AMR/R/first_isolate.R

453 lines
19 KiB
R
Raw Normal View History

2018-02-21 11:52:31 +01:00
# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
# AUTHORS #
# Berends MS (m.s.berends@umcg.nl), Luz CF (c.f.luz@umcg.nl) #
# #
# LICENCE #
# This program is free software; you can redistribute it and/or modify #
# it under the terms of the GNU General Public License version 2.0, #
# as published by the Free Software Foundation. #
# #
# This program is distributed in the hope that it will be useful, #
# but WITHOUT ANY WARRANTY; without even the implied warranty of #
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
# GNU General Public License for more details. #
# ==================================================================== #
#' Determine first (weighted) isolates
#'
2018-04-02 11:11:21 +02:00
#' Determine first (weighted) isolates of all microorganisms of every patient per episode and (if needed) per specimen type.
2018-02-21 11:52:31 +01:00
#' @param tbl a \code{data.frame} containing isolates.
2018-10-23 11:15:05 +02:00
#' @param col_date column name of the result date (or date that is was received on the lab), defaults to the first column of class \code{Date}
#' @param col_patient_id column name of the unique IDs of the patients, defaults to the first column that starts with 'patient' (case insensitive)
2018-11-01 20:50:10 +01:00
#' @param col_mo column name of the unique IDs of the microorganisms (see \code{\link{mo}}), defaults to the first column of class \code{mo}. Values will be coerced using \code{\link{as.mo}}.
#' @param col_testcode column name of the test codes. Use \code{col_testcode = NA} to \strong{not} exclude certain test codes (like test codes for screening). In that case \code{testcodes_exclude} will be ignored. Supports tidyverse-like quotation.
2018-04-02 11:11:21 +02:00
#' @param col_specimen column name of the specimen type or group
#' @param col_icu column name of the logicals (\code{TRUE}/\code{FALSE}) whether a ward or department is an Intensive Care Unit (ICU)
#' @param col_keyantibiotics column name of the key antibiotics to determine first \emph{weighted} isolates, see \code{\link{key_antibiotics}}. Supports tidyverse-like quotation.
2018-02-21 11:52:31 +01:00
#' @param episode_days episode in days after which a genus/species combination will be determined as 'first isolate' again
#' @param testcodes_exclude character vector with test codes that should be excluded (case-insensitive)
2018-02-21 11:52:31 +01:00
#' @param icu_exclude logical whether ICU isolates should be excluded
#' @param filter_specimen specimen group or type that should be excluded
#' @param output_logical return output as \code{logical} (will else be the values \code{0} or \code{1})
#' @param type type to determine weighed isolates; can be \code{"keyantibiotics"} or \code{"points"}, see Details
#' @param ignore_I logical to determine whether antibiotic interpretations with \code{"I"} will be ignored when \code{type = "keyantibiotics"}, see Details
#' @param points_threshold points until the comparison of key antibiotics will lead to inclusion of an isolate when \code{type = "points"}, see Details
2018-02-21 11:52:31 +01:00
#' @param info print progress
2018-08-31 13:36:19 +02:00
#' @param col_bactid (deprecated, use \code{col_mo} instead)
#' @param col_genus (deprecated, use \code{col_mo} instead) column name of the genus of the microorganisms
#' @param col_species (deprecated, use \code{col_mo} instead) column name of the species of the microorganisms
#' @details \strong{WHY THIS IS SO IMPORTANT} \cr
2018-02-27 20:01:02 +01:00
#' To conduct an analysis of antimicrobial resistance, you should only include the first isolate of every patient per episode \href{https://www.ncbi.nlm.nih.gov/pubmed/17304462}{[1]}. If you would not do this, you could easily get an overestimate or underestimate of the resistance of an antibiotic. Imagine that a patient was admitted with an MRSA and that it was found in 5 different blood cultures the following week. The resistance percentage of oxacillin of all \emph{S. aureus} isolates would be overestimated, because you included this MRSA more than once. It would be \href{https://en.wikipedia.org/wiki/Selection_bias}{selection bias}.
2018-07-17 13:02:05 +02:00
#' @section Key antibiotics:
#' There are two ways to determine whether isolates can be included as first \emph{weighted} isolates which will give generally the same results: \cr
2018-03-13 11:57:30 +01:00
#'
2018-03-19 21:23:21 +01:00
#' \strong{1. Using} \code{type = "keyantibiotics"} \strong{and parameter} \code{ignore_I} \cr
2018-07-17 13:02:05 +02:00
#' Any difference from S to R (or vice versa) will (re)select an isolate as a first weighted isolate. With \code{ignore_I = FALSE}, also differences from I to S|R (or vice versa) will lead to this. This is a reliable method and 30-35 times faster than method 2. \cr
#'
2018-03-19 21:23:21 +01:00
#' \strong{2. Using} \code{type = "points"} \strong{and parameter} \code{points_threshold} \cr
2018-07-17 13:02:05 +02:00
#' A difference from I to S|R (or vice versa) means 0.5 points, a difference from S to R (or vice versa) means 1 point. When the sum of points exceeds \code{points_threshold}, an isolate will be (re)selected as a first weighted isolate.
2018-02-21 11:52:31 +01:00
#' @keywords isolate isolates first
2018-07-17 14:48:11 +02:00
#' @seealso \code{\link{key_antibiotics}}
2018-02-26 12:15:52 +01:00
#' @export
2018-02-21 11:52:31 +01:00
#' @importFrom dplyr arrange_at lag between row_number filter mutate arrange
#' @return A vector to add to table, see Examples.
2018-07-29 22:14:51 +02:00
#' @source Methodology of this function is based on: \strong{M39 Analysis and Presentation of Cumulative Antimicrobial Susceptibility Test Data, 4th Edition}, 2014, \emph{Clinical and Laboratory Standards Institute (CLSI)}. \url{https://clsi.org/standards/products/microbiology/documents/m39/}.
2018-02-21 11:52:31 +01:00
#' @examples
#' # septic_patients is a dataset available in the AMR package. It is true, genuine data.
#' ?septic_patients
2018-04-02 11:11:21 +02:00
#'
#' library(dplyr)
2018-08-10 15:01:05 +02:00
#' my_patients <- septic_patients %>%
#' mutate(first_isolate = first_isolate(.,
#' col_date = "date",
#' col_patient_id = "patient_id",
2018-08-31 13:36:19 +02:00
#' col_mo = "mo"))
2018-04-02 11:11:21 +02:00
#'
#' # Now let's see if first isolates matter:
#' A <- my_patients %>%
#' group_by(hospital_id) %>%
2018-08-10 15:01:05 +02:00
#' summarise(count = n_rsi(gent), # gentamicin availability
#' resistance = portion_IR(gent)) # gentamicin resistance
#'
#' B <- my_patients %>%
2018-08-10 15:01:05 +02:00
#' filter(first_isolate == TRUE) %>% # the 1st isolate filter
#' group_by(hospital_id) %>%
2018-08-10 15:01:05 +02:00
#' summarise(count = n_rsi(gent), # gentamicin availability
#' resistance = portion_IR(gent)) # gentamicin resistance
#'
2018-08-10 15:01:05 +02:00
#' # Have a look at A and B.
#' # B is more reliable because every isolate is only counted once.
#' # Gentamicin resitance in hospital D appears to be 5.4% higher than
#' # when you (erroneously) would have used all isolates!
#'
#' ## OTHER EXAMPLES:
#'
2018-02-21 11:52:31 +01:00
#' \dontrun{
#'
2018-02-22 20:48:48 +01:00
#' # set key antibiotics to a new variable
2018-02-21 11:52:31 +01:00
#' tbl$keyab <- key_antibiotics(tbl)
#'
#' tbl$first_isolate <-
#' first_isolate(tbl)
#'
#' tbl$first_isolate_weighed <-
#' first_isolate(tbl,
#' col_keyantibiotics = 'keyab')
#'
#' tbl$first_blood_isolate <-
#' first_isolate(tbl,
#' filter_specimen = 'Blood')
#'
#' tbl$first_blood_isolate_weighed <-
#' first_isolate(tbl,
#' filter_specimen = 'Blood',
#' col_keyantibiotics = 'keyab')
#'
#' tbl$first_urine_isolate <-
#' first_isolate(tbl,
#' filter_specimen = 'Urine')
#'
#' tbl$first_urine_isolate_weighed <-
#' first_isolate(tbl,
#' filter_specimen = 'Urine',
#' col_keyantibiotics = 'keyab')
#'
#' tbl$first_resp_isolate <-
#' first_isolate(tbl,
#' filter_specimen = 'Respiratory')
#'
#' tbl$first_resp_isolate_weighed <-
#' first_isolate(tbl,
#' filter_specimen = 'Respiratory',
#' col_keyantibiotics = 'keyab')
#' }
first_isolate <- function(tbl,
2018-10-23 11:15:05 +02:00
col_date = NULL,
col_patient_id = NULL,
col_mo = NULL,
col_testcode = NULL,
col_specimen = NULL,
col_icu = NULL,
col_keyantibiotics = NULL,
2018-02-21 11:52:31 +01:00
episode_days = 365,
2018-10-23 11:15:05 +02:00
testcodes_exclude = NULL,
2018-02-21 11:52:31 +01:00
icu_exclude = FALSE,
2018-10-23 11:15:05 +02:00
filter_specimen = NULL,
2018-02-21 11:52:31 +01:00
output_logical = TRUE,
type = "keyantibiotics",
ignore_I = TRUE,
2018-02-27 20:01:02 +01:00
points_threshold = 2,
info = TRUE,
2018-10-23 11:15:05 +02:00
col_bactid = NULL,
col_genus = NULL,
col_species = NULL) {
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (!is.data.frame(tbl)) {
stop("`tbl` must be a data frame.", call. = FALSE)
}
# try to find columns based on type
# -- mo
if (!is.null(col_bactid)) {
2018-08-31 13:36:19 +02:00
col_mo <- col_bactid
warning("Use of `col_bactid` is deprecated. Use `col_mo` instead.")
2018-10-23 11:15:05 +02:00
} else if (is.null(col_mo) & "mo" %in% lapply(tbl, class)) {
2018-11-05 15:36:07 +01:00
col_mo <- colnames(tbl)[lapply(tbl, class) == "mo"][1]
2018-10-23 11:15:05 +02:00
message("NOTE: Using column `", col_mo, "` as input for `col_mo`.")
}
# -- date
if (is.null(col_date) & "Date" %in% lapply(tbl, class)) {
2018-11-05 15:36:07 +01:00
col_date <- colnames(tbl)[lapply(tbl, class) == "Date"][1]
2018-10-23 11:15:05 +02:00
message("NOTE: Using column `", col_date, "` as input for `col_date`.")
2018-08-31 13:36:19 +02:00
}
2018-10-23 11:15:05 +02:00
# -- patient id
if (is.null(col_patient_id) & any(colnames(tbl) %like% "^patient")) {
col_patient_id <- colnames(tbl)[colnames(tbl) %like% "^patient"][1]
message("NOTE: Using column `", col_patient_id, "` as input for `col_patient_id`.")
}
2018-04-02 11:11:21 +02:00
# bactid OR genus+species must be available
2018-10-23 11:15:05 +02:00
if (is.null(col_mo) & (is.null(col_genus) | is.null(col_species))) {
2018-08-31 13:36:19 +02:00
stop('`col_mo` or both `col_genus` and `col_species` must be available.')
2018-04-02 11:11:21 +02:00
}
2018-10-23 11:15:05 +02:00
# check if columns exist
2018-02-21 11:52:31 +01:00
check_columns_existance <- function(column, tblname = tbl) {
if (NROW(tblname) <= 1 | NCOL(tblname) <= 1) {
stop('Please check tbl for existance.')
}
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (!is.null(column)) {
2018-02-21 11:52:31 +01:00
if (!(column %in% colnames(tblname))) {
2018-04-02 11:11:21 +02:00
stop('Column `', column, '` not found.')
2018-02-21 11:52:31 +01:00
}
}
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
check_columns_existance(col_date)
2018-02-26 14:06:31 +01:00
check_columns_existance(col_patient_id)
2018-08-31 13:36:19 +02:00
check_columns_existance(col_mo)
2018-02-21 11:52:31 +01:00
check_columns_existance(col_genus)
check_columns_existance(col_species)
check_columns_existance(col_testcode)
check_columns_existance(col_icu)
check_columns_existance(col_keyantibiotics)
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (!is.null(col_mo)) {
2018-07-26 16:30:42 +02:00
# join to microorganisms data set
2018-11-01 20:23:33 +01:00
tbl <- tbl %>%
mutate_at(vars(col_mo), as.mo) %>%
left_join_microorganisms(by = col_mo)
col_genus <- "genus"
col_species <- "species"
}
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (is.null(col_testcode)) {
testcodes_exclude <- NULL
2018-02-21 11:52:31 +01:00
}
# remove testcodes
2018-10-23 11:15:05 +02:00
if (!is.null(testcodes_exclude) & info == TRUE) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Excluded test codes:\n', toString(testcodes_exclude), '\n')
2018-02-21 11:52:31 +01:00
}
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (is.null(col_icu)) {
2018-02-21 11:52:31 +01:00
icu_exclude <- FALSE
} else {
tbl <- tbl %>%
mutate(col_icu = tbl %>% pull(col_icu) %>% as.logical())
}
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (is.null(col_specimen)) {
filter_specimen <- NULL
}
2018-04-02 11:11:21 +02:00
# filter on specimen group and keyantibiotics when they are filled in
2018-10-23 11:15:05 +02:00
if (!is.null(filter_specimen)) {
2018-02-21 11:52:31 +01:00
check_columns_existance(col_specimen, tbl)
if (info == TRUE) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Excluded other than specimen group \'', filter_specimen, '\'\n', sep = '')
2018-02-21 11:52:31 +01:00
}
}
2018-10-23 11:15:05 +02:00
if (!is.null(col_keyantibiotics)) {
2018-02-21 11:52:31 +01:00
tbl <- tbl %>% mutate(key_ab = tbl %>% pull(col_keyantibiotics))
}
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (is.null(testcodes_exclude)) {
2018-02-21 11:52:31 +01:00
testcodes_exclude <- ''
}
2018-04-02 11:11:21 +02:00
# create new dataframe with original row index and right sorting
2018-02-21 11:52:31 +01:00
tbl <- tbl %>%
mutate(first_isolate_row_index = 1:nrow(tbl),
date_lab = tbl %>% pull(col_date),
2018-02-26 14:06:31 +01:00
patient_id = tbl %>% pull(col_patient_id),
species = tbl %>% pull(col_species),
genus = tbl %>% pull(col_genus)) %>%
2018-06-19 10:05:38 +02:00
mutate(species = if_else(is.na(species) | species == "(no MO)", "", species),
genus = if_else(is.na(genus) | genus == "(no MO)", "", genus))
2018-04-02 11:11:21 +02:00
2018-10-23 11:15:05 +02:00
if (is.null(filter_specimen)) {
# not filtering on specimen
2018-02-21 11:52:31 +01:00
if (icu_exclude == FALSE) {
2018-10-23 11:15:05 +02:00
if (info == TRUE & !is.null(col_icu)) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Included isolates from ICU.\n')
2018-02-21 11:52:31 +01:00
}
tbl <- tbl %>%
2018-02-26 14:06:31 +01:00
arrange_at(c(col_patient_id,
2018-02-21 11:52:31 +01:00
col_genus,
col_species,
col_date))
row.start <- 1
row.end <- nrow(tbl)
} else {
if (info == TRUE) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Excluded isolates from ICU.\n')
2018-02-21 11:52:31 +01:00
}
tbl <- tbl %>%
arrange_at(c(col_icu,
2018-02-26 14:06:31 +01:00
col_patient_id,
2018-02-21 11:52:31 +01:00
col_genus,
col_species,
col_date))
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
suppressWarnings(
row.start <- which(tbl %>% pull(col_icu) == FALSE) %>% min(na.rm = TRUE)
)
suppressWarnings(
row.end <- which(tbl %>% pull(col_icu) == FALSE) %>% max(na.rm = TRUE)
)
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
} else {
2018-10-23 11:15:05 +02:00
# filtering on specimen and only analyse these row to save time
2018-02-21 11:52:31 +01:00
if (icu_exclude == FALSE) {
2018-10-23 11:15:05 +02:00
if (info == TRUE & !is.null(col_icu)) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Included isolates from ICU.\n')
2018-02-21 11:52:31 +01:00
}
tbl <- tbl %>%
arrange_at(c(col_specimen,
2018-02-26 14:06:31 +01:00
col_patient_id,
2018-02-21 11:52:31 +01:00
col_genus,
col_species,
col_date))
suppressWarnings(
row.start <- which(tbl %>% pull(col_specimen) == filter_specimen) %>% min(na.rm = TRUE)
)
suppressWarnings(
row.end <- which(tbl %>% pull(col_specimen) == filter_specimen) %>% max(na.rm = TRUE)
)
} else {
if (info == TRUE) {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Excluded isolates from ICU.\n')
2018-02-21 11:52:31 +01:00
}
tbl <- tbl %>%
arrange_at(c(col_icu,
col_specimen,
2018-02-26 14:06:31 +01:00
col_patient_id,
2018-02-21 11:52:31 +01:00
col_genus,
col_species,
col_date))
suppressWarnings(
row.start <- which(tbl %>% pull(col_specimen) == filter_specimen
& tbl %>% pull(col_icu) == FALSE) %>% min(na.rm = TRUE)
)
suppressWarnings(
row.end <- which(tbl %>% pull(col_specimen) == filter_specimen
& tbl %>% pull(col_icu) == FALSE) %>% max(na.rm = TRUE)
)
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
if (abs(row.start) == Inf | abs(row.end) == Inf) {
if (info == TRUE) {
2018-07-19 15:11:23 +02:00
message('No isolates found.')
2018-02-21 11:52:31 +01:00
}
2018-09-01 21:19:46 +02:00
# NAs where genus is unavailable
2018-02-21 11:52:31 +01:00
tbl <- tbl %>%
mutate(real_first_isolate = if_else(genus == '', NA, FALSE))
if (output_logical == FALSE) {
tbl$real_first_isolate <- tbl %>% pull(real_first_isolate) %>% as.integer()
}
return(tbl %>% pull(real_first_isolate))
}
2018-04-02 11:11:21 +02:00
# suppress warnings because dplyr want us to use library(dplyr) when using filter(row_number())
suppressWarnings(
scope.size <- tbl %>%
filter(
2018-05-31 14:19:25 +02:00
row_number() %>% between(row.start,
row.end),
genus != "",
species != "") %>%
nrow()
)
2018-05-31 14:19:25 +02:00
# Analysis of first isolate ----
2018-02-21 11:52:31 +01:00
all_first <- tbl %>%
mutate(other_pat_or_mo = if_else(patient_id == lag(patient_id)
& genus == lag(genus)
& species == lag(species),
FALSE,
TRUE),
days_diff = 0) %>%
mutate(days_diff = if_else(other_pat_or_mo == FALSE,
(date_lab - lag(date_lab)) + lag(days_diff),
0))
2018-04-02 11:11:21 +02:00
weighted.notice <- ''
2018-10-23 11:15:05 +02:00
if (!is.null(col_keyantibiotics)) {
weighted.notice <- 'weighted '
2018-02-21 11:52:31 +01:00
if (info == TRUE) {
if (type == 'keyantibiotics') {
2018-07-19 15:11:23 +02:00
cat('[Criteria] Inclusion based on key antibiotics, ')
if (ignore_I == FALSE) {
2018-07-19 15:11:23 +02:00
cat('not ')
}
2018-07-19 15:11:23 +02:00
cat('ignoring I.\n')
}
if (type == 'points') {
2018-07-19 15:11:23 +02:00
cat(paste0('[Criteria] Inclusion based on key antibiotics, using points threshold of '
, points_threshold, '.\n'))
}
2018-02-21 11:52:31 +01:00
}
2018-03-19 21:03:23 +01:00
type_param <- type
# suppress warnings because dplyr want us to use library(dplyr) when using filter(row_number())
suppressWarnings(
all_first <- all_first %>%
mutate(key_ab_lag = lag(key_ab)) %>%
mutate(key_ab_other = !key_antibiotics_equal(x = key_ab,
y = key_ab_lag,
type = type_param,
ignore_I = ignore_I,
points_threshold = points_threshold,
info = info)) %>%
mutate(
real_first_isolate =
if_else(
between(row_number(), row.start, row.end)
& genus != ""
& species != ""
& (other_pat_or_mo
| days_diff >= episode_days
| key_ab_other),
TRUE,
FALSE))
)
2018-02-21 11:52:31 +01:00
} else {
# suppress warnings because dplyr want us to use library(dplyr) when using filter(row_number())
suppressWarnings(
all_first <- all_first %>%
mutate(
real_first_isolate =
if_else(
between(row_number(), row.start, row.end)
& genus != ""
& species != ""
& (other_pat_or_mo
| days_diff >= episode_days),
TRUE,
FALSE))
)
2018-02-21 11:52:31 +01:00
}
2018-04-02 11:11:21 +02:00
# first one as TRUE
2018-02-21 11:52:31 +01:00
all_first[row.start, 'real_first_isolate'] <- TRUE
# no tests that should be included, or ICU
2018-10-23 11:15:05 +02:00
if (!is.null(col_testcode)) {
2018-02-21 11:52:31 +01:00
all_first[which(all_first[, col_testcode] %in% tolower(testcodes_exclude)), 'real_first_isolate'] <- FALSE
}
if (icu_exclude == TRUE) {
all_first[which(all_first[, col_icu] == TRUE), 'real_first_isolate'] <- FALSE
}
2018-04-02 11:11:21 +02:00
2018-09-01 21:19:46 +02:00
# NAs where genus is unavailable
2018-02-21 11:52:31 +01:00
all_first <- all_first %>%
2018-06-19 10:05:38 +02:00
mutate(real_first_isolate = if_else(genus %in% c('', '(no MO)', NA), NA, real_first_isolate))
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
all_first <- all_first %>%
arrange(first_isolate_row_index) %>%
pull(real_first_isolate)
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
if (info == TRUE) {
2018-07-19 15:11:23 +02:00
message(paste0('Found ',
2018-02-21 11:52:31 +01:00
all_first %>% sum(na.rm = TRUE),
' first ', weighted.notice, 'isolates (',
(all_first %>% sum(na.rm = TRUE) / scope.size) %>% percent(),
' of isolates in scope [where genus was not empty] and ',
(all_first %>% sum(na.rm = TRUE) / tbl %>% nrow()) %>% percent(),
2018-07-19 15:11:23 +02:00
' of total)'))
2018-02-21 11:52:31 +01:00
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
if (output_logical == FALSE) {
all_first <- all_first %>% as.integer()
}
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
all_first
2018-04-02 11:11:21 +02:00
2018-02-21 11:52:31 +01:00
}