In openpharma/simaerep: Find Clinical Trial Sites Under-Reporting Adverse Events

knitr::opts_chunk$set(echo = TRUE, cache = FALSE)

Load

suppressPackageStartupMessages(library(tidyverse))
suppressPackageStartupMessages(library(knitr))
suppressPackageStartupMessages(library(simaerep))

Introduction

Previously we used an unadjusted visit_med75. Simply taking the median of all maximum patient visits at a site and multiplying by 0.75. If the number of patients at a site is reasonably large the smallest maximum visit of the patients making the cut-off will be close or equal to the cut-off point.

set.seed(1)

df_visit <- sim_test_data_study(
  n_pat = 120,
  n_sites = 6,
  frac_site_with_ur = 0.4,
  ur_rate = 0.6)

df_visit$study_id <- "A"
df_site <- site_aggr(df_visit, method = "med75_adj")

plot_visit_med75(df_visit, df_site, study_id_str = "A", n_site = 6)

Only few Patients per Site

If we only have a few patients per site chances will be greater that the smallest maximum visit of those patients that make the cut-off is much greater than the cut-off point and we could thus include more visits into our analysis. For those cases we adjust the evaluation point to the smallest maximum visit of all patients that make the initial cut.

set.seed(1)

df_visit <- sim_test_data_study(
  n_pat = 24,
  n_sites = 6,
  frac_site_with_ur = 0.4,
  ur_rate = 0.6)

df_visit$study_id <- "A"
df_site <- site_aggr(df_visit, method = "med75_adj")

plot_visit_med75(df_visit, df_site, study_id_str = "A", n_site = 6)

Early Starters

It can happen that we have a site that has early starting patients. To avoid an evaluation point for which we will only have a limited patient pool we set a maximum limit for the evaluation point that includes at least 20% of all patients in the study, by taking the 80% quantile of all maximum patient visits at a study.

set.seed(1)

# early starting site
df_visit1 <- sim_test_data_study(
  n_pat = 3,
  n_sites = 1,
  max_visit_sd = 2,
  max_visit_mean = 40
)

# late starting sites
df_visit2 <- sim_test_data_study(
  n_pat = 25,
  n_sites = 5,
  max_visit_sd = 2,
  max_visit_mean = 10
)

df_visit1$site_number <- paste0("A", df_visit1$site_number)
df_visit2$site_number <- paste0("B", df_visit2$site_number)
df_visit1$patnum <- paste0("A", df_visit1$patnum)
df_visit2$patnum <- paste0("B", df_visit2$patnum)


df_visit <- bind_rows(df_visit1, df_visit2)

df_visit$study_id <- "A"

df_site <- site_aggr(df_visit)

plot_visit_med75(df_visit, df_site, study_id_str = "A", n_site = 6)