Essential epidemiology concepts for public health certification exams

Introduction to applied epidemiology in public health

Epidemiology is the quantitative foundation of public health practice, focusing on the distribution and determinants of health-related states or events in specified populations. For professionals preparing for public health certification exams, such as the Certified in Public Health (CPH) credential—a key part of professional certification for career advancement—mastering applied epidemiology is essential. Public health practice relies on understanding how diseases spread, how risk is measured, and how data can be systematically collected and analyzed to design intervention programs. Whether working at local, state, or federal public health agencies, practitioners must be able to use public health data to create strategies for disease prevention and health promotion.

Epidemiological inquiry centers on the concept of a population, defined as a group of people sharing common characteristics such as age, gender, geographic residence, or life events. Populations can be fixed (or closed), meaning membership is permanent and defined by a specific life event, such as passengers on a particular flight or survivors of a specific historical incident. Populations can also be dynamic (or open), characterized by continuous migration where individuals move in and out over time, such as the residents of a major metropolitan city. Understanding population dynamics is important because all epidemiological calculations—including rates, proportions, and ratios—depend on accurately defining the denominator population at risk.

The goal of studying disease distribution within these populations is to establish evidence-based control measures. Certification exams test a candidate's ability to distinguish between various basic mathematical expressions of frequency, such as counts, ratios, proportions, and rates according to CPH exam prep. A count represents the raw number of cases of a condition and is the starting point during novel outbreak investigations where public health officials first tally infected individuals. Ratios divide one quantity by another where the numerator may or may not be part of the denominator, while proportions represent parts of a whole where the numerator is strictly included within the denominator. Rates incorporate an explicit time dimension, measuring the speed at which events occur within a population at risk.

Measures of morbidity, mortality, and disease frequency

Quantifying disease frequency requires distinguishing between prevalence and incidence. Prevalence measures the proportion of existing disease cases—both old and new—in a population at a given point in time, providing a snapshot of disease burden. It is influenced by both the rate at which new cases occur and the duration of the disease, summarized by the relationship that prevalence equals incidence multiplied by duration as described in basic infectious disease concepts. Because prevalence captures all existing cases, it is heavily impacted by factors that prolong survival or cure individuals, making it useful for resource allocation in chronic disease management rather than for establishing acute causal pathways.

Incidence measures the occurrence of new cases of disease in a population at risk over a specified time period, capturing the force of morbidity and the conversion from healthy to diseased. Incidence can be expressed either as an incidence proportion (cumulative incidence or risk) or an incidence rate (incidence density). Incidence proportion calculates the probability or risk that disease-free individuals will develop the disease over a specified observation period, assuming complete follow-up of all participants. When individual follow-up times vary or participants are lost to follow-up, epidemiologists use incidence rate, which divides new cases by total person-time at risk. Person-time accounts for the exact duration each individual remains under observation and disease-free before either developing the outcome, dying, or withdrawing from the study.

Mortality measures follow similar computational logic but focus on the occurrence of death within a population, reflecting disease severity and healthcare effectiveness. Crude mortality rates encompass all deaths divided by the total population at the midpoint of a specified time period per CDC standards, though they can be misleading when comparing populations with different age structures. To resolve this, epidemiologists use age-specific mortality rates, cause-specific mortality rates, and standardized methods like direct and indirect standardization. Additionally, the case fatality rate calculates the proportion of individuals diagnosed with a specific disease who die from it within a defined time frame, which indicates acute disease severity and the efficacy of therapeutic interventions.

Epidemiological study designs: experimental and observational

Selecting the appropriate study design is a key competency tested on public health certification exams according to CPH review sessions. Epidemiological studies are divided into experimental and observational categories. Experimental studies, most notably randomized controlled trials (RCTs), involve active investigator intervention, where participants are randomly assigned to receive either a novel therapeutic agent, a standard control treatment, or a placebo. Randomization ensures that potential confounding variables are distributed evenly between study arms, allowing researchers to attribute differences in outcomes to the intervention. Blinding techniques—single, double, or triple—are frequently used in clinical trials to eliminate subjective assessment bias from patients, clinicians, and data analysts.

When experimental interventions are unfeasible or unethical, public health professionals rely on observational study designs, beginning with cohort studies. Cohort studies classify participants based on their exposure status (exposed versus unexposed) and follow them forward in time to assess the incidence of disease outcomes. Because exposure is documented prior to the development of disease, cohort studies establish a clear temporal sequence and are efficient for investigating rare exposures, such as specific occupational hazards. They also permit the simultaneous evaluation of multiple health outcomes resulting from a single exposure. However, prospective cohort studies can be expensive, time-consuming, and sensitive to participant loss to follow-up, while historical (retrospective) cohort studies rely entirely on pre-existing records and require knowing how to analyze historical disease case studies for research papers.

In contrast, case-control studies begin with the disease outcome rather than the exposure, making them well-suited for rare diseases or conditions with long latency periods. Researchers recruit individuals with the disease (cases) and individuals without the disease (controls), then look backward in time to compare the past proportions of exposure between the two groups. While case-control studies are cost-effective, they are susceptible to selection biases regarding control group selection and recall biases during retrospective exposure assessment. Other key observational designs include cross-sectional (prevalence) studies, which measure exposure and outcome simultaneously, and ecological studies, which use community-level or population-level data to explore broad public health correlations.

Measures of association and effect: risk ratios and odds ratios

Interpreting the mathematical relationship between exposures and health outcomes requires specialized measures of association, primarily relative risk (risk ratio) and the odds ratio. Relative risk is calculated by dividing the incidence rate or proportion in the exposed group by the incidence rate or proportion in the unexposed group. A relative risk of 1.0 indicates no association between exposure and disease, while values greater than 1.0 suggest a harmful exposure that increases risk, and values less than 1.0 point toward a protective factor. Because relative risk requires true incidence data derived from populations with complete follow-up, it is the standard measure of association used in cohort and experimental studies.

For case-control studies, true population incidence is typically unknown because the total number of cases and controls is determined by the investigator. In these cases, the odds ratio is the appropriate substitute. The odds ratio measures the ratio of the odds of exposure among cases to the odds of exposure among controls. In mathematical terms derived from a standard two-by-two contingency table, the odds ratio is calculated as cross-products ($ad/bc$). When studying a rare disease, the odds ratio approximates the relative risk, allowing public health professionals to draw valid etiological inferences from case-control data despite the absence of direct incidence measures.

Beyond relative measures of association that evaluate etiology, absolute measures of effect quantify the public health impact of removing an exposure. The risk difference (attributable risk) calculates the excess risk of disease in the exposed group compared to the unexposed group; this represents the amount of disease incidence that could be eliminated if the exposure were removed. Attributable risk percent and population attributable risk percent scale these absolute differences to proportions, helping public health administrators determine whether limited resources should be allocated toward controlling a specific hazardous exposure or addressing broader population-level determinants of health.

Causation, validity, and error in public health research

Determining whether an observed statistical association represents a true causal relationship is a complex task in epidemiological practice. A causal factor must satisfy strict criteria, most importantly temporal sequence, which mandates that the exposure must precede the outcome in time. To guide causal inference in non-infectious chronic diseases, public health researchers rely on Sir Austin Hill’s criteria, which include the strength of the association, consistency across diverse study populations, biological gradient (dose-response relationships), biological plausibility, and experimental evidence. Complementing Hill's criteria, Kenneth Rothman’s sufficient-component cause model conceptualizes disease development through "sufficient causes"—complete sets of minimal component factors that result in disease—and "necessary causes," which must be present for a disease to occur.

Even with meticulous planning, observational and experimental studies are vulnerable to various forms of error that threaten internal validity, which reflects the degree to which a study accurately portrays the true state of the study population. Random error, or chance, results from natural biological and sampling variability and can be minimized primarily by increasing sample sizes and improving measurement precision. Systematic error, or bias, presents a more severe threat because it introduces flawed estimates of association that can erroneously mask true associations or manufacture false ones. Selection bias arises when the procedures used to select study participants lead to systematic differences between those included and those excluded, violating independence between exposure and outcome.

Observation or information bias occurs after participants are enrolled and stems from systematic inaccuracies in how exposure or disease data are collected. Examples include recall bias, where participants with a disease remember past exposures differently than healthy controls, and interviewer bias, where data collectors probe or record responses differently based on knowledge of participant disease status. Misclassification bias further distorts findings when subjects are incorrectly categorized; non-differential misclassification occurs independently of study groups and usually biases relative measures toward the null, whereas differential misclassification varies across groups and skews risk estimates. Finally, confounding occurs when an extraneous third variable—associated with both the exposure and independently acting as a risk factor for the disease—distorts the true effect. Confounding can be controlled during study design via randomization, restriction, and matching, or managed during analysis using stratification, standardization, and multivariate regression models.

References

• CDC Stacks
• Basic Infectious Disease Concepts in Epidemiology
• CPH Review Session - Epidemiology
• Cph certification in public health exam prep 2025-2026