BackgroundTo assess bias and generalizability of results in randomized controlled trials (RCT), investigators compare participants to non-participants or early- to late-participants. Comparisons can also inform the recruitment approach, especially when working with challenging populations, such as urban adolescents. In this paper, we describe characteristics by participant status of urban teens eligible to participate in a RCT of a school-based, web-based asthma management program.
The denominator for this analysis was all students found to be eligible to participate in the RCT. Data were analyzed for participants and non-participants of the RCT, as well as for students that enrolled during the initially scheduled recruitment period (early-participants) and persons that delayed enrollment until the following fall when recruitment was re-opened to increase sample size (late-participants). Full Time Equivalents (FTEs) of staff associated with recruitment were estimated.
Of 1668 teens eligible for the RCT, 386 enrolled early, and 36 enrolled late, leaving 1246 non-participants. Participants were younger (p < 0.01), more likely to be diagnosed, use asthma medication, and have moderate-to-severe disease than non-participants, odds ratios (95% Confidence Intervals) = 2.1(1.7-2.8), 1.7(1.3-2.1), 1.4(1.0-1.8), respectively. ORs were elevated for the association of late-participation with Medicaid enrollment, 1.9(0.7-5.1) and extrinsic motivation to enroll, 1.7(0.6-5.0). Late-participation was inversely related to study compliance for teens and caregivers, ORs ranging from 0.1 to 0.3 (all p-values < 0.01). Early- and late-participants required 0.45 FTEs/100 and 3.3 FTEs/100, respectively.
Recruitment messages attracted youth with moderate-to-severe asthma, but extending enrollment was costly, resulting in potentially less motivated, and certainly less compliant, participants. Investigators must balance internal versus external validity in the decision to extend recruitment. Gains in sample size and external validity may be offset by the cost of additional staff time and the threat to internal validity caused by lower participant follow-up.