2024-25 Project (Thompson & Leyrat & Bradley)
Improving efficiency in global health trials: accounting for multiple levels of clustering in cluster randomised trials and stepped wedge trials
Dr Jennifer Thompson at LSHTM
Dr Clémence Leyrat at LSHTM
Dr John Bradley at LSHTM
Randomised trials can cost millions of pounds to run, so it is not surprising that there is growing call for improvements in their efficiency. Cluster randomised trials are a type of randomised trial where groups of individuals, such as geographical areas or hospitals, are randomised. These trials are often particular large and expensive to conduct.
There are different cluster randomised trial study designs, and these include parallel cluster randomised trials and stepped wedge trials. In a parallel cluster randomised trial, clusters are randomised to either receive an intervention or a control condition for the duration of a trial. In a stepped wedge trial, they all start the trial receiving the control condition and are randomised to switch to the intervention condition at some time during the trial.
In this research project, you will investigate methods of analysis that that makes these trials more efficient, which will mean fewer patients or less time is needed to learn whether an intervention is effective. You will focus on how to use additional levels of clustering in the data, for example if multiple observations are collected from each patient in a hospital, or multiple observations are made within households within villages.
Project Key Words
Cluster randomised trial Stepped wedge trial Simulation study Correlated data
MRC LID Themes
- Global Health = Yes
- Health Data Science = Yes
- Infectious Disease = Yes
- Translational and Implementation Research = Yes
MRC Core Skills
- Quantitative skills = Yes
- Interdisciplinary skills = No
- Whole organism physiology = No
Skills we expect a student to develop/acquire whilst pursuing this project
Students will develop skills in:
– Literature review
– Appraisal of literature
– Designing cluster randomised trials
– Designing stepped wedge trials
– Analysing cluster randomised trials
– Analysing stepped wedge trials
– Designing and conducting simulation studies
– Correlated data analysis e.g. mixed effect models and generalised estimating equations
Which route/s is this project available for?
- 1+4 = Yes
- +4 = Yes
Possible Master’s programme options identified by supervisory team for 1+4 applicants:
- LSHTM – MSc Health Data Science
- LSHTM – MSc Medical Statistics
Is this project available for full-time study? Yes
Is this project available for part-time study? Yes
Particular prior educational requirements for a student undertaking this project
- LSHTM’s standard institutional eligibility criteria for doctoral study.
- Students will require an MSc in Medical Statistics or closely related subject, and it is desirable to have some training with correlated data analysis.
Other useful information
- Potential CASE conversion? = No
PROJECT IN MORE DETAIL
Scientific description of this research project
In cluster randomised trials, we randomise groups of individuals like hospitals, villages, or schools. Individuals in the same cluster are similar to one another and the design and analysis of these trials must account for this correlation. In a parallel cluster randomised trial, clusters are randomised to either a control or intervention condition for the duration of the trial. In a stepped-wedge cluster randomised trial, clusters all begin in the control condition and are randomised to switch in the intervention condition at one of several time points spread throughout the duration of the trial. There are often additional levels of clustering within the randomisation clusters, for example, households within villages, or wards within hospitals. In this project we will investigate the impact of multiple levels of clustering on the design and analysis of both types of trial.
Objective one: Review existing literature on accounting for multiple levels of clustering in cluster randomised trials and other fields.
Objective two: Determine when there are efficiency gains from accounting for these lower levels of clustering in the analysis of a parallel cluster randomised trial.
Objective three: Determine efficient and robust methods of analysis for stepped wedge trials with individuals that are followed up and observed multiple times throughout the trial.
A systematic review of the literature will be conducted using the three main databases for medical research to achieve objective one. To meet objective two and three, a combination of theoretical development, simulation studies and empirical evaluation on real trial data will be employed.
You will have access to data from a completed parallel cluster randomised trial in Mali looking at the impact of proactive case management on reduction of child mortality, and data from a completed stepped wedge trial in Peru looking at the impact of villages switching to a salt substitute to reduce hypertension. Data sharing for the stepped wedge trial from Peru has been agreed with the trial principal investigator, and we are in contact with the principal investigator of the parallel cluster randomised trial from Mali.
Whilst this is a low risk project, there is a risk that research is published answering these objectives before the start of the project. If this were to happen, the student would develop a new related area of research with the project supervisors. If data sharing for the motivating trials should fall through, the supervisor team have identified alternative data sets.
(Relevant preprints and/or open access articles)
Additional information from the supervisory team
- The supervisory team has provided a recording for prospective applicants who are interested in their project. This recording should be watched before any discussions begin with the supervisory team.