Causal Impact

In Chapter 9, the author discusses the causal impact framework, an approach to epidemiologic methods that can be used to move from internally valid estimates to externally valid estimates to valid estimates of the effects of population interventions. Such work is essential if epidemiologists want the results of their experimental or observational studies to directly inform public health policy decisions. The bulk of this chapter outlines and describes an approach to epidemiologic methods relevant to implementation science. The author summarizes the entire book by briefly addressing the lessons of the previous chapters for the so-called hierarchy of evidence (hierarchy of study designs).

Case-Control Studies

In contrast to an observational cohort study in which participants are identified, exposures are measured, and then outcomes status is measured after follow-up, a case-control study is an observational study in which researchers sample participants based on their outcome status, often only after all outcomes have already occurred. This chapter echoes the structure of the previous two chapters in the discussion of case-control studies. In this chapter, the author’s focus is on understanding the relationship between cohort studies and case-control studies and on how the interpretation of the odds ratio estimated from the case-control study depends on the relationship of the case-control study to a cohort study.

Observational Cohort Studies

In contrast to a randomized trial, an observational cohort study is one in which the investigator observes a group of participants with varying levels of an exposure and then follows-up those participants for a period of time to examine the incidence of one or more specified outcomes. This chapter addresses observational cohort studies in much the same way as the previous chapter addressed randomized trials: discussed are the types of cohort studies, the steps in conducting such a study, and the ways in which such studies meet or do not meet causal identification conditions. Also presented is a brief introduction to analysis. The author expands his discussion of interaction and effect measure modification, as well as generalizability, in this setting.

Randomized Trials

In Chapter 5, the author describes randomized trials. The chapter gives a broad overview of types of trials and the steps in conducting a trial and also describes how trials meet (and fail to meet) core causal identification conditions. The author provides a brief introduction to the analysis of randomized trial data. As well, the chapter introduces factorial trials as well as subgroup analysis of trials as a way of explaining differences between causal interaction and effect measure modification. Finally, the author describes issues in the generalizability and transportability of trials and quantitative approaches to these issues.

Diagnostic Testing, Screening, and Surveillance

Surveillance, diagnostic testing, and screening are central (and related) concepts in epidemiology, and are all addressed together in this chapter. Both description and prediction are also important goals of epidemiology, and these differ in important ways from causal epidemiology. In Chapter 4, the author discusses concepts in diagnostic testing, screening, and active and passive disease surveillance, including concepts of sensitivity, specificity, and positive and negative predictive value. Also in this chapter, the author briefly touches on differences between clinical epidemiology and public health epidemiology. The chapter also briefly discusses very basic bias, or sensitivity, analysis.

Other Study Designs

In Chapter 8, the author discusses several other key study designs, including some of the more “traditional” epidemiologic designs, among them case reports and series, case-crossover studies, and cross-sectional studies, as well as several “hybrid” designs that combine aspects of randomized trials with aspects of observational studies—including systematic reviews and meta-analyses, which are observational studies of study results, often including randomized trial results. Also discussed are quasi-experiments, which are themselves a sort of hybrid in which observational data yields conditions close to what we might see in a randomized experiment. The author’s goal for the chapter is to familiarize the reader with general concepts to form a basis for deciding whether such designs may be worth more of study later.

Causal Inference, Causal Effect Estimation, and Systematic Error

Chapter 3 discusses basic concepts in causal inference, beginning with an introduction to potential outcomes and definitions of causal contrasts (or causal estimates of effect), concepts, terms, and notation. Many concepts introduced here will be developed further in subsequent chapters. The author discusses sufficient conditions for estimation of causal effects (which are sometimes called causal identification conditions), causal directed acyclic graphs (sometimes called causal diagrams), and four key types of systematic error (confounding bias, missing data bias, selection bias, and measurement error/information bias). The author also briefly discusses alternative approaches to causal inference.

Measures of Association

This chapter discusses the measures of contrast between two groups within the study population. Whereas in Chapter 1 the researcher might describe the total number of cases of a disease in a large population as a whole, in this chapter the researcher is interested in (for example) contrasting risk among those exposed to a drug and those unexposed to that drug within a larger population. In this chapter, the author primarily focuses on difference and ratio measures. This chapter introduces the 2 × 2 table, a widely used tool for learning epidemiologic methods. The author also discusses how to communicate these findings for researchers, policy makers, clinicians, and patients, all of whom may need to make decisions based on these data.

Measuring Disease

The first task of epidemiology is to understand in some depth the concepts of prevalence and incidence, how to quantify them, and key types of error that can affect the measurements of each. In Chapter 1, the author describes prevalence and incidence in single samples (a single population), as well as how to quantify these measures. The chapter will focus on the survival curve as the central measure of incidence of disease over time in a population and then describe how simpler measures such as the incidence proportion (that is, the risk), incidence rate, incidence odds, and measures of time can be derived from the survival curve.

Overview

The study of epidemiology is about learning to ask, and answer, good questions in population health. The first section of the book lays the groundwork for understanding how study designs work, what they estimate, and how they can fail. These chapters give an overview of prevalence and incidence, measures of contrast such as risk differences and risk ratios, principles of causal inference and causal effect estimation, diagnostic testing, screening, and surveillance. The second section of the book builds on the core concepts of measuring disease and assessing causality to describe the study designs that are the core tools of epidemiology. This section focuses on randomized trials, observational cohort studies, and case-control studies; the author briefly addresses additional study designs including quasi-experiments. The final section discusses the causal impact approach to epidemiologic methods for moving from internally valid estimates to externally valid estimates to valid estimates of the effects of population interventions