PHLT 103 Epidemiology Exam 2 Seplaki 2026 University of Rochester

work in progress 2/2 lectures covered -- I'll be filling this out after each lecture (before quizzes tbh) and will add perusall at the end

2 vital concerns of epidemiology regarding data

• Quality of available data

• Appropriate applications of data

Big data & defining features + some examples

vast electronic storehouse of information

three Vs = defining features:

• Volume = mass amount of data requiring expandable storage

• Variety = diverse types/formats of data

• Velocity = receiving data quickly for rapid analysis

+two suggested qualities:

• Veracity = trustworthiness/quality of data

• Value = potential applications of data

eg)

• internet search transactions

• social media activities

• health insurance program data

• electronic medical record receipt

Data linkage

joining data elements across databases (like social media, healthcare, pharmacy, etc.) and tying them with a common identifier

Data mining

gathering and exploring masses of data to find patterns and associations; data broker firms specialize in this

• eg) used for targeted advertisements

Pro and con of AI in big data

• Pro = can process large amounts of data quickly to use more sources for improving health

• Con = must be wary of ethical data use and avoiding the perpetuation of racial biases and inequities

Online retrieval

acquiring information for publicly-available and proprietary websites

online resources/databases eg)

• Centers for Disease Control and Prevention (CDC)

• PubMed & MEDLINE made by National Library of Medicine (NLM)

• Google (must be careful in how it’s used—has overestimated flu trends based on searches + sources must be specified)

• American Public Health Association (APHA)

• Society for Epidemiologic Research

• World Health Organization (WHO)

• US Census [every 10 years + annual estimates]

  • helpful in determining federal health programs by clarifying denominators in rates/proportions of epidemiologic measures

  • 2020 Census collected: age, Hispanic/Latino orgin, race, sex, if home is owned/rented, number of people living in home, relationship of other people staying at home

Factors that affect quality of epidemiologic data

looks at sources and reference population coverage (representativeness); quality impacts appropriate application and types of possible statistical analysis

• Nature of the data = sources and content

• Availability of the data = investigator access and ability to deidentify personal information

  • especially due to Health Insurance Portability and Accountability Act of 1996 [HIPAA] protecting confidential medical records with consent)

• Completeness of population coverage = how much the data reflects the population of interest

  • representativeness / external validity = how generalizable the data’s findings are towards its population

• Appropriate uses of the data = what types of questions can be answered depending on measures used

  • eg) you can’t use snapshot data alone to investigate change over time

  • eg) prevalence doesn’t provide information about risk status (incidence does)

Vital events

deaths, births, marriages, divorces, fetal deaths

• this information is routinely collected by the US vital registration system

• National Center for Health Statistics compiles US mortality rates very comprehensively

US death certificate data includes …

• demographics

• data and place of death

• cause of death (immediate and contributing factors)

Why may death certificate data not be entirely accurate?

• Primary cause of death is sometimes unclear

• Lack of standardized diagnostic criteria

  • especially state to state, varying between medical examiner-based and coroner-based systems (anyone elected to investigate a death)

• Stigma of reporting certain diseases

Birth statistics

data on live births, fetal deaths, birth defects, length of gestation, birth weight, mother demographics, etc.

• limited by recall of pregnancy events, conditions not detected at birth, state-to-state certificate requirements, births outside of hospital system

Public health surveillance + specific types

continuous systematic gathering of information about the occurence of diseases and health conditions/events

• limited by incomplete coverage for people that don’t seek metical attention

• syndromic surveillance = identifying illness clusters early before confirmed diagnoses and reports to public health agencies to mobilize rapid responses — early warning system

• wastewater surveillance = helps track infectious disease in communities without easy access to healthcare / clinical testing

Behavioral Risk Factor Surveillance System (BRFSS)

state-level tracking of behavior-related risk factors for chronic disease

Case registry

centralized database for information collection about a disease/condition

• tracks trends, groups in risk + guides program plans + helps find patients for case-control studies

• eg) National Program of Cancer Registries (NPCR); Surveillance, Epidemiology, and End Results (SEER) Program

Data from the National Center for Health Statistics (NCHS)

extensive national information through population surveys

• National Health Interview Survey (NHIS) = tracks nation progress towards specific health goals + healthcare accessibility through thousands of personal interviews

• National Health and Nutrition Examination Survey (NHANES) = continuous collection of physical examinations and interviews with mobile examination centers, releasing data every 2 years

• National Vital Statistics System (NVSS) = NHCS registers vital events and promotes the development of standardized reports

Data from international organizations (examples)

• World Health Organization = gives global infectious disease surveillence and collection of mortality causes

• European Union = includes public health data on social / health inequalities and determinants of health

Descriptive epidemiology

classifies occurrence of disease according to person [affected], place, and time (distributed across a population)

• finds trends in health/disease and provides a foundation for analytics that then plan out health services (like for eg, where telemedicine is most commonly used)

  • ie) provides context/basis for generating hypotheses for analytic epidemiology to then ask the “why?” and “how?”

• cannot be used alone to establish causal relationships in epidemiologic inference

Case report

single occurrence of a noteworthy health-related event [or a small collection of these events]

Case series

larger collection of disease cases, often consecutively with common features [like patient characteristics]

Cross-sectional study

examines relationship between health condition/event and other variables of interest across a population at a single point in time

• can measure prevalence of exposures and/or distributions of outcomes, but all at one time

• eg) BRFSS looking at e-cigarette use across working US adults and comparing state usage

Person variables

characteristics attached to people

• eg) age, sex, race, ethnicity, socioeconomic status, marital status, religion

• may be involved to stratify subgroups or connect to associations

  • eg) age-specific disease rates

What makes measuring race ambiguous?

It’s a social construct rather than biological

Socioeconomic status (SES)

measures position in society; composite of income, education, occupation, food security, housing, healthcare access

• representativeness of each measure varies from place to place; eg) healthcare is equal access in some countries

• strong inverse association with morbidity and mortality

Whitehall studies by Sir Michael Marmot

showed a Social Gradient in which there was an inverse association between British employment grade/rank and rates of heart disease / mortality

Health disparities/inequities

unjust differences in the distribution of diseases, health conditions, and access to healthcare across population subgroups

Social determinants of health (SDoH)

how an individual’s environment, healthcare access/quality, education, socioeconomic status, and social context can impact their health risks/outcomes

• can contribute to health disparities/inequities

• healthcare system doesn’t have the biggest impact on health

Structural variables

large-scale societal context of living that people can’t individually change

• eg) public policies, cultural values

Social capital

connections in society that provide access to resources

Biopsychosocial model of health

responsibility of providers to holistically consider biological, psychological, and social factors impacting health (popularized 1977 shift towards focusing on SDoH)

Modifiable risk factors

behaviors and exposures that individuals can directly change or avoid to lower risk for adverse health outcomes

WHO “Health in All Policies”

idea that every policy must be looked at from a health perspective because they will always connect back to affecting health directly or indirectly

Place variables

geographical characteristics of health distributions

• include climate safety, cultural/societal factors, access to resources

• eg) international, national, urban-rural differences, localized patterns of disease

Localized patterns of disease

diseases associated with specific environmental conditions [in particular geographic areas]

• eg) lung cancer and lingering radioactive gas

Time variables

how disease occurrence varies depending on time period/season

• eg) secular trends, cyclic trends, point epidemics, clustering

Secular trends

gradual changes in frequency of disease / health conditions over long time periods

• eg) rising drug overdose death rates 2000→2020

Cyclic/seasonal trends

health condition/event that increases/decreases in frequency over a period of several years or within a year

• eg) infectious diseases increasing in winter and decreasing in early summer

Point epidemic

response of a group of people from a common place / source of infection / causal factor that they were exposed to simultaneously

• eg) outbreak of Vibrio infections in those impacted by Hurricane Katrina (2005)

Clustering + types

collection of cases of disease / health events that occur closely in time/place; may suggest common exposures or just be due to chance

• Spatial clustering = cases occuring in a common geographic region

• Temporal clustering = cases occuring at a common time

  • eg) maternal postpartum depression developing a few days after birth