Use this URL to cite or link to this record in EThOS:
Title: The impact of adjustment for covariates on meta-analysis of randomised intervention studies for binary outcome
Author: Yu, Ly-Mee
ISNI:       0000 0004 6062 3912
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Background: Covariate adjustment analysis is often used in epidemiological studies but is less common in randomised controlled trials (RCTs) and RCT meta-analyses. There is a lack of consensus on whether the analysis of RCT data should adjust for important baseline covariates. The estimated treatment effect of a binary covariate can differ when logistic regression is carried out, even when the covariate is balanced between treatment groups. Objectives: The objectives of this study were to examine the factors that affect the impact of adjusted analysis in different RCT scenarios and to explore the impact of adjusted analysis in RCT meta-analysis. Methods: Simulation and sampling studies were conducted to identify the factors that affect the impact of using an adjusted logistic regression model. Two covariates, one continuous and one binary, were considered simultaneously. The event rate, treatment effect, binary and continuous variable distributions, covariate prognostic strengths, and correlation between the covariates were varied during the simulations. The impact of adjustment on RCT meta-analysis was investigated using individual participant data obtained from the Perinatal Antiplatelet Review of International Studies. Different methods of performing unadjusted and adjusted meta-analysis were compared. Results: The simulation results suggest that adjustment only has a notable effect in extreme scenarios, such as a very large treatment effect or highly prognostic covariates. The relative difference between the unadjusted and adjusted odds ratios was found to be larger than 50% under these extreme scenarios. Covariate adjustment is likely to have a small effect on meta-analyses with many studies. Summary: Adjusted analysis should be carried out by design. Performing adjusted analysis in a meta-analysis can be challenging as sufficient information about the covariates is often not available.
Supervisor: Altman, Douglas Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available