<p>Sample size and power considerations for ordinary least squares interrupted time series analysis: a simulation study</p>

• DOI: 10.2147/clep.s176723
• Publication Date: 2019-02-24T19:49:20Z
• Authors: Samuel Hawley; M Sanni Ali; Klara Berencsi; Andrew Judge; Daniel Prieto-Alh...
• Keywords: Ordinary least squares; Sample (material)

Interrupted time series (ITS) analysis is being increasingly used in epidemiology. Despite its growing popularity, there a scarcity of guidance on power and sample size considerations within the ITS framework. Our aim this study was to assess statistical detect an intervention effect under various real-life scenarios. datasets were created using Monte Carlo simulations generate cumulative incidence (outcome) values over time. We generated 1,000 per scenario, varying number points, average point, relative reduction post intervention, location series, mediated via 1) slope change 2) step change. Performance measures included percentage bias. found that point had large impact power. Even scenarios with 12 pre-intervention post-intervention points moderate sizes, most analyses underpowered if low. conclude factors need be collectively considered ensure adequate for study. demonstrate means providing insight into underlying requirements ordinary least squares (OLS) measures, based prespecified parameters have developed Stata code estimate this.