public health 224 exam 1

sources of data

existing data:

-hospitals/clinics

-insurance companies

-cancer registries

-national surveys (ex: NHIS, DHS, etc)

-employment records

strengths of existing data

-large populations

-some data are collected as required by law

-standardized form or data collection method

-quicker, easier and cheaper

limitations of existing data

-not collected for research

-incomplete and/or inaccurate

-might miss people of interest

-might not have data on the info u want

-people included might be different from those not included

-other biases: limited generalizability

how to collect new data

-questionnaires

-physical measurments

-laboratory measurements

-surveillance

strengths of collecting new data

-measure variables in which you are interested

-target populations in which you are interested

-more control over quality of data

limitations of collecting data

-higher costs (time, money, resources)

-need to recruit participants

-many potential problems w interviews:

-recall of info

-interviewer quality can vary

-other biases

two categories of epidemiological research

-descriptive

-analytic

descriptive research

-looking for the distribution of disease in terms of person, place and time

-hypothesis generating

analytical research

-evaluating risk factors for disease

-hypothesis testing

-quantifying associations/effects

descriptive statistics in epi

-epidemiology involves counting events

-new or existing cases of disease

-deaths

-persons with certain characteristics

types of descriptive statistics

-proportion

-ratio

-rate

incidence rate

number of new cases in a given population over a defined time period

-population must be at risk for disease

-most widely used tool in epi

-# of new cases/population at risk of becoming a new risk

case report

a detailed account of disease in an individual

-includes symptoms, diagnosis, clinical management and follow up

-unusual one

anecdotal

-can generate hypothesis but cannot determine causal factors

adjusted rates

often need to "adjust" the rates so that dissimilar populations can be compared

-gives population same distribution with respect to adjustment factor

-common ones are age, gender, education level

case series

a detailed account of disease in a small group of patients with similar diagnosis

-includes symptoms, diagnosis, clinical management and follow up

-often for rare diagnoses

anecdotal

-may provide clues in identifying new disease or adverse health effects from exposure

case report and case series

strengths

-document unusual disease occurrences

-identify new diseases and/or epidemics

limitations

-no comparison group

-small sample size

ecological studies

compares disease frequencies in populations based on a factor of interest

-unit of observation in population

do not have person-level data

confounder

A third factor that can make it appear that an observed exposure causes an outcome.

An unobserved exposure that is both associated with an observed exposure and is an actual cause of an outcome.

ecological study limitations

-do not have data on individual level confounders

-prone to "ecologic fallacy"

making false conclusions about individuals based on grouped data

-cannot link exposure directly to disease among i dividuals

-no individual level data

-populations will typically differ on a number of factors

cross sectional studies strengths

-exposure and disease data from individuals

-prove important data for public health planning

-assess prevalence of disease

-useful for generating hypothesis

-can typically be conducted quickly and with limited resources

ecologica study strengths

-quick inexpensive

-may utilize existing data if available

-useful for generating hypothesis

cross sectional studies

-sometimes call "prevalence" studies

-measure exposure and disease at same time

-snapshot of population health at a given time

-we can compare individual level data

-many existing data sets from gov

cross sectional studies limitations

-cannot determine is exposure preceded disease (temporarily)

-cannot determine causality

surveillance

-continuous monitoring of health events in the population

-systematic data collection and analysis

trend analysis

how do things look now vs in the past?

can we predict what will happen in the future?

planning

what are the most important health problems in the population

who is most at risk

hypothesis generating

trends, population groups affected might suggest causes

epidemiology

-the study of what is upon people

-the study of how disease is distributed in populations and the factors that influence or determine this distribution

exposure/risk factor

something that may cause disease (ex. smoking cigarettes)

outcome

health status/disease of interest (ex: lung cancer)

objectives of epidemiology

determine the extent of disease in the community

measures of occurrence:

-counts

-prevalence

-incidence

hippocrates

-greek physician

-lived from 460-377 BC

-"father of western medicine"

-hippocratic oath

-systematic observational

-disease linked to physical environment

james lind

-Observed the effect of time, place, weather, and diet on the prevalence of disease.

-noted symptoms of scurvy occur 4-6 weeks into voyage

-first clinical trial

john snow

-english physician

-solved the cholera outbreak in 1849

-common belief: miasma

-believed in waterborne theory

-mapped cases

epidemiology in the 1850s-1940s

-quantifying disease

-nutritional deficiencies

-infectious diseases

epi in the 1940s-1980s

-chronic diseases

-health related states

-methodological advances

epi currently

-re-emergence of infectious diseases

-genetics

-social factors

-precision medicine

john gaunt

analysis of vital statistics

incidence

how many new cases occur during a given period in a population

-cumulative incidence (aka risk)

-incidence rate

risk

a measure of how often an event occurs in a defined group of people over a defined period of time

-the liklihood or probability of developing a disease

-aka cumulative incidence

risk= # new cases / population at risk

calculating risk

-numerator is new cases only, must exclude people who already have disease

-denominator is at risk population, specified population at the start, must exclude people not at risk

incidence rate

how fast new cases of disease occur within a population

-subjects may be followed for different lengths of time

-rates calculated using person-time

-incidence rate= # new cases/person-time of at-risk population

person-time

sum of the actual time each member of the population is observed during the study period and is at risk for the disease

calculating incidence rate

-numerator is new cases only, exclude existing cases

-denominator is person-time of at-risk population, exclude ppl not at risk, sum of follow up time of subjects

cause-specific mortality rate

# deaths from specific cause/total population

case fatality rate

# deaths from specific cause/population with that disease

proportionate mortality rate

# deaths from specific cause/ # deaths from all causes

risk can be estimated from incidence rate and vice versa

CI = IR x D

CI: cumulative incidence (risk)

IR: incidence rate

D: duration of study in years

prevalence

-measuring existing cases of a disease in the population

-snapshot of health status at a given time

-measured in cross-sectional studies

-prevalence = # new cases of disease in population at specified time / # of ppl in population at specified time

point prevalence

prevalence of a disease at a single point in time

-ex: prevalence of breast cancer in amherst on january 1, 2018

period prevalence

prevalence of disease over a specified period of time

-ex: prevalence of breast cancer in amherst from january 1, 2017- december 31,2017

types of analytic studies

case control

prospective cohort

randomized controlled trial

prospective cohort studied

-observational

-people selected before have disease, people with and without exposure to measure future disease

-both groups followed to see who gets disease

case control study

-participants enrolled after disease status is known

-classified based on presence or absence of disease

-case group has disease

-control group does not

-exposure history is determined in each group, ask qs about exposure before had disease

randomized controlled trial (experimental)

-recruit participant w out disease

-assign exposure (treatment/intervention) by random assignment

-measure association in groups to determine the effectiveness of exposure

sampling error

studies use samples of all people and use statistics to make inferences

-groups may appear different due to random variability

-sampling error quantified by p-values and reduced by increasing sample size

measurement error

exposures are sometimes hard to assess and measurements may be imprecise

-results in groups appearing more similar to each other than in reality

selection bias

when groups appear to be different due to differences in the selection process

information bias

groups appear to be different due to differences in the measurement process

confounding

the effect of one effect is mistakenly attributes to another cause of disease (ex: smoking, coffee and cancer)

risk ratio and rate ratio

a ratio comparing incidence of disease in exposed group to incidence of disease in unexposed group

RR = incidence of disease in exposed group/ incidence of disease in unexposed group

-measures the strength or size of the relationship between exposure and disease

RR of greater than 1

exposure appears to increase risk of disease

RR= 1

no association

RR is less than 1

exposure appears to decrease risk of disease

validity

how well you are measuring what you are trying to measure

internal validity

ability of study to attribute the results to the effects of the studied exposure

external validity

ability of the results to be applied to the general population "generalizability"

-studies cannot be externally valid if they are not internally valid