Online Program

Return to main conference page
Wednesday, January 8
Wed, Jan 8, 10:30 AM - 12:15 PM
West Coast Ballroom
Quasi-Experimental Methods in Health Policy

The use of segmented regression for evaluation of an interrupted time series study involving complex intervention: The CaPSAI Project Experience (306653)

Victoria Boydell, Global Health Centre 
Joanna Paula Cordero, The World Health Organization 
*Ndema Abu Habib, The World Health Organization 
James Kiarie, The World Health Organization 
Dela Nai, Population Council 
My Huong Nguyen, The World Health Organization 
Donat Shamba, Ifakara Health Institute 
Petrus S Steyn, The World Health Organization 
Soe Soe Thwin, The World Health Organization 

Keywords: complex intervention, quasi-experiment, interrupted time series, segmented regression, community-driven intervention, modern contraception uptake.

Interrupted time series with a parallel control group (ITS-CG) design is the most powerful quasi-experimental design for evaluation of effectiveness of community-driven complex public health interventions. Using this design, the effectiveness of a social accountability intervention on increasing uptake of modern contraception was studied. The intervention was rolled-out gradually over time to intervention communities in Ghana and Tanzania; with control communities receiving standard of care. The ITS Poisson segmented model applying a generalized estimating equations (GEE) is proposed for evaluation of the level and rate of uptake of modern contraception. Two approaches are demonstrated to account for the lag in intervention roll-out;(1) through a two-segmented ITS model; and (2) through a three-segmented ITS model. Parameters interpretation and study design strengths and limitations are highlighted. In a well-planned ITS study design with a parallel control group, the segmented regression analysis that applies advanced statistical techniques to ensure the internal validity of the results remains the preferred approach to evaluating the impact of community-based interventions.