Missing Data in Paediatric Clinical Trials

: why does it happen and what can we do about it?

Abstract

Retention of participants, therefore reducing missing data, in randomised
controlled trials (RCTs) is one of the most important issues currently concerning
clinical trialists. There are several research projects which are investigating
retention in clinical trials, however none of these projects are specifically
exploring retention in paediatric trials.

Analyses of RCTs are often carried assuming that the probability of
being missing only depends on the observed data (Missing At Random, MAR),
but sensitivity analyses using statistical methods that allow the assumption
that the missingness is related to the actual value of the missing observation
(Missing Not At Random, MNAR) are rarely carried out, although advised.

In this thesis, I begin by conducting a systematic review of retention
of participants to reporting the primary outcome in paediatric RCTs published
between January 2015 to December 2019 within six high impact-factor journals.
I conduct meta-regressions of trial and participant factors which may be
associated with retention to the primary outcome. I conduct a systematic
review and narrative synthesis of qualitative studies exploring participant
retention in paediatric trials, and a qualitative study exploring clinical trialists
experience of conducting paediatric RCTS. I review methods which are suitable
for sensitivity analyses to the MAR assumption for normally-distributed
missing outcome data in RCTs. In a simulation study, I compare the Mean
Score, Delta-shift after multiple imputation, Selection Model with inverse
probability weighting and Stacked multiple imputation methods. I apply
these to a trial data example, the Bristol Girls Dance Project. I conclude with a
discussion and suggestions for future research.

Date of Award	5 Dec 2023
Original language	English
Awarding Institution	University of Bristol
Supervisor	Chris Metcalfe (Supervisor), Jeremy Horwood (Supervisor), Rach Hughes (Supervisor) & Lucy Beasant (Supervisor)