The Half-Cycle Correction: Banish Rather Than Explain It

Abstract

The half-cycle correction is often used in discrete Markov models to estimate state membership. This article shows that the correction, in addition to being unintuitive, actually produces the wrong results in many circumstances. These include quality-adjusted life year (QALY) weights and unit costs that differ by cycle. The half-cycle correction is also incompatible with discounting of the obtained stream of state membership. It is furthermore shown that the life table method of estimating state membership obtains correct results under these circumstances and is also much more transparent. The article concludes that the half-cycle correction should be dropped in favor of the life table method.