This thesis is based on a worldwide overview (meta-analysis) of prospective observational studies of blood pressure and cholesterol, involving a centralised collection of data on over one million individuals from 59 studies, which I have co- ordinated since its inception. Analytically, the aim has been to develop and to use appropriate statistical techniques to assess the age- and sex-specific associations of usual blood pressure and of usual cholesterol with cause-specific mortality. Since the data set is uniquely large, and because appropriate methods of analysis (with full account taken of the time-dependent nature of the regression dilution bias) have been developed and used, these associations have been established more reliably. An integral part of the methodological element of the thesis has been to investigate the systematic underestimation of associations between risk factor and disease that are obtained when only a single baseline measurement is used to assess levels of such risk factors (the regression dilution bias). The extent of this bias has been investigated in each study that had repeat measurements of risk factors during follow-up. One particularly novel aspect has been the emphasis on, and methods developed to account for, the regression dilution bias in several studies simultaneously and in an appropriately time-dependent way. This thesis illustrates the extent to which random error and inappropriate statistical analysis lead to misleading conclusions concerning the importance of blood pressure and blood cholesterol, particularly in premature death. Only by studying adequate numbers of deaths (136,000 deaths among 1 million adults during 13 million person- years of follow-up) and by using appropriate statistical techniques - taking proper account of (a) the regression dilution bias; (b) the full range of blood pressure and cholesterol; (c) the opposing effects of HDL.and the remaining non-HDL cholesterol; and (d) age at death - did it become possible to provide reliable results on the true relationships between blood pressure, cholesterol fractions and vascular and other causes of death. These analyses have demonstrated reliably that, as causes of IHD death in early middle age, blood pressure and blood lipids are three to five times more important than suggested by inappropriate analyses, with no clinically relevant inverse associations with cancer or other non-vascular mortality (except, surprisingly, COPD).
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:418648 |
| Date | January 1999 |
| Creators | Lewington, Sarah |
| Publisher | University of Oxford |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://ora.ox.ac.uk/objects/uuid:517a1b6c-4752-46e7-868b-48a4ea078e69 |
Page generated in 0.0019 seconds