In CC-HIC/inspectEHR: Data Quality Reporting for CC-HIC

Events in Detail

This section reviews each event in detail. It is advisable to study each event in detail before embarking on a research endevour. If you find a problem with anything here, please submit an issue request on github.

load("./_bookdown_files/data/working_data.RData")

for(i in seq_along(code_names)) {
  if (dir.exists(
    paste0(root, code_names[i]))) {

    event_files <- list.files(paste0(root, code_names[i]))

    main <- event_files[grepl(pattern = "main", x = event_files)]
    coverage <- event_files[grepl(pattern = "covarage", x = event_files)]
    ks <- event_files[grepl(pattern = "ks", x = event_files)]
    period <- event_files[grepl(pattern = "periodicity", x = event_files)]

    ## Event Title
    cat("## ",
        substr(
          code_names[i], nchar(code_names[i])-3, nchar(code_names[i])),
        ": ",
        names(code_names)[i],
        "  \n", sep = "")

    if (nrow(hic_event_summary[[code_names[i]]]$error_checks) != 0) {

      errors <- hic_event_summary[[code_names[i]]]$error_checks %>%
        group_by(site) %>%
        spread(key = error_type, value = n) %>%
        select(site, duplicate_events, early_event, late_event, low_value, high_value) %>%
        rename(`duplicate events` = duplicate_events,
             `early event` = early_event,
             `late event` = late_event,
             `low value` = low_value,
             `high value` = high_value)

    this_cap <- paste0("Errors for ", names(code_names)[i],
                       ": duplicate events = events that appear to be duplicates based on exact replication of a value and a datetime stamp (to the second), ",
                       "early event = events that occurred before the patient arrived in ICU, ",
                       "late event = events that occurred after the patient left ICU, ",
                       "low value = events below a possible range, ",
                       "high value = events above a possible range.")

    knitr::kable(errors, caption = this_cap)
    }

    if (code_names[i] %in% one_d) {
        knitr::kable(hic_event_summary[[code_names[i]]]$missingness)
    }

    # check to see if there is anything to show here
    if (length(event_files) == 0) {
      cat("There are no plots for the dataitem. This is usually for reasons of confidentiality.  \n")
    } else {

    ## eCDF/Histograms
    if (length(main) != 0) {
    cat("### Distribution  \n")
      for (j in seq_along(main)) {
        cat("![](", root, code_names[i], "/", main[j], ")  \n", sep = "")
        cat("\n")
      }
    rm(j)
    }
    ## Coverage
    if (length(coverage) != 0) {
    cat("### Coverage  \n")
      for (j in seq_along(coverage)) {
        cat("![](", root, code_names[i], "/", coverage[j], ")  \n", sep = "")
        cat("\n")
      }
    rm(j)
    }
    ## Periodicity
    if (length(period) != 0) {
    cat("### Periodicity  \n")
      cat("![](", root, code_names[i], "/", period, ")  \n", sep = "")
      cat("\n")
    }
    ## KS
    if (length(ks) != 0) {
    cat("### KS Distance  \n")
      cat("![](", root, code_names[i], "/", ks, ")  \n", sep = "")
      cat("\n")
    }

    }
  }
}