Public Health & Epidemiology

Public Health Foundations

• Public health — organized efforts to prevent disease, prolong life, and improve health at the population level. Its tools include surveillance, policy, environmental protection, health promotion, and delivery of preventive services.
• Population health — the distribution of health outcomes within a defined population and the factors that shape that distribution. It concerns both average health and inequalities between groups.
• Epidemiology — the study of the distribution and determinants of health-related states or events in specified populations and the application of that study to health problems.
• Biostatistics — statistical reasoning and methods applied to biological, clinical, and public-health data. It supports design, measurement, inference, prediction, and communication of uncertainty.
• Social determinants of health — social, economic, political, and environmental conditions that influence exposure, vulnerability, resources, and access to care across the life course.
• Health inequity — a systematic difference in health judged to be avoidable and unjust. Description of a disparity alone does not establish its cause or identify the intervention that will reduce it.
• One Health — a collaborative approach recognizing connections among human, animal, and environmental health, especially for zoonoses, antimicrobial resistance, food systems, and ecosystem change.

• Health in all policies — evaluates health consequences across sectors such as housing, transportation, education, agriculture, labor, and climate rather than treating health as the concern of medical care alone.

• Essential public-health services — a framework for assessing population health, investigating hazards, communicating, building partnerships, creating policy, assuring equitable access, maintaining infrastructure, and improving performance.
• Health equity — a condition in which everyone has a fair opportunity to attain health and no group is systematically disadvantaged by avoidable social arrangements.
• Population strategy of prevention — shifts the distribution of risk across a whole population. A small average change can prevent more cases than an intensive intervention limited to a small high-risk group.
• Health literacy — the ability of people and organizations to find, understand, evaluate, and use health information and services. It depends on communication and system design, not individual skill alone.
• Community health assessment — combines quantitative data, lived experience, assets, needs, and stakeholder priorities to describe local health and select action areas.
• Planetary health — studies how human health depends on natural systems and how environmental change driven by human activity feeds back into health, livelihoods, and inequality.

Measuring Disease and Health

• Case definition — a standard set of clinical, laboratory, person, place, and time criteria used to classify cases for surveillance or investigation. Surveillance definitions prioritize consistency and need not equal clinical diagnostic rules.
• Count, proportion, ratio, and rate — a count is a number of events; a proportion has a numerator contained in its denominator; a ratio compares quantities; a rate incorporates population at risk and, in a true incidence rate, person-time.
• Incidence proportion — the proportion of an initially disease-free population that develops an outcome during a specified period. It is also called cumulative incidence or risk.
• Incidence rate — new events divided by accumulated person-time at risk. Its unit is events per unit of person-time rather than a probability.
• Prevalence — the proportion of a population with a condition at a point or over a period. It reflects both incidence and duration and can rise when survival improves even if incidence does not.
• Mortality rate — deaths in a defined population over time. Cause-specific rates use deaths from a stated cause in the numerator; case fatality compares deaths among diagnosed cases.
• Risk difference — risk in one group minus risk in another. It measures absolute excess or reduction and is often central to population impact.
• Risk ratio — risk in an exposed or treated group divided by risk in a comparison group. A relative effect must be interpreted alongside baseline risk.
• Odds ratio — the ratio of two odds. It equals the exposure-outcome cross-product ratio in a two-by-two table and can approximate a risk ratio when the outcome is rare under suitable sampling.
• Rate ratio — incidence rate in one group divided by the rate in another. It compares event occurrence per person-time.
• Attributable fraction — estimates the proportion of events among an exposed group, or in a population, associated with an exposure under causal assumptions. It combines effect and exposure prevalence.
• Age standardization — applies age-specific rates to a common reference population so groups with different age structures can be compared. Standardized rates are comparative summaries, not observed crude rates.
• Life expectancy — the mean remaining years implied by a life table’s age-specific mortality rates. Period life expectancy applies current rates hypothetically and is not a forecast for an actual birth cohort.

• Disability-adjusted life year — a population burden measure combining years of life lost and years lived with disability. Its weights and assumptions make it a policy metric, not a direct biological quantity.

• Person-time — sum of time each eligible person remains under observation and at risk. It accommodates unequal follow-up but does not describe how risk changes within that time.
• Crude rate — the observed event rate in an entire population without standardization. It represents actual population burden but may be misleading for comparisons when group compositions differ.
• Standardized mortality ratio — observed deaths divided by deaths expected after applying reference rates to the study population’s strata. It compares with a reference but is not directly comparable across differently structured study populations.
• Survival function — probability of remaining free of the event beyond a given time. Censoring methods require assumptions about why follow-up ended.
• Hazard — instantaneous event rate among those still event-free at a given time. A hazard ratio is a relative rate, not a ratio of cumulative risks or survival probabilities.

Descriptive Epidemiology and Surveillance

• Person, place, and time — the core dimensions used to describe who is affected, where events occur, and how patterns change. Description generates hypotheses but does not alone identify causes.
• Public-health surveillance — ongoing, systematic collection, analysis, interpretation, and dissemination of health data for action.
• Passive surveillance — relies on routine reporting by clinicians, laboratories, or institutions. It is relatively inexpensive but often incomplete and sensitive to reporting practices.
• Active surveillance — public-health staff seek cases directly from reporting sources. It can improve completeness during outbreaks or elimination programs but requires more resources.
• Sentinel surveillance — selected sites provide detailed or timely information. It can detect trends but may not represent an entire population without a suitable design.
• Syndromic surveillance — monitors symptom patterns or proxies before diagnoses are confirmed, such as emergency-department complaints. Speed is gained at the cost of specificity.
• Notifiable disease — a condition that law or regulation requires designated parties to report to health authorities. Lists and reporting duties vary by jurisdiction.
• Epidemic curve — plots cases by time of symptom onset or another event. Its shape can suggest point-source, continuous, intermittent, or propagated transmission but must be combined with exposure and contact evidence.

• Endemic, epidemic, and pandemic — endemic describes expected ongoing occurrence in a population; epidemic is occurrence above an expected level; pandemic is an epidemic spreading across countries or continents with sustained transmission.

• Case finding — systematic identification of people meeting a case definition through reports, records, laboratory data, interviews, or active outreach.
• Line list — one-row-per-case table containing identifiers or codes, classification, onset, exposures, laboratory results, location, contacts, and outcomes needed for an investigation.
• Surveillance sensitivity — proportion of events meeting the surveillance target that the system detects. Changes in testing or reporting can alter sensitivity and create artificial trends.
• Surveillance timeliness — speed from event occurrence through detection, reporting, analysis, and action. A complete system can still fail its purpose if information arrives too late.
• Surveillance representativeness — how well reported events describe occurrence across people, places, and time. Participation and care access can produce systematic gaps.
• Wastewater surveillance — measures biological or chemical markers in community wastewater to track population-level trends without requiring individual clinical testing. Catchment, flow, decay, and laboratory methods affect interpretation.

Study Designs

• Case report and case series — detailed descriptions of one or several patients without a comparison group. They can identify new phenomena but cannot estimate comparative effects reliably.
• Cross-sectional study — measures exposure and outcome at one time or over a short window. It estimates prevalence and associations but often cannot establish temporal order.
• Case-control study — begins with cases and controls and compares prior exposures. It is efficient for rare outcomes; the odds ratio estimates the sampling-compatible association when controls represent the exposure distribution in the source population.
• Cohort study — follows exposed and unexposed groups, prospectively or through records, to compare outcome occurrence. Loss to follow-up and time-varying exposures require careful handling.
• Randomized controlled trial — assigns interventions by a random mechanism. Randomization supports exchangeability in expectation, while allocation concealment, adherence, attrition, blinding, and outcome measurement still affect validity.
• Cluster-randomized trial — randomizes groups such as clinics, schools, or communities. Analysis and sample-size calculations must account for correlation within clusters.
• Ecologic study — compares group-level exposures and outcomes. Associations at the group level need not hold among individuals, a problem known as the ecological fallacy.
• Natural experiment — studies an exposure change driven by policy, environment, or another process outside investigator assignment. Causal strength depends on the assignment mechanism and comparison strategy.

• Systematic review — uses prespecified methods to identify, appraise, and synthesize evidence addressing a focused question. A meta-analysis is a statistical synthesis and is neither required nor automatically valid when studies differ materially.

• Nested case-control study — samples cases and controls from an established cohort, preserving the cohort’s temporal structure while reducing the cost of measuring stored exposures.
• Case-cohort study — compares all cohort cases with a randomly selected subcohort that can serve as the comparison group for multiple outcomes.
• Case-crossover study — compares a person’s exposure shortly before an acute event with that same person’s exposure during control periods, controlling stable individual characteristics by design.
• Interrupted time-series study — estimates whether an outcome’s level or trend changes after an intervention using repeated observations before and after it. Concurrent events and time-varying measurement remain threats.
• Stepped-wedge trial — clusters cross from control to intervention at randomized times until all receive it. Secular trends and within-cluster correlation require explicit modeling.

Bias, Confounding, and Causal Reasoning

• Selection bias — the analyzed relationship differs from the target relationship because inclusion, retention, or observation depends on causes of exposure and outcome.
• Information bias — systematic error in measuring exposure, outcome, or covariates. Recall bias, interviewer effects, diagnostic suspicion, and imperfect records are common mechanisms.
• Confounding — distortion of an exposure-outcome comparison by shared causes. Adjustment should follow a causal model; automatically controlling every measured variable can introduce bias.
• Collider bias — conditioning on a common effect of two variables can create a noncausal association between them. Selection into hospitals, tested populations, or complete-case samples can act as collider conditioning.
• Mediation — a mediator lies on a causal path from exposure to outcome. Estimating direct and indirect effects requires stronger assumptions than estimating a total effect.
• Reverse causation — an outcome or its preclinical process influences the measured exposure. Temporal information and design are needed to distinguish this from forward causation.
• Immortal-time bias — a period during which an outcome could not occur under the exposure definition is incorrectly credited to the exposed group.
• Healthy-worker effect — employed populations often have lower overall mortality than the general population because severely ill people are less likely to enter or remain employed.
• Negative control — an exposure or outcome chosen because a proposed causal mechanism predicts no effect. An observed association can reveal residual bias, though a null result does not prove the main estimate unbiased.

• Sensitivity analysis — quantifies how estimates change under alternative assumptions about measurement, missingness, model choice, or unmeasured confounding.

• Nondifferential misclassification — measurement error has the same classification probabilities across comparison groups under a specified condition. It does not universally bias effects toward the null.
• Differential misclassification — measurement quality differs according to exposure, outcome, or another analysis variable, allowing bias in either direction.
• Recall bias — accuracy or completeness of remembered exposure differs between comparison groups, often because outcome experience changes attention or reporting.
• Loss-to-follow-up bias — the observed cohort differs causally from the target because retention depends on variables related to exposure and outcome.
• Residual confounding — confounding that remains after adjustment because causes were unmeasured, poorly measured, incorrectly modeled, or grouped too coarsely.
• Effect-measure modification — an effect differs across levels of another variable on a stated scale. It is a property to describe, not a bias to remove automatically.

Screening and Diagnostic Testing

• Sensitivity — the probability a test is positive among people who have the target condition. It is a property of the test in a population and protocol, not an unconditional constant.
• Specificity — the probability a test is negative among people who do not have the target condition.
• Positive predictive value — the probability of the condition among people with a positive result. It depends strongly on prevalence as well as sensitivity and specificity.
• Negative predictive value — the probability of not having the condition among people with a negative result. It also changes with prevalence and setting.
• Likelihood ratio — compares the probability of a test result among people with and without a condition. Positive and negative likelihood ratios update pretest odds to post-test odds.
• Receiver operating characteristic curve — plots sensitivity against one minus specificity over thresholds. Area under the curve measures ranking discrimination and does not determine calibration, clinical value, or a preferred threshold.
• Screening — testing people without recognized symptoms to identify elevated risk or early disease. Benefit depends on disease natural history, test performance, effective follow-up, treatment benefit, harms, access, and overdiagnosis.
• Lead-time bias — earlier detection lengthens measured survival from diagnosis without delaying death.
• Length bias — periodic screening preferentially detects slower-progressing disease because it remains detectable longer.

• Overdiagnosis — detection of a condition that would not have caused symptoms or death during a person’s lifetime. It is distinct from a false-positive test.

• Pretest probability — estimated probability of a condition before the new test result, based on prevalence, symptoms, history, and setting. It anchors interpretation of likelihood ratios and predictive values.
• Diagnostic threshold — score or measurement boundary used to classify results. Moving it usually trades sensitivity against specificity and changes downstream workload and harm.
• Verification bias — the reference diagnosis is obtained preferentially for people with certain index-test results, distorting estimated test performance.
• Spectrum effect — test performance changes with disease severity, comorbidity, demographics, and alternative diagnoses, so estimates from selected cases and healthy controls may not transport to practice.
• Screening interval — time between repeated screens. It should reflect disease progression, test sensitivity, prior results, treatment benefit, burden, and competing harms.

Outbreak Investigation and Infectious Disease

• Outbreak investigation — commonly includes verifying the diagnosis, defining and finding cases, describing person-place-time patterns, developing and testing hypotheses, implementing controls, and communicating findings. Control measures may begin before every analytic step is complete.
• Index case and primary case — an index case is the first case recognized by investigators; a primary case is the person who introduces infection into a group. They need not be the same person.
• Incubation period — time from infection to symptom onset. The latent period runs from infection to becoming infectious and can be shorter or longer than incubation.
• Basic reproduction number — the expected secondary cases generated by a typical infectious person in a wholly susceptible population under specified conditions. It is not a fixed biological constant.
• Effective reproduction number — expected secondary cases under current immunity, behavior, interventions, and contact patterns. Values below one imply declining transmission on average.
• Attack rate — cumulative incidence during an outbreak, often within a defined exposed group. A secondary attack rate measures spread among susceptible contacts of primary cases.
• Reservoir — the habitat in which an infectious agent normally lives and multiplies. A vector transmits an agent between hosts; a vehicle is a nonliving medium such as food or water.
• Herd protection — indirect protection produced when immune individuals reduce opportunities for transmission. Threshold calculations from simple homogeneous models can fail under clustered contact, waning immunity, or variant change.

• Antimicrobial resistance — survival and spread of microbes under antimicrobial selection. Stewardship, infection prevention, diagnostics, vaccination, sanitation, and surveillance are complementary controls.

• Serial interval — time between symptom onset in an infector and symptom onset in a person they infect. It can be negative when presymptomatic transmission occurs.
• Generation interval — time between infection of an infector and infection of a secondary case. It is usually unobserved and inferred from linked cases and timing data.
• Infectious period — interval during which an infected host can transmit an agent. It need not match either symptom duration or test positivity.
• Contact tracing — identifies and supports people exposed to a case so they can test, monitor, receive prophylaxis, isolate, or quarantine as appropriate.
• Genomic epidemiology — combines pathogen sequence variation with time, place, contact, and sampling information to investigate transmission and evolution. Genetic similarity alone does not prove direct transmission.
• Superspreading — highly uneven transmission in which a minority of infections generate many secondary cases. Biology, behavior, timing, setting, and network structure can all contribute.

Prevention and Intervention

• Primordial prevention — prevents emergence of social or environmental conditions that create risk, such as policies shaping tobacco use, unsafe streets, or air pollution.
• Primary prevention — acts before disease begins, including vaccination, safer environments, protective equipment, and reduction of harmful exposures.
• Secondary prevention — detects and treats disease early, often through screening or case finding.
• Tertiary prevention — reduces complications, disability, or recurrence after disease is established.
• Quaternary prevention — seeks to prevent unnecessary medical intervention and harm from overmedicalization.
• Vaccination — induces immune protection without the risks of natural infection. Program impact depends on effectiveness, coverage, timing, durability, access, and pathogen ecology.
• Isolation and quarantine — isolation separates people known or suspected to be infected; quarantine restricts movement of exposed people during a possible incubation period. Legal authority and terminology vary by jurisdiction.
• Health promotion — supports people and communities in increasing control over determinants of health through information, environments, policy, and collective action.
• Harm reduction — reduces adverse consequences of a behavior or exposure without requiring its immediate elimination. Examples include sterile-syringe services, naloxone access, and safer-use interventions.

• Program evaluation — asks whether an intervention was implemented, reached its intended population, changed intermediate outcomes, improved health, and produced unintended effects.

• Efficacy and effectiveness — efficacy estimates benefit under controlled or ideal conditions; effectiveness estimates performance in routine settings with real uptake, adherence, delivery, and population diversity.
• Number needed to treat — reciprocal of an absolute risk reduction over a specified time under compatible assumptions. It changes with baseline risk and should include uncertainty.
• High-risk prevention strategy — targets people with elevated individual risk. It may offer favorable benefit-to-harm tradeoffs for recipients while preventing fewer total cases than a broad population shift.
• Health communication — designs and exchanges information so audiences can understand risk, options, uncertainty, and action. Trust, messenger, language, access, and two-way engagement affect impact.
• Implementation evaluation — examines reach, adoption, fidelity, adaptation, dose, acceptability, cost, context, and sustainability to explain how an intervention operated in practice.

Environmental and Occupational Health

• Hazard and risk — a hazard can cause harm; risk combines the probability and consequence of harm under a specified exposure scenario.
• Dose-response relationship — describes how outcome frequency or severity changes with exposure dose. Thresholds, latency, susceptible subgroups, and mixtures can complicate the relationship.
• Exposure assessment — estimates contact with biological, chemical, physical, or social agents by route, intensity, frequency, and duration.
• Environmental justice — examines unequal environmental burdens, benefits, participation, and enforcement, especially where structural disadvantage shapes exposure and vulnerability.
• Occupational hierarchy of controls — prioritizes elimination, substitution, engineering controls, administrative controls, and personal protective equipment, in that order.
• Built environment — housing, land use, transportation, public space, and infrastructure shape injury, physical activity, heat, pollution, access, and social connection.

• Climate and health — climate change influences heat illness, air quality, infectious-disease ecology, food and water security, disasters, displacement, and mental health, with unequal vulnerability.

• Source-pathway-receptor model — traces a hazard from its origin through environmental transport and an exposure route to an affected person or population, identifying points for control.
• Cumulative exposure — combined exposure to multiple agents or repeated exposure over time. Mixtures, timing, vulnerability, and social conditions can matter beyond a single-agent dose.
• Biomonitoring — measures chemicals, metabolites, or biological markers in human specimens. Detection establishes internal exposure but not automatically its source, timing, or health effect.
• Risk characterization — integrates hazard, dose-response, exposure, and uncertainty into a decision-relevant account of expected harm and its limits.
• Health impact assessment — prospectively evaluates how a policy, plan, or project could affect health distribution and recommends ways to enhance benefit and reduce harm.
• Environmental epidemiology — studies health effects of physical, chemical, biological, and social environmental exposures using designs that must address spatial, temporal, and measurement uncertainty.

Health Systems, Policy, and Ethics

• Health system — the people, institutions, financing, information, products, and governance involved in maintaining or improving health. Clinical care is one component of a broader system.
• Primary care — first-contact, accessible, continuous, comprehensive, and coordinated care. It differs from primary prevention, which describes timing of preventive action.
• Universal health coverage — access to needed quality services without financial hardship. Coverage on paper does not guarantee timely, acceptable, or effective care.
• Health insurance risk pooling — combines financial risk across people so unpredictable individual costs are shared. Pool size, selection, benefits, prices, and subsidies affect protection.
• Cost-effectiveness analysis — compares incremental costs and health consequences across options. It informs efficiency but does not by itself decide distribution, affordability, rights, or political feasibility.
• Quality of care — commonly includes safety, effectiveness, patient-centeredness, timeliness, efficiency, and equity. Measurement can create incentives and must be checked for gaming and case-mix differences.
• Public-health ethics — balances population benefit, individual liberty, justice, reciprocity, transparency, proportionality, and the least restrictive effective means.
• Community engagement — includes affected communities in setting priorities, designing interventions, interpreting evidence, and sharing authority. Consultation without influence is not meaningful participation.

• Data privacy — limits collection, access, linkage, use, retention, and disclosure of identifiable information. De-identification reduces but may not eliminate re-identification risk.

• Access to care — the practical ability to obtain appropriate services when needed. Availability, affordability, geography, language, trust, disability access, and administrative burden all contribute.
• Health-care utilization — use of services such as preventive, outpatient, emergency, inpatient, pharmacy, and long-term care. Higher use can indicate either better access or greater unmet health need.
• Health technology assessment — evaluates clinical effects, safety, cost, ethics, organizational consequences, and social implications of a health technology for a defined policy decision.
• Priority setting — allocates limited resources among competing health needs using evidence, values, opportunity costs, equity, transparency, and appeal or revision processes.
• Continuity of care — coherent care over time across clinicians and settings through stable relationships, shared information, coordinated plans, and reliable transitions.

Sources and Further Study

• CDC, Principles of Epidemiology in Public Health Practice
• CDC, Introduction to Epidemiology
• CDC, Field Epidemiology Manual
• World Health Organization, health topics
• Hernán and Robins, Causal Inference: What If