Skip to contents

Anchor-based MCID

Source: R/clinimetrics-estimate.R
estimateMCID_anchor.Rd

Estimates the MCID from change scores anchored to an external improvement indicator. "roc" returns the change cut-off that maximizes the Youden index for classifying the anchor; "mean_change" the mean change among the (minimally) improved; "predictive" the change at which a logistic model of the anchor crosses probability 0.5.

Usage

estimateMCID_anchor(
  change,
  anchor,
  method = c("roc", "mean_change", "predictive"),
  direction = c("increase", "decrease")
)

Arguments

change

Numeric change scores (follow-up minus baseline).

anchor

Improvement indicator aligned to change: logical, or 0/1, where TRUE/1 marks an (minimally) improved case.

method

"roc" (default), "mean_change" or "predictive".

direction

"increase" (default) if a higher change means improvement, or "decrease" if a lower change does.

Value

The MCID estimate (on the change scale).

References

Jaeschke et al. (1989); Copay et al. (2007).

See also

estimateMDC(), estimateMCID_distribution()

Examples

set.seed(1)
change <- c(rnorm(50, 1, 2), rnorm(50, 8, 2))
anchor <- rep(c(0, 1), each = 50)
estimateMCID_anchor(change, anchor)
#> [1] 4.290322