Research Designs for Intervention Research with Small Samples II: Stepped Wedge and Interrupted Time-Series Designs

Abstract

The stepped wedge design (SWD) and the interrupted time-series design (ITSD) are two alternative research designs that maximize efficiency and statistical power with small samples when contrasted to the operating characteristics of conventional randomized controlled trials (RCT). This paper provides an overview and introduction to previous work with these designs, and compares and contrasts them with the dynamic wait-list design (DWLD) and the regression point displacement design (RPDD), which were presented in a previous article (Wyman, Henry, Knoblauch, and Brown, 2015) in this Special Section. The SWD and the DWLD are similar in that both are intervention implementation roll-out designs. We discuss similarities and differences between the SWD and DWLD in their historical origin and application, along with differences in the statistical modeling of each design. Next, we describe the main design characteristics of the ITSD, along with some of its strengths and limitations. We provide a critical comparative review of strengths and weaknesses in application of the ITSD, SWD, DWLD, and RPDD as small samples alternatives to application of the RCT, concluding with a discussion of the types of contextual factors that influence selection of an optimal research design by prevention researchers working with small samples.

Stepped wedge cluster randomised trials: a review of the statistical methodology used and available

“…Moulton et al [ 26 ] outline an approach to the analysis of a SWD using a Cox proportional hazards model adjusting for clustering by either bootstrapping or using the RVE. Other authors also suggested the use of a GLMM or a GEE [ 24 , 26 , 27 , 31 , 32 , 35 ] but some had different views on how the fixed or random effects should be specified. For example both Wyman et al [ 32 ] and Van den Heuvel et al [ 31 ] advocate the use of random effects for time in addition to the random intercept whereas Fok et al [ 27 ] outline a random intercept model with an interaction term between time and intervention as a fixed effect.…”

“…Other authors also suggested the use of a GLMM or a GEE [ 24 , 26 , 27 , 31 , 32 , 35 ] but some had different views on how the fixed or random effects should be specified. For example both Wyman et al [ 32 ] and Van den Heuvel et al [ 31 ] advocate the use of random effects for time in addition to the random intercept whereas Fok et al [ 27 ] outline a random intercept model with an interaction term between time and intervention as a fixed effect. Gruber et al take a causal modelling approach to the analysis and outline how to estimate the complier average causal effect from a stepped wedge trial [ 28 ].…”

Multi-Level Cultural Intervention for the Prevention of Suicide and Alcohol Use Risk with Alaska Native Youth: a Nonrandomized Comparison of Treatment Intensity

“…This allowed comparison of one community to another at a point earlier in its intervention process. This approach, a variation of a stepped-wedge design (Fok, Henry, & Allen, 2015), avoided a direct comparison of quality of the intervention work across two close-knit communities, because the unit of analysis was intensity of intervention.…”

“…Wyman et al (2015) describe logistical and ethical issues that can arise, and note inefficiencies in how the RCT uses information can result in low power and external validity. They propose the DWLD and regression point displacement design as alternative designs that maintain rigor while optimizing power, and Fok et al (2015) propose the stepped wedge design (SWD) and interrupted time-series design. Roll out designs such as the DWLD and SWD allow for the additional element of randomization of intervention start time.…”

A better start to literacy learning: findings from a teacher-implemented intervention in children’s first year at school

“…A stepped wedge research design (Fok, Henry, & Allen, 2015) was utilized to examine the intervention effects for children with lower levels of oral language where the intervention was rolled out sequentially in Group A (72 children across 3 schools) and then Group B (69 children across 4 schools) following a baseline monitoring phase. In New Zealand, the school year is divided into four teaching terms of 10 weeks in duration.…”