Modern Epidemiologic Principles and Concepts: Guidelines for Study Conceptualization, Design, Conduct and Application

We often conceive epidemiology in either simplistic or complex terms, and neither of these is accurate. To illustrate this, the complexities in epidemiology could be achieved by considering a study to determine the correlation between serum lipid profile as total cholesterol, HDL, LDL, triglyceride, and total body fatness or obesity measured by BMI in children. Two laboratories measured serum lipid profiles, and one observed a correlation with BMI, while the other did not. Which is the reliable finding? To address this question, one needs to examine the context of blood drawing since fasting blood level may provide a better indicator of serum lipid. Epidemiologic studies could be easily derailed given the inability to identify and address possible confounding. Therefore, understanding the principles and concepts used in epidemiologic studies designed and conducted to answer clinical research questions facilitates e accurate and reliable findings in these areas. Another similar example in a health fair setting involves geography and health, termed health-o-graphy. The risk of dying in one zip code A was 59.5 per 100,000, and in the other zip code B was 35.4 per 100,000. There is a common sense and non-epidemiologic tendency to conclude that there is an increased risk of dying in zip code A. To arrive at such inference, one must first find out the age distribution of these two zip codes since advancing age is associated with increased mortality. Indeed, zip code A is comparable to the United States population while, zip code B is the Mexican population. These two examples are indicative of the need to understand epidemiologic concepts such as confounding by age or effect measure modification prior to undertaking clinical research.

This textbook describes the basics of research in medical and clinical settings, as well as the concepts and application of epidemiologic designs in research. Design transcends statistical techniques, and no matter how sophisticated statistical modeling, errors of design/sampling cannot be corrected. The author of this textbook has presented a complex field in a very simplified and reader-friendly manner with the intent that such a presentation will facilitate the understanding of the design process and epidemiologic thinking in clinical research. Additionally, this book provides a very basic explanation of how to examine the data collected for research conduct for the possibility of confounders and how to address such confounders, thus disentangling such effects for reliable and valid inference. 

Research is presented as an exercise around measurement, with measurement error inevitable in its conduct, hence the inherent uncertainties of all findings in clinical and medical research. Modern Epidemiologic Principles and Concepts for Clinicians covers research conceptualization, namely research objectives, questions, hypothesis, design, implementation, data collection, analysis, results, and interpretation. While the primary focus of epidemiology is to assess the relationship between exposure (risk or predisposing factor) and outcome (disease or health-related event), the causal association is presented in a simplified manner, including the role of quantitative evidence synthesis (QES) in causal inference. Epidemiology has evolved over the past three decades, resulting in several fields being developed. This text presents, in brief, the perspectives and future of epidemiology in the era of the molecular basis of medicine, “big data,” “3Ts,” and systems science.

Epidemiologic evidence is more reliable if conceptualized and conducted within the context of translational, transdisciplinary, and team science. With molecular epidemiology, we are better equipped with tools to identify molecular biologic indicators of risk as well as biologic alterations in the early stages of disease, and with 3 Ts and systems science, we are more capable of providing accurate and reliable inference on causality and outcomes research. Further, the author argues that unless sampling error and confounding are identified and addressed, clinical research findings will remain largely inconsistent, implying an inconsequential epidemiologic approach.  

Appropriate knowledge of research conceptualization, design, and statistical inference is essential for conducting clinical and biomedical research. This knowledge is acquired through the understanding of epidemiologic/observational (non-experimental) and experimental designs and the choice of the appropriate test statistic for statistical inference. However, regardless of how sophisticated the statistical technique employed for statistical inference is, study conceptualization and design are the building blocks of valid scientific evidence. Since clinical research is performed to improve patients’ care, it remains relevant to assess not only the statistical significance but the clinical and biologic importance of the findings, for clinical decision-making in the care of an individual patient. Therefore, the aim of this book is to provide clinicians, biomedical researchers, graduate students in research methodology, students of public health, and all those involved in clinical/biomedical research with a simplified but concise overview of the principles and practice of epidemiology. In addition, the author stresses common flaws in the conduct, analysis, and interpretation of epidemiologic studies.

Valid and reliable scientific research is that which considers the following elements in arriving at the truth from the data, namely biological relevance, clinical importance, and statistical stability and precision (statistical inference based on the p-value and the 90, 95, and 99 percent confidence interval). The interpretation of results of new research must rely on factual association or effect and the alternative explanation, namely systematic error, random error (precision), confounding, and effect measure modifier. Therefore, unless these perspectives are disentangled, the results from any given research cannot be considered reliable. However, even with this disentanglement, all study findings remain inconclusive with some degree of uncertainty.

This book presents a comprehensive guide on how to conduct clinical and medical research—mainly research question formulation, study implementation, hypothesis testing using appropriate test statistics to analyze the data, and results interpretation. In so doing, it attempts to illustrate the basic concepts used in study conceptualization, epidemiologic design, and appropriate test statistics for statistical inference from the data. Therefore, though statistical inference is emphasized throughout the presentation in this text, equal emphasis is placed on clinical relevance or importance and biological relevance in the interpretation of the study results.

Specifically, this book describes in basic terms and concepts how to conduct clinical and medical research using epidemiologic designs. The author presents epidemiology as the main profession in the trans-disciplinary approach to the understanding of complex ecologic models of disease and health. Clinicians, even those without preliminary or infantile knowledge of epidemiologic designs, could benefit immensely from what, when, where, who, and how studies are conceptualized, data collected as planned with the scale of measurement of the outcome and independent variables, data edited, cleaned and processed prior to analysis, appropriate analysis based on statistical assumptions and rationale, results tabulation for scientific appraisal, results interpretation and inference. Unlike most epidemiologic texts, this is the first book that attempts to simplify complex epidemiologic methods for users of epidemiologic research, namely clinicians and allied health researchers. Additionally, it is rare to find a book with integrates of basic research methodology into epidemiologic designs. Finally, research innovation and the current challenges of epidemiology are presented in this book to reflect the currency of the materials and the approach, as well as the responses to the challenges of epidemiology today namely, “big data”, accountability, and policy.

A study could be statistically significant but biologically and clinically irrelevant since the statistical stability of a study does not rule out bias and confounding. The p-value is deemphasized, while the use of effect size or magnitude and confidence intervals in the interpretation of results for application in clinical decision-making is recommended. The use of p-value could lead to an erroneous interpretation of the effectiveness of treatment. For example, studies with large sample sizes and very little or insignificant effects of no clinical importance may be statistically significant, while studies with small samples though a large magnitude of effects are labeled “negative result.” Such results are due to low statistical power and increasing variability, hence the inability to pass the arbitrary litmus test of the 5 percent significance level.  

Epidemiology Conceptualized

Epidemiologic investigation and practice are as old as the history of modern medicine. It dates back to Hippocrates (circa 2,400 years ago). In recommending the appropriate practice of medicine, Hippocrates appealed to the physicians’ ability to understand the role of environmental factors in predisposition to disease and health in the community. During the Middle Ages and the Renaissance, epidemiologic principles continued to influence the practice of medicine, as demonstrated in De Morbis Artificum (1713) by Ramazinni and the works on scrotal cancer in relation to chimney sweeps by Percival Pott in 1775.

With the works of John Snow, a British physician (1854), on cholera mortality in London, the era of scientific epidemiology began. By examining the distribution/pattern of mortality and cholera in London, Snow postulated that cholera was caused by contaminated water.

 Epidemiology Today – Epigenomic Epidemiology  

There are several definitions of epidemiology, but a practical definition is necessary for the understanding of this science and art. Epidemiology is the basic science of public health. The objective of this profession is to assess the distribution and determinants of disease, disabilities, injuries, natural disasters (tsunamis, hurricanes, tornados, and earthquakes), and health-related events at the population level. Epidemiologic investigation or research focuses on a specific population. The basic issue is to assess the groups of people at higher risk: women, children, men, pregnant women, teenagers, whites, African Americans, Hispanics, Asians, poor, affluent, gay, lesbians, married, single, older individuals, etc. Epidemiology also examines how the frequency of the disease or the event of interest changes over time. In addition, epidemiology examines the variation of the disease of interest from place to place. Simply, descriptive epidemiology attempts to address the distribution of disease with respect to “who,” “when,” and “where.” For example, cancer epidemiologists attempt to describe the occurrence of prostate cancer by observing the differences in populations by age, socioeconomic status, occupation, geographic locale, race/ethnicity, etc.

Epidemiology also attempts to address the association between the disease and exposure. For example, why are some men at high risk for prostate cancer? Does race/ethnicity increase the risk for prostate cancer? Simply, is the association causal or spurious? This process involves the effort to determine whether a factor (exposure) is associated with the disease (outcome). In the example of prostate cancer, such exposure includes a high-fat diet, race/ethnicity, advancing age, pesticides, family history of prostate cancer, and so on. Whether or not the association is factual or a result of chance remains the focus of epidemiologic research. The questions to be raised are as follows: Is prostate cancer associated with pesticides? Does pesticide cause prostate cancer?

Epidemiology often goes beyond disease-exposure association or relationship to establish a causal association. In this process of causal inference, it depends on certain criteria, one of which is the strength or magnitude of association, leading to the recommendation of preventive measures. However, complete knowledge of the causal mechanism is not necessary prior to preventive measures for disease control. Further, findings from epidemiologic research facilitate the prioritization of health issues and the development and implementation of intervention programs for disease control and health promotion.

Epidemiology today reflects the application of gene and environment interaction in disease causation, morbidity, prognosis, survival, and mortality in subpopulation health outcomes. The knowledge and understanding of subpopulation differentials in DNA methylation of specific genes and histone modification allows for the application of abnormal transcriptomes, impaired gene expression, protein synthesis dysfunctionality, and abnormal cellular functionality.

This book is conceptually organized into three sections. Section I deals with research methods, section II epidemiologic designs, as well as causal inference and perspectives in epidemiology, while section III delves into perspectives, epidemiologic challenges, and special topics in epidemiology, namely epidemiologic tree, challenges, emerging fields, the consequentialist perspective of epidemiology and epidemiologic role in health and healthcare policy formulation, as well as epigenomic epidemiology and epigenomic determinants of health (EDH). Throughout this book, attempts are made to describe the research methods and non-experimental as well as experimental designs. Section I comprises research methods with an attempt to describe the following: Research objectives and purposes, Research questions, Hypothesis statements: null and alternative, Rationales for research, clinical reasoning, and diagnostic tests, as well as Study conceptualization and conduct—research question, data collection, data management, hypothesis testing, data analysis.

Section II comprises the epidemiologic study designs with an attempt to describe the basic notion of epidemiology and the designs used in clinical research:  The notion of epidemiology and the measures of disease occurrence and frequency and the measure of disease association, Ecologic and cross-sectional designs, Case-control studies, Cohort studies: prospective, retrospective, and am bidirectional, Clinical trials or experimental designs, and, Quantitative evidence synthesis (QES), systematic review, scientific study appraisal, and causal inference.

Section III consists of perspectives, challenges, and special topics in epidemiology to illustrate the purposive role of epidemiology in facilitating the goal of public health, mainly disease control and health promotion. Additionally, this section presents the integrative dimension of epidemiology as well as novel epidemiology as epigenomic epidemiology: Epidemiologic perspectives: advances, challenges, emerging fields and the future, Consequentialism epidemiology, and Role of epidemiology in health and healthcare policy formulation. Specifically, this section addresses the gene and environment interaction in disease causation, prognosis, and survival.

Significantly, section I chapters deals with the basic descriptions of scientific research at the clinical and population levels and how the knowledge gained from the population could be applied to the understanding of individual patients in the future. In these two chapters, an attempt is made to discuss clinical reasoning and the use of diagnostic tests (sensitivity and specificity) in clinical decision-making. The notions, numbers needed to treat (NNT), and numbers needed to harm (NNH) are discussed later in the chapter on causal inference. The last chapter in this section delves into clinical research conceptualization, design involving subject recruitment, variable ascertainment, data collection, data management, data analysis, and the outline of the research proposal.

In section II, epidemiologic principles and methods are presented with the intent to stress the importance of careful design in conducting clinical and biomedical research. Epidemiology remains the basic science of clinical medicine and public health that deals with disease, disabilities, injury, and health-related events distributions and determinants and the application of this knowledge to the control and prevention of disease, disabilities, injuries, and related health events at the population level. Depending on the research question and whether or not the outcome (disease or event of interest) has occurred prior to the commencement of the study or if the investigator assigns subjects to treatment or control, an appropriate design is selected for the clinical research. The measures of effects or point estimates are discussed with concrete examples to illustrate the application of epidemiologic principles in arriving at a reliable and valid result. Designs are illustrated with flow charts, figures, and boxes for distinctions and similarities. The hierarchy of study design is demonstrated with randomized clinical trials (RCT) and the associated Meta-Analysis and quantitative evidence synthesis as the design that yields the most reliable and valid evidence from data. Though RCTs are considered the “gold standard” of clinical research, it is sometimes not feasible to use this design because of ethical considerations, hence the alternative need for prospective cohort design.

Interpreting research findings is equally as essential as conducting the study itself. Interpretation of research findings must be informative and constructive in order to identify future research needs. A research result cannot be considered valid unless we disentangle the role of bias and confounding from a statistically significant finding, as a result, can be statistically significant and yet driven by measurement, selection, and information bias as well as confounding. While my background in basic medical sciences and clinical medicine (internal medicine) allows me to appreciate the importance of biologic and clinical relevance in the interpretation of research findings, biostatisticians without similar training must look beyond random variation (p-value and confidence interval) in the interpretation and utilization of clinical research findings. Therefore, quantifying the random error with a p-value (a meaningful null hypothesis with a strong case against the null hypothesis requires the use of a significance level) without a confidence interval deprives the reader of the ability to assess the clinical importance of the range of values in the interval. Using Fisher’s arbitrary p-value cutoff point for type I error (alpha level) tolerance, a p-value of 0.05 need not provide strong evidence against the null hypothesis, but p less than 0.0001 does.[i] The precise p-value should be presented without reference to arbitrary thresholds. Therefore, results of clinical and biomedical research should not be presented as “significant” or “non-significant” but should be interpreted in the context of the type of study and other available evidence. Secondly, systematic error and confounding should always be considered for findings with low p-values, as well as the potential for effect measure modifiers (if any) in the explanation of the results. Neyman and Pearson describe their accurate observation:

No test based upon a theory of probability can by itself provide any valuable evidence of the truth or falsehood of a hypothesis. But we may look at the purpose of tests from another viewpoint. Without hoping to know whether each separate hypothesis is true or false, we may search for rules to govern our behavior with regard to them, in following which we ensure that, in the long run of experience, we shall not often be wrong.

This text is expected to provide practical knowledge to clinicians, biomedical researchers, and public health scientists, implying all researchers use biological and biochemical specimens or samples, in an attempt to understand health and disease processes at cellular, clinical, and population levels. Additionally, all those who translate such data from bench to clinics in an attempt to improve the health and well-being of the patients seen by healthcare providers.

Further, this book describes in basic terms and concepts how to conduct clinical and biomedical research using epidemiologic designs. The author presents epidemiology as the main discipline, so to speak, in the trans-disciplinary approach to the understanding of complex ecologic models of disease and health. Clinicians, even those without preliminary or infantile knowledge of epidemiologic designs, could benefit immensely from what, when, where, who, and how studies are conceptualized, data collected as planned with the scale of measurement of the outcome and independent variables, data edited, cleaned and processed prior to analysis, appropriate analysis based on statistical assumptions and rationale, results tabulation for scientific appraisal, results interpretation and inference. Unlike most epidemiologic texts, this is one of the few books that attempts to simplify complex epidemiologic methods for users of epidemiologic research, namely clinicians. Additionally, it is extremely rare to access a book with an integration of basic research methodology into epidemiologic designs. Finally, research innovation and the current challenges of epidemiology are presented in this book to reflect the currency of the materials and the approach.