Design, analysis, power, and sample size calculation for three‐phase interrupted time series analysis in evaluation of health policy interventions

Summary

This paper presents a methodological framework and simulation-based tool for designing and analysing three-phase interrupted time series studies used to evaluate health policy and systems interventions. The authors provide ready-to-use computer programs to calculate appropriate sample sizes whilst accounting for autocorrelation and varying effect sizes, addressing a gap in statistical guidance for policy evaluation research. The work enables investigators to prospectively design adequately powered ITS studies by simulating power across realistic scenarios.

UK applicability

The statistical methods and software tools presented are broadly applicable to UK health policy evaluation regardless of geography, as they address fundamental design and analysis challenges in interrupted time series studies. UK researchers and policy evaluators implementing three-phase ITS designs for National Health Service or public health interventions would benefit from these power calculation approaches.

Key measures

Statistical power, sample size requirements, segmented autoregressive error models, autocorrelation effects, effect size detection across level changes, trend changes, and combined scenarios

Outcomes reported

The study developed simulation-based methods to calculate statistical power and sample size requirements for three-phase interrupted time series (ITS) studies evaluating health policy, systems, or environmental interventions. Power estimates were generated across varying autocorrelation levels (−0.9 to 0.9) and effect sizes for segmented autoregressive error models.

Topic tags