Using an antedependence test to analyse post-operative pain measurements

Post-operative pain is frequently measured by a visual analogue scale (VAS). The most common methods for analysing such data are the so-called time-by-time method and repeated measures analysis of variance. In the first, due to correlation between time points, successive tests are not independent and cannot measure the treatment effects over time. In the second, a rather unlikely covariance structure has to be assumed. In a simulation study we have shown that the power of the antedependence test is nearly the same as that of repeated measures of analysis of variance and it does not suffer from some of the latter's shortcomings. We also present an analysis of real data using the antedependence test.