scripts/plot--tb_pop_year.R

#
#
#
#
#

SCREENED <- c(0,1)
# SCREENED <- 1

# subset by covariate?
cohort1 <-
  IMPUTED_sample %>%
  dplyr::filter(IMPUTED_sample$who_inc_Pareek2011 %in% "(150,250]",
                screen %in% SCREENED)
cohort2 <-
  IMPUTED_sample %>%
  dplyr::filter(IMPUTED_sample$who_inc_Pareek2011 %in% "(250,350]",
                screen %in% SCREENED)
cohort3 <-
  IMPUTED_sample %>%
  dplyr::filter(IMPUTED_sample$who_inc_Pareek2011 %in% "(350,1e+05]",
                screen %in% SCREENED)


# histograms with different bin sizes -------------------------------------

hist(cohort1$notif_issdt.years[!is.na(cohort1$notif_issdt.years)],
     breaks = c(0,1,2,3,4,5,6,7,8,9,15,50,100),
     col = "black",
     prob = T,
     main = "",
     xlab = "year")
hist(cohort2$notif_issdt.years[!is.na(cohort2$notif_issdt.years)],
     breaks = c(0,1,2,3,4,5,6,7,8,9,15,50,100),
     prob = TRUE,
     add = T,
     col = "blue",
     border = "blue")
hist(cohort3$notif_issdt.years[!is.na(cohort3$notif_issdt.years)],
     breaks = c(0,1,2,3,4,5,6,7,8,9,15,50,100),
     prob = TRUE,
     add = T,
     col = "white",
     border = "red")


# density plots -----------------------------------------------------------

plot(density(cohort1$notif_issdt.years, from = 0, bw = 1, na.rm = TRUE),
     lwd = 2,
     ylim = c(0, 0.004),
     main = "")
lines(density(cohort2$notif_issdt.years, from = 0, bw = 1, na.rm = TRUE),
      lwd = 2, col = "blue")
lines(density(cohort3$notif_issdt.years, from = 0, bw = 1, na.rm = TRUE),
      lwd = 2, col = "red")

legend('topright',
       legend = c("(150,250]","(250,350]","(350,1e+05]"),
       col = c("black","blue","red"),
       lty = 1)


# survival plots ----------------------------------------------------------

library(survminer)
library(survival)

# include all-cause death in risk set
IMPUTED_sample$notif_issdt.years[is.na(IMPUTED_sample$notif_issdt.years)] <- Inf

fit_tb <- survfit(Surv(notif_issdt.years) ~ who_inc_Pareek2011,
                  # data = IMPUTED_sample[IMPUTED_sample$screen == 1, ])
                  data = IMPUTED_sample)

ggsurvplot(fit_tb,
           data = IMPUTED_sample,
           risk.table = TRUE,
           ylim = c(0.95, 1),
           # xlim = c(0, 20),
           conf.int = TRUE,
           main = "TB notification in EWNI")#,
# ggtheme = theme_minimal(),
# risk.table.y.text.col = T,
# risk.table.y.text = FALSE)

fit_exit <- survfit(Surv(IMPUTED_sample$date_exit_uk1_issdt.years) ~ who_inc_Pareek2011,
                    data = IMPUTED_sample)

ggsurvplot(fit_exit,
           data = IMPUTED_sample,
           risk.table = TRUE,
           main = "Exit EWNI")
n8thangreen/LTBIscreeningproject documentation built on May 23, 2019, 12:01 p.m.