Comparison of six statistical methods for interrupted time series studies: empirical evaluation of 190 published series

Abstract Background The Interrupted Time Series (ITS) is a quasi-experimental design commonly used in public health to evaluate the impact of interventions or exposures. Multiple statistical methods are available analyse data from ITS studies, but no empirical investigation has examined how different compare when applied real-world datasets. Methods A random sample 200 studies identified previous review were included. series each these was sought. Each dataset re-analysed using six methods. Point and confidence interval estimates for level slope changes, standard errors, p- values autocorrelation compared between Results From including 230 time series, 190 datasets obtained. We found that choice method can importantly affect change point estimates, their width intervals values. Statistical significance (categorised at 5% level) often differed across pairwise comparisons methods, ranging 4 25% disagreement. Estimates depending on length series. Conclusions lead substantially conclusions about interruption. Pre-specification encouraged, naive based should be avoided.