Use this URL to cite or link to this record in EThOS:
Title: Statistical analysis of quantitative seroepidemiological data
Author: Kafatos, George
ISNI:       0000 0004 2710 8599
Awarding Body: Open University
Current Institution: Open University
Date of Award: 2011
Availability of Full Text:
Access from EThOS:
Access from Institution:
The material for this thesis is based on the European Sero-Epidemiology Network 2 (ESEN2), a study funded by the European Commission. As part of the ESEN2 project, age-specific population seroprevalence of 8 antigens tested in 22 national laboratories was estimated and compared between countries. To achieve harmonised serological results, a reference panel was tested by each participant laboratory. Each laboratory's panel results were regressed against the reference centre, thus obtaining standardisation equations. These equations were used to convert the quantitative measurements of the serosurveys into common units that were subsequently classified into negative (susceptible) or positive (protected) according to a serological cut-off. The aim of this thesis was to further develop and validate the methodology for standardising serological outcomes, and to propose alternative methods for achieving comparable population seroprevalence. As part of this thesis, a statistical algorithm was established to standardise serological results. Censored regression methods were considered to account for measurements outside the assay detection range. The impact of standardisation on seroprevalence was examined. Mixture modelling of the serological results was proposed as an alternative method to standardisation for estimating seroprevalence. Although mixture modelling may provide better seroprevalence estimates in certain situations, it is heavily dependent on model assumptions, mainly of well-separated underlying distributions. In terms of seroprevalence estimation using standardisation, the validity of the assay cut-off point was examined. A method for re-estimating cut-offs was proposed based on mixture modelling that improved seroprevalence estimates under certain distributional assumptions. The impact of variability occurring due to serum testing in batches (plate-to-plate variability) on seroprevalence was assessed. The method currently used by the laboratories was examined, and a new method was proposed to adjust for this based on mixture models. In conclusion, the standardisation method used for the ESEN2 project was validated and some improvements were proposed.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral