Search Results

The previous chapter described a prospective population cohort study as a prototype. By modifying one or another element of this prototype, this chapter constructs natural experiments, historical cohorts, and special exposure cohorts. To these three major variants a fourth is added—the clinical cohort—the purpose of which is to discover the determinants of the course and outcome of a disorder. This variant, close kin to the etiologic cohort, has been prominent in psychiatric epidemiology. The chapter gives an account of a cohort study that has evolved over a period of more than fifty years. Finally, it introduces a data set, the Prenatal Determinants of Schizophrenia Example (PDSE) study, for illustration in this and other parts of the book.

Environmental modifiers of the effects of genetic variants, or gene-environment interactions, have received increased attention in recent years due to the recognition that genetic variants alone are unlikely to explain most of the recent increases in chronic diseases. Such increases are more likely due to environmental and behavioral changes interacting with a genetic predisposition, suggesting that failing to identify and control environmental modifiers of disease risk could mask important associations with genetic variants or misestimate the magnitude of their effects. Identifying environmental modifiers of these variants may also be essential in mitigating the risk conferred by these variants. Population-based genetic association studies with detailed characterization of environmental exposures are critical and underused resources for identifying potential interacting factors. This chapter explores the substantial and complementary strengths offered by the two main approaches to these studies — case-control and cohort designs — in the search for the genetic and environmental influences on common diseases.

Cohort studies compare individuals exposed and unexposed to a hypothesized risk factor, and the respective proportions who develop the disease over a specified period of time. This chapter introduces the main elements of the cohort design. To facilitate exposition, it describes as a prototype a prospective population cohort study.

Cohort studies of occupational populations generally provide the most comprehensive approach for evaluating overall patterns of health and disease. This chapter describes the design features of prospective and historical cohort studies, cohort definition approaches and options, practical considerations for implementation, and the relative advantages and limitations of historical and prospective cohort designs. Also included are descriptions of some standard methods for analyzing cohort data, including overall standardized mortality (or morbidity) ratio (SMR) analysis, stratified analysis approaches, and subcohort analysis. Other topics addressed in this chapter are the choice of reference populations, treatment of missing exposure data, and disease latency analysis.

Epidemiology is defined by its objective: to study determinants of diseases and ill health in human populations. To do so, epidemiologists use experimental and non-experimental methods. This chapter provides guidelines for teaching the most common design used for epidemiology — the randomized controlled trial — other follow-up studies, case-control studies, and ecological designs.

This chapter explains the logic of the case-control strategy. It shows how to conceptualize a case-control study as a condensed version of a cohort study. This relationship between the two designs is well established in epidemiology. It is not familiar, however, to many researchers in other fields.

This chapter provides a framework for the prevention and control of bias due to unequal attrition in cohort studies. Like the effects of third-variable confounding, the effects of unequal attrition can be limited by the design, as well as the analysis, of a cohort study. The most problematic kind of attrition is differential. This kind of attrition causes a bias that may not be remediable. The chapter uses sensitivity analysis to gauge the potential bias.

This chapter focuses on genetic association studies. The goal of genetic association studies is to investigate the effects of genotypic variation on disease risk. These designs are built on a concept of a genetic cause that closely resembles the concept of a risk factor in epidemiology. Moreover, they correspond well to designs previously articulated by risk factor epidemiologists. Thus, the concepts and methods of risk factor epidemiology are particularly useful in this context.

This chapter examines the contributions of several large prospective studies conducted over the second half of the 20th century to our understanding of the health hazards of tobacco use. Beginning in the early 1950s, large cohort studies played a major role in helping to identify the multitude of adverse health effects caused by tobacco use, particularly manufactured cigarettes. They demonstrated that the harmful effects applied to women as well as men, that cigarettes with low machine-measure tar and nicotine were no less hazardous with respect to lung cancer than filter-tip ‘regular’ tar cigarettes, and that the burden of disease caused by smoking increased over time as smokers initiated regular cigarette smoking at progressively earlier ages. Large cohort studies will continue to be important for monitoring the course of the epidemic as it evolves in different cultures and for sustaining the political resolve to end it.

The choice of epidemiologic study design is largely dictated by the health outcomes of primary interest, feasibility, and availability of valid exposure data. This chapter provides brief descriptions of study design options, including historical and prospective cohort studies; case-control studies nested within defined occupational cohorts; population-based (registry-based) case-control studies; proportionate mortality studies; cross-sectional studies; and, repeated measures studies. The relative strengths and limitations of the various designs are discussed. In addition, the connections between study designs are explained in the context of the common goal of every epidemiologic study, which is to characterize the incidence of illness and injury, and their relations with exposures, in the source population. This chapter provides the background for more in-depth presentations of study designs given in Chapters 4–7.

This chapter describes different risk factor study designs. Study designs can be viewed as techniques for organizing observations in ways that can reveal the causes of disease. The risk factor study designs attempt to do this by mimicking the counterfactual, which is the true test of causal effects. When the mimicry is successful, studies isolate the effects of the exposure on the disease. In regard to this mimicry, the randomized controlled trial is the simplest and most direct, but practical for only a small number of exposures. The cohort study is much more widely applicable, but selection into exposure is not random. The case-control study is more complex and indirect but often the most useful. Although the specific techniques differ among designs, the underlying strategy is the same—the examination of fully comparable groups of exposed and unexposed individuals to isolate the effects of specific risk factors. The chapter illustrates how each study was used to investigate the causal effect of low prenatal folate intake on neural tube defects (NTDs).

A cohort study is always a kind of ‘thought experiment’. One needs to imagine what results ‘would have been’ if the exposed and the unexposed groups had in fact been fully comparable. This chapter sets up a framework for conducting this thought experiment in a systematic way. The exposed and unexposed may differ on (1) other causes of the disease (confounding); (2) the length of time over which they are observed (unequal attrition); and (3) the accuracy with which cases are ascertained (differential misclassification). Each of these factors may account for part, or even all, of an apparent association between the exposure and the disease. They can also suppress an association or reverse it. The process of causal inference depends upon the identification and evaluation of these alternative explanations for results. This chapter shows how confounding, unequal attrition, and differential misclassification can produce an association between exposure and disease.

This chapter discusses cohort studies. Topics covered include cohort identification, methods of ascertainment of exposure status, estimating the expected occurrence of disease among “exposed” cohort members, follow-up of cohort members, nature of the illness outcome, and issues in the analysis and interpretation of cohort studies. Exercises are provided at the end of the chapter.

This chapter addresses misclassification of the disease that is linked to the exposure, that is, misclassification of disease that is unequal for the exposed and the unexposed groups. This is the type of misclassification most relevant to cohort studies. It begins by describing common scenarios of differential misclassification in psychiatric research. It then considers the resulting bias and how this can be limited. The chapter also discusses two special cases: third-variable effects and dependent nondifferential misclassification.

This chapter addresses the principles of research study design. It provides an in-depth description of commonly used study designs to address research objectives. It presents information about study designs for descriptive epidemiology (i.e., studies of disease incidence, prevalence, and mortality rates) and analytic epidemiology (i.e., randomized trials, prospective cohort studies, retrospective cohort studies, cross-sectional studies, case-control studies). It describes the measures of effect that reflect the associations between health predictors and neurologic outcomes (odds ratios, relative risks). The second part of the chapter describes common study biases that can adversely affect study validity (sampling error, selection bias, confounding) and how to avoid these biases in the design and conduct of clinical and epidemiological studies.

This chapter focuses on observational methods in epidemiology. It presents a fourfold table that serves in field studies to test the precision of measures, either of exposure or of outcome, and in public health applications in screening for affected individuals. It discusses cohort studies and case-control studies.

Epidemiologic studies make comparisons between two or more groups to draw inferences about possible causal relations between some attribute that may affect health, generically termed exposure, and some health outcome or state, generically termed disease. This chapter discusses study designs, the purpose of comparison groups, selection bias and confounding, evaluation of selection bias in cohort studies, and integrated assessment of potential for selection bias in cohort studies.

Many life course hypotheses can only be addressed using observational studies which follow the same individuals through time, and a birth cohort is perhaps the ideal study in which to investigate life course influences on ageing. Careful consideration of study design is vital to the delivery of good quality research on healthy ageing. The design of a study not only determines the research questions that can be addressed, but also various sources of error and bias. This chapter reviews approaches to data collection, highlighting the importance of mode and frequency of data collection and discusses the future opportunities afforded by innovations in data collection technology and record linkage. It argues that strong study management, a consideration of ethical concerns, and engagement of study participants are vital to the delivery of high quality data and maintenance of response rates. With increasing cross-cohort collaboration, the current resource of studies provides exciting opportunities.