Epidemiology Flashcards Midterm

Goal of Descriptive Epidemiology

How a disease or risk factor is distributed in a community (Who, What, When, Where)

Goal of Analytic Epidemiology

Causal agents (Why, How)

Sources of Gathering Quantitative Data

Individuals, health Care Facilities, Environment, Public Health Agencies

Methods for Gathering Quantitative Data

Notification/Reporting, Registries, Survey of Interview Data

Sentinel Reporting…

relies on a pre-arranged sample of healthcare providers who agree to report certain disease/conditions

Benefits of Sentinel reporting

can be less costly and more efficient than traditional reporting as well as more specifically targeted geographically

Passive reporting…

is provider initiated reporting

Active reporting…

involves local or state health departments soliciting reports from health care entities or reviewing records to find cases

Registries are…

methods for tracking events or persons diagnosed OVER TIME

Tracking over time allows us to …

understand diagnosis, treatment, prognosis and outcome information

Surveys are a method of data collection to understand…

behaviors that impact health

Counts -

number of cases, events, individuals with certain characteristics P

Proportions -

what fraction of the population is affected

Rates -

how fast the outcome is occuring

Numerator of frequency -

new or existing outcomes

Denominator of Frequency -

persons at risk for the outcome

Prevalence is the…

amount of disease present at a defined time

Prevalence is a…

proportion, boundaries are 0-100%

Point vs. Period Prevalence

specific point in time vs. over a certain time period

Survivor bias makes prevalence less useful for…

studies relating to cause (etiologic studies)

Incidence rate or Density rate is…

how fast disease is occurring in a population over a specified time interval

Incidence/Density Rate is synonymous to…

Mortality rate

Incidence/Density Rate Formula

(#new cases/total person * time) *K)

Incidence Rate assumes…

uniformity of risk over time

Fixed Population -

nobody can enter or leave outside of death

Dynamic Population -

people can enter or exit over time

Formula for calculating Incidence rate with aggregate population data

[#new cases/(midpoint population*length of study time)]

Cumulative Incidence (Risk)…

Chances of becoming a case over a defined interval

Cumulative Incidence Formula

New cases / Count of persons at risk at start of interval

Cumulative Incidence/Risk is a…

proportion, boundaries 0-100%

Cumulative Incidence/Risk is synonymous to

Incidence Proportion

Bathtub Figure

Incident Cases (Faucet) > Prevalent Cases (Tub Water) > Deaths, Emigrations & Recovery (Drained Water)

Case Fatality (proportion) is the…

Cumulative Incidence of death among those who have a disease

No Denominator >

Count

Frequency of existing disease at a single point in time →

Prevalence

Occurrence of New disease events over time (fatal or not) →

Incidence

Denominator of initial risk >

Cumulative Incidence

Incidence Density Rate Denominator →

Person Time at Risk

Occurrence of New disease events over time (fatal) >

Mortality

Ratio of Risks is used to understand…

exposed vs unexposed risks

Proportionate Mortality Ratio does not tell us about…

underlying risk

Proportionate Mortality Ratio tells us…

proportion of mortality across defined population (ex. deaths to cause / deaths of all causes)

Incidence can be used to measure the occurrence of…

any kind of event

Three components of incidence:

Population, Event, Observation Period

Cohort studies…

group people with common characteristics followed over time for outcome development

In a cohort study, exposure is observed…

before outcome

Cohort study design is ______ but data collection can be _______

prospective; retrospective

In a cohort study exposure is observed ____ the outcome.

before

t/f Cohort studies can be observational or experimental

True

t/f Cohort study population is selected after exposure is identified

false

Cohort studies usually require ______ follow up times

long

Study ID: 

• Random sample of adult population 30-59 in Framingham, MA in 1948 (n=5209, 2/3 of target pop)

• Data Collection: prospective with in-person exams conducted every 2 years through 2014

  • Medical history, CV physical exam, blood, urine, ecg, other heart imaging

• Milestones: Smoking as rick for heart disease, hypertension increase risk for stroke, obesity risk factor for heart disease…

Prospective Cohort

Study ID:

• Cohort – children ~10 years with exposure to diabetes in utero (past exposure) and those that didn’t

• Every perinatal medical record reviewed to determine exposure status (blinded) outcomes obtained by MRI

• Hypothesis was exposure would increase centralized obesity in childhood

Retrospective Cohort

Cohort Study Strengths

• Direct observation of incidence

• Allows changes in exposure

• Allows study of rare/multiple outcomes

  • Temporal sequencing

Cohort Study Weakness

• Impractical for rare diseases (may not have enough occurrence)

• Require large sample size

• Long time follow up

• Limited to measured exposures

Potential Biases of Cohort Study

• Selection Bias

• Information Bias (misclassification & info quality)

When is Cohort study Appropriate?

Prior evidence of association from another study design or RCT not possible

Hierarchy of evidence for Causal Relationship

Systematic Review → RCT → Cohort → Case-Control → Cross-Sectional → Case Series / Expert opinion

Case-Control studies…

study groups selected on the basis of outcome status

Case-Control data collection is ALWAYS _______

retrospective

Case-Control study design is always ______

Observational

Reverse Causation -

having disease alters exposure

If knowledge of being a case results in more information on exposure for individual or record sources → becoming a case results in exposure recording

This is an example of ____

Reverse Causation

Controls represent the ______

base population that gave rise to the cases

Control sampling is valid if ________

prevalence of determinant is same as the base population

Overmatching occurs when

cases and controls are too similar to eacother (ex. twins) and become non-representative of the base population

Advantages of Multiple Controls

• increased study power (n) up to 1:4 case:control

(Ex. of Multiple controls) Role of aspirin in occurrence of Reye Syndrome in children groups

Children at same hospital + Children visiting emergency room same hospital + Children attending same school

Cases & Controls match by…

factors you want to remain similar (age, time of diagnosis, etc.)

Efficient matching is important in that it provides…

similar distribution of confounders in cases and controls.

Group Matching -

Proportion of controls w/ similar characteristics to cases

Individual Matching -

Each individual case gets a control with similar characteristics (mostly used in hospital studies)

t/f Matching factors cannot be studied as determinants

true

Strengths of Case-Control

• good for rare disease

• Quick and Cheap

  • Can use existing data

Limitations of Case-Control

• Cannot establish temporal sequence

• Cannot calculate incidence

• More bias than Cohort

• Hard to study disease mechanisms

Study ID:

• Question: Is maternal smoking in 1^st trimester associated with CHD

• Outcome

  • Cases – singleton births with CHD identified from hospital discharge data in Washington state 1989-2001

  • Controls – randomly selected live-born singleton births matched by year of delivery (3 controls per case)

• Exposure – maternal self-report of smoking on birth certificate

• Design – retrospective

  • Data on determinants/exposure collected retrospectively

Case-Control

Section bias is _______

differential probability being included in the study based on exposure status

Recall Bias is when _________

Cases may have been more or less likely to report exposure than controls

Case-Control Studies can be used as a _______

preliminary step to justify a cohort study

Cross Sectional Studies are useful for:

• patterns of exposure and outcome

• burden of illness & risk factors

• determining service needs

• hypothesis generation

Probability Sampling methods:

random stratified, random systematic, random clustered

Non-probability sampling methods:

convenience, judgmental, snowball method (cannot be used for statistical conclusions)

Cross-Sectional Study Advantages

• quick and cheap

  • descriptive

Cross-Sectional Study Dis-advantages

• non-temporal

• association but NOT causality

• selection bias (healthy people will participate more)

• survivor bias

Study ID: NHANES

• Examines nationally representative sample of 5,000 persons. Each year across the US, sample is selected using 4 stage sample design

  • Sample weights are used to produce estimates for entire target population

Cross-Sectional

Ecological Studies are…

Observation studies assembled from aggregate data across populations with different characteristics

• data is collected from summary measures rather than individuals

Study ID:

per capita chocolate consumption to Nobel laureate population

Ecological study

Cross-sectional

Nested Case-Control take advantage of ________

defined cohort with prospective exposure and outcome assessment

Nested Case-Control cases are selected _________ and controls are matched to each case or by calendar time.

from cohort as they occur

Reasons to conduct a Nested Case-Control Study

• rare outcome

• minimized bias (recall/information

• tight control of confoundind

Selecting from the same underlying cohort means _____

cases & controls have greater comparability

Case-Cohort Cases are selected ________

as they occur form a parent cohort study

Case-Cohort Controls are ________

randomly sampled from parent cohort rather than being matched

Reasons to Conduct Case-Cohort

• clear sequence of events

• eliminates resale and reverse causation biases

• more economical (possibility of multiple studies in one)

  • lower confounding → prospective exposure measurement

Study ID:

Was event triggered by an immediately preceding unusual event.

Case-Crossover

Advantages of Case-Crossover

• Cases serve as own control

  • eliminates non-variable confounding (genetics, location, gender)

Study ID:

Does folic acid antagonist use during pregnancy increase risk for cardiovascular birth defects

• Case window: month 2-3 of pregnancy (cardiovascular development)

Case-Crossover

Assumption of Case-Crossover

• exposure/confounders don’t change systematically over time

• timing of exposure doesn’t affect reporting

• exposing is intermittent (temporary variable)

• induction periods are short

Measures of Occurrence -

indicate the magnitude of occurrence in a group, do not indicate association