"0.36%" (rounded)

TIME, POPULATIONS & DISEASE OCCURRENCE

  • This set of notes covers key concepts for measuring disease occurrence in populations, with emphasis on definitions, study design terminology (population types), time-at-risk concepts, and common epidemiologic measures (prevalence, incidence, rates, and related calculations).

WHAT IS A POPULATION?

  • A population is a specific group defined by three dimensions:

    • Person (e.g., age, sex, race/ethnicity, job)

    • Place (e.g., country, state, county, city, workplace)

    • Time (e.g., calendar year, life-course stage such as adolescence or emerging adulthood)

  • These three dimensions together define the group to study or make inferences about.

WHAT IS A TARGET POPULATION?

  • Target population: Individuals about whom inferences are to be made.

  • The goal of epidemiologic research is to identify disease causes in the target population.

  • Example: The causes of AIDS in men who have sex with men (MSM) in San Francisco and New York in the early 1980s.

  • Target population is the group for whom you want to draw conclusions.

WHAT IF WE CAN’T STUDY THE ENTIRE TARGET POPULATION?

  • When enumeration of the entire target population isn’t feasible, identify a group of individuals expected to have the same exposure-disease association as the target population and that can be enumerated.

  • Source population: the group from which the study population is drawn.

  • Example: MSM enrolled in health clinics, frequenting other venues, or participating in outreach programs in San Francisco in 1984.

  • Diagrammatic idea: Target Population ← Source Population ← Study Population (when full enumeration isn’t possible)

TARGET & SAMPLE POPULATIONS

  • If every individual in the target population can be enumerated and the study is feasible: you may use the full Target Population or select a sample from the Source Population.

  • Conceptual relation: Target Population = Source Population (if you study everyone) OR Target Population is inferred from a Sample drawn from the Source Population.

STUDY POPULATION

  • If it is NOT feasible to enumerate and study every individual in the target or source population: you study a Study Population drawn from the Source Population.

  • Key terms:

    • Source Population: group from which the study population is drawn

    • Target Population: group to whom inferences will be made

    • Study Population: group actually studied in your research

  • A central concern is representativeness: how accurately the study population represents the target population.

EXAMPLE: STUDY OF TECHNOLOGY-FACILITATED VIOLENCE

  • Dr. Lambert’s CDC-funded study on digital dating violence among LGBTQIA+ youth aged 13–17 in eight Deep South states.

  • Recruited 400 youth via social media who completed an online survey.

  • A purposefully sampled subset of 35 youth participated in an in-depth qualitative interview.

  • Topics: experiences seeking healthcare, sexual health, mental health, experiences of violence (online and offline; IPV and non-IPV), identity development, impact of COVID-19.

  • Analysis finding: 320 of the 400 youth reported prior experiences of digital dating abuse.

  • Questions to classify populations:

    • What is the study population?

    • What is the sample population?

    • What is the target population?

  • Answers:

    • Target Population: LGBTQIA+ youth aged 13–17 in eight Deep South states.

    • Source Population: Youth reachable via social media recruitment (the population from which the sample is drawn).

    • Study Population: The 400 youth who completed the online survey.

    • Sample Population: The 35 youth who participated in the qualitative interviews.

    • Outcome of interest (digital dating abuse) reported by 320/400 participants.

COHORTS

  • Cohorts are populations of individuals moving through time together.

  • Components:

    • Target Population

    • Source Population

    • Study Population

  • Time dimension (t) is central to cohort definitions.

COHORT FOLLOW-UP

  • Visual: a sequence showing the study population over time with follow-up points.

  • Key idea: follow-up captures whether individuals experience outcomes over time.

MEASURING THE AMOUNT OF TIME A PERSON IS AT RISK

  • At risk: an individual who can experience the endpoint of interest.

  • Individuals may be at risk for different lengths of time.

  • You need a method to measure, for each individual, the time during which they are at risk.

MEASURING PERSON-TIME AT RISK

  • Example setup: Individual 1 with Entry1, Exit1, and Total time at risk T1.

  • Total person-time at risk is the sum across individuals: extTotalperson−time=(extT<em>1+extT</em>2+extT<em>3+extT</em>4+extT<em>5+extT</em>6+extT7+)ext{Total person-time} = \big( ext{T}<em>1 + ext{T}</em>2 + ext{T}<em>3 + ext{T}</em>4 + ext{T}<em>5 + ext{T}</em>6 + ext{T}_7 + \big)

  • Person-time can be summed across all individuals in the cohort to yield the denominator for rate calculations.

DIAGRAMS OF INDIVIDUALS IN POPULATIONS MOVING THROUGH TIME

  • Key ideas: entry into follow-up, development of the endpoint, censoring events, start and end of follow-up.

  • Censoring occurs when follow-up ends before the endpoint is observed for some individuals (e.g., administrative end, loss to follow-up, death from other causes).

COHORT TYPE: CLOSED VS OPEN

  • Closed cohort:

    • A group of at-risk individuals followed over time with no additions

    • No exits except for the endpoint of interest

  • Open (dynamic) cohort:

    • Individuals enter at different times

    • Individuals may exit for reasons other than the endpoint of interest

EXAMPLES OF COHORT TYPES

  • Closed cohort example: Enter together; exit only for the endpoint of interest (e.g., a birth cohort followed until all have died).

  • Closed cohort with administrative censoring: Enter together; exit for endpoint or end of follow-up (administrative censoring).

  • Open cohort example: Participants entering at different times and exiting for various reasons (e.g., CLUE II cohort followed for colorectal cancer diagnosis with censored observations).

MORE ABOUT COHORTS

  • Closed cohorts (entering together, exiting only for the endpoint) are uncommon in practice but are useful for teaching the concept of disease frequency measures.

COHORT EXAMPLES

  • Classic cohorts that entered at the same time and exited for endpoint(s):

    • Framingham Heart Study (1948)

    • British Doctors Study (1951)

    • CLUE I, CLUE II (1974, 1989)

    • Multicenter AIDS Cohort Study (1984, 1987, 2001)

  • Cohorts entering at different times with varying exit reasons: US cancer incidence, Medicare (age 65+)

WHAT KIND OF COHORT?

  • Examples to illustrate different entry patterns:

    • All patients admitted to the NICU during a single week and followed until death or NICU discharge

    • All UGA students admitted in Fall 2023 and followed until graduation

    • All 2022-2023 UGA students

PERSON-TIME AT RISK: EQUIVALENCE

  • Example shown: Individual-by-individual entries with T1, T2, …, T7 summing to total time at risk (e.g., 26 person-years in the example).

  • Purpose: demonstrate how person-time equivalence captures varying follow-up durations.

PERSON-TIME AT RISK: THE DATA

  • Data structure typically includes: entry time, exit time, status at end of follow-up, and computed person-time at risk for each individual.

  • This forms the basis for calculating incidence rates using person-time as the denominator.

WHY CALCULATE PERSON-TIME AT RISK?

  • Important denominator for disease occurrence rates.

  • Rates are a measure of disease occurrence and are estimated as:

  • extIncidencerate=racextNumberofeventsduringaperiodextTotalperson−timeatriskduringthatperiodext{Incidence rate} = rac{ ext{Number of events during a period}}{ ext{Total person-time at risk during that period}}

  • Units example: per person-year, per 100,000 person-years, etc.

MEASURING DISEASE OCCURRENCE

  • Types of measures: Count, Ratio, Proportion, Rate

  • Warning: Many epidemiology terms are used incorrectly; clarity about what numerator and denominator represent is essential.

TYPES OF MEASURES

  • Count: Number of events or cases (e.g., number of food-poisoning cases in two cities).

  • Ratio: One quantity divided by another quantity (e.g., men to women, oranges to apples). extRatio=racextNumeratorextDenominatorext{Ratio} = rac{ ext{Numerator}}{ ext{Denominator}}

  • Proportion: A type of ratio where the numerator is part of the denominator (e.g., infected individuals among a group). extProportion=racextPartextWholeext{Proportion} = rac{ ext{Part}}{ ext{Whole}}

  • Rate: A ratio in which time is part of the denominator (e.g., events per person-time). extRate=racextEventsextTimeatriskext{Rate} = rac{ ext{Events}}{ ext{Time at risk}}

  • Takeaway: Proportions and counts are not the same as rates; rates incorporate time.

PREVALENCE & INCIDENCE

  • Prevalence measures the existence of current disease at a point in time or over a period.

  • Incidence measures the occurrence of new disease events over a period.

  • Intuition: prevalence reflects burden at a moment; incidence reflects the risk of developing disease over time.

  • Common relationships: Cures and deaths affect prevalence over time.

PREVALENCE

  • Definition: Prevalence = Number of EXISTING cases / Size of the population

  • extPrevalence=racextNumberofEXISTINGcasesextSizeofthepopulationext{Prevalence} = rac{ ext{Number of EXISTING cases}}{ ext{Size of the population}}

  • Notes:

    • Prevalence is a proportion, not a rate (though people often misuse the term “prevalence rate”).

    • Population of interest (POI) is typically the total population or the population at risk.

POINT VS PERIOD PREVALENCE

  • Point prevalence: the number of existing cases at a given date divided by the population at that date.

  • Period prevalence: the number of cases in a population during a given period divided by the population size during that period.

  • Denominator considerations:

    • If the population changes, defining the denominator for period prevalence is challenging.

    • Solutions: use mid-point population or average population over the period.

CALCULATING POINT & PERIOD PREVALENCE (EXAMPLES)

  • Example setup (simplified):

    • Jan 1, 2024: 1,000 prevalent cases; population = 1,000,000

    • Dec 31, 2024: 1,000 prevalent cases cured; 1,000 incident cases cured; 1,000 incident cases with disease; population = 1,200,000 (start) / 1,100,000 (mid-year)

  • Point prevalence on Jan 1, 2024:

    • ext{Point Prev}_{Jan1,2024} = rac{2000}{1000000} = 0.002 = 0.2 ext{%}

  • Point prevalence on Dec 31, 2024 (using Dec31 existing cases):

    • If there are 2,000 existing cases and population is 1,200,000: ext{Point Prev}_{Dec31,2024} = rac{2000}{1200000} = 0.0017 = 0.17 ext{%}

  • Period prevalence on Dec 31, 2024 (with 4,000 cases with disease over the period and population 1,100,000):

    • $$ ext{Period Prev}_{Dec31,2024} = rac{4000}{1100000} \