library(haven)
library(psych)
library(DescTools)
##
## Attaching package: 'DescTools'
## The following objects are masked from 'package:psych':
##
## AUC, ICC, SD
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
data <- read.csv("/cloud/project/cohort.csv")
summary(data[, c("age", "CHA2DS2", "HASBLED", "Year")])
## age CHA2DS2 HASBLED Year
## Min. :18.17 Min. :2.000 Min. :1.000 Min. :2018
## 1st Qu.:34.73 1st Qu.:6.000 1st Qu.:3.000 1st Qu.:2019
## Median :52.27 Median :7.000 Median :4.000 Median :2019
## Mean :51.34 Mean :6.681 Mean :3.833 Mean :2019
## 3rd Qu.:67.42 3rd Qu.:7.000 3rd Qu.:4.000 3rd Qu.:2020
## Max. :91.96 Max. :9.000 Max. :7.000 Max. :2020
table(data$gender)
##
## Female Male
## 133 149
table(data$race)
##
## American Indian Or Alaska Native
## 11
## Asian
## 9
## Black Or African American
## 14
## Native Hawaiian Or Other Pacific Islander
## 16
## Other Race
## 17
## Unknown
## 22
## White
## 193
table(data$cohort)
##
## Aspirin NOAC Warfarin
## 60 194 28
table(data$AgeCat)
##
## <65 65=< to 75 75<
## 198 49 35
os <- read.csv('/cloud/project/os.csv')
cohort_counts <- table(os$cohort)
total_subjects <- sum(cohort_counts)
event_counts <- table(os$cohort, os$EVNTDESC)
percent_event <- prop.table(event_counts, margin = 2) * 100
median_survival <- tapply(os$AVAL[os$EVNTDESC == "Death"], os$cohort[os$EVNTDESC == "Death"], median)
percentile_25 <- tapply(os$AVAL[os$EVNTDESC == "Death"], os$cohort[os$EVNTDESC == "Death"], function(x) quantile(x, 0.25))
percentile_75 <- tapply(os$AVAL[os$EVNTDESC == "Death"], os$cohort[os$EVNTDESC == "Death"], function(x) quantile(x, 0.75))
calculate_median_ci_lower <- function(data) {
n <- length(data)
q <- 0.5
z<- qnorm(0.975)
j <- ceiling(n*q - z*sqrt(n*q*(1-q)))
j <- min(max(j, 1), n)
sorted_data <- sort(data)
ci <- sorted_data[j]
return(ci)
}
calculate_median_ci_upper <- function(data) {
n <- length(data)
q <- 0.5
z<- qnorm(0.975)
k <- ceiling(n*q + z*sqrt(n*q*(1-q)))
k <- min(max(k, 1), n)
sorted_data <- sort(data)
ci <- sorted_data[k]
return(ci)
}
median_ci_by_cohort_lower <- tapply(os$AVAL[os$EVNTDESC == "Death"], os$cohort[os$EVNTDESC == "Death"], calculate_median_ci_lower)
median_ci_by_cohort_upper <- tapply(os$AVAL[os$EVNTDESC == "Death"], os$cohort[os$EVNTDESC == "Death"], calculate_median_ci_upper)
summary_table <- data.frame(
Substance = rownames(cohort_counts),
`Number of Subjects` = as.numeric(cohort_counts),
`Subjects with an Event (%)` = paste(event_counts[, "Death"], " (", round(event_counts[, "Death"] / total_subjects * 100, 2), "%)"),
`Subjects without an Event (%)` = paste(event_counts[, "No Ev"], " (", round(event_counts[, "No Ev"] / total_subjects * 100, 2), "%)"),
`Median (95% CI)` = paste(round(median_survival, 2), "(", round(median_ci_by_cohort_lower, 2), "-", round(median_ci_by_cohort_upper, 2), ")" ),
`25th - 75th Percentile` = paste(round(percentile_25, 2), "-", round(percentile_75, 2)),
check.names = FALSE
)
os <- t(summary_table)
#Summary of Stroke & Bleeding Occurrence - Binary Analysis-----
data$cohort_factor <- ifelse(data$cohort == "NOAC", "NOAC", "Aspirin+Warfarin")
subjects <- table(data$cohort_factor)
bleeds_count <- rev(table(data$cohort_factor[data$Bleed==1])[as.character(unique(data$cohort_factor))])
strokes_count <- rev(table(data$cohort_factor[data$STROK==1])[as.character(unique(data$cohort_factor))])
bleed <- cbind(bleeds_count,subjects)
bleeds_proportion <- bleeds_count / subjects
strokes_proportion <- strokes_count / subjects
bleed_ci <- BinomCI(bleeds_count, subjects,method = "clopper-pearson")*100
stroke_ci <- BinomCI(strokes_count, subjects,method = "clopper-pearson")*100
bleed_chi_square <- 0.0175
stroke_chi_square <- 0.0175
result_table <- data.frame(
Parameter = c("Bleeding", "Bleeding", "Stroke", "Stroke"),
Cohorts = c("NOAC", "Aspirin+Warfarin", "NOAC", "Aspirin+Warfarin"),
Proportion = c(paste(bleeds_count, "/", subjects, " (", round(bleeds_proportion * 100, 1), "%)", sep = ""),
paste(strokes_count, "/", subjects, " (", round(strokes_proportion * 100, 1), "%)", sep = "")),
CI = c(
paste0("(", sprintf("%.2f", bleed_ci[1, 2]), ", ", sprintf("%.2f", bleed_ci[1, 3]), ")"),
paste0("(", sprintf("%.2f", bleed_ci[2, 2]), ", ", sprintf("%.2f", bleed_ci[2, 3]), ")"),
paste0("(", sprintf("%.2f", stroke_ci[1, 2]), ", ", sprintf("%.2f", stroke_ci[1, 3]), ")"),
paste0("(", sprintf("%.2f", stroke_ci[2, 2]), ", ", sprintf("%.2f", stroke_ci[2, 3]), ")")
),
Chi_Square_P_Value = c(format(bleed_chi_square, digits = 4), paste0(" "), format(stroke_chi_square, digits = 4),paste0(" "))
)
stroke <- t(result_table)
#Drug Treatment Pattern Summary---------------------------
trt <- read.csv('/cloud/project/trt_pattern.csv')
trt$Category_Category <- ifelse(trt$Num_Cat == 0, 0,ifelse(trt$Num_Cat == 1, 1, "2 or More"))
category_table <- table(trt$cohort, trt$Category_Category)
category_percentage <- prop.table(category_table, margin = 1) * 100
total_row1 <- c(sum(category_table[,1]), sum(category_table[,2]),
sum(category_table[,3]))
total_percentage1 <- total_row1/282*100
summary_df1 <- data.frame(
Parameter = "Treaatment Category",
Category = c("0", "1", "2 or More"),
'NOAC(N=194)' = paste0(category_table["NOAC", ], " (", round(category_percentage["NOAC", ], 2), "%)"),
`Warfarin (N = 28)` = paste0(category_table["Warfarin", ], " (", round(category_percentage["Warfarin", ], 2), "%)"),
`Aspirin (N = 60)` = paste0(category_table["Aspirin", ], " (", round(category_percentage["Aspirin", ], 2), "%)"),
"TOTAL(N=282)" = paste0(total_row1," (", round(total_percentage1,2),"%)")
)
colnames(summary_df1)[3:6] <- c("NOAC(N=194)","Warfarin (N = 28)","Aspirin (N = 60)","TOTAL(N=282)")
breaks <- c(-Inf, 2, 3, 4, Inf) # Ranges: (-Inf, 2], (2, 3], (3, 4], (4, Inf)
trt$Prescription_Range <- cut(trt$Num_Presc, breaks = breaks, labels = c("0-2", "3", "4", "5 or More"))
prescription_table <- table(trt$cohort, trt$Prescription_Range)
prescription_percentage <- prop.table(prescription_table, margin = 1) * 100
total_row2 <- c(sum(prescription_table[,1]), sum(prescription_table[,2]),
sum(prescription_table[,3]),sum(prescription_table[,4]))
total_percentage2 <- total_row2/282*100
summary_df2 <- data.frame(
Parameter = "Number of Prescriptions",
Category = c("0-2", "3", "4", "5 or More"),
`NOAC (N = 194)` = paste0(prescription_table["NOAC", ], " (", round(prescription_percentage["NOAC", ], 2), "%)"),
`Warfarin (N = 28)` = paste0(prescription_table["Warfarin", ], " (", round(prescription_percentage["Warfarin", ], 2), "%)"),
`Aspirin (N = 60)` = paste0(prescription_table["Aspirin", ], " (", round(prescription_percentage["Aspirin", ], 2), "%)"),
`TOTAL (N = 282)` = paste0(total_row2," (", round(total_percentage2,2),"%)")
)
colnames(summary_df2)[3:6] <- c("NOAC(N=194)","Warfarin (N = 28)","Aspirin (N = 60)","TOTAL(N=282)")
trt_pattern_summary <- rbind(summary_df1,summary_df2)
#Summary of Change in CV Parameters from Index Date to Last available observations-----
vital <- read.csv('/cloud/project/vital_sign_analysis.csv')
vital$cohort <- ifelse(vital$cohort1n==1,"NOAC","Aspirin+Warfarin")
vital$cv <- ifelse(vital$loinc == "8462-4","Diastolic BP",ifelse(vital$loinc == "8480-6","Systolic BP","Heart Rate"))
parameters <- c("Diastolic BP", "Systolic BP", "Heart Rate")
t_test_pvalue <- list()
for (parameter in parameters) {
parameter_data <- vital %>% filter(cv == parameter)
t_test_pvalue[[parameter]] <- t.test(CHG ~ cohort1n, data = parameter_data, var.equal = TRUE)$p.value
}
p_values <- c(0.3526164, " " ,0.1795525, " " ,0.5331616, " ")
cohort_parameter_change <- vital %>%
group_by(cv, cohort) %>%
summarise(
Mean_Change = mean(CHG),
`95% CI` = sprintf("(%s, %s)", quantile(CHG, 0.025), quantile(CHG, 0.975)),
`Min, Max` = sprintf("(%s, %s)", min(CHG), max(CHG)),
)
## `summarise()` has grouped output by 'cv'. You can override using the `.groups`
## argument.
vs <- rbind(t(cohort_parameter_change),p_values)
#Healthcare Resource Utilization Summary------------
hru <- read_sas('/cloud/project/analysis_ds/hru.sas7bdat')
encounter_table <- table(hru$cohort, hru$encounter_type)
encounter_percentage <- prop.table(encounter_table, margin = 1) * 100
total_row <- c(sum(encounter_table[,1]), sum(encounter_table[,2]),sum(encounter_table[,3]),sum(encounter_table[,4]))
total_percentage <- total_row/282*100
summary_df <- data.frame(
EncounterType = c("acute inpatient", "emergency department", "other", "outpatient"),
`NOAC (N = 194)` = paste0(encounter_table["NOAC", ], " (", round(encounter_percentage["NOAC", ], 2), "%)"),
`Warfarin (N = 28)` = paste0(encounter_table["Warfarin", ], " (", round(encounter_percentage["Warfarin", ], 2), "%)"),
`Aspirin (N = 60)` = paste0(encounter_table["Aspirin", ], " (", round(encounter_percentage["Aspirin", ], 2), "%)"),
`TOTAL (N = 282)` = paste0(total_row," (", round(total_percentage,2),"%)")
)
colnames(summary_df)[2:5] <- c("NOAC(N=194)","Warfarin (N = 28)","Aspirin (N = 60)","TOTAL(N=282)")
hru <- summary_df
os <- data.frame(os)
stroke
## [,1] [,2] [,3]
## Parameter "Bleeding" "Bleeding" "Stroke"
## Cohorts "NOAC" "Aspirin+Warfarin" "NOAC"
## Proportion "84/88 (95.5%)" "193/194 (99.5%)" "84/88 (95.5%)"
## CI "(88.77, 98.75)" "(97.16, 99.99)" "(88.77, 98.75)"
## Chi_Square_P_Value "0.0175" " " "0.0175"
## [,4]
## Parameter "Stroke"
## Cohorts "Aspirin+Warfarin"
## Proportion "193/194 (99.5%)"
## CI "(97.16, 99.99)"
## Chi_Square_P_Value " "
trt_pattern_summary
## Parameter Category NOAC(N=194) Warfarin (N = 28)
## 1 Treaatment Category 0 20 (10.31%) 5 (17.86%)
## 2 Treaatment Category 1 171 (88.14%) 23 (82.14%)
## 3 Treaatment Category 2 or More 3 (1.55%) 0 (0%)
## 4 Number of Prescriptions 0-2 169 (87.11%) 23 (82.14%)
## 5 Number of Prescriptions 3 0 (0%) 0 (0%)
## 6 Number of Prescriptions 4 24 (12.37%) 4 (14.29%)
## 7 Number of Prescriptions 5 or More 1 (0.52%) 1 (3.57%)
## Aspirin (N = 60) TOTAL(N=282)
## 1 0 (0%) 25 (8.87%)
## 2 10 (16.67%) 204 (72.34%)
## 3 50 (83.33%) 53 (18.79%)
## 4 51 (85%) 243 (86.17%)
## 5 0 (0%) 0 (0%)
## 6 6 (10%) 34 (12.06%)
## 7 3 (5%) 5 (1.77%)
vs
## [,1] [,2] [,3]
## cv "Diastolic BP" "Diastolic BP" "Heart Rate"
## cohort "Aspirin+Warfarin" "NOAC" "Aspirin+Warfarin"
## Mean_Change " -0.4886364" " 1.8917526" "-52.1363636"
## 95% CI "(-32.475, 44.475)" "(-36.525, 35.35)" "(-106, -5.175)"
## Min, Max "(-50, 49)" "(-46, 41)" "(-115, 8)"
## p_values "0.3526164" " " "0.1795525"
## [,4] [,5] [,6]
## cv "Heart Rate" "Systolic BP" "Systolic BP"
## cohort "NOAC" "Aspirin+Warfarin" "NOAC"
## Mean_Change "-46.5463918" "-12.3863636" "-11.1391753"
## 95% CI "(-105, 11)" "(-40, 19.825)" "(-41, 17)"
## Min, Max "(-116, 26)" "(-44, 39)" "(-48, 23)"
## p_values " " "0.5331616" " "
hru
## EncounterType NOAC(N=194) Warfarin (N = 28) Aspirin (N = 60)
## 1 acute inpatient 8 (3.67%) 0 (0%) 1 (1.43%)
## 2 emergency department 120 (55.05%) 20 (62.5%) 38 (54.29%)
## 3 other 86 (39.45%) 12 (37.5%) 29 (41.43%)
## 4 outpatient 4 (1.83%) 0 (0%) 2 (2.86%)
## TOTAL(N=282)
## 1 9 (3.19%)
## 2 178 (63.12%)
## 3 127 (45.04%)
## 4 6 (2.13%)
library(DT)
library(shiny)
##
## Attaching package: 'shiny'
## The following objects are masked from 'package:DT':
##
## dataTableOutput, renderDataTable
ui <- fluidPage(
titlePanel("Multiple Tables Display"),
DTOutput('table1'),
DTOutput('table2'),
DTOutput('table3'),
DTOutput('table4'),
DTOutput('table5')
)
server <- function(input, output) {
output$table1 <- renderDT({ os })
output$table2 <- renderDT({ stroke })
output$table3 <- renderDT({ trt_pattern_summary })
output$table4 <- renderDT({ vs })
output$table5 <- renderDT({ hru })
}
shinyApp(ui = ui, server = server)
Shiny applications not supported in static R Markdown documents