Causal Inference

A comprehensive set of Q&A flashcards covering core concepts from the lecture notes on causal inference in epidemiology. Cards include definitions of association vs causation, levels of evidence, models of causation, Hill’s criteria, Rothman’s causal pies, critiques, and pragmatic approaches to identifying true causes.

What term describes a relationship where an exposure is linked with an outcome in epidemiology?

Association.

What phrase asserts that the exposure truly causes the outcome?

Exposure causes outcome.

What phrase describes a relationship observed between exposure and outcome without implying causation?

Exposure is associated with the outcome.

What are the six categories used to describe possible exposure-disease relationships in the notes?

Unworthy of study; Worthy of study; Basis of public policy; Limited evidence of a relationship; Good evidence of a relationship (more research warranted); Strong evidence of a relationship (accepted as causal).

Which exposure-disease pair is listed as an example of a strong evidence (accepted as causal) relationship?

Lung Cancer & Smoking.

Name the exposure-disease pair that involves oral contraceptives and ovarian cancer.

Ovarian Cancer & Oral Contraceptives.

Which drug combination is mentioned in relation to COVID-19?

Hydroxychloroquine & Azithromycin.

What is the exposure-outcome pair related to academic performance?

Bedtime for high school students & academic performance.

What are the two basic interpretations of an exposure-outcome relationship shown on pages 7-8?

Association vs Causation.

What are the three possibilities for the relationship between Exposure and Outcome (as per the Fundamental Question)?

Exposure associated with outcome; Exposure causes outcome; Both exist.

What is the fundamental question of epidemiology?

Is there an association or causal relationship between exposure and outcome?

Why is epidemiology described as an 'art' in this course?

To distinguish associational and causal relationships and to assess whether observed results are valid.

Name a non-causal explanation for an observed association.

Reverse causation.

Name another non-causal explanation for an observed association.

Bias.

What does 'chance' mean in explaining associations?

There is no true association; observed due to random variation.

What does the phrase 'the exposure is part of the causal pie' mean?

The exposure is a component of a sufficient cause that leads to disease.

What is the Web of Causation?

A model where causation arises from a complex set of interconnected factors, including social, environmental, and biological factors.

Name two types of factors included in the Web of Causation.

Social factors; Environmental factors (also genetics, culture, access to care, etc.).

What is a key example used to illustrate Web of Causation in the slides?

Lead poisoning (Web of Causation diagram).

Give an example of a precipitating condition for lead poisoning.

Lead paint used in older housing.

What is a major environmental source contributing to lead exposure in the home?

Soil and dust; Lead pipes; Gasoline/air pollution.

Who criticized the Web of Causation with the question 'has anyone seen the spider'?

Nancy Krieger.

What is one fundamental criticism of the Web of Causation?

Lack of causal mechanisms and theory behind the model.

What is another fundamental criticism related to the Web of Causation?

Focus on proximal factors and lacking differentiation between individual and population determinants.

What are the three key features of the Biomedical Model criticized in the slides?

Focus on biological determinants; Social determinants are secondary; Populations are a sum of individuals.

List Hill’s causal criteria (the full set).

Experimental evidence; Temporal relationship; Strength of the association; Dose-response; Biological plausibility; Consistency; Analogy; Specificity; Coherence.

What does 'Temporal relationship' mean in Hill’s criteria?

Exposure precedes the development of the outcome.

What does 'Strength of the association' mean in Hill’s criteria?

Stronger associations are more likely to be causal.

What does 'Dose-response' mean in Hill’s criteria?

Greater exposure is associated with a greater effect.

What does 'Consistency' mean in Hill’s criteria?

The association is observed in different contexts with similar results.

What does 'Biological plausibility' mean in Hill’s criteria?

There is a reasonable biological mechanism linking the exposure to the outcome.

What does 'Coherence' mean in Hill’s criteria?

Findings are coherent with existing biological and epidemiological knowledge; lab and field results align.

What does 'Analogy' mean in Hill’s criteria?

There are similarities to other established causal relationships.

What does 'Specificity' mean in Hill’s criteria?

A specific exposure leads to a specific outcome.

What does 'Experimental evidence' mean in Hill’s criteria?

There is scientific evidence of an association from experiments.

What is Modified Determinism (causal pies)?

Ken Rothman’s model using pies to represent sufficient causes composed of component causes.

What is a 'pie' in Rothman’s model?

A sufficient set of causes that can produce disease.

What is a 'component cause' in Rothman’s model?

A factor needed in some but not all sufficient causes.

What is a 'necessary cause' in Rothman’s model?

Found in all cases of the disease.

In the TB example, what is a sufficient cause composed of?

Exposure to TB bacteria, poor nutrition, crowding, AIDS, absence of BCG, poor ventilation.

What does Pie #3 illustrate in the TB example?

Poor nutrition is needed in addition to other factors to be part of a sufficient cause.

What is one pro of Modified Determinism?

Conceptually sensible; helps explain mechanisms; explains patterning; flexible.

What is one con of Modified Determinism?

Difficult to operationalize; hard to identify all causes; does not address quantitative risk.

What is Reductionism, and what is proposed as an alternative?

Reductionism: explain complex causation with simplest components; alternative: consider distal, intermediate, and proximal upstream causes.

What are Susser’s Pragmatic Causal Criteria?

Association; Time order; Direction.

What extra factors help evaluate causal direction in Susser’s framework?

Consistency; Strength; Specificity; Predictive performance; Coherence.

What is the Fundamental Question of Epidemiology summarized on slide 71?

Exposure associated with outcome; Exposure causes outcome; Both exist.

What is the overall takeaway about causation from the course summary?

Causation is the Holy Grail but hard to prove; often more important to rule out alternative explanations than to prove causation.