu2<- factor(mer$edu2,
levels = c(1,2),
labels=c("Completed Secondary", "Less than Secondary"))
freq(mer$edu2)
#Marital status
mer$mar<- NA
mer$mar[mer$v501==1]<-1
mer$mar[mer$v501==2]<-2
mer$mar[mer$v501==3|mer$v501==4|mer$v501==5]<-1
mer$mar<- factor(mer$mar,
levels = c(1,2),
labels=c("Ever married", "Never married"))
freq(mer$mar)
#Gainfully employed in the past year or not
mer$work<-NA
mer$work[mer$v731==1]<-1
mer$work[mer$v731==2]<-1
mer$work[mer$v731==3]<-1
mer$work[mer$v731==0]<-2
mer$work<- factor(mer$work,
levels = c(1,2),
labels=c("Yes", "No"))
#Wealth Index
mer$wealth<-NA
mer$wealth[mer$v190==5]<-1
mer$wealth[mer$v190==4]<-2
mer$wealth[mer$v190==3]<-3
mer$wealth[mer$v190==2]<-4
mer$wealth[mer$v190==1]<-5
mer$wealth<- factor(mer$wealth,
levels = c(1,2,3,4,5),
labels=c("Richest", "Rich", "Middle", "Poor","Poorest"))
freq(mer$wealth)
#Has bank account
mer$bank<-NA
mer$bank[mer$v170==1]<-1
mer$bank[mer$v170==0]<-2
mer$bank<- factor(mer$bank,
levels = c(1,2),
labels=c("Yes", "No"))
freq(mer$bank)
#Earns more than partner
mer$earn<-NA
mer$earn[mer$v746==1]<-1
mer$earn[mer$v746==2]<-2
mer$earn[mer$v746==3]<-2
mer$earn[mer$v746==4]<-1
mer$earn[mer$v746==8]<-2
mer$earn[is.na(mer$earn)]<-2
mer$earn<- factor(mer$earn,
levels = c(1,2),
labels=c("Yes", "No"))
freq(mer$earn)
#Independently decides how her earnings are spent
mer$dec<-NA
mer$dec[mer$v739==1]<-1
mer$dec[mer$v739==2]<-2
mer$dec[mer$v739==4]<-2
mer$dec[is.na(mer$dec)]<-2
mer$dec<- factor(mer$dec,
levels = c(1,2),
labels=c("Yes", "No"))
freq(mer$dec)
#Age-disparate relationship with last partner
mer$disp<- NA
mer$disp<- ifelse(mer$v834a-mer$age >=10, 2,1)
mer$disp[is.na(mer$disp)]<-1
mer$disp<- factor(mer$disp,
levels = c(1,2),
labels=c("No", "Yes"))
freq(mer$disp)
#Number of sexual partners in past year
mer$partners<-mer$v766b
#Multiple sexual partners in past year
mer$multi<-NA
mer$multi[mer$v766b==0]<-1
mer$multi[mer$v766b==1]<-1
mer$multi[mer$v766b==2]<-2
mer$multi[mer$v766b==3]<-2
mer$multi[mer$v766b==4]<-2
mer$multi[mer$v766b==95]<-2
mer$multi<- factor(mer$multi,
levels = c(1,2),
labels=c("No", "Yes"))
freq(mer$multi)
#Age at sexual debut <15yrs
mer$debut<-ifelse(mer$v531<15, 1,0)
freq(mer$debut)
#Can refuse sex
mer$ref<-NA
mer$ref[mer$v850a==1]<-1
mer$ref[is.na(mer$ref)]<-2
mer$ref<- factor(mer$ref,
levels = c(1,2),
labels=c("Yes", "No"))
freq(mer$ref)
#Can ask partner to use a condom
mer$ask<-NA
mer$ask[mer$v850b==1]<-1
mer$ask[is.na(mer$ask)]<-2
mer$ask<- factor(mer$ask,
levels = c(1,2),
labels=c("Yes", "No"))
freq(mer$ask)
#Experienced Intimate Partner Violence
mer$IPV<-NA
mer$IPV[mer$d104==0&mer$d106==0&mer$d107==0&mer$d108==0]<-1
mer$IPV[mer$d104==1&mer$d106==0&mer$d107==0&mer$d108==0]<-2
mer$IPV[mer$d106==1]<-3
mer$IPV[mer$d107==1]<-3
mer$IPV[mer$d108==1]<-4
mer$IPV<- factor(mer$IPV,
levels = c(1,2,3,4),
labels=c("None", "Emotional only", "Physical", "Sexual"))
freq(mer$IPV)
mer$cluster<-mer$v001 #Cluster variable
mer$hivwt<- mer$hiv05/1000000 #HIV Weights
mer<- subset(mer, select = -c(v012,v025,v133,v149,v731,v190,v170,v169a,v745a,v502,
v746,v739,v834a,v833a,v201,v850a,v850b,v766b,
v001,v002,v003,v005,v013,v501, v502,v531,v836,v714,v473b,
hivclust,hivnumb,hivline,hiv01,hiv02,hiv03,hiv07,hiv08,
d005,hiv05, hiv06)) #remove redundant variables
rm(wom,hiv) #remove redundant datasets
#Create Women's Financial Autonomy Index
mer$fin<- NA
mer$fin[is.na(mer$fin)]=0
mer$fin[mer$work=="Yes"]<-(mer$fin+1)
mer$fin[mer$bank=="Yes"]<-(mer$fin+1)
mer$fin[mer$earn=="Yes"]<-(mer$fin+1)
mer$fin[mer$dec=="Yes"]<-(mer$fin+1)
freq(mer$fin)
#Create tertiles of financial autonomy index
quantile(mer$fin, probs = seq(0, 1, 1/3))
mer$fin2<-NA
mer$fin2[mer$fin==0]<-3
mer$fin2[mer$fin==1]<-2
mer$fin2[mer$fin>=2]<-1
mer$fin2<- factor(mer$fin2,
levels = c(1,2,3),
labels=c("High", "Medium", "Low"))
freq(mer$fin2)
#Create Women's Sexual Autonomy Index
mer$sex<- NA
mer$sex[is.na(mer$sex)]=0
mer$sex[mer$ref=="Yes"]<-(mer$sex+1)
mer$sex[mer$ask=="Yes"]<-(mer$sex+1)
freq(mer$sex)
#Create tertiles of sexual autonomy index
quantile(mer$sex, probs = seq(0, 1, 1/3))
mer$sex2<-NA
mer$sex2[mer$sex==0]<-3
mer$sex2[mer$sex==1]<-2
mer$sex2[mer$sex==2]<-1
mer$sex2<- factor(mer$sex2,
levels = c(1,2,3),
labels=c("High", "Medium", "Low"))
mer1<-subset(mer, mer$IPV!="NA")
# SECTION 2: ANALYSIS
#TABLE 1: Descriptive Stats
library(expss) #This package enables the computation of frequencies using sampling weights
fre(mer1$edu1, weight = mer1$hivwt)
fre(mer1$ageCAT, weight = mer1$hivwt)
fre(mer1$rsd, weight = mer1$hivwt)
fre(mer1$mar, weight = mer1$hivwt)
fre(mer1$wealth, weight = mer1$hivwt)
fre(mer1$work, weight = mer1$hivwt)
fre(mer1$bank, weight = mer1$hivwt)
fre(mer1$earn, weight = mer1$hivwt)
fre(mer1$dec, weight = mer1$hivwt)
fre(mer1$ref, weight = mer1$hivwt)
fre(mer1$ask, weight = mer1$hivwt)
fre(mer1$IPV, weight = mer1$hivwt)
fre(mer1$debut, weight = mer1$hivwt)
fre(mer1$multi, weight = mer1$hivwt)
fre(mer1$disp, weight = mer1$hivwt)
fre(mer1$hiv, weight = mer1$hivwt)
#TABLE 2: Relationships between main explanatory varaibles
crosstab(mer1$fin2, mer1$IPV, prop.c = T, chisq = T) #crosstab of IPV vs financial autonomy
crosstab(mer1$sex2, mer1$IPV, prop.r = T, chisq = T) #crosstab of IPV vs sexual autonomy
crosstab(mer1$sex2, mer1$fin2, prop.r = T, chisq = T)#crosstab of sexual and financial autonomy
cor(mer1$fin, mer1$sex) #correlation of sexual and financial autonomy indices
#Stacked bar graph in fig 1
#Table below was prepared from the crosstabulation of financial and sexual autonomy above)
sex<- c(rep("cLow", 3), rep("bMedium", 3), rep("aHigh", 3))
fin<- c("cLow", "bMedium", "aHigh", "cLow", "bMedium", "aHigh", "cLow", "bMedium", "aHigh")
frequency<- c(72,92,54,69,83,86,98,90,85)
tab<-data.frame(sex,fin,frequency)
library(ggplot2)
library(RColorBrewer)
ggplot(tab, aes(fill=fin, y=frequency, x=sex)) +
geom_bar(position="fill", stat="identity")+
labs(x="Sexual autonomy", y="Proportion", col="Site")+
scale_fill_brewer(palette = "Dark2")+
theme_bw()
#TABLE 3, Section A: Bivariate associations of HIV with women's characteristics
crosstab(mer1$hiv, mer1$edu1, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$ageCAT, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$rsd, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$mar, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$wealth, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$fin2, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$work, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$bank, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$earn, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$dec, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$sex2, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$ref, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$ask, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$IPV, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$debut, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$disp, prop.c = T, chisq = T)
crosstab(mer1$hiv, mer1$multi, prop.c = T, chisq = T)
#TABLE 3, Section B: Bivariate logistic regressions of HIV with women's characteristics
library(miceadds) #Executes cluster-robust standard errors for general linear models
library(sandwich) #Required by miceadds
lm1<-glm.cluster(hiv~edu1, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs education
lm1<-glm.cluster(hiv~ageCAT, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs age
lm1<-glm.cluster(hiv~rsd, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs residence
lm1<-glm.cluster(hiv~mar, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs marital status
lm1<-glm.cluster(hiv~wealth, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs household wealth
lm1<-glm.cluster(hiv~fin2, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs financial autonomy tertiles
lm1<-glm.cluster(hiv~work, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs working status
lm1<-glm.cluster(hiv~bank, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs personal bank account
lm1<-glm.cluster(hiv~earn, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs earns more than partner
lm1<-glm.cluster(hiv~dec, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs decides how earnings are spent
lm1<-glm.cluster(hiv~sex2, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs sexual autonomy
lm1<-glm.cluster(hiv~ref, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs can refuse sex
lm1<-glm.cluster(hiv~ask, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs can ask partner to use condom
lm1<-glm.cluster(hiv~IPV, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs IPV
lm1<-glm.cluster(hiv~debut, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs age at sexual debut <=15 years
lm1<-glm.cluster(hiv~disp, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs age disparate relationship
lm1<-glm.cluster(hiv~multi, data=mer1, cluster = mer1$cluster, family=binomial)
exp(cbind(OR = coef(lm1), confint.default(lm1))) #HIV vs multiple sexual partners
# TABLE 4: Multivariate logistic regression model
#Quadratic term for age
mer1$ages<-mer1$age^2
#Model
freq(mer1$mar)
lm1<-glm.cluster(hiv~sex2+fin2+IPV+edu2+age+ages+rsd+mar+wealth+partners, data=mer1, cluster = mer1$cluster, family=binomial)
summary(lm1)
exp(cbind(OR = coef(lm1), confint.default(lm1)))
plot(lm1$glm_res)
# SECTION 3: LOGISTIC REGRESSION MODEL DIAGNOSTICS
# Predict the probability (p) of HIV seropositivity
mer1$prob <- predict(lm1$glm_res, type = "response")
mer1$predicted.classes <- ifelse(prob > 0.5, "pos", "neg")
head(mer1$predicted.classes)
mer1$logit = log(mer1$prob/(1-mer1$prob))
#1 The dependent variable is binary
#This is satisfied, as the outcome (HIV serostatus) is binary
#2 Linear relationship with continuous variables
#Logistic regression assumes a linear relationship between the logit of the
#outcome and each continuous predictor variable (in this case, age).
ggplot(mer1, aes(logit, age))+
geom_point(size = 0.5, alpha = 0.5) +
geom_smooth(method = "loess") +
theme_bw()+
labs(x="Log Odds of HIV", y="Age in Years")
#3 Independent observations
#This assumption is satisfied, as each row of the dataset represents a unique woman.
#While observations within DHS sampling clusters are likely to be correlated, this was
#taken into account in the analysis by using cluster analysis (glm.cluster).
#4 No multicollinearity between independent variables.
#For the categorical varaibles, multicollinearity can be detected with
#Spearman rank correlation coefficient (ordinal variables) and chi-square test (nominal variables).
#The chi-squared tests in table 2 show no significant correlation between financial
#autonomy, sexual autonomy, and intimate partner violence.
#5 No influential outliers
#Data points with an absolute standardized residuals above 3 represent possible
#outliers and may deserve closer attention.
plot(lm1$glm_res, which = 4)
plot(lm1$glm_res, which = 5)
#Remove outliers and re-run analysis
mer2<- mer1[(mer1$ID!=172412), ]
mer2<- mer2[(mer2$ID!=368601), ]
mer2<- mer2[(mer2$ID!=605632), ]
#Run regression
lm1<-glm.cluster(hiv~sex2+fin2+IPV+edu2+age+ages+rsd+mar+wealth+partners, data=mer2, cluster = mer2$cluster, family=binomial)
summary(lm1)
exp(cbind(OR = coef(lm1), confint.default(lm1)))
#Outlier does not influence the results significantly
#The model appears to satisfy the assumptions of logistic regression