*Geek Box: Interrupted Time-Series Analysis

*Geek Box: Interrupted Time-Series Analysis

Interrupted time series [ITS] analysis is a method used in epidemiology to evaluate the effect of population-wide public health interventions. Breaking down the name may help to understand what this analysis is evaluating. Imagine there is a public health policy that came into place in January 2022; this policy is the ‘interruption’ from the previous status quo. And let’s say we were interested in the effect of the intervention 1-year later; we could establish a ‘time series’ of 1-year prior to the intervention and 1-year post intervention.

It is important to note that this type of analysis is not looking at individual-level effects and averages [i.e., means] calculated from individual-level effects, as would be done in a randomised controlled trial [e.g., measuring every participant’s cholesterol levels] or prospective cohort study [e.g., assessing every participant’s diet with a food-frequency questionnaire]. Rather, ITS is specifically looking at population-level effects, e.g., does the introduction of a vaccine reduce prevalence and incidence of a disease in the population compared to before.

While this is a strength of ITS, it also opens the limitations of this analytical approach unless other methods are considered in the analysis. A good ITS analysis compares a population who received an intervention to a population that did not receive it, which can be the same population compared before and after an intervention was introduced.

This allows for the framing of a “counterfactual” scenario, which describes the effect of an exposure on an outcome contrasting two potential outcomes. For example, the effect of a vaccine on a disease can be contrasted between those who received the vaccine and those who did not; it is not possible for someone to have had both outcomes. In effect, the counterfactual scenario estimates what would have happened in the absence of the intervention.

This allows for effective use of ITS analysis to compare trends in the exposure-outcome relationship of interest in the group exposed to the intervention compared to the counterfactual situation.