AMR/R/portion.R

334 lines
15 KiB
R
Raw Normal View History

2018-08-10 15:01:05 +02:00
# ==================================================================== #
# TITLE #
# Antimicrobial Resistance (AMR) Analysis #
# #
2019-01-02 23:24:07 +01:00
# SOURCE #
# https://gitlab.com/msberends/AMR #
2018-08-10 15:01:05 +02:00
# #
# LICENCE #
2019-01-02 23:24:07 +01:00
# (c) 2019 Berends MS (m.s.berends@umcg.nl), Luz CF (c.f.luz@umcg.nl) #
2018-08-10 15:01:05 +02:00
# #
2019-01-02 23:24:07 +01:00
# This R package is free software; you can freely use and distribute #
# it for both personal and commercial purposes under the terms of the #
# GNU General Public License version 2.0 (GNU GPL-2), as published by #
# the Free Software Foundation. #
# #
# This R package was created for academic research and was publicly #
# released in the hope that it will be useful, but it comes WITHOUT #
# ANY WARRANTY OR LIABILITY. #
# Visit our website for more info: https://msberends.gitab.io/AMR. #
2018-08-10 15:01:05 +02:00
# ==================================================================== #
#' Calculate resistance of isolates
#'
2018-08-23 00:40:36 +02:00
#' @description These functions can be used to calculate the (co-)resistance of microbial isolates (i.e. percentage S, SI, I, IR or R). All functions support quasiquotation with pipes, can be used in \code{dplyr}s \code{\link[dplyr]{summarise}} and support grouped variables, see \emph{Examples}.
2018-08-10 15:01:05 +02:00
#'
#' \code{portion_R} and \code{portion_IR} can be used to calculate resistance, \code{portion_S} and \code{portion_SI} can be used to calculate susceptibility.\cr
2018-08-23 00:40:36 +02:00
#' @param ... one or more vectors (or columns) with antibiotic interpretations. They will be transformed internally with \code{\link{as.rsi}} if needed. Use multiple columns to calculate (the lack of) co-resistance: the probability where one of two drugs have a resistant or susceptible result. See Examples.
2018-10-16 09:59:31 +02:00
#' @param minimum the minimal amount of available isolates. Any number lower than \code{minimum} will return \code{NA} with a warning. The default number of \code{30} isolates is advised by the Clinical and Laboratory Standards Institute (CLSI) as best practice, see Source.
#' @param as_percent a logical to indicate whether the output must be returned as a hundred fold with \% sign (a character). A value of \code{0.123456} will then be returned as \code{"12.3\%"}.
#' @param also_single_tested a logical to indicate whether (in combination therapies) also observations should be included where not all antibiotics were tested, but at least one of the tested antibiotics contains a target interpretation (e.g. S in case of \code{portion_S} and R in case of \code{portion_R}). \strong{This would lead to selection bias in almost all cases.}
2018-08-22 00:02:26 +02:00
#' @param data a \code{data.frame} containing columns with class \code{rsi} (see \code{\link{as.rsi}})
2018-08-13 16:42:37 +02:00
#' @param translate_ab a column name of the \code{\link{antibiotics}} data set to translate the antibiotic abbreviations to, using \code{\link{abname}}. This can be set with \code{\link{getOption}("get_antibiotic_names")}.
2018-10-16 09:59:31 +02:00
#' @param combine_IR a logical to indicate whether all values of I and R must be merged into one, so the output only consists of S vs. IR (susceptible vs. non-susceptible)
2018-08-10 15:01:05 +02:00
#' @details \strong{Remember that you should filter your table to let it contain only first isolates!} Use \code{\link{first_isolate}} to determine them in your data set.
#'
2018-12-14 07:23:25 +01:00
#' These functions are not meant to count isolates, but to calculate the portion of resistance/susceptibility. Use the \code{\link[AMR]{count}} functions to count isolates. \emph{Low counts can infuence the outcome - these \code{portion} functions may camouflage this, since they only return the portion albeit being dependent on the \code{minimum} parameter.}
2018-08-22 00:02:26 +02:00
#'
#' \code{portion_df} takes any variable from \code{data} that has an \code{"rsi"} class (created with \code{\link{as.rsi}}) and calculates the portions R, I and S. The resulting \emph{tidy data} (see Source) \code{data.frame} will have three rows (S/I/R) and a column for each variable with class \code{"rsi"}.
2018-08-11 21:30:00 +02:00
#'
2018-08-10 15:01:05 +02:00
#' The old \code{\link{rsi}} function is still available for backwards compatibility but is deprecated.
#' \if{html}{
#' \cr\cr
#' To calculate the probability (\emph{p}) of susceptibility of one antibiotic, we use this formula:
#' \out{<div style="text-align: center">}\figure{mono_therapy.png}\out{</div>}
#' To calculate the probability (\emph{p}) of susceptibility of more antibiotics (i.e. combination therapy), we need to check whether one of them has a susceptible result (as numerator) and count all cases where all antibiotics were tested (as denominator). \cr
#' \cr
#' For two antibiotics:
#' \out{<div style="text-align: center">}\figure{combi_therapy_2.png}\out{</div>}
#' \cr
2018-08-23 00:40:36 +02:00
#' For three antibiotics:
2018-08-10 15:01:05 +02:00
#' \out{<div style="text-align: center">}\figure{combi_therapy_3.png}\out{</div>}
2018-08-23 00:40:36 +02:00
#' \cr
#' And so on.
2018-08-10 15:01:05 +02:00
#' }
#' @source \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/}.
#'
#' Wickham H. \strong{Tidy Data.} The Journal of Statistical Software, vol. 59, 2014. \url{http://vita.had.co.nz/papers/tidy-data.html}
2018-12-14 07:48:12 +01:00
#' @seealso \code{\link[AMR]{count}_*} to count resistant and susceptible isolates.
2018-08-10 15:01:05 +02:00
#' @keywords resistance susceptibility rsi_df rsi antibiotics isolate isolates
#' @return Double or, when \code{as_percent = TRUE}, a character.
#' @rdname portion
#' @name portion
#' @export
2019-01-02 23:24:07 +01:00
#' @inheritSection AMR Read more on our website!
2018-08-10 15:01:05 +02:00
#' @examples
#' # septic_patients is a data set available in the AMR package. It is true, genuine data.
#' ?septic_patients
#'
2018-08-10 15:01:05 +02:00
#' # Calculate resistance
#' portion_R(septic_patients$amox)
#' portion_IR(septic_patients$amox)
#'
#' # Or susceptibility
#' portion_S(septic_patients$amox)
#' portion_SI(septic_patients$amox)
#'
2018-08-23 00:40:36 +02:00
#' # Do the above with pipes:
2018-08-10 15:01:05 +02:00
#' library(dplyr)
2018-08-23 00:40:36 +02:00
#' septic_patients %>% portion_R(amox)
#' septic_patients %>% portion_IR(amox)
#' septic_patients %>% portion_S(amox)
#' septic_patients %>% portion_SI(amox)
#'
2018-08-10 15:01:05 +02:00
#' septic_patients %>%
#' group_by(hospital_id) %>%
#' summarise(p = portion_S(cipr),
#' n = n_rsi(cipr)) # n_rsi works like n_distinct in dplyr
#'
#' septic_patients %>%
#' group_by(hospital_id) %>%
#' summarise(R = portion_R(cipr, as_percent = TRUE),
#' I = portion_I(cipr, as_percent = TRUE),
#' S = portion_S(cipr, as_percent = TRUE),
#' n = n_rsi(cipr), # works like n_distinct in dplyr
#' total = n()) # NOT the amount of tested isolates!
#'
#' # Calculate co-resistance between amoxicillin/clav acid and gentamicin,
#' # so we can see that combination therapy does a lot more than mono therapy:
2018-10-12 16:35:18 +02:00
#' septic_patients %>% portion_S(amcl) # S = 67.1%
#' septic_patients %>% count_all(amcl) # n = 1576
2018-08-10 15:01:05 +02:00
#'
2018-08-23 00:40:36 +02:00
#' septic_patients %>% portion_S(gent) # S = 74.0%
2018-10-12 16:35:18 +02:00
#' septic_patients %>% count_all(gent) # n = 1855
2018-08-23 00:40:36 +02:00
#'
2018-10-12 16:35:18 +02:00
#' septic_patients %>% portion_S(amcl, gent) # S = 92.0%
#' septic_patients %>% count_all(amcl, gent) # n = 1517
2018-08-10 15:01:05 +02:00
#'
#'
#' septic_patients %>%
#' group_by(hospital_id) %>%
#' summarise(cipro_p = portion_S(cipr, as_percent = TRUE),
2018-10-12 16:35:18 +02:00
#' cipro_n = count_all(cipr),
2018-08-10 15:01:05 +02:00
#' genta_p = portion_S(gent, as_percent = TRUE),
2018-10-12 16:35:18 +02:00
#' genta_n = count_all(gent),
2018-08-10 15:01:05 +02:00
#' combination_p = portion_S(cipr, gent, as_percent = TRUE),
2018-10-12 16:35:18 +02:00
#' combination_n = count_all(cipr, gent))
2018-08-10 15:01:05 +02:00
#'
#' # Get portions S/I/R immediately of all rsi columns
#' septic_patients %>%
#' select(amox, cipr) %>%
#' portion_df(translate = FALSE)
#'
#' # It also supports grouping variables
#' septic_patients %>%
#' select(hospital_id, amox, cipr) %>%
#' group_by(hospital_id) %>%
#' portion_df(translate = FALSE)
#'
#'
2018-08-10 15:01:05 +02:00
#' \dontrun{
#'
#' # calculate current empiric combination therapy of Helicobacter gastritis:
#' my_table %>%
#' filter(first_isolate == TRUE,
#' genus == "Helicobacter") %>%
#' summarise(p = portion_S(amox, metr), # amoxicillin with metronidazole
2018-10-12 16:35:18 +02:00
#' n = count_all(amox, metr))
2018-08-10 15:01:05 +02:00
#' }
2018-08-23 00:40:36 +02:00
portion_R <- function(...,
2018-08-10 15:01:05 +02:00
minimum = 30,
as_percent = FALSE,
also_single_tested = FALSE) {
2018-08-23 00:40:36 +02:00
rsi_calc(...,
type = "R",
2018-08-10 15:01:05 +02:00
include_I = FALSE,
minimum = minimum,
2018-08-22 00:02:26 +02:00
as_percent = as_percent,
also_single_tested = also_single_tested,
2018-08-22 00:02:26 +02:00
only_count = FALSE)
2018-08-10 15:01:05 +02:00
}
#' @rdname portion
#' @export
2018-08-23 00:40:36 +02:00
portion_IR <- function(...,
2018-08-10 15:01:05 +02:00
minimum = 30,
as_percent = FALSE,
also_single_tested = FALSE) {
2018-08-23 00:40:36 +02:00
rsi_calc(...,
type = "R",
2018-08-10 15:01:05 +02:00
include_I = TRUE,
minimum = minimum,
2018-08-22 00:02:26 +02:00
as_percent = as_percent,
also_single_tested = also_single_tested,
2018-08-22 00:02:26 +02:00
only_count = FALSE)
2018-08-10 15:01:05 +02:00
}
#' @rdname portion
#' @export
2018-08-23 00:40:36 +02:00
portion_I <- function(...,
2018-08-10 15:01:05 +02:00
minimum = 30,
as_percent = FALSE,
also_single_tested = FALSE) {
2018-08-23 00:40:36 +02:00
rsi_calc(...,
type = "I",
2018-08-10 15:01:05 +02:00
include_I = FALSE,
minimum = minimum,
2018-08-22 00:02:26 +02:00
as_percent = as_percent,
also_single_tested = also_single_tested,
2018-08-22 00:02:26 +02:00
only_count = FALSE)
2018-08-10 15:01:05 +02:00
}
#' @rdname portion
#' @export
2018-08-23 00:40:36 +02:00
portion_SI <- function(...,
2018-08-10 15:01:05 +02:00
minimum = 30,
as_percent = FALSE,
also_single_tested = FALSE) {
2018-08-23 00:40:36 +02:00
rsi_calc(...,
type = "S",
2018-08-10 15:01:05 +02:00
include_I = TRUE,
minimum = minimum,
2018-08-22 00:02:26 +02:00
as_percent = as_percent,
also_single_tested = also_single_tested,
2018-08-22 00:02:26 +02:00
only_count = FALSE)
2018-08-10 15:01:05 +02:00
}
#' @rdname portion
#' @export
2018-08-23 00:40:36 +02:00
portion_S <- function(...,
2018-08-10 15:01:05 +02:00
minimum = 30,
as_percent = FALSE,
also_single_tested = FALSE) {
2018-08-23 00:40:36 +02:00
rsi_calc(...,
type = "S",
2018-08-10 15:01:05 +02:00
include_I = FALSE,
minimum = minimum,
2018-08-22 00:02:26 +02:00
as_percent = as_percent,
also_single_tested = also_single_tested,
2018-08-22 00:02:26 +02:00
only_count = FALSE)
2018-08-10 15:01:05 +02:00
}
2018-08-12 17:44:06 +02:00
#' @rdname portion
2018-08-22 00:02:26 +02:00
#' @importFrom dplyr %>% select_if bind_rows summarise_if mutate group_vars select everything
2018-08-12 17:44:06 +02:00
#' @export
2018-08-22 00:02:26 +02:00
portion_df <- function(data,
translate_ab = getOption("get_antibiotic_names", "official"),
minimum = 30,
2018-10-16 09:59:31 +02:00
as_percent = FALSE,
combine_IR = FALSE) {
2018-08-22 00:02:26 +02:00
2018-09-16 22:11:17 +02:00
if (!"data.frame" %in% class(data)) {
stop("`portion_df` must be called on a data.frame")
}
2018-08-22 00:02:26 +02:00
if (data %>% select_if(is.rsi) %>% ncol() == 0) {
stop("No columns with class 'rsi' found. See ?as.rsi.")
}
2018-08-13 16:42:37 +02:00
if (as.character(translate_ab) == "TRUE") {
translate_ab <- "official"
}
options(get_antibiotic_names = translate_ab)
resS <- summarise_if(.tbl = data,
.predicate = is.rsi,
2018-08-22 00:02:26 +02:00
.funs = portion_S,
minimum = minimum,
as_percent = as_percent) %>%
mutate(Interpretation = "S") %>%
select(Interpretation, everything())
2018-10-16 09:59:31 +02:00
if (combine_IR == FALSE) {
resI <- summarise_if(.tbl = data,
.predicate = is.rsi,
.funs = portion_I,
minimum = minimum,
as_percent = as_percent) %>%
mutate(Interpretation = "I") %>%
select(Interpretation, everything())
2018-10-16 09:59:31 +02:00
resR <- summarise_if(.tbl = data,
.predicate = is.rsi,
.funs = portion_R,
minimum = minimum,
as_percent = as_percent) %>%
mutate(Interpretation = "R") %>%
select(Interpretation, everything())
2018-10-16 09:59:31 +02:00
data.groups <- group_vars(data)
2018-08-12 17:44:06 +02:00
2018-10-16 09:59:31 +02:00
res <- bind_rows(resS, resI, resR) %>%
mutate(Interpretation = factor(Interpretation, levels = c("R", "I", "S"), ordered = TRUE)) %>%
tidyr::gather(Antibiotic, Value, -Interpretation, -data.groups)
} else {
resIR <- summarise_if(.tbl = data,
.predicate = is.rsi,
.funs = portion_IR,
minimum = minimum,
as_percent = as_percent) %>%
mutate(Interpretation = "IR") %>%
select(Interpretation, everything())
data.groups <- group_vars(data)
res <- bind_rows(resS, resIR) %>%
mutate(Interpretation = factor(Interpretation, levels = c("IR", "S"), ordered = TRUE)) %>%
tidyr::gather(Antibiotic, Value, -Interpretation, -data.groups)
}
2018-08-13 16:42:37 +02:00
if (!translate_ab == FALSE) {
if (!tolower(translate_ab) %in% tolower(colnames(AMR::antibiotics))) {
stop("Parameter `translate_ab` does not occur in the `antibiotics` data set.", call. = FALSE)
}
res <- res %>% mutate(Antibiotic = abname(Antibiotic, from = "guess", to = translate_ab))
2018-08-12 17:44:06 +02:00
}
2018-08-13 16:42:37 +02:00
2018-08-12 17:44:06 +02:00
res
}
#' Calculate resistance of isolates
#'
#' This function is deprecated. Use the \code{\link{portion}} functions instead.
#' @inheritParams portion
#' @param ab1,ab2 vector (or column) with antibiotic interpretations. It will be transformed internally with \code{\link{as.rsi}} if needed.
#' @param interpretation antimicrobial interpretation to check for
#' @param ... deprecated parameters to support usage on older versions
#' @importFrom dplyr tibble case_when
#' @export
rsi <- function(ab1,
ab2 = NULL,
interpretation = "IR",
minimum = 30,
as_percent = FALSE,
...) {
2018-10-12 16:35:18 +02:00
.Deprecated(new = paste0("portion_", interpretation))
if (all(is.null(ab2))) {
df <- tibble(ab1 = ab1)
} else {
df <- tibble(ab1 = ab1,
ab2 = ab2)
}
2018-10-12 16:35:18 +02:00
if (!interpretation %in% c("S", "SI", "IS", "I", "RI", "IR", "R")) {
stop("invalid interpretation")
}
result <- case_when(
interpretation == "S" ~ portion_S(df, minimum = minimum, as_percent = FALSE),
interpretation %in% c("SI", "IS") ~ portion_SI(df, minimum = minimum, as_percent = FALSE),
interpretation == "I" ~ portion_I(df, minimum = minimum, as_percent = FALSE),
interpretation %in% c("RI", "IR") ~ portion_IR(df, minimum = minimum, as_percent = FALSE),
2018-10-12 16:35:18 +02:00
interpretation == "R" ~ portion_R(df, minimum = minimum, as_percent = FALSE))
if (as_percent == TRUE) {
percent(result, force_zero = TRUE)
} else {
result
}
}