Two-by-two factorial randomised study within a trial (SWAT) to evaluate strategies for follow-up in a randomised prevention trial

Lucy E Bradshaw,Rachel H Haines,Alan A Montgomery,Joanne R Chalmers,Hywel C Williams

doi:10.1186/s13063-020-04373-4

Lucy E Bradshaw, Rachel H Haines + Show 3 more

Open Access

https://doi.org/10.1186/s13063-020-04373-4

Copy DOI

Abstract

BackgroundFailure to collect outcome data in randomised trials can result in bias and loss of statistical power. Further evaluations of strategies to increase retention are required. We assessed the effectiveness of two strategies for retention in a randomised prevention trial using a two-by-two factorial randomised study within a trial (SWAT).MethodsParents of babies included in the host trial were randomised to (1) short message service (SMS) notification prior to sending questionnaires at 3, 6, 12 and 18 months versus no SMS notification and (2) a £10 voucher sent with the invitation letter for the primary follow-up visit at 24 months or given at the visit. The two co-primary outcomes were collection of host trial (1) questionnaire data at interim follow-up times and (2) primary outcome at 24 months during a home/clinic visit with a research nurse.ResultsBetween November 2014 and November 2016, 1394 participants were randomised: 350 to no SMS + voucher at visit, 345 to SMS + voucher at visit, 352 to no SMS + voucher before visit and 347 to SMS + voucher before visit. Overall questionnaire data was collected at interim follow-up times for 75% in both the group allocated to the prior SMS notification and the group allocated to no SMS notification (odds ratio (OR) SMS versus none 1.02, 95% CI 0.83 to 1.25). Host trial primary outcome data was collected at a visit for 557 (80%) allocated to the voucher before the visit in the invitation letter and for 566 (81%) whose parents were allocated to receive the voucher at the visit (OR before versus at visit 0.89, 95% CI 0.69 to 1.17).ConclusionThere was no evidence of a difference in retention according to SMS notification or voucher timing. Future synthesis of SWAT results is required to be able to detect small but important incremental effects of retention strategies.Trial registrationISRCTN registry, ID: ISRCTN21528841. Registered on 25 July 2014. SWAT Repository Store ID 25.

Highlights

Failure to collect outcome data in randomised trials can result in bias and loss of statistical power
There was some evidence that monetary incentives for postal questionnaires and offer of monetary incentives for electronic questionnaires were effective in increasing response, but there were no trials that compared a monetary incentive given in advance with an incentive conditional on questionnaire completion
Between November 2014 and November 2016, 1394 families were randomised to the host trial and study within a trial (SWAT) (Fig. 1) at 12 hospitals and four general practice sites in the United Kingdom (UK)

Summary

Introduction

Failure to collect outcome data in randomised trials can result in bias and loss of statistical power. Failure to collect outcome data in randomised trials is inefficient, and can result in bias and loss of statistical power. Most of the 38 trials included were concerned with collection of outcome data via postal or electronic questionnaires rather than the return of participants to study sites. The authors encouraged trialists to consider conducting adequately powered evaluations of retention strategies within their trials. Organisations such as Trial Forge have pointed out that even small incremental gains of 1%, when added together, can make a substantial difference to trial performance [5]

Methods

Results

Discussion

Conclusion