epi imp

Exposure Variable

An independent variable or predictor of primary interest in an epidemiologic study; a variable that an investigator wants to assess as to whether or not it predicts a health outcome.

Covariate

A variable that is measured or observed in a study; often considered as a control variable in the data analysis.

Variable

Any quantity that varies; any attribute, phenomenon, or event that can have different values.

Health Outcome

The possible health-related states or events of interest; the dependent variable of interest; in epidemiology, usually the disease variable being studied, although the outcome of interest may alternatively be mortality status, recovery from illness, or any health condition other than a disease.

Epidemiology

The branch of medicine that deals with the incidence, distribution, and possible control of diseases and other factors relating to health.

Controlling

Adjusting for the influence of variables other than the exposure variable that have potential influence on the health outcome.

Association

The relationship between two or more variables; in epidemiology, an association is quantified using a measure of effect, e.g., an odds ratio, a risk ratio; a popular statistical measure of association is a correlation coefficient.

Bias

The presence of systematic error; a flaw in the study design, the methods of data collection, or the methods of data analysis that may lead to spurious conclusions about a exposure-disease relationship. Three general sources of bias occur in: the selection of study subjects, incorrect information gathered on study subjects and the failure to adjust for variables other than the exposure, commonly called confounders.

Causation

Relationship between a cause and an effect, i.e., knowledge or measurement of one variable (i.e., the cause) leads to knowledge of another variable (i.e., the effect). Several criteria for causation have been proposed by research scientists and philosophers, including widely quoted criteria proposed by Bradford Hill in 1971.

Causal inferences

Drawing conclusion about causality involving one variable as a “cause” (i.e., determinant) of another variable “the effect”, (usually a health outcome in epidemiology) where the latter is usually a health outcome.

Meta analysis

Statistical methods for contrasting and combining results from different studies in the hope of identifying patterns among study results, sources of disagreement among those results, or other interesting relationships that may come to light in the context of multiple studies.

Confounder

A variable that needs to be controlled in order to avoid a bias in the assessment of an exposure-disease relationship.

Control variable

A variable other than the exposure variable that has a potential effect on the outcome variable and is therefore subject to “control” (i.e., being taken into account) in the analysis.

Outbreak

An occurrence of disease greater than expected at a particular time and place. It may affect a small and localized group or impact thousands of people across an entire continent.

Cohort Study

An observational study in which subjects are sampled based on presence (exposed) or absence (unexposed) of an exposure variable of interest; these subjects are followed over time for the development of a health outcome of interest; often used as a synonym for follow-up study.

Study Design

The plan of the procedures and methods for implementing a research project.

Conceptual hypothesis

A statement or conjuncture about a possible exposure-disease relationship that is based on insight obtained from research literature or conceptual theory.

OR

An observational study in which subjects are sampled based on presence (exposed) or absence (unexposed) of an exposure variable of interest; these subjects are followed over time for the development of a health outcome of interest; often used as a synonym for follow-up study.

Operational hypothesis

A statement about a possible exposure-disease relationship that translates a conceptual hypothesis into a testable form involving measurable variables and a clearly specified study plan.

Randomization

An allocation procedure for assigning subjects to exposure/study groups by chance that attempts to make groups comparable with respect to other predictors of the health outcome.

Experimental

A study in which randomization is used to allocate subjects to exposure/treatment/study groups.

Observational

A study that does not involve randomization; the investigator monitors but does not influence the exposure status of individual subjects and their subsequent health outcome status.

Timing

A design option that answers the question, “has the information being studied already occurred before the study actually began?”; can be retrospective, prospective, and ambispective.

Retrospective

A study in which the data are derived from past experience; both the study factor and the health outcome occur before the onset of the study.

Prospective

A study in which both the study factor and the health outcome occur after the study onset; new data are used.

Efficacy

The effect that a treatment or other clinical procedure has on the well- being of individuals.

Forwards (directionality)

The directionality of a study in which the health outcome is observed after (in time) the exposure variable is observed; the study proceeds “forwards” over time; a cohort study and a clinical trial always have forwards directionality.

Non-directional

A study in which the health outcome and the exposure variable are observed at the same time; the study is essentially a snapshot time, i.e., a survey; a cross-sectional study is always non-directional.

Measurement error

Incorrect measure or observation of a variable of interest; can lead to information bias or misclassification bias.

RR

Incorrect measure or observation of a variable of interest; can lead to information bias or misclassification bias.

Directionality

The directionality of a study in which the health outcome is observed after (in time) the exposure variable is observed; the study proceeds “forwards” over time; a cohort study and a clinical trial always have forwards directionality.

Clinical Trial

An epidemiologic study design that most closely resembles a laboratory experiment; the major objective is to test the efficacy of a therapeutic or preventive intervention.

Blinding

A feature of a clinical trial in which the patient/subject and/or the investigator is unaware of the treatment assigned.

Intention-to-treat analysis

Analysis of the results of a clinical trial that is based on the initial treatment assignment regardless of whether or not subjects completed the full course of treatment, i.e., “you analyze what you randomize”

Selection Bias

Systematic error that results from the way subjects are selected into the study and/or because there are selective losses of subjects prior to data analysis.

Effectiveness

The effect that a treatment or other clinical procedure has on the well- being of individuals under ordinary circumstances.

Effect

The effect that a treatment or other clinical procedure has on the well- being of individuals under ordinary circumstances.

Cross-sectional

A basic observational study design in which all variables studied are observed at a single point or a very short period in time.

Nested Case-Control

A hybrid design that is a variation of the case-cohort design in which controls are sampled by density sampling; that is, controls are matched to cases on the time of the case diagnosis.

Ecologic

Observational studies in which the unit of analysis is a group, often defined geographically, such as a census tract, a state, or a country; primary criticism: there is no data on individuals.

Observational study design

A study design that does not involve randomization; the investigator monitors but does not influence the exposure status of individual subjects and their subsequent health outcome status.

Case-Cohort study

A hybrid design that combines features of both case control and cohort designs; controls are sampled from the original cohort.

Case-crossover

An epidemiologic study design that is a variant of a matched case-control study in which each case serves as its own control; evaluates the effect of brief exposures with transient effects on acute health outcomes when a traditional control group is not readily available.

Population-at-risk

The population or group that is eligible to have or develop a health outcome of interest, also called the candidate population, e.g., the disease-free population.

Population

The population or group that is eligible to have or develop a health outcome of interest, also called the candidate population, e.g., the disease-free population.

Prevalence

The proportion of persons in a defined population who have a particular health outcome, e.g., a disease, at a point or during a period of time; existing cases of a disease.

Risk

The probability that an individual will experience an event of interest within a specified period of follow-up.

Cumulative Incidence

A population-based estimate of individual risk; the proportion of persons that develop a health outcome out of those candidates (e.g., disease-free) available for developing the health outcome.

Incidence

A population-based estimate of individual risk; the proportion of persons that develop a health outcome out of those candidates (e.g., disease-free) available for developing the health outcome.

Disease Frequency

A measure of how often a disease or other health outcome occurs in a population.

Simple Cumulative Incidence

A population-based estimate of risk that has the formula CI=I/N, where N denotes the size of a disease-free fixed cohort at the start of follow-up and I denotes the number of cases of a health outcome that develop over the follow-up period.

Fixed Cohort

A group (i.e., cohort) of subjects identified at some point of time and followed for a given period of detection of new cases. The cohort is “fixed” in the sense that no entries are permitted into the study after the onset of follow-up, although subsequent losses of subjects may occur for various reasons such as withdrawal, migration, and death.

Dynamic cohort

Synonym: Dynamic population. A group of subjects that is continually changing, allowing for both the addition of new members and the loss of previously entered members during a period of follow-up.

Probability

A number between (and possibly including) 0 and 1 that represents how likely and event of interest will occur.

Proportion

The ratio of two quantities in which the numerator is always less than or equal to the denominator; ranges between 0 and 1, inclusive.

Percentage

A proportion scaled to a range from 0 to 100, i.e., a ratio of two quantities multiplied by 100, where the numerator is always less than or equal to the denominator.

Rate

A measure of the rapidity with which health events such as new cases of disease or deaths occur; also called the hazard and incidence density.

Hazard

A synonym for rate; used primarily in survival analysis; also called incidence density.

Person-time

The total amount of disease-free follow-up time experienced by all persons in a cohort that is being followed over a specified time period.

Average rate

Number of new cases in (T0,T1)} divided by PT where T0 and T1 denote the starting and ending time points of follow-up, and PT denotes the amount of disease-free person- time accumulated during the interval from T0 to T1.

Cohort

A group of persons who share a common attribute, such as birth in a particular year, or residence in a particular town, who are followed over time; an observational study design involving such a group is called a cohort study.

Incidence rate

Synonym for rate and incidence density; a measure of the rapidity with which health events such as new cases of disease or deaths occur; also called hazard and incidence density.

Point prevalence

The proportion of persons in a defined population who have a particular health outcome, e.g., disease, at a point or during a period of time; existing cases of a disease at point in time.

Period prevalence

The proportion of persons in a defined population who have a particular health outcome, e.g., disease, during a period of time; existing cases of a disease during a period of time.

Disease-specific mortality

A risk or rate measure of the incidence of dying from a particular disease.

Mortality

A risk or rate measure of the incidence of dying from a particular disease.

Case-fatality

A risk or rate measure of persons with a particular disease that die of that disease over a period of follow- up.

Mortality risk

A measure of an individual’s probability of dying during a period of follow-up.

Mortality rate

A measure of the rapidity at which persons are dying in a cohort followed over time.

Age-adjusted

Summary (i.e., overall) rates, risks, or prevalence measures for comparing different populations in which underlying differences in the age distribution are removed. Also called age-standardized.

Crude Death Rate

Summary (i.e., overall) estimate of a death rate, risk, or prevalence measure that ignores the control of other variables.

Age-adjusted Rates

Summary (i.e., overall) rates, risks, or prevalences for comparing different populations in which underlying differences in the age distribution are removed. Also called age- standardized rates.

Age-adjustment

A procedure used to calculate summary rates, risks, or prevalences, for comparing different populations in which underlying differences in the age distributions are removed. Also called age standardization.

Standard Population

A population used for computing age- adjusted rates that provides a common age distribution that allows the comparison of two or more groups with differing age distributions, e.g., if death rates for Arizona and Alaska are to be compared, the standard population would typically be the US population; can also be used for any kind of rate adjustment.

Confounding

Bias resulting from failure to control for other variables in assessing an exposure-disease relationship.

Direct method

A method of adjusting a rate, risk or prevalence measure for a confounding factor that substitutes the distribution of the confounder in a standard population for the separate distributions of the confounder in groups being compared.

Indirect method

Synonym for indirect adjustment or indirect standardization. Typically a weighted average of (age-) specific rates for a select standard population using the distribution of the study population as weights.

SMR

Notatation for the standardized morbidity or mortality ratio. Number of observed cases (O) divided by the number of expected cases (E); typically used for indirect adjustment when category-specific rates in the study population are unknown or unreliable.

Measure of Effect

Synonyms: Effect Measure, Measure of Association. A quantity that expresses the strength of association between two variables, often referred to as the exposure variable and disease variable in epidemiologic research; a comparison of two or more measures of disease frequency.

Odds Ratio

The ratio of two odds; the exposed odds ratio is an odds ratio derived from a Case-Control study; the risk odds ratio is derived from a cohort study; the prevalence odds ratio is derived from a cross-sectional study. Computational formula: ad/bc, the cross product ratio from a 2 x 2 table.

Risk Ratio

Measure of effect used in follow- up/cohort studies defined as the risk for one group divided by the risk for another group.

Null Value

The value of a parameter under the null hypothesis; the null value of a ratio effect measure is 1, whereas the null value of a difference effect measure is 0.

Odds

The odds of an event is defined as P/(1-P) where P is the probability of the event, e.g., if P=1/3, then the odds equals 1/3 divided by 2/3, which equals 1/2, or 1 to 2.

Exposure Odds for Cases

The ratio of the number of cases who are exposed divided by the number of cases who are unexposed.

Exposure Odds for Controls

The ratio of the number of controls who are exposed divided by the number of controls who are unexposed.

Attributable risk

Synonym for risk difference. Sometimes used to denote the attributable risk percent, which has a different meaning.

Conditional Probability

The probability of an event given that another event has occurred, e.g., P(A | B) is the probability that the event A occurs given that B has occurred. If A is the event of having lung cancer and B is the event of being a smoker, then P(A | B) is the number of persons with lung cancer who are smokers divided by the total number of persons who are smokers.

Bayes Theorem

Synonym: Bayes Rule. A mathematical theorem about conditional probabilities that allows P(D|E) and P(D|not E) to be expressed in terms of P(E|D) and P(E|not D). This rule is needed to show that EOR equals ROR under certain conditions (see Lesson 5, lesson page 3); also used to express predictive value in terms of sensitivity, specificity and prevalence parameters under rare disease assumptions.

Rate Ratio

A measure of effect defined as the ratio of two rates; synonyms: incidence density ratio (IDR); hazard ratio.

Incidence density

IDR; measure of effect defined as the ratio of two (average) rates. Also called rate ratio or hazard ratio.

Risk difference

Measure of effect obtained by subtracting the risk from one group from the risk for another group, e.g., RD = R1 - R0, Ri denotes the risk for group i.

Potential Impact

Quantifies the expected effect of changing the distribution of one or more risk factors in a particular population; gives the proportion of new cases that can be attributed to a risk factor, e.g., etiologic fraction (EF), prevented fraction (PF)

Etiologic fraction

Synonym for population attributable risk percent, population attributable risk, and attributable risk percent; the proportion or corresponding percentage of new cases attributable to exposure; EF = I/I, where I is the number of new cases occurring during a period of follow-up and I denotes the number of new cases that occurred minus the number of new cases that would have occurred in the absence of exposure; can also be used with rate (i.e., incidence density) estimates instead of risk estimates.

Validity

The absence of systematic error; concerns methodologic imperfections in the study design and/or analysis.

Precision

Inverse of the variance; the higher the precision the narrower is the width of the corresponding confidence interval.

Random Error

A difference between an estimate computed from study data and the effect measure actually being estimated.

Statistical Inference

Drawing conclusions about a population parameter based on an estimate obtained from a sample.

Sample

A subset of a larger group (i.e., a population) on which study data is obtained and analyzed.

Selection Ratio

The number of subjects from one of the four cells in the study population divided by the corresponding number of subjects in the source population.

Selection Probability

If the study population is a subset of the source population, gives the likelihood that a person from one of the four cells in the source population will be a member of the study population.