riseR:

#Lets try doing a little bit of logistic regression! - Jeff F

#create data set made up of symptom data, healthcare data, and antibiotic data
regression <- child %>% filter (survey_check2 == 1) %>%
  mutate (infection = ifelse(cough == 1 | breathing == 1 | diarrhea1 == 1 |
                               fever == 1 | skin == 1, 1, 0),
          age = ((date("2019-10-01") - child_dob) / 365),
          healthcare = ifelse (doctor_yn > 0 | hospital_yn > 0, 1, 0),
          antibiotics = ifelse(doctor_antibiotics == 1, 1, 0)) %>%
  rename (settlement = settlement_barcode, diarrhea = diarrhea1, sex = gender_pl) %>%
#  mutate(cough = lapply(cough, as.logical), #turn variables into boolean
#         infection = lapply(infection, as.logical),
#         toothache = lapply(toothache, as.logical),
#         breathing = lapply(breathing, as.logical),
#         diarrhea = lapply(diarrhea, as.logical),
#         fever = lapply(fever, as.logical),
#         skin = lapply(skin, as.logical),
#         injury = lapply(injury, as.logical),
#         healthcare = lapply(healthcare, as.logical),
#         antibiotics = lapply(antibiotics, as.logical)) %>%
  select (settlement, age, sex, infection, cough, toothache, breathing,
          diarrhea, fever,skin, injury, healthcare, antibiotics)

#chisquare test:
chisq.test(regression$healthcare, regression$cough)
chisq.test(regression$healthcare, regression$age)
chisq.test(regression$healthcare, regression$settlement)
chisq.test(regression$healthcare, regression$sex)
chisq.test(regression$healthcare, regression$fever) #statistically significant
chisq.test(regression$healthcare, regression$diarrhea) #statistically significant
chisq.test(regression$healthcare, regression$breathing)
chisq.test(regression$healthcare, regression$skin)
chisq.test(regression$healthcare, regression$injury)
chisq.test(regression$healthcare, regression$antibiotics)








#logistic regression - create tables
healthcare_cough <- glm(formula = healthcare ~ cough + age + sex + settlement,
                     data = regression,
                     family = "binomial")
healthcare_fever <- glm(formula = healthcare ~ fever + age + sex + settlement,
                        data = regression,
                        family = "binomial")
healthcare_skin <- glm(formula = healthcare ~ skin + age + sex + settlement,
                        data = regression,
                        family = "binomial")
healthcare_breathing <- glm(formula = healthcare ~ breathing + age + sex + settlement,
                        data = regression,
                        family = "binomial")
healthcare_diarrhea <- glm(formula = healthcare ~ diarrhea + age + sex + settlement,
                            data = regression,
                            family = "binomial")

#summary data
summary(healthcare_cough)
exp(cbind(OR = coef(healthcare_cough), confint(healthcare_cough)))

summary(healthcare_fever)
exp(cbind(OR = coef(healthcare_fever), confint(healthcare_fever)))

summary(healthcare_skin)
exp(cbind(OR = coef(healthcare_skin), confint(healthcare_skin)))

summary(healthcare_breathing)
exp(cbind(OR = coef(healthcare_breathing), confint(healthcare_breathing)))

summary(healthcare_diarrhea)
exp(cbind(OR = coef(healthcare_diarrhea), confint(healthcare_diarrhea)))