SeLECT Thrombolysis
Authors:
0.0.1 Download libraries
library(survminer)
library(survival)
library(readxl)
library(MatchIt)
library(stddiff)
library(tableone)
library(mice)
library(VIM)
library(table1)
library(pROC)
library(sensitivity)0.0.2 Load Data and Handle Missings (Multiple Imputation)
datawithmissings <- read_excel("data_for_analysis.xlsx")
tempdata <- mice(datawithmissings, m = 5, maxit = 50, meth = 'pmm', seed = 500, printFlag = FALSE)## Warning: Number of logged events: 251data <- complete(tempdata,1)1 Thrombolysis
1.1 Table One
data2 <- data
data2$male <-
factor(data2$male, levels=c(0,1),
labels=c("Female ",
"Male"))
data2$nihss_middle <-
factor(data2$nihss_middle, levels=c(0,1),
labels=c("No",
"Yes"))
data2$nihss_high <-
factor(data2$nihss_high, levels=c(0,1),
labels=c("No",
"Yes"))
data2$macro <-
factor(data2$macro, levels=c(0,1),
labels=c("No",
"Yes"))
data2$micro <-
factor(data2$micro, levels=c(0,1),
labels=c("No",
"Yes"))
data2$cortical <-
factor(data2$cortical, levels=c(0,1),
labels=c("No",
"Yes"))
data2$mca <-
factor(data2$mca, levels=c(0,1),
labels=c("No",
"Yes"))
data2$thrombolysis <-
factor(data2$thrombolysis, levels=c(0,1),
labels=c("No thrombolysis",
"Thrombolysis"))
data2$iv_thrombolysis <-
factor(data2$iv_thrombolysis, levels=c(0,1),
labels=c("No",
"Yes"))
data2$ia_thrombolysis <-
factor(data2$ia_thrombolysis, levels=c(0,1),
labels=c("No",
"Yes"))
data2$early <-
factor(data2$early, levels=c(0,1),
labels=c("No",
"Yes"))
data2$early_treatment <-
factor(data2$early_treatment, levels=c(0,1),
labels=c("No",
"Yes"))
data2$los <-
factor(data2$los, levels=c(0,1),
labels=c("No late seizures",
"Late seizures"))
label(data2$male) <- "Sex"
label(data2$age) <- "Age"
label(data2$source) <- "Cohort"
label(data2$nihss_middle) <- "NIHSS Middle"
label(data2$nihss_high) <- "NIHSS High"
label(data2$macro) <- "Large-artery artherosclerosis(embolus/thrombosis)"
label(data2$micro) <- "Small-vessel occulsion(lacune)"
label(data2$cortical) <- "Cortical compromise"
label(data2$mca) <- "Middle cerebral artery compromise"
label(data2$iv_thrombolysis) <- "IV thrombolysis"
label(data2$ia_thrombolysis) <- "Intraarterial thrombolysis"
label(data2$thrombolysis) <- "Thrombolysis"
label(data2$early) <- "Early seizures"
label(data2$los) <- "Late seizures"
label(data2$early_treatment) <- "Early AED treament"
label(data2$time) <- "Time to late seizures"
units(data$age) <- "Days"
rndr <- function(x, name, ...) {
if (length(x) == 0) {
y <- data2[[name]]
s <- rep("", length(render.default(x=y, name=name, ...)))
if (is.numeric(y)) {
p <- fisher.test(y ~ data2$thrombolysis)$p.value
} else {
p <- chisq.test(table(y, droplevels(data2$thrombolysis)))$p.value
}
s[2] <- sub("<", "<", format.pval(p, digits=3, eps=0.001))
s
} else {
render.default(x=x, name=name, ...)
}
}
rndr.strat <- function(label, n, ...) {
ifelse(n==0, label, render.strat.default(label, n, ...))
}
table1(~source +male + age + nihss_middle+ nihss_high + macro+ micro +mca +cortical +iv_thrombolysis +ia_thrombolysis +early +early_treatment +los +time| thrombolysis, data=data2, render=rndr, render.strat=rndr.strat)| No thrombolysis (n=2501) |
Thrombolysis (n=700) |
Overall (n=3201) |
|
|---|---|---|---|
| Cohort | |||
| bentes | 46 (1.8%) | 93 (13.3%) | 139 (4.3%) |
| hebron | 275 (11.0%) | 236 (33.7%) | 511 (16.0%) |
| kssg | 1061 (42.4%) | 139 (19.9%) | 1200 (37.5%) |
| linz | 357 (14.3%) | 102 (14.6%) | 459 (14.3%) |
| muenster | 238 (9.5%) | 73 (10.4%) | 311 (9.7%) |
| saar | 136 (5.4%) | 46 (6.6%) | 182 (5.7%) |
| udine | 388 (15.5%) | 11 (1.6%) | 399 (12.5%) |
| Sex | |||
| Female | 1114 (44.5%) | 292 (41.7%) | 1406 (43.9%) |
| Male | 1387 (55.5%) | 408 (58.3%) | 1795 (56.1%) |
| Age | |||
| Mean (SD) | 71.3 (13.6) | 70.0 (13.6) | 71.0 (13.6) |
| Median [Min, Max] | 74.0 [19.0, 109] | 73.0 [18.0, 99.0] | 73.0 [18.0, 109] |
| NIHSS Middle | |||
| No | 1746 (69.8%) | 405 (57.9%) | 2151 (67.2%) |
| Yes | 755 (30.2%) | 295 (42.1%) | 1050 (32.8%) |
| NIHSS High | |||
| No | 2119 (84.7%) | 356 (50.9%) | 2475 (77.3%) |
| Yes | 382 (15.3%) | 344 (49.1%) | 726 (22.7%) |
| Large-artery artherosclerosis(embolus/thrombosis) | |||
| No | 2048 (81.9%) | 519 (74.1%) | 2567 (80.2%) |
| Yes | 453 (18.1%) | 181 (25.9%) | 634 (19.8%) |
| Small-vessel occulsion(lacune) | |||
| No | 1829 (73.1%) | 628 (89.7%) | 2457 (76.8%) |
| Yes | 672 (26.9%) | 72 (10.3%) | 744 (23.2%) |
| Middle cerebral artery compromise | |||
| No | 915 (36.6%) | 170 (24.3%) | 1085 (33.9%) |
| Yes | 1586 (63.4%) | 530 (75.7%) | 2116 (66.1%) |
| Cortical compromise | |||
| No | 1374 (54.9%) | 318 (45.4%) | 1692 (52.9%) |
| Yes | 1127 (45.1%) | 382 (54.6%) | 1509 (47.1%) |
| IV thrombolysis | |||
| No | 2501 (100%) | 32 (4.6%) | 2533 (79.1%) |
| Yes | 0 (0%) | 668 (95.4%) | 668 (20.9%) |
| Intraarterial thrombolysis | |||
| No | 2501 (100%) | 634 (90.6%) | 3135 (97.9%) |
| Yes | 0 (0%) | 66 (9.4%) | 66 (2.1%) |
| Early seizures | |||
| No | 2387 (95.4%) | 648 (92.6%) | 3035 (94.8%) |
| Yes | 114 (4.6%) | 52 (7.4%) | 166 (5.2%) |
| Early AED treament | |||
| No | 2421 (96.8%) | 650 (92.9%) | 3071 (95.9%) |
| Yes | 80 (3.2%) | 50 (7.1%) | 130 (4.1%) |
| Late seizures | |||
| No late seizures | 2368 (94.7%) | 626 (89.4%) | 2994 (93.5%) |
| Late seizures | 133 (5.3%) | 74 (10.6%) | 207 (6.5%) |
| Time to late seizures | |||
| Mean (SD) | 971 (1110) | 936 (759) | 963 (1040) |
| Median [Min, Max] | 731 [0.00, 43800] | 724 [0.00, 3660] | 731 [0.00, 43800] |
1.2 Unmatched data
tab1 <- CreateTableOne(data = data, strata = "thrombolysis", testApprox = chisq.test, testExact = fisher.test)
cat <- c("source", "los", "early", "early_treatment", "male", "nihss_middle", "nihss_high", "macro", "micro", "mca", "cortical", "thrombolysis", "iv_thrombolysis", "ia_thrombolysis", "early_treatment")
cont <- c("time","age")
tab2 <- print(tab1, nonnormal = cont, smd = TRUE, missing = TRUE)## Stratified by thrombolysis
## 0
## n 2501
## ID (mean (SD)) 1490.60 (932.63)
## source (%)
## bentes 46 ( 1.8)
## hebron 275 (11.0)
## kssg 1061 (42.4)
## linz 357 (14.3)
## muenster 238 ( 9.5)
## saar 136 ( 5.4)
## udine 388 (15.5)
## time (median [IQR]) 731.00 [401.00, 1563.00]
## los (mean (SD)) 0.05 (0.22)
## male (mean (SD)) 0.55 (0.50)
## age (median [IQR]) 74.00 [63.00, 81.00]
## nihss_middle (mean (SD)) 0.30 (0.46)
## nihss_high (mean (SD)) 0.15 (0.36)
## macro (mean (SD)) 0.18 (0.39)
## cardio (mean (SD)) 0.29 (0.45)
## micro (mean (SD)) 0.27 (0.44)
## mca (mean (SD)) 0.63 (0.48)
## cortical (mean (SD)) 0.45 (0.50)
## thrombolysis (mean (SD)) 0.00 (0.00)
## iv_thrombolysis (mean (SD)) 0.00 (0.00)
## ia_thrombolysis (mean (SD)) 0.00 (0.00)
## early (mean (SD)) 0.05 (0.21)
## early_treatment (mean (SD)) 0.03 (0.18)
## Stratified by thrombolysis
## 1 p test
## n 700
## ID (mean (SD)) 2114.18 (909.08) <0.001
## source (%) <0.001
## bentes 93 (13.3)
## hebron 236 (33.7)
## kssg 139 (19.9)
## linz 102 (14.6)
## muenster 73 (10.4)
## saar 46 ( 6.6)
## udine 11 ( 1.6)
## time (median [IQR]) 723.50 [365.00, 1733.50] 0.037 nonnorm
## los (mean (SD)) 0.11 (0.31) <0.001
## male (mean (SD)) 0.58 (0.49) 0.183
## age (median [IQR]) 73.00 [62.00, 80.00] 0.048 nonnorm
## nihss_middle (mean (SD)) 0.42 (0.49) <0.001
## nihss_high (mean (SD)) 0.49 (0.50) <0.001
## macro (mean (SD)) 0.26 (0.44) <0.001
## cardio (mean (SD)) 0.37 (0.48) <0.001
## micro (mean (SD)) 0.10 (0.30) <0.001
## mca (mean (SD)) 0.76 (0.43) <0.001
## cortical (mean (SD)) 0.55 (0.50) <0.001
## thrombolysis (mean (SD)) 1.00 (0.00) <0.001
## iv_thrombolysis (mean (SD)) 0.95 (0.21) <0.001
## ia_thrombolysis (mean (SD)) 0.09 (0.29) <0.001
## early (mean (SD)) 0.07 (0.26) 0.002
## early_treatment (mean (SD)) 0.07 (0.26) <0.001
## Stratified by thrombolysis
## SMD Missing
## n
## ID (mean (SD)) 0.677 0.0
## source (%) 0.999 0.0
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## time (median [IQR]) 0.037 0.0
## los (mean (SD)) 0.195 0.0
## male (mean (SD)) 0.057 0.0
## age (median [IQR]) 0.092 0.0
## nihss_middle (mean (SD)) 0.251 0.0
## nihss_high (mean (SD)) 0.777 0.0
## macro (mean (SD)) 0.188 0.0
## cardio (mean (SD)) 0.167 0.0
## micro (mean (SD)) 0.436 0.0
## mca (mean (SD)) 0.270 0.0
## cortical (mean (SD)) 0.191 0.0
## thrombolysis (mean (SD)) Inf 0.0
## iv_thrombolysis (mean (SD)) 6.457 0.0
## ia_thrombolysis (mean (SD)) 0.456 0.0
## early (mean (SD)) 0.121 0.0
## early_treatment (mean (SD)) 0.179 0.0write.csv(tab2, 'tableonedata.csv')
listvars <- c(
"source",
"age",
"male",
"nihss_middle",
"nihss_high",
"macro",
"micro",
"cardio",
"mca",
"cortical",
"early_treatment")
tabUnmatched <- CreateTableOne(vars = listvars, strata = "thrombolysis", data = data, test = TRUE, testApprox = chisq.test )
tabUnmatched2 <- print(tabUnmatched, smd = TRUE)## Stratified by thrombolysis
## 0 1 p test
## n 2501 700
## source (%) <0.001
## bentes 46 ( 1.8) 93 (13.3)
## hebron 275 (11.0) 236 (33.7)
## kssg 1061 (42.4) 139 (19.9)
## linz 357 (14.3) 102 (14.6)
## muenster 238 ( 9.5) 73 (10.4)
## saar 136 ( 5.4) 46 ( 6.6)
## udine 388 (15.5) 11 ( 1.6)
## age (mean (SD)) 71.28 (13.64) 70.03 (13.61) 0.031
## male (mean (SD)) 0.55 (0.50) 0.58 (0.49) 0.183
## nihss_middle (mean (SD)) 0.30 (0.46) 0.42 (0.49) <0.001
## nihss_high (mean (SD)) 0.15 (0.36) 0.49 (0.50) <0.001
## macro (mean (SD)) 0.18 (0.39) 0.26 (0.44) <0.001
## micro (mean (SD)) 0.27 (0.44) 0.10 (0.30) <0.001
## cardio (mean (SD)) 0.29 (0.45) 0.37 (0.48) <0.001
## mca (mean (SD)) 0.63 (0.48) 0.76 (0.43) <0.001
## cortical (mean (SD)) 0.45 (0.50) 0.55 (0.50) <0.001
## early_treatment (mean (SD)) 0.03 (0.18) 0.07 (0.26) <0.001
## Stratified by thrombolysis
## SMD
## n
## source (%) 0.999
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## age (mean (SD)) 0.092
## male (mean (SD)) 0.057
## nihss_middle (mean (SD)) 0.251
## nihss_high (mean (SD)) 0.777
## macro (mean (SD)) 0.188
## micro (mean (SD)) 0.436
## cardio (mean (SD)) 0.167
## mca (mean (SD)) 0.270
## cortical (mean (SD)) 0.191
## early_treatment (mean (SD)) 0.179addmargins(table(ExtractSmd(tabUnmatched) > 0.1))##
## FALSE TRUE Sum
## 2 9 111.3 Matched data
set.seed(12347)
psm <- matchit(thrombolysis ~
source+
age +
male+
nihss_middle +
nihss_high+
macro+
micro+
cardio+
mca+
cortical+
early_treatment,
data = data, method = "nearest", distance = "logit" ,replace = FALSE, caliper = 0.2)
psm_matchdata <- match.data(psm)
tabMatched <- CreateTableOne(vars = listvars, strata = "thrombolysis",data = psm_matchdata, test = TRUE, testApprox = chisq.test, smd = TRUE)
tabMatched2 <- print(tabMatched, smd = TRUE)## Stratified by thrombolysis
## 0 1 p test
## n 598 598
## source (%) 0.218
## bentes 45 ( 7.5) 67 (11.2)
## hebron 154 (25.8) 172 (28.8)
## kssg 154 (25.8) 139 (23.2)
## linz 108 (18.1) 98 (16.4)
## muenster 71 (11.9) 70 (11.7)
## saar 51 ( 8.5) 41 ( 6.9)
## udine 15 ( 2.5) 11 ( 1.8)
## age (mean (SD)) 70.25 (14.60) 69.44 (13.73) 0.320
## male (mean (SD)) 0.56 (0.50) 0.59 (0.49) 0.293
## nihss_middle (mean (SD)) 0.48 (0.50) 0.46 (0.50) 0.487
## nihss_high (mean (SD)) 0.42 (0.49) 0.44 (0.50) 0.521
## macro (mean (SD)) 0.24 (0.42) 0.26 (0.44) 0.316
## micro (mean (SD)) 0.11 (0.31) 0.11 (0.31) 0.853
## cardio (mean (SD)) 0.39 (0.49) 0.36 (0.48) 0.403
## mca (mean (SD)) 0.73 (0.45) 0.75 (0.43) 0.431
## cortical (mean (SD)) 0.57 (0.50) 0.56 (0.50) 0.684
## early_treatment (mean (SD)) 0.06 (0.24) 0.07 (0.26) 0.355
## Stratified by thrombolysis
## SMD
## n
## source (%) 0.167
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## age (mean (SD)) 0.058
## male (mean (SD)) 0.061
## nihss_middle (mean (SD)) 0.040
## nihss_high (mean (SD)) 0.037
## macro (mean (SD)) 0.058
## micro (mean (SD)) 0.011
## cardio (mean (SD)) 0.048
## mca (mean (SD)) 0.046
## cortical (mean (SD)) 0.024
## early_treatment (mean (SD)) 0.054addmargins(table(ExtractSmd(tabMatched) > 0.1))##
## FALSE TRUE Sum
## 10 1 111.3.1 Examine matching performance
plot(psm, type = "hist")
resCombo <- cbind(print(tabUnmatched, printToggle = FALSE),
print(tabMatched, printToggle = FALSE))
resCombo <- rbind(Group = rep(c("No Thrombolysis", "Thrombolysis", "P-value","", "No Thrombolysis", "Thrombolysis", "P-value", ""),6), resCombo)
colnames(resCombo) <- c("Unmatched","","","", "Matched","","", "")
print(resCombo, quote = FALSE)## Unmatched
## Group No Thrombolysis Thrombolysis P-value
## n 2501 700
## source (%) <0.001
## bentes 46 ( 1.8) 93 (13.3)
## hebron 275 (11.0) 236 (33.7)
## kssg 1061 (42.4) 139 (19.9)
## linz 357 (14.3) 102 (14.6)
## muenster 238 ( 9.5) 73 (10.4)
## saar 136 ( 5.4) 46 ( 6.6)
## udine 388 (15.5) 11 ( 1.6)
## age (mean (SD)) 71.28 (13.64) 70.03 (13.61) 0.031
## male (mean (SD)) 0.55 (0.50) 0.58 (0.49) 0.183
## nihss_middle (mean (SD)) 0.30 (0.46) 0.42 (0.49) <0.001
## nihss_high (mean (SD)) 0.15 (0.36) 0.49 (0.50) <0.001
## macro (mean (SD)) 0.18 (0.39) 0.26 (0.44) <0.001
## micro (mean (SD)) 0.27 (0.44) 0.10 (0.30) <0.001
## cardio (mean (SD)) 0.29 (0.45) 0.37 (0.48) <0.001
## mca (mean (SD)) 0.63 (0.48) 0.76 (0.43) <0.001
## cortical (mean (SD)) 0.45 (0.50) 0.55 (0.50) <0.001
## early_treatment (mean (SD)) 0.03 (0.18) 0.07 (0.26) <0.001
## Matched
## Group No Thrombolysis Thrombolysis P-value
## n 598 598
## source (%) 0.218
## bentes 45 ( 7.5) 67 (11.2)
## hebron 154 (25.8) 172 (28.8)
## kssg 154 (25.8) 139 (23.2)
## linz 108 (18.1) 98 (16.4)
## muenster 71 (11.9) 70 (11.7)
## saar 51 ( 8.5) 41 ( 6.9)
## udine 15 ( 2.5) 11 ( 1.8)
## age (mean (SD)) 70.25 (14.60) 69.44 (13.73) 0.320
## male (mean (SD)) 0.56 (0.50) 0.59 (0.49) 0.293
## nihss_middle (mean (SD)) 0.48 (0.50) 0.46 (0.50) 0.487
## nihss_high (mean (SD)) 0.42 (0.49) 0.44 (0.50) 0.521
## macro (mean (SD)) 0.24 (0.42) 0.26 (0.44) 0.316
## micro (mean (SD)) 0.11 (0.31) 0.11 (0.31) 0.853
## cardio (mean (SD)) 0.39 (0.49) 0.36 (0.48) 0.403
## mca (mean (SD)) 0.73 (0.45) 0.75 (0.43) 0.431
## cortical (mean (SD)) 0.57 (0.50) 0.56 (0.50) 0.684
## early_treatment (mean (SD)) 0.06 (0.24) 0.07 (0.26) 0.355write.csv(resCombo, 'selectmatched.csv')1.4 Analysis
1.4.1 Time to first seizure unmatched data
formula <- glm(formula = early ~ thrombolysis, data = data, family=binomial)
predicted <- predict(formula, data=data, type="response")
summary(formula)##
## Call:
## glm(formula = early ~ thrombolysis, family = binomial, data = data)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.3929 -0.3055 -0.3055 -0.3055 2.4853
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -3.04159 0.09586 -31.730 < 2e-16 ***
## thrombolysis 0.51895 0.17310 2.998 0.00272 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1305.7 on 3200 degrees of freedom
## Residual deviance: 1297.3 on 3199 degrees of freedom
## AIC: 1301.3
##
## Number of Fisher Scoring iterations: 51.4.2 Time to first seizure matched data
formula <- glm(formula = early ~ thrombolysis, data = psm_matchdata, family=binomial)
predicted <- predict(formula, data=psm_matchdata, type="response")
summary(formula)##
## Call:
## glm(formula = early ~ thrombolysis, family = binomial, data = psm_matchdata)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.4001 -0.4001 -0.3624 -0.3624 2.3478
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.6904 0.1676 -16.049 <2e-16 ***
## thrombolysis 0.2054 0.2273 0.904 0.366
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 608.15 on 1195 degrees of freedom
## Residual deviance: 607.33 on 1194 degrees of freedom
## AIC: 611.33
##
## Number of Fisher Scoring iterations: 51.4.3 Time to late seizure unmatched data
formula <- coxph(Surv(time,los) ~ thrombolysis, data = data)
summary(formula)## Call:
## coxph(formula = Surv(time, los) ~ thrombolysis, data = data)
##
## n= 3201, number of events= 207
##
## coef exp(coef) se(coef) z Pr(>|z|)
## thrombolysis 0.7315 2.0781 0.1453 5.034 4.8e-07 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## exp(coef) exp(-coef) lower .95 upper .95
## thrombolysis 2.078 0.4812 1.563 2.763
##
## Concordance= 0.562 (se = 0.017 )
## Likelihood ratio test= 23.26 on 1 df, p=1e-06
## Wald test = 25.34 on 1 df, p=5e-07
## Score (logrank) test = 26.48 on 1 df, p=3e-07km <- Surv(time = as.numeric(data$time), event = data$los)
fit <- survfit(km ~ thrombolysis, data = data)
ggsurvplot(fit, data = data, pval = TRUE, risk.table = TRUE, tables.theme = clean_theme(), conf.int = TRUE, tables.height = 0.25, title = "Time to late seizure (Unmatched)", legend.labs = c("No Thrombolysis", "Thrombolysis"), legend = c(0.8, 0.2), font.main = c(13, "bold"), font.x = c(11, "bold" ),font.y = c(11, "bold"),font.tickslab = c(10, "plain"), xlim = c(0, 2500), ylim = c(0.5, 1), pval.size = 4, break.time.by = 500, risk.table.fontsize = 4, risk.table.title.fontsize = 10, xlab = "Time (days)", palette = c("#808080", "#90ab09"))
1.4.4 Time to late seizure matched data
formula2 <- coxph(Surv(as.numeric(psm_matchdata$time),psm_matchdata$los) ~ psm_matchdata$thrombolysis, data = psm_matchdata)
summary(formula2)## Call:
## coxph(formula = Surv(as.numeric(psm_matchdata$time), psm_matchdata$los) ~
## psm_matchdata$thrombolysis, data = psm_matchdata)
##
## n= 1196, number of events= 120
##
## coef exp(coef) se(coef) z Pr(>|z|)
## psm_matchdata$thrombolysis 0.2298 1.2583 0.1840 1.249 0.212
##
## exp(coef) exp(-coef) lower .95 upper .95
## psm_matchdata$thrombolysis 1.258 0.7947 0.8773 1.805
##
## Concordance= 0.529 (se = 0.025 )
## Likelihood ratio test= 1.57 on 1 df, p=0.2
## Wald test = 1.56 on 1 df, p=0.2
## Score (logrank) test = 1.57 on 1 df, p=0.2km2 <- Surv(time = as.numeric(psm_matchdata$time), event =psm_matchdata$los)
fit2 <- survfit(km2 ~ thrombolysis, data = psm_matchdata)
ggsurvplot(fit2, data = psm_matchdata, pval = TRUE, risk.table = TRUE, tables.theme = clean_theme(), conf.int = TRUE, tables.height = 0.25, title = "Time to late seizure (Matched)", legend.labs = c("No Thrombolysis", "Thrombolysis"), legend = c(0.8, 0.2), font.main = c(13, "bold"), font.x = c(11, "bold" ),font.y = c(11, "bold"),font.tickslab = c(10, "plain"), xlim = c(0, 2500), ylim = c(0.5, 1), pval.size = 4, break.time.by = 500, risk.table.fontsize = 4, risk.table.title.fontsize = 10, xlab = "Time (days)", palette = c("#808080", "#90ab09"))
2 IV Thrombolysis
2.1 Unmatched data
data <- read_excel("data_for_analysis_iv.xlsx")
tabiv <- CreateTableOne(data = data, strata = "iv_thrombolysis")
cat <- c("source", "los", "early", "early_treatment", "male", "nihss_middle", "nihss_high", "macro", "micro", "mca", "cortical", "thrombolysis", "iv_thrombolysis", "ia_thrombolysis")
cont <- c("time","age")
tab2iv <- print(tabiv, nonnormal = cont, smd = TRUE, missing = TRUE)## Stratified by iv_thrombolysis
## 0
## n 2506
## ID (mean (SD)) 1492.72 (933.89)
## source (%)
## bentes 46 ( 1.8)
## hebron 275 (11.0)
## kssg 1061 (42.3)
## linz 359 (14.3)
## muenster 238 ( 9.5)
## saar 139 ( 5.5)
## udine 388 (15.5)
## time (median [IQR]) 731.00 [401.75, 1563.75]
## los (mean (SD)) 0.05 (0.23)
## male (mean (SD)) 0.56 (0.50)
## age (median [IQR]) 74.00 [63.00, 81.00]
## nihss_middle (mean (SD)) 0.30 (0.46)
## nihss_high (mean (SD)) 0.15 (0.36)
## macro (mean (SD)) 0.18 (0.39)
## cardio (mean (SD)) 0.29 (0.45)
## micro (mean (SD)) 0.27 (0.44)
## mca (mean (SD)) 0.63 (0.48)
## cortical (mean (SD)) 0.45 (0.50)
## thrombolysis (mean (SD)) 0.00 (0.04)
## iv_thrombolysis (mean (SD)) 0.00 (0.00)
## ia_thrombolysis (mean (SD)) 0.00 (0.04)
## early (mean (SD)) 0.05 (0.21)
## early_treatment (mean (SD)) 0.03 (0.18)
## Stratified by iv_thrombolysis
## 1 p test
## n 622
## ID (mean (SD)) 2144.96 (953.63) <0.001
## source (%) <0.001
## bentes 93 (15.0)
## hebron 236 (37.9)
## kssg 139 (22.3)
## linz 100 (16.1)
## muenster 0 ( 0.0)
## saar 43 ( 6.9)
## udine 11 ( 1.8)
## time (median [IQR]) 723.50 [365.00, 1743.00] 0.045 nonnorm
## los (mean (SD)) 0.10 (0.31) <0.001
## male (mean (SD)) 0.59 (0.49) 0.175
## age (median [IQR]) 73.00 [63.00, 80.75] 0.464 nonnorm
## nihss_middle (mean (SD)) 0.41 (0.49) <0.001
## nihss_high (mean (SD)) 0.50 (0.50) <0.001
## macro (mean (SD)) 0.26 (0.44) <0.001
## cardio (mean (SD)) 0.37 (0.48) <0.001
## micro (mean (SD)) 0.12 (0.32) <0.001
## mca (mean (SD)) 0.74 (0.44) <0.001
## cortical (mean (SD)) 0.59 (0.49) <0.001
## thrombolysis (mean (SD)) 1.00 (0.00) <0.001
## iv_thrombolysis (mean (SD)) 1.00 (0.00) <0.001
## ia_thrombolysis (mean (SD)) 0.05 (0.22) <0.001
## early (mean (SD)) 0.08 (0.26) 0.003
## early_treatment (mean (SD)) 0.07 (0.26) <0.001
## Stratified by iv_thrombolysis
## SMD Missing
## n
## ID (mean (SD)) 0.691 0.0
## source (%) 1.185 0.0
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## time (median [IQR]) 0.034 0.0
## los (mean (SD)) 0.190 0.0
## male (mean (SD)) 0.061 0.0
## age (median [IQR]) 0.026 0.0
## nihss_middle (mean (SD)) 0.230 0.0
## nihss_high (mean (SD)) 0.795 0.0
## macro (mean (SD)) 0.198 0.0
## cardio (mean (SD)) 0.173 0.0
## micro (mean (SD)) 0.394 0.0
## mca (mean (SD)) 0.224 0.0
## cortical (mean (SD)) 0.278 0.0
## thrombolysis (mean (SD)) 31.623 0.0
## iv_thrombolysis (mean (SD)) Inf 0.0
## ia_thrombolysis (mean (SD)) 0.316 0.0
## early (mean (SD)) 0.124 0.0
## early_treatment (mean (SD)) 0.186 0.0write.csv(tab2iv, 'tableonedata_iv.csv')
listvars <- c(
"source",
"age",
"male",
"nihss_middle",
"nihss_high",
"macro",
"micro",
"cardio",
"mca",
"cortical",
"early_treatment")
tabUnmatchediv <- CreateTableOne(vars = listvars, strata = "iv_thrombolysis", data = data, test = TRUE, testApprox = chisq.test )
tabUnmatched2iv <- print(tabUnmatchediv, smd = TRUE)## Stratified by iv_thrombolysis
## 0 1 p test
## n 2506 622
## source (%) <0.001
## bentes 46 ( 1.8) 93 (15.0)
## hebron 275 (11.0) 236 (37.9)
## kssg 1061 (42.3) 139 (22.3)
## linz 359 (14.3) 100 (16.1)
## muenster 238 ( 9.5) 0 ( 0.0)
## saar 139 ( 5.5) 43 ( 6.9)
## udine 388 (15.5) 11 ( 1.8)
## age (mean (SD)) 71.28 (13.63) 70.94 (12.83) 0.573
## male (mean (SD)) 0.56 (0.50) 0.59 (0.49) 0.175
## nihss_middle (mean (SD)) 0.30 (0.46) 0.41 (0.49) <0.001
## nihss_high (mean (SD)) 0.15 (0.36) 0.50 (0.50) <0.001
## macro (mean (SD)) 0.18 (0.39) 0.26 (0.44) <0.001
## micro (mean (SD)) 0.27 (0.44) 0.12 (0.32) <0.001
## cardio (mean (SD)) 0.29 (0.45) 0.37 (0.48) <0.001
## mca (mean (SD)) 0.63 (0.48) 0.74 (0.44) <0.001
## cortical (mean (SD)) 0.45 (0.50) 0.59 (0.49) <0.001
## early_treatment (mean (SD)) 0.03 (0.18) 0.07 (0.26) <0.001
## Stratified by iv_thrombolysis
## SMD
## n
## source (%) 1.185
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## age (mean (SD)) 0.026
## male (mean (SD)) 0.061
## nihss_middle (mean (SD)) 0.230
## nihss_high (mean (SD)) 0.795
## macro (mean (SD)) 0.198
## micro (mean (SD)) 0.394
## cardio (mean (SD)) 0.173
## mca (mean (SD)) 0.224
## cortical (mean (SD)) 0.278
## early_treatment (mean (SD)) 0.186addmargins(table(ExtractSmd(tabUnmatchediv) > 0.1))##
## FALSE TRUE Sum
## 2 9 112.2 Matched data
set.seed(122457)
psmiv <- matchit(iv_thrombolysis ~
source +
age +
male +
nihss_middle +
nihss_high +
macro +
micro +
cardio +
mca +
cortical +
early_treatment,
data = data, method = "nearest", distance = "logit" ,replace = FALSE, caliper = 0.2)
psm_matchdataiv <- match.data(psmiv)
tabMatchediv <- CreateTableOne(vars = listvars, strata = "iv_thrombolysis", data = psm_matchdataiv, test = TRUE, testApprox = chisq.test, smd = TRUE)
tabMatched2iv <- print(tabMatchediv, smd = TRUE)## Stratified by iv_thrombolysis
## 0 1 p test
## n 515 515
## source (%) 0.215
## bentes 46 ( 8.9) 64 (12.4)
## hebron 159 (30.9) 165 (32.0)
## kssg 145 (28.2) 139 (27.0)
## linz 107 (20.8) 95 (18.4)
## muenster 3 ( 0.6) 0 ( 0.0)
## saar 48 ( 9.3) 41 ( 8.0)
## udine 7 ( 1.4) 11 ( 2.1)
## age (mean (SD)) 71.17 (14.31) 70.25 (13.05) 0.281
## male (mean (SD)) 0.58 (0.49) 0.60 (0.49) 0.448
## nihss_middle (mean (SD)) 0.45 (0.50) 0.44 (0.50) 0.707
## nihss_high (mean (SD)) 0.43 (0.50) 0.45 (0.50) 0.415
## macro (mean (SD)) 0.26 (0.44) 0.25 (0.43) 0.775
## micro (mean (SD)) 0.13 (0.34) 0.12 (0.33) 0.638
## cardio (mean (SD)) 0.37 (0.48) 0.38 (0.49) 0.700
## mca (mean (SD)) 0.70 (0.46) 0.73 (0.45) 0.336
## cortical (mean (SD)) 0.60 (0.49) 0.60 (0.49) 0.849
## early_treatment (mean (SD)) 0.07 (0.25) 0.08 (0.27) 0.470
## Stratified by iv_thrombolysis
## SMD
## n
## source (%) 0.181
## bentes
## hebron
## kssg
## linz
## muenster
## saar
## udine
## age (mean (SD)) 0.067
## male (mean (SD)) 0.047
## nihss_middle (mean (SD)) 0.023
## nihss_high (mean (SD)) 0.051
## macro (mean (SD)) 0.018
## micro (mean (SD)) 0.029
## cardio (mean (SD)) 0.024
## mca (mean (SD)) 0.060
## cortical (mean (SD)) 0.012
## early_treatment (mean (SD)) 0.045addmargins(table(ExtractSmd(tabMatchediv) > 0.1))##
## FALSE TRUE Sum
## 10 1 112.2.1 Examine matching performance
plot(psmiv, type = "hist")
resComboiv <- cbind(print(tabUnmatchediv, printToggle = FALSE),
print(tabMatchediv, printToggle = FALSE))
resComboiv <- rbind(Group = rep(c("No IV Thrombolysis", "IV Thrombolysis", "P-value","", "No IV Thrombolysis", "IV Thrombolysis", "P-value", ""),6), resComboiv)
colnames(resComboiv) <- c("Unmatched","","","", "Matched","","", "")
print(resComboiv, quote = FALSE)## Unmatched
## Group No IV Thrombolysis IV Thrombolysis P-value
## n 2506 622
## source (%) <0.001
## bentes 46 ( 1.8) 93 (15.0)
## hebron 275 (11.0) 236 (37.9)
## kssg 1061 (42.3) 139 (22.3)
## linz 359 (14.3) 100 (16.1)
## muenster 238 ( 9.5) 0 ( 0.0)
## saar 139 ( 5.5) 43 ( 6.9)
## udine 388 (15.5) 11 ( 1.8)
## age (mean (SD)) 71.28 (13.63) 70.94 (12.83) 0.573
## male (mean (SD)) 0.56 (0.50) 0.59 (0.49) 0.175
## nihss_middle (mean (SD)) 0.30 (0.46) 0.41 (0.49) <0.001
## nihss_high (mean (SD)) 0.15 (0.36) 0.50 (0.50) <0.001
## macro (mean (SD)) 0.18 (0.39) 0.26 (0.44) <0.001
## micro (mean (SD)) 0.27 (0.44) 0.12 (0.32) <0.001
## cardio (mean (SD)) 0.29 (0.45) 0.37 (0.48) <0.001
## mca (mean (SD)) 0.63 (0.48) 0.74 (0.44) <0.001
## cortical (mean (SD)) 0.45 (0.50) 0.59 (0.49) <0.001
## early_treatment (mean (SD)) 0.03 (0.18) 0.07 (0.26) <0.001
## Matched
## Group No IV Thrombolysis IV Thrombolysis P-value
## n 515 515
## source (%) 0.215
## bentes 46 ( 8.9) 64 (12.4)
## hebron 159 (30.9) 165 (32.0)
## kssg 145 (28.2) 139 (27.0)
## linz 107 (20.8) 95 (18.4)
## muenster 3 ( 0.6) 0 ( 0.0)
## saar 48 ( 9.3) 41 ( 8.0)
## udine 7 ( 1.4) 11 ( 2.1)
## age (mean (SD)) 71.17 (14.31) 70.25 (13.05) 0.281
## male (mean (SD)) 0.58 (0.49) 0.60 (0.49) 0.448
## nihss_middle (mean (SD)) 0.45 (0.50) 0.44 (0.50) 0.707
## nihss_high (mean (SD)) 0.43 (0.50) 0.45 (0.50) 0.415
## macro (mean (SD)) 0.26 (0.44) 0.25 (0.43) 0.775
## micro (mean (SD)) 0.13 (0.34) 0.12 (0.33) 0.638
## cardio (mean (SD)) 0.37 (0.48) 0.38 (0.49) 0.700
## mca (mean (SD)) 0.70 (0.46) 0.73 (0.45) 0.336
## cortical (mean (SD)) 0.60 (0.49) 0.60 (0.49) 0.849
## early_treatment (mean (SD)) 0.07 (0.25) 0.08 (0.27) 0.470write.csv(resComboiv, 'selectmatched_iv.csv')2.3 Analysis
2.3.1 Time to first seizure unmatched data
formula <- glm(formula = early ~ iv_thrombolysis, data = data, family=binomial)
predicted <- predict(formula, data=data, type="response")
summary(formula)##
## Call:
## glm(formula = early ~ iv_thrombolysis, family = binomial, data = data)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.3964 -0.3065 -0.3065 -0.3065 2.4825
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -3.03453 0.09546 -31.790 < 2e-16 ***
## iv_thrombolysis 0.53031 0.17924 2.959 0.00309 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1274.7 on 3127 degrees of freedom
## Residual deviance: 1266.5 on 3126 degrees of freedom
## AIC: 1270.5
##
## Number of Fisher Scoring iterations: 52.3.2 Time to first seizure matched data
formula <- glm(formula = early ~ iv_thrombolysis, data = psm_matchdataiv, family=binomial)
predicted <- predict(formula, data=psm_matchdataiv, type="response")
summary(formula)##
## Call:
## glm(formula = early ~ iv_thrombolysis, family = binomial, data = psm_matchdataiv)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.4125 -0.4125 -0.3969 -0.3969 2.2718
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.50186 0.16656 -15.021 <2e-16 ***
## iv_thrombolysis 0.08043 0.23166 0.347 0.728
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 567.40 on 1029 degrees of freedom
## Residual deviance: 567.28 on 1028 degrees of freedom
## AIC: 571.28
##
## Number of Fisher Scoring iterations: 52.3.3 Time to late seizure unmatched data
formula <- coxph(Surv(time,los) ~ iv_thrombolysis, data = data)
summary(formula)## Call:
## coxph(formula = Surv(time, los) ~ iv_thrombolysis, data = data)
##
## n= 3128, number of events= 199
##
## coef exp(coef) se(coef) z Pr(>|z|)
## iv_thrombolysis 0.7143 2.0427 0.1515 4.715 2.41e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## exp(coef) exp(-coef) lower .95 upper .95
## iv_thrombolysis 2.043 0.4895 1.518 2.749
##
## Concordance= 0.553 (se = 0.017 )
## Likelihood ratio test= 20.19 on 1 df, p=7e-06
## Wald test = 22.23 on 1 df, p=2e-06
## Score (logrank) test = 23.19 on 1 df, p=1e-06km <- Surv(time = as.numeric(data$time), event = data$los)
fit <- survfit(km ~ iv_thrombolysis, data = data)
ggsurvplot(fit, data = data, pval = TRUE, risk.table = TRUE, tables.theme = clean_theme(), conf.int = TRUE, tables.height = 0.25, title = "Time to late seizure (Unmatched)", legend.labs = c("No IV Thrombolysis", "IV Thrombolysis"), legend = c(0.8, 0.2), font.main = c(13, "bold"), font.x = c(11, "bold" ),font.y = c(11, "bold"),font.tickslab = c(10, "plain"), xlim = c(0, 2500), ylim = c(0.5, 1), pval.size = 4, break.time.by = 500, risk.table.fontsize = 4, risk.table.title.fontsize = 10, xlab = "Time (days)", palette = c("#808080", "#90ab09"))
2.3.4 Time to late seizure matched data
formula2 <- coxph(Surv(as.numeric(psm_matchdataiv$time),psm_matchdataiv$los) ~ psm_matchdataiv$iv_thrombolysis, data = psm_matchdataiv)
summary(formula2)## Call:
## coxph(formula = Surv(as.numeric(psm_matchdataiv$time), psm_matchdataiv$los) ~
## psm_matchdataiv$iv_thrombolysis, data = psm_matchdataiv)
##
## n= 1030, number of events= 98
##
## coef exp(coef) se(coef) z Pr(>|z|)
## psm_matchdataiv$iv_thrombolysis 0.1508 1.1627 0.2031 0.742 0.458
##
## exp(coef) exp(-coef) lower .95 upper .95
## psm_matchdataiv$iv_thrombolysis 1.163 0.86 0.781 1.731
##
## Concordance= 0.512 (se = 0.028 )
## Likelihood ratio test= 0.55 on 1 df, p=0.5
## Wald test = 0.55 on 1 df, p=0.5
## Score (logrank) test = 0.55 on 1 df, p=0.5km2 <- Surv(time = as.numeric(psm_matchdataiv$time), event =psm_matchdataiv$los)
fit2 <- survfit(km2 ~ iv_thrombolysis, data = psm_matchdataiv)
ggsurvplot(fit2, data = psm_matchdataiv, pval = TRUE, risk.table = TRUE, tables.theme = clean_theme(), conf.int = TRUE, tables.height = 0.25, title = "Time to late seizure (Matched)", legend.labs = c("No IV Thrombolysis", "IV Thrombolysis"), legend = c(0.8, 0.2), font.main = c(13, "bold"), font.x = c(11, "bold" ),font.y = c(11, "bold"),font.tickslab = c(10, "plain"), xlim = c(0, 2500), ylim = c(0.5, 1), pval.size = 4, break.time.by = 500, risk.table.fontsize = 4, risk.table.title.fontsize = 10, xlab = "Time (days)", palette = c("#808080", "#90ab09"))