Etiological studies in complex diseases

Abstract

The incidence of a number of site-specific cancers is increased among patients with type 2 diabetes mellitus (T2DM) as compared to non-diabetic individuals. Similarly, the incidence of venous thromboembolism (VTE) is higher in patients with multiple sclerosis (MS) when compared to patients without MS. These observation statements are said to be of a descriptive nature in that they merely state the frequency and distribution of disease among specific populations. However, merely knowing that a disease occurs in certain patient groups is not the issue of ultimate interest. Observing an association between two health-related conditions raises the more substantial question as to why they are related. In other words, we want to know whether the observed association has a causal nature. Scientific explanations invariably relate to knowledge on how particular phenomena can be explained in terms of the factors that cause them. In epidemiology, however, reaching an unambiguous verdict regarding the nature of an association is hampered, mainly because of the inherently flawed comparisons that are made as a result of its non-experimental nature. Consequently, alternative explanations to a causal link between the exposure and outcome are not systematically remedied in epidemiologic studies. That is to say, the difference observed between groups in non-experimental comparison studies is not necessarily attributable to the variable under study but could be the result of systematic, or non-random, variability. Against this complex background, the thesis concentrates on the assessment of the influence of specific risk factors in the etiology of (particular types of) cancer in patients with T2DM and in the etiology of VTE in patients with MS. More specifically, issues related to the exclusion of alternative explanations for observed exposure-effect relationships in order to facilitate the assessment of causality from epidemiologic evidence are addressed. With respect to the increased incidence of certain types of cancer in patients with type 2 diabetes mellitus, the research covers the effect of exogenous insulin use on cancer risk, the influence of choices in study design on the observed association between exogenous insulin exposure and the risk of several site-specific cancers, the effect of body mass-index and duration of obesity on colorectal cancer incidence, and whether metformin use affects survival among female type 2 diabetic patients with breast cancer. Regarding the increased incidence of VTE among patients with MS, focus lies on the influence of several general and MS-related risk factors that might explain why these two health-related conditions appear to be associated with each other. General risk factors include a prior VTE, varicose veins, major trauma, pregnancy, and infection, while disability, spasticity, inflammation, and exposure to glucocorticoids are considered MS-related risk factors. This thesis concludes with a theoretical investigation into whether epidemiologic evidence ultimately can contribute to our understanding of why certain health-related conditions occur. Based on a deliberation of the foundation of knowledge and the abstract concept of causality, the controversies surrounding the attribution of causality from epidemiologic evidence are put into perspective. Central to this discussion is the question: what is good epidemiologic knowledge?