Cutoff designs for community-based intervention studies

Abstract

Public health interventions are often designed to target communities defined either geographically (e.g. cities, counties) or socially (e.g. schools or workplaces). The group randomized trial (GRT) is regarded as the gold standard for evaluating these interventions. However, community leaders may object to randomization as some groups may be denied a potentially beneficial intervention. Under a regression discontinuity design (RDD), individuals may be assigned to treatment based on the levels of a pretest measure, thereby allowing those most in need of the treatment to receive it. In this article, we consider analysis, power, and sample size issues in applying the RDD and related cutoff designs in community-based intervention studies. We examine the power of these designs as a function of intraclass correlation, number of groups, and number of members per group and compare results to the traditional GRT.

Copyright © 2011 John Wiley & Sons, Ltd.

Figures

Figure 1

Hypothetical Regression Discontinuity Experiments.

Figure 2

Efficiency of the Cutoff-Based GRT (CO-GRT) relative to the Group Randomized Trial (GRT) and the Regression Discontinuity Design (RDD).

Figure 3

Comparison of Power of Three Designs for a Nested Cross-Sectional Study: the Group Randomized Trial (GRT), Regression Discontinuity Design (RDD), and Cutoff Group Randomized Trial (CO-GRT) with 50% randomized. Plots assume a balanced design, normally distributed pretest measures, a difference in condition means (Δ) of 0.25, variance σy2=1, and σg:c2∕(σg:c2+σtg:c2)=0.76.

Figure 4

Comparison of Power of Cutoff-Based Nested Cross-Sectional GRTs with Different Randomization Proportions (q). Plots assume a balanced design, normally distributed pretest measures, a difference in condition means (Δ) of 0.25, variance σy2=1, and σg:c2∕(σg:c2+σtg:c2)=0.76.

Figure 5

Comparison of Power of Three Designs for a Nested Cohort Study: the Group Randomized Trial (GRT), Regression Discontinuity design (RDD), and Cutoff Group Randomized Trial (CO-GRT) with 50% Randomized. Plots assume a balanced design, normally distributed pretest measures, a difference in condition means (Δ) of 0.25, variance σy2=1, and σm:g:c2∕(σm:g:c2+σe12)=0.26.

Figure 6

Comparison of Power of Cutoff-Based Nested Cohort GRTs with Different Randomization Proportions (q). Plots assume a balanced design, normally distributed pretest measures, a difference in condition means (Δ) of 0.25, variance σy2=1, and σm:g:c2∕(σm:g:c2+σe12)=0.26.