Chapter 6 - Introduction to epidemiological methods for studying effects of exposure to pesticides

In this chapter, we review the basics of epidemiological concepts, necessary for understanding more advanced issues. We start by defining the key variables of epidemiological studies. Then, we sketch the three classical study designs (cohort, case-control, and cross-sectional), while describing an alternative classification. The meaning of measures of disease occurrence (prevalence, risk, and incidence rate) and their relationships are illustrated in an intuitive, nonstatistical way. Then, we show the respective relevance of absolute and relative measures of effect, calculated as differences or ratios of measures of disease occurrence, respectively. In study design and conduction, errors may be made that could affect study validity and precision. Regarding validity, systematic errors (selection, information, and confounding biases) are reviewed. Classical methods to control confounding in the study design phase or in the statistical analysis phase (standardization, stratification, and regression models) are illustrated, with also a gentle introduction to new advanced topics, including directed acyclic graphs and g methods. Random error is presented in the context of precision. Finally the role of statistical analysis in epidemiology is addressed, with emphasis on abuse and misuse of hypothesis testing (P-values and statistical significance); instead, presentation of absolute and/or relative effect estimates and their confidence intervals is recommended.