Rasch (interval) analysis of a clinical ADL/IADL instrument

Fits a polytomous Rasch model to a cohort's responses on a registered instrument and returns the psychometric summary that published ADL/IADL analyses report: the item difficulty hierarchy, category threshold ordering (with disordered-threshold flags), person and item measures on an interval (logit) scale, item fit (infit/outfit), person/item separation reliability, and the raw-score to interval-measure conversion. The Partial Credit Model is the default (it accommodates the Barthel Index's mixed item lengths); pass model = "RSM" for a uniform-format scale.

Usage

raschAnalyze(instrument, responses, model = c("PCM", "RSM"), ...)

Arguments

instrument: A ClinicalInstrument or an instrument id (e.g. "barthel", "lawton_iadl").
responses: A persons x items matrix or data.frame of responses on the instrument's native scale (see raschRecode()).
model: "PCM" (default) or "RSM".
...: Passed to PhysioAppKit::pcm_measure().

Value

The poly_rasch fit (see PhysioAppKit::pcm_measure()) with an added class clin_rasch and extra fields: instrument (id), item_hierarchy (items ordered hardest-to-easiest by calibrated location) and recoded (the category matrix analysed).

Details

Requires the suggested package PhysioAppKit (the domain-neutral engine).

Examples

# \donttest{
set.seed(1)
items <- getInstrument("barthel")@items
# a small synthetic cohort of weighted Barthel responses
resp <- t(replicate(50, {
  ability <- stats::rnorm(1)
  vapply(c(2, 1, 1, 2, 2, 2, 2, 3, 3, 2), function(m)
    c(0, 5, 10, 15)[min(m, max(0, round(ability + m / 2))) + 1], numeric(1))
}))
colnames(resp) <- items
if (requireNamespace("PhysioAppKit", quietly = TRUE)) {
  fit <- raschAnalyze("barthel", resp)
  fit$item_hierarchy
}
#>          item   location        se      infit     outfit disordered
#> 1     bathing  0.5295996 0.8043889 0.09322405 0.04509349      FALSE
#> 2    grooming  0.5295996 0.8043889 0.09322405 0.04509349      FALSE
#> 3   transfers  0.2249286 0.5335115 0.11378499 0.10444891      FALSE
#> 4    mobility  0.2249286 0.5335115 0.11378499 0.10444891      FALSE
#> 5     feeding -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
#> 6    dressing -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
#> 7      bowels -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
#> 8     bladder -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
#> 9  toilet_use -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
#> 10     stairs -0.2007309 0.6806820 0.06327283 0.05849089      FALSE
# }