Distribution-based Minimal Detectable Change from test-retest data

Estimates the MDC from a test-retest reliability study: the intraclass correlation gives the reliability, the standard error of measurement is \(SD\sqrt{1 - ICC}\), and \(MDC = SEM \times z \times \sqrt{2}\) (delegating to PhysioCore::icc/sem/mdc).

Usage

estimateMDC(
  test_retest,
  method = "distribution",
  confidence = 0.95,
  model = c("twoway", "oneway")
)

Arguments

test_retest

A subjects-by-occasions numeric matrix (or data.frame) of repeated measurements.

method

Estimation method; currently "distribution".

confidence

Confidence level for the MDC (default 0.95).

model

ICC model passed to PhysioCore::icc (default "twoway").

Value

A list with mdc, sem, icc and confidence.

References

Shrout & Fleiss (1979); de Vet et al. (2006).

See also

estimateMCID_distribution(), estimateMCID_anchor()

Examples

set.seed(1)
base <- rnorm(40, 50, 10)
retest <- base + rnorm(40, 0, 3)
estimateMDC(cbind(base, retest))
#> $mdc
#> [1] 5.389172
#> 
#> $sem
#> [1] 1.944281
#> 
#> $icc
#> [1] 0.9565712
#> 
#> $confidence
#> [1] 0.95
#>