Interrupted Time Series

Interrupted Time Series

What is an interrupted time series?

What questions does it help me answer?

What are its strengths and weaknesses?



Interrupted Time Series

Time Series Design











What is an interrupted time series design?  Researchers seek to establish causal relationships by conducting experiments. Often circumstances will not permit meeting all conditions of a ‘true experiment’. Thus, quasi-experimental is chosen. Among quasi-experimental design is one that rivals the true experiment: the interrupted time series design.  It has become a standard method of causal analysis in evaluating the impacts of interventions, health programs, and state/national policies.

Time series designs can be used in prospective or retrospective evaluations of program impacts. Multiple repeated observations are made at regular intervals before and after an intervention.  Statistical analyses are used to determine whether there is change in scores or trend in scores of the observations before and after the intervention.

What type of questions does it help to answer?This design method is used to draw causal inferences about interventions or any ‘interruption’ (policy change, natural disasters) effects. It requires data collect for many consecutive points in time before and after the intervention is introduced.

Strengths: The advantage of this method is that it can draw causal inferences about the program, detect delayed or intermittent program effects, and determine if these change are temporary or permanent.

Limitations:  One is unable to determine whether the change noted is due to the intervention or to other factors such as other concurrent events during the time of the intervention. However, this may be alleviated by adding a similar comparison group (control group) or by replicating the intervention/program in a variety of settings and populations (to obtain similar results).

Examples of studies using this technique:

Soumerai, S.B., et al., Payment Restrictions for Prescription Drugs under Medicaid. New England Journal of Medicine, 1987. 317(9): p. 550-556.

Click here to go in depth.

  1. Fang Zhang, PhD DPM says:

    In longitudinal observational studies, it is very important to adjust for the baseline trend for desired outcomes. The usual pre-post analyses will introduce various biases including history, maturation, instrumentation, regression to the mean and etc. The interrupted time-series design aims to minimize such biases, and provide the closest design to a randomized clinical trial(RCT) in a practical setting.

Leave a Reply

× eight = 40