Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.722096
Title: Multivariate meta-analysis methods for bias reduction in systematic reviews
Author: Frosi, G.
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Background: Missing treatment effect estimates for particular outcomes in a study have the potential to affect the conclusions in a meta-analysis (MA), especially if missingness is a result of outcome reporting bias (ORB). ORB comes from the results-based selection for publication of a subset of the original measured outcome variables and can result in overestimating treatment effects, resulting in bias. As well as missing treatment effect estimates at the study level, outcome data may also be missing within studies at the individual participant level. Multivariate meta-analysis (MVMA) of individual participant data (IPD) has the potential to overcome the impact of both these problems, by utilising the correlation between outcomes. Methods: An assessment of ORB was carried out in a cohort of systematic reviews (SR) with a core set of outcomes investigating pharmacological treatments for rheumatoid arthritis (RA). Novel IPD- MVMA methods to borrow strength across correlated outcomes were applied and evaluated through simulation, with the aim to show and quantify how this approach can reduce bias and improve precision of MA results, compared to traditional univariate methods, when there is missing outcome data as a result of both ORB and missing participant data. ORB assessments and the MVMA methods were applied to examples in RA. Results: Of the 167 assessable trials from 21 Cochrane RA reviews, 23% contained high suspicion of ORB in at least one of the core outcomes. Results from the simulations showed that the ‘borrowing of strength’ (BoS) in a multivariate model can reduce the magnitude of bias and increase precision in the pooled estimates. Results showed that when an ORB mechanism is introduced or there was missing IPD, MVMA tends to reduce the bias, increase the precision and improve the coverage when compared to a univariate analysis. In some instances, these benefits observed were also found in applying MVMA to the RA reviews. Conclusions: MVMA is not the solution to all missing data related problems within review meta- analyses, but, informed by this thesis, it can unquestionably be seen as a route to address missing outcome data in SRs. The BoS, reduction in bias and increase in precision can all be seen as promising.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.722096  DOI: Not available
Share: