compare_mic: Compare and validate MIC values

In MIC: Analysis of Antimicrobial Minimum Inhibitory Concentration Data

Compare and validate MIC values

Description

This function compares an vector of MIC values to another. Generally, this is in the context of a validation experiment – an investigational assay or method (the "test") is compared to a gold standard. The rules used by this function are in line with "ISO 20776-2:2021 Part 2: Evaluation of performance of antimicrobial susceptibility test devices against reference broth micro-dilution."

There are two levels of detail that are provided. If only the MIC values are provided, the function will look for essential agreement between the two sets of MIC. If the organism and antibiotic arguments are provided, the function will also calculate the categorical agreement using EUCAST breakpoints (or, if breakpoint not available and accept_ecoff = TRUE, ECOFFs).

The function returns a special dataframe of results, which is also an mic_validation object. This object can be summarised using summary() for summary metrics, plotted using plot() for an essential agreement confusion matrix, and tabulated using table().

Usage

compare_mic(
  gold_standard,
  test,
  ab = NULL,
  mo = NULL,
  accept_ecoff = FALSE,
  simplify = TRUE,
  ea_mode = "categorical",
  tolerate_censoring = "gold_standard",
  tolerate_matched_censoring = "both",
  ...
)

Arguments

gold_standard	vector of MICs to compare against.
test	vector of MICs that are under investigation
ab	character vector (same length as MIC) of antibiotic names (optional)
mo	character vector (same length as MIC) of microorganism names (optional)
accept_ecoff	if TRUE, ECOFFs will be used when no clinical breakpoints are available
simplify	if TRUE, MIC values will be coerced into the closest halving dilution (e.g., 0.55 will be converted to 0.5)
ea_mode	"categorical" or "numeric", see essential_agreement
tolerate_censoring	"strict", "gold_standard", "test", or "both" - how to handle censored data (see essential_agreement for details). Generally, this should be left as "gold_standard" since this setting "tolerates" a test that has higher granularity (i.e., less censoring) than the gold standard. Setting to "test" or "both" should be used with caution but may be appropriate in some cases where the test also produces censored results.
tolerate_matched_censoring	"strict", "gold_standard", "test", or "both" - how to handle situations where one of the values is censored, but both values match (e.g., gold_standard = ">2", test = "2"). Generally, this should be left as "both", since these values are considered to be in essential agreement. For more details, see essential_agreement.
...	additional arguments to be passed to AMR::as.sir

Value

S3 mic_validation object

Examples

# Just using MIC values only
gold_standard <- c("<0.25", "8", "64", ">64")
test <- c("<0.25", "2", "16", "64")
val <- compare_mic(gold_standard, test)
summary(val)

# Using MIC values and antibiotic and organism names
gold_standard <- c("<0.25", "8", "64", ">64")
test <- c("<0.25", "2", "16", "64")
ab <- c("AMK", "AMK", "AMK", "AMK")
mo <- c("B_ESCHR_COLI", "B_ESCHR_COLI", "B_ESCHR_COLI", "B_ESCHR_COLI")
val <- compare_mic(gold_standard, test, ab, mo)
"error" %in% names(val)  # val now has categorical agreement

MIC documentation built on June 10, 2025, 9:14 a.m.