An interrupted vaping time series: read the trend before the change

Mark the intervention and the observation series

UK digital-health evaluation guidance distinguishes analysis of trends across repeated measurements from simply reporting pre- and post-intervention periods. It also identifies seasonal patterns, random variation, and correlation between nearby observations as analytic issues. Only those methods concepts are used here; its clinical examples do not establish a vaping effect.

Find what each observation counts, its interval, and when the intervention actually began. A launch announcement, service opening, and gradual availability could have different dates. Use the report’s documented interruption rather than selecting the most impressive point on the graph. If the date or definition is unclear, keep that as an unresolved design question.

UK Office for Health Improvement and Disparities: Interrupted time series study

Sketch two different hypothetical changes

In an invented diagram, one series drops immediately at an intervention point and then continues its earlier slope. Another changes slope gradually with no immediate drop. Label the first a possible level change and the second a possible trend change. These are illustrative shapes, not data or proof of a support program’s impact.

Now draw a line that was already falling before the intervention. A lower later average may partly reflect continuation of that earlier pattern. The task is not to infer the real explanation from the sketch. It is to see why the report needs repeated earlier observations and an analysis matching the question about change.

Read the other events on the timeline

Look for the authors’ discussion of measurement changes, seasonality, and other events occurring around the interruption. If a comparison series was used, identify what it represents and why it was chosen. Do not assume any unrelated series automatically controls every competing explanation.

A densely plotted graph can look persuasive even when its underlying observation definition changed. Read the method and uncertainty rather than counting points as independent proof. This introductory timeline exercise cannot perform a statistical model check or certify a causal effect. It identifies questions a fuller appraisal needs to address.

Describe the observed change and the inference separately

A useful note names the series, intervention point, modeled change, and acknowledged limitations. If the report only compares one before value with one after value, do not relabel it an interrupted time series merely because a line connects them. Keep what you verified separate from what the headline claims caused it.

This guide does not evaluate a named policy or recommend starting and withdrawing support to test yourself. Keep agreed support and professional instructions intact. Your personal record can show when you changed a routine without establishing the unobserved course that would otherwise have occurred. For treatment-related interpretation of a study, seek qualified help with the actual methods and findings.