Foundations, Measures of Disease Frequency & Mortality

Epidemiology

The study of how often diseases happen in different groups of people and why, used to help prevent and control health problems.

5 Big Questions of Epidemiology

Who gets sick? How many? Where and when? Why (what causes it)? What can we do to stop it?

Study Design vs. Study Purpose

Design = HOW you run the study (the setup). Purpose = WHY you run it (the question you want to answer).

PICO

Population, Intervention, Comparator, Outcome (also Time and Setting). A checklist for writing a clear research question about a treatment.

PECO

Population, Exposure, Comparator, Outcome. Same idea as PICO, but for questions about what causes disease.

6 Study Purposes

Intervention (does a treatment work?), Etiology (what causes it?), Diagnosis (testing), Screening (early detection), Prognosis (what happens over time), Prevalence (how common is it?).

Study Purposes That Are ALWAYS Non-Randomized

Etiology, prognosis, and prevalence. You can't assign people to get a disease or to have certain traits.

Study Purposes That Can Be Randomized OR Not

Intervention, diagnosis, and screening. They can be tested in an experiment or just observed.

Best Design for an Intervention Question

Randomized controlled trial (RCT). It is the fairest test of whether a treatment works.

Etiology Study

Looks for the causes or risk factors of a disease. Usually observational because you can't force people to be exposed to something harmful.

Diagnostic Study

Checks how well a test finds a disease by comparing tests (looks at sensitivity, specificity, predictive value).

Prognostic Study

Follows people who have a disease over time to see what usually happens to them and what predicts a better or worse outcome.

Prevalence Study

A snapshot that counts how many people have a condition right now.

Randomized vs. Observational Studies

Randomized: researchers decide who gets what, by chance. Observational: researchers only watch and record what happens naturally.

Why Do Observational Studies?

1) You can't ethically assign harmful exposures. 2) Results reflect real life. 3) They are cheaper and faster for slow-developing diseases.

Ratio

One number divided by another where the top number is NOT part of the bottom number. E.g., 10 men : 5 women = 2:1.

Proportion

A fraction where the top number IS part of the bottom number. E.g., 3 sick students out of 10 = 0.3. Always between 0 and 1.

Percentage

A proportion multiplied by 100 (so it goes from 0% to 100%).

Rate

How often something happens in a group over a set amount of time, so time is built into the measure. E.g., 5 per 1,000 people per year.

Incidence

The number of NEW cases of a disease in a group that started out disease-free, over a set time.

Prevalence

ALL existing cases (old and new) in a population at one point in time or over a period.

Candidate (At-Risk) Population

People who don't have the disease yet but COULD get it. Once someone gets sick, they leave this group.

Cumulative Incidence (CI) Formula

CI = new cases during the period / disease-free people at risk at the start.

Cumulative Incidence (CI) Meaning

Risk: the chance that a person gets the disease over a set time. A proportion from 0 to 1, with no units.

Cumulative Incidence Assumption

Everyone is followed for the whole period, with no dropouts.

Incidence Density (ID) Formula

ID = new cases / total person-time at risk.

Incidence Density (ID) Meaning

Incidence rate: how fast new cases appear per unit of person-time (e.g., per 100 person-years). It can be bigger than 1.

Person-Time

The total time each person was followed while still at risk, added together. A person stops counting when they get the disease, die, drop out, or the study ends.

Why Use Incidence Density Instead of CI?

People are followed for different lengths of time (dropouts, late joiners). Person-time handles that fairly, but CI can't.

CI vs. Incidence Density

CI is a proportion (0 to 1, no units). ID is a rate (per person-time, can be above 1).

Who Adds More Person-Time: Healthy for 365 Days or Diagnosed on Day 50?

The healthy person (365 vs. 50), because the sick person stops being at risk once diagnosed.

Point Prevalence Formula

Existing cases at one moment / total population at that moment.

Period Prevalence Formula

(Cases at the start + new cases during the period) / total population.

What Does Prevalence Depend On?

How fast new cases appear (incidence) AND how long people stay sick (duration).

Prevalence-Incidence Formula

P / (1 - P) = IR x D, where IR = incidence rate and D = average duration of the disease.

Assumptions of P / (1 - P) = IR x D

A stable population (steady state), constant incidence and duration, and a rare disease (prevalence under 10%).

Prevalence Goes UP When…

More new cases appear, OR people live longer with the disease without being cured (longer duration).

Prevalence Goes DOWN When…

Fewer new cases appear, OR people are cured or die faster (shorter duration).

Denominator for Cumulative Incidence

People at risk at the start. Remove anyone who already has the disease, because they can't become a NEW case.

Denominator for Period Prevalence

The WHOLE population. Prevalence counts existing cases, so no one is removed.

Denominator for a Later Time Window (Outbreak Questions)

Remove everyone who is still sick (existing + new cases). If the illness lasts long enough, they are still ill and can't be new cases.

Crude Mortality Rate

Total deaths from all causes / total population (per 100,000 per year). Not adjusted for age.

Cause-Specific Mortality Rate

Deaths from one cause / total population (per 100,000 per year).

Age-Specific Mortality Rate

Deaths in one age group / population of that same age group (per 100,000 per year).

Years of Potential Life Lost (YPLL)

Counts the years of life lost to early deaths (before a set age like 75). It gives more weight to deaths of young people.

Live Birth Rate

Live births / total population (per 1,000 per year).

Infant Mortality Rate

Deaths of babies under 1 year / live births in the same time (per 1,000 live births).

Birth Defect Rate

Babies born with defects / total births (per 10,000 births).

Morbidity Rate

How common illness is in a population (existing plus new cases), often per 100 people per year.

Attack Rate

People who got sick / people at risk during an outbreak. It is calculated like cumulative incidence.

Case Fatality Rate (CFR)

Deaths from a disease / people who have the disease x 100. Shows how deadly the disease is.

Survival Rate

1 - case fatality rate. The share of people with the disease who stay alive.

Why Are Crude Rates Misleading for Comparisons?

Populations have different age mixes. An older population has more deaths even if its health care is just as good.

Purpose of Standardization

Removes the effect of age (or another factor) so death rates from different populations can be compared fairly.