library(tidyverse)
library(dplyr)
library(ggplot2)
library(survival)
library(emmeans)
library(foreign)
library(gtsummary)foo
dat <- read.csv("./unos.txt", sep = "\t")
head(dat) hlamat age age.1 cold_isc death year sex txtype fu
1 2 25 15 6 0 2002 0 1 0
2 1 42 10 11 1 1999 0 1 0
3 1 9 14 8 0 2002 1 1 0
4 0 35 12 26 0 2002 1 1 0
5 2 17 12 28 1 1997 1 1 0
6 3 44 13 1 0 2002 1 0 0
names(dat) <- c("hla.match", "age.donor", "age.rec", "cold.isc", "death",
"year", "sex", "tx.type", "follow.up")Complete case
nrow(dat)[1] 9775
sapply(names(dat), function(col) {
is.na(dat[, col]) |> sum()
})hla.match age.donor age.rec cold.isc death year sex tx.type
234 113 9 2250 0 0 0 0
follow.up
0
1 Exercise
1.1 Exercise 1
Illustrate in a table the characteristics of the population (age, sex, race, donor, . . . ).
hist(dat$age.donor)
hist(dat$age.rec)
hist(dat$cold.isc)
hist(dat$year)
hist(dat$follow.up)
unique(dat$hla.match)[1] 2 1 0 3 4 NA 5 6
unique(dat$year) [1] 2002 1999 1997 2001 1994 1995 1996 1993 1990 1991 1992 2000 1998
Change the order!!
dat.table1 <- dat |>
mutate(
sex = factor(sex, levels = c(0,1), labels = c("Female","Male")),
tx.type = factor(tx.type, levels = c(0,1), labels = c("Cadaveric","Living")),
hla.match = factor(hla.match),
year = factor(year)
)table1 <- dat.table1 |>
select(hla.match, age.donor, age.rec, cold.isc, year, sex, tx.type) |>
tbl_summary(
statistic = list(
all_continuous() ~ "{median} ({p25}, {p75})"
),
label = list(
hla.match ~ "HLA matches, n(%)",
age.donor ~ "Donor age, median (IQR)",
age.rec ~ "Recipient age, median (IQR)",
cold.isc ~ "Cold ischemic time (hours), median (IQR), ",
year ~ "Year of transplant",
sex ~ "Sex, n(%)",
tx.type ~ "Transplant type, n(%)"
),
missing = "ifany"
) |>
modify_footnote(all_stat_cols() ~ NA)
table1| Characteristic | N = 9,775 |
|---|---|
| HLA matches, n(%) | |
| 0 | 784 (8.2%) |
| 1 | 1,504 (16%) |
| 2 | 1,553 (16%) |
| 3 | 3,778 (40%) |
| 4 | 1,278 (13%) |
| 5 | 365 (3.8%) |
| 6 | 279 (2.9%) |
| Unknown | 234 |
| Donor age, median (IQR) | 33 (21, 41) |
| Unknown | 113 |
| Recipient age, median (IQR) | 13 (8, 16) |
| Unknown | 9 |
| Cold ischemic time (hours), median (IQR), | 7 (1, 19) |
| Unknown | 2,250 |
| Year of transplant | |
| 1990 | 728 (7.4%) |
| 1991 | 734 (7.5%) |
| 1992 | 666 (6.8%) |
| 1993 | 736 (7.5%) |
| 1994 | 787 (8.1%) |
| 1995 | 813 (8.3%) |
| 1996 | 789 (8.1%) |
| 1997 | 745 (7.6%) |
| 1998 | 740 (7.6%) |
| 1999 | 842 (8.6%) |
| 2000 | 706 (7.2%) |
| 2001 | 834 (8.5%) |
| 2002 | 655 (6.7%) |
| Sex, n(%) | |
| Female | 4,014 (41%) |
| Male | 5,761 (59%) |
| Transplant type, n(%) | |
| Cadaveric | 5,148 (53%) |
| Living | 4,627 (47%) |
table.split <- dat.table1 |>
select(hla.match, age.donor, age.rec, cold.isc, year, sex, tx.type) |>
tbl_summary(
by = tx.type,
statistic = list(
all_continuous() ~ "{median} ({p25}, {p75})"
),
label = list(
hla.match ~ "HLA matches, n(%)",
age.donor ~ "Donor age, median (IQR)",
age.rec ~ "Recipient age, median (IQR)",
cold.isc ~ "Cold ischemic time (hours), median (IQR), ",
year ~ "Year of transplant",
sex ~ "Sex, n(%)"
),
missing = "ifany"
) |>
add_overall() |>
modify_footnote(all_stat_cols() ~ NA)
table.split| Characteristic | Overall N = 9,775 |
Cadaveric N = 5,148 |
Living N = 4,627 |
|---|---|---|---|
| HLA matches, n(%) | |||
| 0 | 784 (8.2%) | 125 (2.5%) | 659 (14%) |
| 1 | 1,504 (16%) | 160 (3.2%) | 1,344 (29%) |
| 2 | 1,553 (16%) | 228 (4.6%) | 1,325 (29%) |
| 3 | 3,778 (40%) | 3,039 (61%) | 739 (16%) |
| 4 | 1,278 (13%) | 1,007 (20%) | 271 (5.9%) |
| 5 | 365 (3.8%) | 241 (4.9%) | 124 (2.7%) |
| 6 | 279 (2.9%) | 153 (3.1%) | 126 (2.7%) |
| Unknown | 234 | 195 | 39 |
| Donor age, median (IQR) | 33 (21, 41) | 37 (31, 42) | 22 (15, 37) |
| Unknown | 113 | 112 | 1 |
| Recipient age, median (IQR) | 13 (8, 16) | 13 (7, 16) | 14 (9, 16) |
| Unknown | 9 | 7 | 2 |
| Cold ischemic time (hours), median (IQR), | 7 (1, 19) | 1 (0, 1) | 19 (13, 25) |
| Unknown | 2,250 | 1,490 | 760 |
| Year of transplant | |||
| 1990 | 728 (7.4%) | 318 (6.2%) | 410 (8.9%) |
| 1991 | 734 (7.5%) | 387 (7.5%) | 347 (7.5%) |
| 1992 | 666 (6.8%) | 345 (6.7%) | 321 (6.9%) |
| 1993 | 736 (7.5%) | 403 (7.8%) | 333 (7.2%) |
| 1994 | 787 (8.1%) | 355 (6.9%) | 432 (9.3%) |
| 1995 | 813 (8.3%) | 427 (8.3%) | 386 (8.3%) |
| 1996 | 789 (8.1%) | 416 (8.1%) | 373 (8.1%) |
| 1997 | 745 (7.6%) | 408 (7.9%) | 337 (7.3%) |
| 1998 | 740 (7.6%) | 412 (8.0%) | 328 (7.1%) |
| 1999 | 842 (8.6%) | 436 (8.5%) | 406 (8.8%) |
| 2000 | 706 (7.2%) | 392 (7.6%) | 314 (6.8%) |
| 2001 | 834 (8.5%) | 483 (9.4%) | 351 (7.6%) |
| 2002 | 655 (6.7%) | 366 (7.1%) | 289 (6.2%) |
| Sex, n(%) | |||
| Female | 4,014 (41%) | 2,105 (41%) | 1,909 (41%) |
| Male | 5,761 (59%) | 3,043 (59%) | 2,718 (59%) |
Calculate median follow-up (reverse Kaplan-Meier method), with 95% CI
rev_km <- survfit(Surv(follow.up, death == 0) ~ 1, data = dat)
summary(rev_km)$table records n.max n.start events rmean se(rmean)
9.775000e+03 9.775000e+03 9.775000e+03 9.310000e+03 4.055027e+00 3.202711e-02
median 0.95LCL 0.95UCL
3.380822e+00 3.227397e+00 3.567123e+00
median_followup <- summary(rev_km)$table["median"]
confidence_interval <- c(summary(rev_km)$table["0.95LCL"], summary(rev_km)$table["0.95UCL"])
median_followup median
3.380822
confidence_interval 0.95LCL 0.95UCL
3.227397 3.567123
g <- ggplot(dat)g + geom_point(aes(x = follow.up, y = death))
g + geom_density(aes(x = follow.up))
g + geom_density(aes(x = age.donor))Warning: Removed 113 rows containing non-finite outside the scale range
(`stat_density()`).

g + geom_bar(aes(x = age.rec))Warning: Removed 9 rows containing non-finite outside the scale range
(`stat_count()`).

g + geom_boxplot(aes(x = age.rec, y = age.donor, group = age.donor))Warning: Removed 113 rows containing missing values or values outside the scale range
(`stat_boxplot()`).
Warning: Removed 7 rows containing non-finite outside the scale range
(`stat_boxplot()`).

# dat$age.1 |> table()g + geom_boxplot(aes(x = year, y = follow.up, group = year))
1.2 Exercise 2
Plot the Kaplan-Meier overall survival curve for pediatric kid- ney transplant recipients for the first 12 years after transplantation.
!!!!!!!! just set xlim to 12, do not remove those individuals !!!!!!!
km <- survfit(Surv(follow.up, death) ~ 1, data = dat[dat$follow.up <= 12, ])
plot(km)
1.3 Exercise 3
We are going to compare mortality rates (hazard functions) between children whose transplanted kidney was provided by a living donor (in general a family member) and those whose source was recently deceased (variable donor type: txtype). Use the life table method to calculate the death rates for the first 5 years for each group (take in the first year intervals of 4 months and then look at each year) and show the results in a table. Estimate the hazard ratio in each time interval as the ratio between the mortality rates in the two groups. What do you notice?
dat.5 <- dat[dat$follow.up <= 5, ]
head(dat.5) hla.match age.donor age.rec cold.isc death year sex tx.type follow.up
1 2 25 15 6 0 2002 0 1 0
2 1 42 10 11 1 1999 0 1 0
3 1 9 14 8 0 2002 1 1 0
4 0 35 12 26 0 2002 1 1 0
5 2 17 12 28 1 1997 1 1 0
6 3 44 13 1 0 2002 1 0 0
classify_time_interval = function(fu) {
if (fu <= 1/3) {
return(1/3)
} else if (fu <= 2/3) {
return(2/3)
} else if (fu <= 1) {
return(1)
}
ceiling(fu)
}
dat.5$fu.interval <- sapply(dat.5$follow.up, classify_time_interval)
table(dat.5$fu.interval)
0.333333333333333 0.666666666666667 1 2
982 528 640 1226
3 4 5
1184 1093 950
head(dat.5) hla.match age.donor age.rec cold.isc death year sex tx.type follow.up
1 2 25 15 6 0 2002 0 1 0
2 1 42 10 11 1 1999 0 1 0
3 1 9 14 8 0 2002 1 1 0
4 0 35 12 26 0 2002 1 1 0
5 2 17 12 28 1 1997 1 1 0
6 3 44 13 1 0 2002 1 0 0
fu.interval
1 0.3333333
2 0.3333333
3 0.3333333
4 0.3333333
5 0.3333333
6 0.3333333
dat.5.life <- dat.5 |>
group_by(fu.interval) |>
summarize(
n.censored = sum(death == 0),
n.event = sum(death),
n.at.risk = nrow(dat),
)
for (i in 2:nrow(dat.5.life)) {
j <- i - 1
n.censored.pre <- dat.5.life$n.censored[j]
n.event.pre <- dat.5.life$n.event[j]
n.at.risk.pre <- dat.5.life$n.at.risk[j]
n.at.risk <- n.at.risk.pre - n.event.pre - n.censored.pre
dat.5.life$n.at.risk[i] <- n.at.risk
}
print(nrow(dat))[1] 9775
dat.5.life# A tibble: 7 × 4
fu.interval n.censored n.event n.at.risk
<dbl> <int> <int> <int>
1 0.333 830 152 9775
2 0.667 491 37 8793
3 1 613 27 8265
4 2 1164 62 7625
5 3 1143 41 6399
6 4 1053 40 5215
7 5 919 31 4122
n.removed.acc
dat.5.life <- dat.5.life |>
mutate(
hazard.rate = n.event / n.at.risk
)get_life_table = function(dat.sub, dat) {
dat.sub <- dat.sub |>
group_by(fu.interval) |>
summarize(
n.censored = sum(death == 0),
n.event = sum(death),
n.at.risk = nrow(dat),
)
for (i in 2:nrow(dat.sub)) {
j <- i - 1
n.censored.pre <- dat.sub$n.censored[j]
n.event.pre <- dat.sub$n.event[j]
n.at.risk.pre <- dat.sub$n.at.risk[j]
n.at.risk <- n.at.risk.pre - n.event.pre - n.censored.pre
dat.sub$n.at.risk[i] <- n.at.risk
}
dat.sub <- dat.sub |>
mutate(
hazard.rate = n.event / n.at.risk
)
return(dat.sub)
}dat.5.tx0 = dat.5[dat.5$tx.type == 0, ]
dat.5.tx1 = dat.5[dat.5$tx.type == 1, ]tx0.life <- get_life_table(dat.5.tx0, dat[dat$tx.type == 0, ])
tx0.life# A tibble: 7 × 5
fu.interval n.censored n.event n.at.risk hazard.rate
<dbl> <int> <int> <int> <dbl>
1 0.333 463 49 5148 0.00952
2 0.667 253 13 4636 0.00280
3 1 340 11 4370 0.00252
4 2 591 24 4019 0.00597
5 3 568 17 3404 0.00499
6 4 521 16 2819 0.00568
7 5 501 11 2282 0.00482
tx1.life <- get_life_table(dat.5.tx1, dat[dat$tx.type == 1, ])
tx1.life# A tibble: 7 × 5
fu.interval n.censored n.event n.at.risk hazard.rate
<dbl> <int> <int> <int> <dbl>
1 0.333 367 103 4627 0.0223
2 0.667 238 24 4157 0.00577
3 1 273 16 3895 0.00411
4 2 573 38 3606 0.0105
5 3 575 24 2995 0.00801
6 4 532 24 2396 0.0100
7 5 418 20 1840 0.0109
tx1.life$hazard.rate / tx0.life$hazard.rate[1] 2.338731 2.058881 1.631929 1.764675 1.604557 1.764816 2.254941
hazard.df <- data.frame(
fu.interval = tx1.life$fu.interval,
hazard.rate.0 = tx0.life$hazard.rate,
hazard.rate.1 = tx1.life$hazard.rate,
hazard.ratio = tx1.life$hazard.rate / tx0.life$hazard.rate
)
ggplot(hazard.df, aes(x = fu.interval)) +
geom_line(aes(y = hazard.rate.0), color = "blue") +
geom_line(aes(y = hazard.rate.1), color = "orange")
ggplot(hazard.df, aes(x = fu.interval)) +
geom_line(aes(y = hazard.ratio))
General Solution:
get.life.table <- function(dat, time.intervals) {
n.pop <- nrow(dat)
dat |>
recode.dat(time.intervals) |>
group_by(fu.interval) |>
summarize(
n.censored = sum(.data$death == 0),
n.event = sum(.data$death),
) |>
ungroup() |>
calculate.hazard(n.pop)
}
get.life.table.by.groups <- function(dat, time.intervals, grps) {
grps |>
lapply(function(grp) {
dat |>
get.life.table.by.group(time.intervals, grp) |>
mutate(
grp.name = grp,
grp.value = pick(1)[[1]]
) |>
select(-1)
}) |>
bind_rows()
}
get.life.table.by.group <- function(dat, time.intervals, grp) {
dat |>
recode.dat(time.intervals) |>
group_by(fu.interval, .data[[grp]]) |>
summarize(
n.censored = sum(.data$death == 0),
n.event = sum(.data$death),
.groups = "keep"
) |>
ungroup(fu.interval) |>
group_modify(function(df.sub, grp) {
grp.name <- names(grp)
grp.value <- grp[[1]]
n.pop <- (dat[[grp.name]] == grp.value) |> sum()
calculate.hazard(df.sub, n.pop)
}) |>
ungroup()
}
calculate.hazard <- function(life.table, n.pop) {
n.removed <- life.table$n.event + life.table$n.censored
n.removed.accum <- c(0, cumsum(n.removed)[-length(n.removed)])
life.table |>
mutate(
n.at.risk = n.pop - n.removed.accum,
# TODO: how to account for censored? How do we adjust for uneven interval?
hazard.rate = n.event / n.at.risk
)
}
recode.dat <- function(dat, time.intervals) {
df <- dat[dat$follow.up <= sum(time.intervals), ]
time.points <- cumsum(time.intervals)
df$fu.interval <- sapply(df$follow.up, function(time) {
time.points[time <= time.points][1]
})
df
}time.intervals <- c(1/3, 1/3, 1/3, 1, 1, 1, 1)
grps <- c("tx.type", "sex", "hla.match", "cold.isc")
get.life.table.by.groups(dat, time.intervals, grps)# A tibble: 593 × 7
fu.interval n.censored n.event n.at.risk hazard.rate grp.name grp.value
<dbl> <int> <int> <dbl> <dbl> <chr> <dbl>
1 0.333 463 49 5148 0.00952 tx.type 0
2 0.667 253 13 4636 0.00280 tx.type 0
3 1 340 11 4370 0.00252 tx.type 0
4 2 591 24 4019 0.00597 tx.type 0
5 3 568 17 3404 0.00499 tx.type 0
6 4 521 16 2819 0.00568 tx.type 0
7 5 501 11 2282 0.00482 tx.type 0
8 0.333 367 103 4627 0.0223 tx.type 1
9 0.667 238 24 4157 0.00577 tx.type 1
10 1 273 16 3895 0.00411 tx.type 1
# ℹ 583 more rows
grps <- c("tx.type", "sex", "hla.match", "cold.isc")
grps <- c("tx.type")
plot.hazard.rate.by.groups <- function(dat, time.intervals, grps) {
dat |>
get.life.table.by.groups(time.intervals, grps) |>
ggplot(aes(x = fu.interval, y = hazard.rate,
color = paste(grp.name, grp.value, sep = " = "))) +
geom_point() +
geom_line() +
labs(
color = "Group"
) +
theme(
legend.position = "bottom"
)
}time.intervals <- c(rep(1/3, 3), rep(1, 4))
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- rep(1, 5)
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- rep(1/3, 15)
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- c(rep(1/3, 3*3), rep(1, 2))
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- c(rep(1/4, 4), rep(1, 4))
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- c(rep(1/4, 4*2), rep(1, 3))
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
time.intervals <- c(rep(1/4, 4*2), rep(1/3, 3*3), rep(1, 6))
plot.hazard.rate.by.groups(dat, time.intervals, c("tx.type"))
Compare
time.intervals <- c(1/3, 1/3, 1/3, 1, 1, 1, 1)
get.life.table.by.groups(dat, time.intervals, grps)# A tibble: 14 × 7
fu.interval n.censored n.event n.at.risk hazard.rate grp.name grp.value
<dbl> <int> <int> <dbl> <dbl> <chr> <int>
1 0.333 463 49 5148 0.00952 tx.type 0
2 0.667 253 13 4636 0.00280 tx.type 0
3 1 340 11 4370 0.00252 tx.type 0
4 2 591 24 4019 0.00597 tx.type 0
5 3 568 17 3404 0.00499 tx.type 0
6 4 521 16 2819 0.00568 tx.type 0
7 5 501 11 2282 0.00482 tx.type 0
8 0.333 367 103 4627 0.0223 tx.type 1
9 0.667 238 24 4157 0.00577 tx.type 1
10 1 273 16 3895 0.00411 tx.type 1
11 2 573 38 3606 0.0105 tx.type 1
12 3 575 24 2995 0.00801 tx.type 1
13 4 532 24 2396 0.0100 tx.type 1
14 5 418 20 1840 0.0109 tx.type 1
hazard.df fu.interval hazard.rate.0 hazard.rate.1 hazard.ratio
1 0.3333333 0.009518260 0.022260644 2.338731
2 0.6666667 0.002804142 0.005773394 2.058881
3 1.0000000 0.002517162 0.004107831 1.631929
4 2.0000000 0.005971635 0.010537992 1.764675
5 3.0000000 0.004994125 0.008013356 1.604557
6 4.0000000 0.005675772 0.010016694 1.764816
7 5.0000000 0.004820333 0.010869565 2.254941
1.4 Exercise 4
Show a plot with Kaplan-Meier survival curves for the two donor types.
km.tx <- survfit(Surv(follow.up, death) ~ tx.type, data = dat[dat$follow.up <= 12, ])
plot(km.tx, col = c("blue", "orange"))
legend(legend = c("cadaveric", "living"), "bottomleft", lwd = 2, col = c("blue", "orange"))
1.5 Exercise 5
Fit a univariate Cox model with predictor donor type. Report the hazard ratio and 95% confidence interval and interpret the result obtained.
cox.tx <- coxph(Surv(follow.up, death) ~ tx.type, data = dat)
summary(cox.tx)Call:
coxph(formula = Surv(follow.up, death) ~ tx.type, data = dat)
n= 9775, number of events= 465
coef exp(coef) se(coef) z Pr(>|z|)
tx.type 0.64469 1.90539 0.09558 6.745 1.53e-11 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
tx.type 1.905 0.5248 1.58 2.298
Concordance= 0.586 (se = 0.012 )
Likelihood ratio test= 47.1 on 1 df, p=7e-12
Wald test = 45.5 on 1 df, p=2e-11
Score (logrank) test = 47.09 on 1 df, p=7e-12
log.hazard.ratio.hat <- cox.tx$coefficient
hazard.ratio.hat <- log.hazard.ratio.hat |> exp()log.hazard.ratio.se <- summary(cox.tx)$coefficients[3]
log.hazard.ratio.hat.lower.bound <- log.hazard.ratio.hat - qnorm(0.975) * log.hazard.ratio.se
log.hazard.ratio.hat.upper.bound <- log.hazard.ratio.hat + qnorm(0.975) * log.hazard.ratio.se
c(log.hazard.ratio.hat.lower.bound, log.hazard.ratio.hat.upper.bound) |> exp() tx.type tx.type
1.579899 2.297932
Patients receiving kidney transplants from living donors (tx.type = 1) has 1.91 higher hazard rate than those receiving from cadaveric donors (tx.type = 0).
Since 1 is not in the 95% confidence interval, the result is statistically significant.
2 Exercise 6
Research shows that an important determinant of mortality after kidney transplant is the age of the recipient. Fit a Cox model with age as predictor and estimate the hazard ratio and its confidence interval. Consider age first as continuous variable and then divide into categories.
2.1 Continuous
cox.age.rec <- coxph(Surv(follow.up, death) ~ age.rec, data = dat)
cox.age.rec |> summary()Call:
coxph(formula = Surv(follow.up, death) ~ age.rec, data = dat)
n= 9766, number of events= 464
(9 observations deleted due to missingness)
coef exp(coef) se(coef) z Pr(>|z|)
age.rec -0.042550 0.958342 0.008434 -5.045 4.53e-07 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
age.rec 0.9583 1.043 0.9426 0.9743
Concordance= 0.586 (se = 0.016 )
Likelihood ratio test= 24.81 on 1 df, p=6e-07
Wald test = 25.45 on 1 df, p=5e-07
Score (logrank) test = 25.75 on 1 df, p=4e-07
2.2 Categorical (Case 1)
recode.age <- function(dat, ages) {
df$age.group <- sapply(df$follow.up, function(time) {
time.points[time <= time.points][1]
})
df
}dat.age.rec <- dat[!is.na(dat$age.rec), ]
class.age <- function(age) {
if (age < 2) {
return("0,1")
} else if (age < 6) {
return("2,3,4,5")
} else if (age < 11) {
return("6,7,8,9,10")
} else if (age < 19) {
return("11-18")
}
}dat.age.rec$age.group <- sapply(dat.age.rec$age.rec, class.age) |>
factor(levels = c(
"0,1",
"2,3,4,5",
"6,7,8,9,10",
"11-18"
))
cox.age.rec <- coxph(Surv(follow.up, death) ~ age.group, data = dat.age.rec)
cox.age.rec |> summary()Call:
coxph(formula = Surv(follow.up, death) ~ age.group, data = dat.age.rec)
n= 9766, number of events= 464
coef exp(coef) se(coef) z Pr(>|z|)
age.group2,3,4,5 -0.5947 0.5517 0.1767 -3.366 0.000764 ***
age.group6,7,8,9,10 -1.0578 0.3472 0.1776 -5.956 2.59e-09 ***
age.group11-18 -1.0192 0.3609 0.1513 -6.736 1.63e-11 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
exp(coef) exp(-coef) lower .95 upper .95
age.group2,3,4,5 0.5517 1.813 0.3902 0.7801
age.group6,7,8,9,10 0.3472 2.880 0.2452 0.4918
age.group11-18 0.3609 2.771 0.2683 0.4855
Concordance= 0.584 (se = 0.015 )
Likelihood ratio test= 44.93 on 3 df, p=1e-09
Wald test = 53.97 on 3 df, p=1e-11
Score (logrank) test = 57.87 on 3 df, p=2e-12
emmeans(cox.age.rec, pairwise ~ age.group, type = "response")$emmeans
age.group response SE df asymp.LCL asymp.UCL
0,1 1.000 0.0000 Inf 1.000 1.000
2,3,4,5 0.552 0.0975 Inf 0.390 0.780
6,7,8,9,10 0.347 0.0617 Inf 0.245 0.492
11-18 0.361 0.0546 Inf 0.268 0.485
Confidence level used: 0.95
Intervals are back-transformed from the log scale
$contrasts
contrast ratio SE df null z.ratio p.value
0,1 / 2,3,4,5 1.813 0.320 Inf 1 3.366 0.0043
0,1 / 6,7,8,9,10 2.880 0.511 Inf 1 5.956 <.0001
0,1 / (11-18) 2.771 0.419 Inf 1 6.736 <.0001
2,3,4,5 / 6,7,8,9,10 1.589 0.251 Inf 1 2.928 0.0179
2,3,4,5 / (11-18) 1.529 0.196 Inf 1 3.314 0.0051
6,7,8,9,10 / (11-18) 0.962 0.124 Inf 1 -0.298 0.9908
P value adjustment: tukey method for comparing a family of 4 estimates
Tests are performed on the log scale
3 Exercise 7
Fit a multivariate Cox model by using other predictors and describe your results.
sapply(names(dat), function(col) {
is.na(dat[, col]) |> sum()
})hla.match age.donor age.rec cold.isc death year sex tx.type
234 113 9 2250 0 0 0 0
follow.up
0
names(dat)[1] "hla.match" "age.donor" "age.rec" "cold.isc" "death" "year"
[7] "sex" "tx.type" "follow.up"
Waldtest <- function(cox.fit, i.betas){
q <- length(i.betas)
coefs <- cox.fit$coefficients
# print(cox.fit$var)
var <- cox.fit$var[i.betas, i.betas]
Wald <- coefs[i.betas]%*%solve(var)%*%coefs[i.betas]
p.value <- 1-pchisq(as.numeric(Wald), df = q)
return(cbind(Wald, p.value))
}fit.cox.model <- function(surv, i.betas) {
fit.tx <- coxph(surv, data = dat)
Waldtest(fit.tx, i.betas)
}Start with tx.type
surv <- Surv(follow.up, death) ~ tx.type + age.rec
fit.cox.model(surv, 1:2) # NOTE: Why is this 1??? p.value
[1,] 77.3778 0
Iteration 1: tx.type + age + ?
names(dat)[1] "hla.match" "age.donor" "age.rec" "cold.isc" "death" "year"
[7] "sex" "tx.type" "follow.up"
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match
fit.cox.model(surv, 3) p.value
[1,] 8.146717 0.004313919
surv <- Surv(follow.up, death) ~ tx.type + age.rec + age.donor
fit.cox.model(surv, 3) p.value
[1,] 6.509223 0.01073164
surv <- Surv(follow.up, death) ~ tx.type + age.rec + cold.isc
fit.cox.model(surv, 3) p.value
[1,] 0.7057259 0.4008664
surv <- Surv(follow.up, death) ~ tx.type + age.rec + year
fit.cox.model(surv, 3) p.value
[1,] 4.680593 0.03050521
surv <- Surv(follow.up, death) ~ tx.type + age.rec + sex
fit.cox.model(surv, 3) p.value
[1,] 1.921176 0.1657271
Add hla.match
Iteration 2: tx.type + age.rec + hla.match + ?
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor
fit.cox.model(surv, 4) p.value
[1,] 5.211567 0.0224371
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + cold.isc
fit.cox.model(surv, 4) p.value
[1,] 0.9157965 0.3385811
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + year
fit.cox.model(surv, 4) p.value
[1,] 4.94216 0.02620927
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + sex
fit.cox.model(surv, 4) p.value
[1,] 0.981911 0.3217275
Add age.donor
Iteration 2: tx.type + age.rec + hla.match + age.donor + ?
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor +
cold.isc
fit.cox.model(surv, 5) p.value
[1,] 0.4450261 0.5047065
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor +
year
fit.cox.model(surv, 5) p.value
[1,] 4.367751 0.03662531
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor +
sex
fit.cox.model(surv, 5) p.value
[1,] 0.9343691 0.3337302
Add year
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor +
year + cold.isc
fit.cox.model(surv, 6) p.value
[1,] 0.1614349 0.6878388
surv <- Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor +
year + sex
fit.cox.model(surv, 6) p.value
[1,] 0.9641705 0.3261383
Next, we test for each remaining variables
4 Exercise 8
Estimate the survival function for specific covariate patterns. Based on the previous results choose the best predictors.
Exercise 9 — Check the proportional hazards assumption. You may use the function cox.zph. Discuss the result and possible implications.
H0: PH-assumption holds H1: PH-assumption doesn’t hold
# different cox model!!
cox.final <- coxph(Surv(follow.up, death) ~ tx.type + age.rec + hla.match + age.donor, data = dat)
cox.zph(cox.final) chisq df p
tx.type 2.56 1 0.10948
age.rec 35.98 1 2.0e-09
hla.match 8.22 1 0.00414
age.donor 11.05 1 0.00089
GLOBAL 47.37 4 1.3e-09
Exercise 10 — Plot the Schoenfeld residuals and comment. A non-random pattern or slope in a plot of scaled residuals against time indicates a violation, suggesting the covariate’s impact changes, while a horizontal line around zero supports the assumption.
plot(cox.zph(cox.final))



NOTES: left censoring: event happened before time zero -> not applicable left truncation: not applicable?? right censoring: we saw this in the follow up times, administrative censoring high
! No left truncation, since our target population is patients who received a kidney transplant. So we don’t care about patients who died before receiving a transplant.
! ADD: we used a Cox model but we already saw that the PH assumption doesn’t hold !
? remove year from prediction model, since it is not a patient characteristic and may not be stable for future predictions. It mainly captures changes in clinical practice rather than individual risk.