Sample size for a baseline-adjusted (ANCOVA) continuous endpoint

The recommended pre-post rehabilitation analysis: adjusting the follow-up outcome for its baseline reduces the residual variance by a factor \((1-\rho^2)\), where \(\rho\) is the baseline-outcome correlation - shrinking the required sample versus an unadjusted comparison.

Usage

sampleSizeANCOVA(
  delta,
  sd,
  baseline_cor,
  power = 0.8,
  sig_level = 0.05,
  ratio = 1,
  sides = 2L,
  dropout = 0,
  cluster_size = NULL,
  icc = NULL
)

Arguments

delta, sd, power, sig_level, ratio, sides, dropout, cluster_size, icc

As in sampleSizeContinuous(). sd is the raw (unadjusted) outcome SD.

baseline_cor

Correlation \(\rho\) between baseline and follow-up outcome (0-1); the variance is scaled by \((1-\rho^2)\).

Value

A trial_power object.

References

Frison & Pocock (1992), Stat Med; Borm et al. (2007), J Clin Epi.

See also

sampleSizeChangeScore(), estimateBaselineCorrelation()

Examples

sampleSizeANCOVA(delta = 5, sd = 10, baseline_cor = 0.6)   # < 64/arm
#> <trial_power> ANCOVA (baseline-adjusted)
#>   effect: delta=5, sd=10, baseline_cor=0.6, d_adjusted=0.625
#>   power 0.80 | alpha 0.05 (2-sided) | allocation 1:1
#>   n per arm: control 42, treatment 42  (total 84)