Mediation analysis is a popular approach in the social an biomedical sciences to examine the extent to which the effect of an exposure on an outcome is through an intermediate variable (mediator) and the extent to which the effect is direct. We first develop statistical methods and software for the estimation of direct and indirect causal effects in generalized linear models when exposure-mediator interaction may be present. We then study the bias of direct and indirect effects estimators that arise in this context when a continuous mediator is measured with error or a binary mediator is misclassified. We develop methods of correction for measurement error and misclassification coupled with sensitivity analyses for which no auxiliary information on the mediator measured with error is needed. The proposed methods are applied to a lung cancer study to evaluate the effect of genetic variants mediated through smoking on lung cancer risk and to a perinatal epidemiological study on the determinants of preterm birth.

Diagnostic tests are widely used in medicine and epidemiology. Most diagnostic tests areimperfect at distinguishing subjects with and without the condition of interest, and manythat provide results on a continuous scale have overlapping densities from diseased andnon-diseased subjects. For the se continuous tests, most statistical techniques developed todate assume a parametric (e.g. normal) family for the distribution of the continuousoutcomes within groups, an often unverifiable but convenient distributional assumption.In addition, evaluating the properties of these tests typically requires or assumes a perfectgold standard test is available. [...] / Les tests diagnostiques sont abondamment utilisés en médecine et en épidémiologie. Laplupart d'entre eux ne distinguent pas parfaitement les sujets qui ont ou non la conditiond'intérêt, et ceux qui fournissent des résultats sur une échelle continue ont souvent desdensités des résultats des sujets malades et non-malades qui se chevauchent. Pour cestests continus, la plupart des techniques statistiques développées jusqu'à présentprésument une famille de distributions paramétriques des résultats dans les 2 groupes, unehypothèse pratique mais souvent non vérifiable. De plus, l'évaluation de leurs propriétésrequiert typiquement qu'un test étalon d'or soit disponible. [...]

No description available.

This study analyses the data from medical records from a former tuberculous sanatorium for children at Stannington, Morpeth, Northumberland. It focuses on the demographic profile of the sanatorium and also examines the changes in therapy that occurred between 1937 and 1953. The objective of the study was to understand the patterning of tuberculosis in the sanatorium by considering the differences between male and female patients, ages affected, and the socio-economic backgrounds of patients from a sample of patient records taken from pre- and post-antibiotic eras, pre- and post-Second World War, and pre- and post-NHS years. 1,897 patient records were utilized in this study, all held at the Northumberland Record Office at Morpeth. This study was followed in accordance to the limitations given by the Medical Ethics Committee which was to ensure that patient confidentiality would be maintained. A limited database is included with this reasearch, but a complete database will be held in the future at the Northumberland Health Authority, in Morpeth. Overall more females than males were admitted to the sanatorium and all patients from various types of tuberculosis. The majority of the children (over 60%) were suffering from pulmonary tuberculosis, but there were a large number also suffering from tuberculosis of the bones and joints (230 cases or 12%). Most of the children came from poor backgrounds and originated from the Newcastle and Gateshead areas. The introduction of chemotherapy, the end of the Second World War and the implementation of the NHS did not have a great effect on who was being treated at the sanatorium. In conclusion these records hold a wealth of information that may help build an epidemiological model of tuberculosis in the North-East of England. Future work on the records is suggested and limitations of the research outlined.

This epidemiological thesis firstly reviews the disease Multiple Sclerosis (MS): its history, pathology, clinical expression, and the current views on immunopathogenesis, aetiology and treatment. A separate review on ultraviolet radiation (UVR) and MS indicates that recent work in photoimmunology provides evidence that UVR can attenuate T helper 1 cell mediated processes through several mechanisms, and that epidemiological features of MS, such as the striking latitudinal gradient and seasonal variation in month of birth, MS onset and disease activity, are at least in part consistent with the hypothesis that UVR exposure may reduce the risk of MS. An ecological analysis was conducted as part of the PhD to demonstrate that regional variation in MS prevalence in the continent of Australia could be closely predicted by regional UVR levels, but analytical epidemiological studies are required to further investigate the UVR hypothesis. The project central to this thesis was a population based case-control study on Multiple Sclerosis, conducted in Tasmania, Australia. It examined: (i) whether high past sun exposure was associated with a reduced risk of MS, (ii) whether sibship structure and past infections influenced the risk of MS and (iii) whether having had children and differences in prevalence and strength of MS risk factors between men and women could explain the sex difference in MS. Interviews were conducted with 136 cases with MS and 272 controls randomly drawn from the community and matched on sex and birth year. In one of the methodology chapters, a measure-retest and method comparison was conducted to examine aspects of reliability of the sun exposure measures used in the case-control study. A separate study on 104 healthy volunteers was carried out to examine the effect of seasonal variation and body hair on melanin density estimates based on skin reflectance. The case-control study showed that higher past sun exposure, particularly during childhood and early adolescence, was associated with a reduced risk of MS, which is compatible with UVR having a protective role against MS. Having younger siblings, but not older siblings, was also associated with a reduced risk of MS, while having had glandular fever or having high antibody titers against the Epstein-Barr virus was associated with an increased risk of MS. Among women, a negative association was found between having had children and MS. The finding of an inverse association between sun exposure during childhood and early adolescence and MS, if confirmed in future work, will have important public health implications.

Diss. (sammanfattning) Stockholm : Karol. inst. / Härtill 5 uppsatser.

Diss. (sammanfattning) Stockholm : Karol. inst. / Härtill 5 uppsatser.

Thesis (Ph. D.)--University of New South Wales, 2002. / Also available online.

Thesis (Ph. D.)--University of Hong Kong, 1999. / Includes bibliographical references (leaves 234-245).

"Major Subject: Epidemiology" Title from author supplied metadata (automated record created 2010-03-12 12:08:51). Includes bibliographical references.