Midterm Review

Question 1

What are the 9 Bradford Hill critera?

Answer 1

1: Strength of Association
2: Consistency
3: Specificity
4: Temporality
5: Biological Gradient
6: Plausibility
7: Coherence
8: Experiment
9: Analogy

Mnemonic: “Some Assholes Constantly Specify That Bitchy Pigeons Collect Every Ant”

Question 2

Bradford Hill: Strength

Answer 2

Strong associations compelling because weaker
associations could be more easily explained by
confounders etc.

Question 3

Bradford Hill: Consistency

Answer 3

Repeated observation in different populations under different circumstances

Question 4

Bradford Hill: Specificity

Answer 4

A cause leads to a single effect, not multiple effects
An effect has one cause, not multiple causes

Question 5

Bradford Hill: Temporality

Answer 5

Cause precedes effect in time

Question 6

Bradford Hill: Biologic Gradient

Answer 6

Dose-response or exposure-response curve with an expected shape

Question 7

Bradford Hill: Plausibility

Answer 7

Scientific plausibility of an association

Question 8

Bradford Hill: Coherence

Answer 8

Cause-and effect interpretation does not conflict with what is known of natural history and biology of disease
“it makes sense with what we knew before; it is coherent with it”

Question 9

Bradford Hill: Experimental Evidence

Answer 9

Effect of reducing or eliminating putatively
harmful exposure and seeing if there is a resulting decrease in disease

Question 10

Bradford Hill: Analogy

Answer 10

Extension to other similar conditions, exposures.
“when one causal agent is known, the standards of evidence are lowered for a second causal agent that is similar in some way”

Question 11

Incidence Times

Answer 11

Times after a common reference event, when new cases occur among population

Question 12

Incidence Rate

Answer 12

Occurrences of new cases per unit of PERSON-TIME

follow each member of our population to see
specifically how long each person is observed for the outcome of interest…

Note you could also look at this as a time-weighted average of individual rates

IT ISN’t A PROPORTION!

Question 13

Incidence Proportion

Answer 13

Also called Cumulative Incidence
Proportion of people who develop new disease during specified time period

This is a PROPORTION

Question 14

Prevalence

Answer 14

Measures the total number of cases (both new and existing) in a population at a specific point in time (point prevalence) or over a period (period prevalence).

Question 15

Incidence Time

Answer 15

T0 to the point the event occurs

Used in time-to-event (survival) analysis to track when cases occur rather than just counting how many occur.

Question 16

Censoring

Answer 16

Occurs when a study does not observe the exact time of an event (e.g., disease onset, death) for an individual, either because they leave the study early or the study ends before the event occurs.

Question 17

Post Hoc Fallacy

Answer 17

A logical fallacy that assumes “because X happened before Y, X must have caused Y.

Question 18

Reason to use Kaplan-Meier Curve

Answer 18

Cumulative survival will not work when censored observations are in the sample. Kaplan-Meier curves (or other actuarial methods like life tables) account for this

Question 19

Assumption of Life Tables

Answer 19

Uniformity of events and losses within each interval

Question 20

4 Assumptions of Kaplan-Meier Curves

Answer 20

1: Independence between censoring and survival
The probability of being censored is not related to the probability of experiencing the event

2: Event Occurs at the Recorded Time
No measurement errors

3: Survival Probabilities are the Same for All Individuals at a Given Time
Every participant who remains at risk has the same probability of survival as others still in the study.

4: No Changes in Risk Over Time (no secular changes)
The conditions affecting survival remain constant throughout the study period.

Question 21

Secular Trends

Answer 21

Gradual changes in the occurrence of disease over a long period of time, usually years or decades.They are also known as temporal trends

Question 22

Direction of bias if those censored have HIGHER risk than those not censored

Answer 22

Towards null (if outcome is death/illness): study sample is progressively healthy over time (lower risk patients remain in study)

Question 23

Direction of bias if those censored have LOWER risk than those not censored

Answer 23

Away from null (if outcome is death/illness): study sample is progressively sicker over time (higher risk patients remain in study)

Question 24

Odds

Answer 24

The ratio of the (probability of event) to (probability of non-event)

Question 25

Odds Ratio

Answer 25

A measure of association that compares the odds of an outcome occurring in an exposed group to the odds in an unexposed group. It is commonly used in case-control studies but can also be applied in cohort and cross-sectional studies.

Question 26

Risk

Answer 26

The probability that an individual in a population will develop a disease or experience an event over a specified time period.

Question 27

Risk Ratio

Answer 27

Compares the risk of developing a disease in an exposed group to the risk in an unexposed group.

Question 28

Prevalence Ratio

Answer 28

A measure that compares the prevalence of a disease or condition between two groups—typically an exposed group and an unexposed group. It is similar to Relative Risk (RR) but is used in cross-sectional studies, where disease status is measured at a single point in time rather than over a period.

Question 29

Prevalence Odds Ratio

Answer 29

Similar to the Odds Ratio (OR) but is specifically used in cross-sectional studies. It compares the odds of having a disease (prevalence) between an exposed and unexposed group.

Question 30

Age effect

Answer 30

Change in the rate of a condition according to age, irrespective of birth cohort and calendar time ## Footnote e.g. increased heart disease risk with age

Question 31

Cohort Effect

Answer 31

Change in the rate of a condition according to year of birth, irrespective of birth cohort and calendar time ## Footnote ie: for zika virus, didnt have to do with the cohort of the mom or baby, just the time the baby was born (ie during the outbreak) different from Period Effect bc this only effects birth cohort

Question 32

Period Effect

Answer 32

Change in the rate of a condition affecting an entire population at some point in time, irrespective of age and birth cohort ## Footnote e.g., a pandemic, economic crisis, Introduction of Antibiotics (1940s) → Sharp decline in infectious disease deaths for all age groups. different from Cohort effect bc effects EVERONE in the period

Question 33

3 thing cross sectional studies mask that Birth Cohort Analysis shows

Answer 33

Age effect Period effect Cohort effect

Question 34

Rare Disease Assumption

Answer 34

As a disease becomes more rare the OR more closely approximates the RR

Question 35

Conditions when OR will approximate RR

Answer 35

* Controls are representative of individuals in the base population * Cases are representative of all individuals in the base population with the disease of interest * The disease is** relatively rare (generally < 5%** of the base population)

Question 36

True or false: OR of the event is the reciprocal of the non-event

Answer 36

True! This is because it is a counterfactual; you either did or didn't die NOT SO FOR RR or other risk estimates | Ratio or Risks, NOT complements

Question 37

Incidence density sampling

Answer 37

Also called risk set sampling, is a method used in case-control studies nested within a cohort. It selects controls from individuals who were at risk at the same time cases occurred, preserving the timing of exposure and ensuring comparability between cases and controls.

Question 38

Attributable Risk

Answer 38

A measure of association based on the *absolute difference between two risk estimates*

Question 39

What are the 4 assumptions for Attributable Risk?

Answer 39

1. Causality 2. No Bias 3. No Confounding 4. Reversibility ## Footnote Also Independence, according to sources

Question 40

Population Attributable Risk (Levin's)

Answer 40

Estimates the proportion of the disease risk in the underlying population associated with exposure of interest

Question 41

Ecologic Fallacy, aka Aggregation Bias

Answer 41

Bias that can occur because an association between variables on an aggregate level does not necessarily represent the association that exists on the individual level

Question 42

4 Cohort Study Hightlights

Answer 42

1. Categorized according to exposure 2. All disease free at baseline 3. Exposure occurred first 4. Exposure categorization (exposed vs not exposed) documented prior to outcome ## Footnote Allows for temporality!!!

Question 43

Incidence Denstity Sampling (aka risk set sampling)

Answer 43

A method used in **nested case-control studies** where **controls are selected from individuals who are still at risk at the time each case occurs.** This approach ensures that the control group represents the population at risk at the same time the cases arise, maintaining comparability between cases and controls.

Question 44

Directon of ITT Bias

Answer 44

Intention to Treat will always **underestimate the true impact**

Question 45

Randomization vs Random Allocation

Answer 45

Randomization = Deciding who enters the study (random sampling). Random Allocation = Deciding which group they go into (random assignment).