Epidemiological Studies Essay

Core public health disciplines

Epidemiology - study of distribution and determinants of disease frequency in human populations

Data/biostatistics - mathematical science of data analysis and the numbers that describe health in populations

environmental health - study of how the environment - natural or built — impacts human health and well-being

health management - strategic approach to mobilize societal resources to improve the health of populations

social/behavioral - study of the social determinants of health and behavioral modification to improve health

Prevalence (%)

total number of cases in a specified population at a specific time

# of cases/population at risk

Incidence

rate of new cases in a specified population over a defined period

# of new cases/population at risk per unit of time (per year)

Morbidity (%)

proportion of persons affected by a disease in a particular population

also refers to degree of pathology

Morality (%)

number of deaths per population at risk

ratio of the number of deaths to the number of population during a specified time-period (death-rate)

_____ used to measure disease. Why?

Mortality. - easy to determine, indicates severity, vital health statistics

YLL (Years of Potential Life Lost)

number of additional years of life that would have taken place in absence of disease

calculated by subtracting age of death from reference age

emphasizes diseases that kill younger persons

DALY (Disability adjusted life years)

measure of overall disease burden

years of healthy life lost due to ill-health, disability or early death

DALY = YLL + YLD (years lived with disability)

Epidemiology

studies patterns of disease occurrence in populations and the factors influening the pattern

derived from word epidemic

Who? When?, Where? = describe disease as well to control or prevent disease

Endemic

health conditions present in a population at a defined rate

Epidemic

sudden rise in endemic rate clearly exceeding normal levels

Facets of epidemiological investigations

1. counting the cases and determining their distribution by persons (who), place (where) and time (when)

2. use this information to determine possible cause of disease or outbreak

3. institute interventions to prevent or control disease based on findings

How did John Snow demonstrate epidemiological investigations

1. Cases - Who? cholera victims Where? London households When? 1853 cholera outbreak

2. Possible causation - Southwark/Vauxhall Water company, water being drawn form area of Thames contaminated with human fecal waste

3. Intervention - removed pump handle at Broad Street, cases decreased

Epidemiological studies used to __________

determine cause of disease and to test interventions

Descriptive epidemiology

who, where, and when questions (observational)

analytical epidemiology

examines relationships between exposures and outcomes in populations

cross-sectional

study that analyzes exposure and outcome data from a population at the ‘present’ time or limited period

prospective

follows individuals over time to observe and collect data on exposure and development of specific outcomes

retrospective

examines past exposures in individuals and their current outcomes

Cohort studies

healthy individuals are enrolled and observed over time to look for associations between exposure and outcomes

Case-control study

Compares people who have a specific outcome or disease (cases) to a similar group of people who do not have that outcome (controls) to identify past risk factors

Intervention studies

determine efficacy and safety of interventions (drugs, vaccines, behavior modification)

Nurses Health Study

cohort study that enrolled 120,000 married female nurses

send questionnaire every 2 years asking about diet, drinking, smoking, drugs, oral contraceptives (factors related to development of breast cancer)

Doll-Hill Study

smokers vs non-smoker mortality (10 years earlier) - analytical analysis

Strength of association

ratio of incidence in expose individual/ incidence of non-exposed individual

T/F Cohort studies are more efficient than case-control studies

False; case-control studies more efficient due to less people and less time

How are case-control studies conducted?

two groups (cases + controls)

determine exposure to factors that are possibly related to disease occurrence and progression

strength of association is determined with odds-ratio

how to calculate odds ratio (OR)

What do intervention studies test?

laboratory experiment: drugs or other intervention in individuals with disease, tests prevention

Error

Analytical epidemiology attempts to determine association between exposure and outcome

ALL studied are subject to error

Biological Variation

Fluctuation of biological processes in the same individual over time

Sampling Error

Random influences on who or what is selected for the study

Measurement error

random fluctuations in measurement (including medical services)

systematic error

not randomly distributed between exposed and unexposed subjects in a study —> consistent deviation from the true value

affect validity of study

Sampling error

random sampling error: potential source of variation between samples

systematic sampling error: the test group does not reflect the population being studied

Human behavior and experience are difficult to control

true

What study used to follow individuals and observe their diet

Cohort study - not require people to change their behavior

What study used individuals with heart disease vs healthy individuals

case control study - questions about diet over past 5 years

What biases are present?

Reporting + recall bias (diet intake unreliable)

Investigator/observer vias (investigators may favor certain outcomes)

Confounding variables

Systematic error

another factor is associated with exposure which affects the outcome

Example: Researchers notice that people who drink more coffee tend to have higher rate of lung cancer

Trap: coffee causes lung cancer

Confounding variable: smokers drink more coffee and also get lung cancer much more often

effect modification

occurs when the effect of one exposure on an outcome changes across the levels of a third background variable

Interaction

happens when two distinct exposures have a combined effect that is different from the sum of their individual effects

Can we believe epidemiological studies

Interpreting epidemiological studies is difficult

Biases and confounders must be controlled

Epidemiological studies of the risks of daily life are frequently and eagerly reported by the popular press

Correlation does not prove causation

Epidemiological studies identify associations or correlations between exposures and outcomes (roosters crow associated with sunrise, but does not cause it)

statistics used to quantify these associations.

Uncertainty

integral part of science

contradictory results from studies are common

statistics used to assist in decision making

sources of uncertainty

random error: unpredictable fluctuations in measurement or observation

systemic error: a bias pushes results in one direction including confounding variables

Mean, median, mode, range, variance, standard deviation

mean -average

median - middle value

mode - most common value

range - highest + lowest difference

variance - squared difference average of mean

standard deviation - square root of variance

probability

hypothesis testing

P value

scientist quantify uncertainty by measure probabilities

p value of .05 means that if an experiment was repeated 100 times, 95 times would yield same result and 5 times would yield different result

Confidence interval

range of values which likely contains true result (reliability)

T/F statistical significance means that there is a biological and clinical significance

FALSE - does not always mean

death rates are often adjusted

P value misconceptians

Low p values do not prove the hypothesis

p values suggests an effect exists, but says nothing about the size of the effect or its clinical or practical importance

sample size sensitivity —> in large datasets trivial differences can yield significant p-values

Cause-effect + 4 attributes that increase support

epidemiological studies cannot prove cause-and-effect

a large sample size is more likely to yield valid results than a small study

high relative risk or odds ratio values indicate a stronger association between exposure and outcome

A dose-response relationship between exposure and outcome

a biological explanation makes argument more convincing

John Graunt

publishes landmark analysis of morality data

detect new threats, identify special risk groups, plan public health programs, evaluate efficacy of programs, and prepare government budgets

Government agencies

National center for health statistics, part of CDC

Collects data from local governments, repository for vital statistics, conducts periodic surveys from representative populations

National Health and Nutrition Examination Survey - every year, 5000 individuals from 15 countries are visited and detailed data collected

behavioral risk factor surveillance survey

Data collected

vital statistics - births and deaths are most reliable data collected

The census - conducted every 10 years. denominator to calculate rates (information on age, sex, race, socioeconomic characteristics