Epidemiology quiz

Causal Inference in Epidemiology

Jennifer Deal, PhD Johns Hopkins University

The material in this video is subject to the copyright of the owners of the material and is being provided for educational purposes under rules of fair use for registered students in this course only. No additional copies of the copyrighted work may be made or distributed.

Definition and Properties of Risk Factors

3

Recall… (1)

► Public health activities ► Assess the burden of disease ► Determine risk factors for disease ► Evaluate interventions ► Make policy ► Communicate effectively with the public

4

Recall… (2)

► Public health activities ► Assess the burden of disease ► Determine risk factors for disease ► Evaluate interventions ► Make policy ► Communicate effectively with the public

5

What Is a Risk Factor? (1)

► A factor that is causally related to a change in the risk of a relevant health process, outcome, or condition

Source: Porta. (2014). A dictionary of epidemiology (6th ed.).

6

What Is a Risk Factor? (2)

► If the relationship is non-causal, the factor is just a risk marker

Source: Porta. (2014). A dictionary of epidemiology (6th ed.).

7

Categorization of Risk Factors

► Non-modifiable

► Modifiable

► Individual vs. Societal level

8

Properties of Risk Factors—1

► Direct vs. Indirect

► Necessary vs. Sufficient

9

Properties of Risk Factors—2

► Direct vs. Indirect

► Necessary vs. Sufficient

10

Direct vs. Indirect—1

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

11

Direct vs. Indirect—2

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

12

Direct vs. Indirect—3

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► If Indirect Factor 1 is the exposure of interest, what is Direct Factor 1 in regards to the exposure-disease relationship?

13

Direct vs. Indirect—4

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► If Indirect Factor 1 is the exposure of interest, what is Direct Factor 1 in regards to the exposure-disease relationship?

14

Direct vs. Indirect—5

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► What is this Factor?

15

Direct vs. Indirect—6

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► What is this Factor? ► Direct Risk Factor

16

Direct vs. Indirect—7

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► What is this Factor? ► Direct Risk Factor ► Confounder of the Director Factor 1 → Outcome

Relationship

17

Direct vs. Indirect—8

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► What is this Factor?

18

Direct vs. Indirect—9

► Direct Risk Factor ► Directly causes a disease without any

intermediate step

► Indirect Risk Factor ► Causes the disease but only through an

intermediate step

► What is this Factor? ► Direct Risk Factor ► Mediator of the Director Factor 1 → Outcome

Relationship

19

Properties of Risk Factors

► Direct vs. Indirect

► Necessary vs. Sufficient

20

Necessary vs. Sufficient Cause

► Necessary—the factor must be present for disease to develop

► Sufficient—the factor alone can produce the disease, but so can other factors (so the factor is not required to be present for the disease to develop)

● In the presence of that factor, the disease will always develop

21

Necessary Cause, but Not Sufficient

► Factors A, B, and C must all be present to cause the disease ► Yet no factor may produce the disease on its own ► Factors A, B, and C are therefore necessary, but not sufficient

22

Sufficient Cause, but Not Necessary

► Factors A, B, and C all cause the disease ► Each factor causes the disease on its own ► Factors A, B, and C are therefore sufficient, but not necessary

23

Both Necessary and Sufficient Cause

► Factor A is the only cause of the disease ► Factor A causes the disease on its own ► Factor A is therefore both necessary and sufficient

24

Neither Necessary nor Sufficient Cause

► Factor A and Factor B together can cause the disease, but neither can cause the disease on its own

► Factor C and Factor D together can cause the disease, but neither can cause the disease on its own

► Factors A, B, C, and D are therefore neither necessary nor sufficient

25

Recall… Phenylalanine in the Diet— Necessary or Sufficient Cause?—1

26

Recall… Phenylalanine in the Diet— Necessary or Sufficient Cause?—2

► Required for the development of PKU-related neurocognitive development effects

27

Recall… Phenylalanine in the Diet— Necessary or Sufficient Cause?—3

► Required for the development of PKU-related neurocognitive development effects ► Necessary

28

Recall… Phenylalanine in the Diet— Necessary or Sufficient Cause?—4

► Required for the development of PKU-related neurocognitive development effects ► Necessary

► Babies will not develop neurocognitive outcomes unless they are born with the PKU gene

29

Recall… Phenylalanine in the Diet— Necessary or Sufficient Cause?—5

► Required for the development of PKU-related neurocognitive development effects ► Necessary

► Babies will not develop neurocognitive outcomes unless they are born with the PKU gene ► Not sufficient

30

Smoking— Necessary or Sufficient Cause of Lung Cancer?

► Not all smokers develop lung cancer ► Not sufficient

► Some nonsmokers develop lung cancer ► Not necessary

► Smoking is neither a necessary nor sufficient cause of lung cancer

31

Examples— Necessary vs. Sufficient Cause

32

Why Assess Risk Factors for Disease?—1

1. To identify groups at high risk for disease

2. To direct preventive efforts to appropriate populations ► Individual ► Societal

33

Why Assess Risk Factors for Disease?—2

1. To identify groups at high risk for disease ► Factors associated with increased risk may or may not be causal (risk marker)

2. To direct preventive efforts to appropriate populations

34

Why Assess Risk Factors for Disease?—3

1. To identify groups at high risk for disease ► Factors associated with increased risk may or may not be causal (risk marker)

2. To direct preventive efforts to appropriate populations ► To prevent disease, the factor must be…

● A cause of disease ● Modifiable

35

Recall Three Types of Prevention—1

Type of prevention Definition Examples

Primary Preventing the initial development of a disease Immunization, reducing exposure to a risk factor

Secondary Early detection of existing disease to reduce severity and complications

Screening for cancer

Tertiary Reducing the impact of the disease Rehabilitation for stroke

36

Recall Three Types of Prevention—2

Type of prevention Definition Examples

Primary Preventing the initial development of a disease Immunization, reducing exposure to a risk factor

Secondary Early detection of existing disease to reduce severity and complications

Screening for cancer

Tertiary Reducing the impact of the disease Rehabilitation for stroke

37

Recall Public Health Activities

► Public health activities ► Assess the burden of disease ► Determine risk factors for disease—Cause! ► Evaluate interventions ► Make policy ► Communicate effectively with the public

The material in this video is subject to the copyright of the owners of the material and is being provided for educational purposes under rules of fair use for registered students in this course only. No additional copies of the copyrighted work may be made or distributed.

What Is a Cause?

2

Cause

3

What Is a Cause? “Something That Brings about an Effect or a Result”

Source: Merriam-Webster. Accessed February 1, 2019, at https://www.merriam-webster.com/dictionary/cause

4

Problems with Defining Cause

► “When we look about us towards external objects, and consider the operation of causes, we are never able, in a single instance, to discover any power or necessary connexion; any quality, which binds the effect to the cause, and renders the one an infallible consequence of the other. We only find, that the one does actually, in fact, follow the other.…”

David Hume (1711–1776)

Source: An Enquiry Concerning Human Understanding. Accessed February 1, 2019, at https://www.bartleby.com/37/3/9.html

5

Translation

► Our understanding of cause and effect can only be based on observations where the putative cause precedes the putative effect

► So can we ever know what a cause is?

David Hume (1711–1776)

Source: An Enquiry Concerning Human Understanding. Accessed February 1, 2019, at https://www.bartleby.com/37/3/9.html

6

How Can We Evaluate Cause?—1

► Animal models

► In vitro models (e.g., cell culture)

► However, we must extrapolate these findings to human populations

7

How Can We Evaluate Cause?—2

► Observations in human populations

► Randomized experiments in humans

8

Steps in Epidemiologic Causal Reasoning—1

1. Determine whether there is an association between a factor and the disease

2. If an association is found, determine whether the observed association is likely to be a causal one

9

Steps in Epidemiologic Causal Reasoning—2

1. Determine whether there is an association between a factor and the disease

2. If an association is found, determine whether the observed association is likely to be a causal one

10

Recall the Main Types of Epidemiologic Study Designs

Study type Characteristics

Experimental Studies prevention and treatment of diseaseInvestigator actively manipulates which groups receive the study agent/intervention

Observational Studies causes, prevention, and treatment for diseasesInvestigator watches as natures takes its course

Cohort Examines multiple health effects of an exposureSubjects defined by exposure levels and followed for disease occurrence

Case-control Typically examines multiple exposures in relation to a diseaseSubjects are defined as cases and controls, and exposure histories compared

Cross-sectional Examine relationship between exposure and disease prevalence in a defined population at one point in time

Ecological Examines relationship between exposure and disease with population-level data rather than individual data

Clinical observations

Physician might notice more cases than usual or perhaps a sequence of events suggesting a correlation

11

Recall Possible Explanations for Associations

► Cause

► Chance—non-causal

► Bias—non-causal

► Confounding—non-causal

12

Correlation ≠ Causation

Source: Spurious Correlations. Accessed February 1, 2019, at http://tylervigen.com/spurious-correlations

13

Non-causal Associations

► Primarily a concern in observational studies

► Chance associations (Biostatistics)

► Bias related to… ► Data generation (information bias) ► Participant selection (selection bias) ► Other factors (confounding)

14

Randomized Trials—1

► Temporality is established

► Exposure is randomized

► Masking

15

Randomized Trials—2

► Exposure is randomized ► Primary: removes bias in the allocation of treatment ► Secondary: increases the likelihood that treatment groups are balanced on factors

related to the outcome

16

Randomized Trials—3

► Exposure is randomized ► Primary: removes bias in the allocation of treatment ► Secondary: increases the likelihood that treatment groups are balanced on factors

related to the outcome

► Masking ► Minimizes bias related to prior knowledge or beliefs about treatment

17

Why Randomized Trials Are the Gold Standard—1

► Exposure is randomized ► Primary: removes bias (selection bias) in the allocation of treatment ► Secondary: increases the likelihood that treatment groups are balanced on factors

related to the outcome

► Masking ► Minimizes bias related to prior knowledge or beliefs about treatment

18

Why Randomized Trials Are the Gold Standard—2

► Exposure is randomized ► Primary: removes bias (selection bias)

in the allocation of treatment ► Secondary: increases the likelihood

that treatment groups are balanced on factors related to the outcome (confounding is minimized)

► Masking ► Minimizes bias related to prior

knowledge or beliefs about treatment

19

Why Randomized Trials Are the Gold Standard—3

► Exposure is randomized ► Primary: removes bias (selection bias)

in the allocation of treatment ► Secondary: increases the likelihood

that treatment groups are balanced on factors related to the outcome (confounding is minimized)

► Masking ► Minimizes bias related to prior

knowledge or beliefs about treatment (information bias)

20

Randomized Trials and Cause

Main types of epidemiologic study designs

Study type Characteristics

Experimental Studies prevention and treatment of disease Investigator actively manipulates which groups receive the study agent/intervention

Observational Studies causes, prevention, and treatment for diseasesInvestigator watches as natures takes its course

Cohort Examines multiple health effects of an exposureSubjects defined by exposure levels and followed for disease occurrence

Case-control Typically examines multiple exposures in relation to a diseaseSubjects are defined as cases and controls, and exposure histories compared

Cross-sectional Examine relationship between exposure and disease prevalence in a defined population at one point in time

Ecological Examines relationship between exposure and disease with population-level data rather than individual data

Clinical observations Physician might notice more cases than usual or perhaps a sequence of events suggesting a correlation

► A→ B (if A changes, then B changes)

21

Why Can’t We Just Use Randomized Trials?

► Generalizability of the study findings?

► Unethical to randomize to a harmful treatment

22

Steps in Epidemiologic Reasoning—1

► Determine whether there is an association between a factor and the disease

► If an association is found, determine whether the observed association is likely to be a causal one

23

Steps in Epidemiologic Reasoning—2

► Determine whether there is an association between a factor and the disease

► If an association is found, determine whether the observed association is likely to be a causal one (“causal inference”)

The material in this video is subject to the copyright of the owners of the material and is being provided for educational purposes under rules of fair use for registered students in this course only. No additional copies of the copyrighted work may be made or distributed.

Causal Inference and Guidelines

2

Causal Inference—1

► The thought processes and methods that assess whether a relation of cause to effect exists

Source: Porta. (2014). A dictionary of epidemiology (6th ed.).

3

Causal Inference—2

► Neither simple nor straightforward

► Requires judgment based on the totality of evidence

► We can *never* make a causal inference from the results of one single, observational study

4

Causal Guidelines—Infectious Disease

► Koch’s postulates 1. The organism is present in every case 2. The organism is never found in cases

of other diseases 3. The organism, when isolated from a

case, can produce the disease in experimental animals

Robert Koch (1843–1910)

Photo of Robert Koch in the public domain. Accessed February 1, 2019, at https://en.wikipedia.org/wiki/Robert_Koch#/media/File:RobertKoch_cropped.jpg

5

Causal Guidelines—Chronic Disease

► 1964 Surgeon General’s Report ► “Statistical methods cannot establish proof of a causal relationship in

an association. The causal significance of an association is a matter of judgment which goes beyond any statement of statistical probability. To judge or evaluate the causal significance of the association between the attribute or agent and the disease, or effect upon health, a number of criteria must be utilized, no one of which is an all-sufficient basis for judgment. These criteria include: 1. The consistency of the association 2. The strength of the association 3. The specificity of the association 4. The temporal relationship of the association 5. The coherence of the association”

Source: 1964 Surgeon General’s Report, p. 20.

6

Austin Bradford Hill (1897–1991)

► With Richard Doll, first to demonstrate the connection of smoking with lung cancer

► Developed theories of association and causation still used today to characterize causal relationships between exposure and disease

Photo by unknown. Creative Commons BY 4.0. Accessed January 31, 2019, at Wikipedia.

7

Hill and Consideration of Cause

► “Disregarding then any such problem in semantics we have this situation. Our observations reveal an association between two variables, perfectly clear-cut and beyond what we would care to attribute to the play of chance. What aspects of that association should we especially consider before deciding that the most likely interpretation of it is causation?”

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

8

Hill’s Guidelines for Causation—1

1. Strength of the association ► How strong is the measure of association (e.g., OR, RR) between a factor and an

outcome? ► Compared to weak associations, strong associations may be less likely to be explained

only by bias, confounding, or chance

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

9

Strength of the Association

Lung cancer and CHD mortality in male British physicians by smoking status—age adjusted mortality rates per 100,000:

Disease Smokers Non-smokers RR AR AR%

Lung cancer 140 10 14.0 130 92.9

CHD 669 413 1.6 256 38.3

10

Hill’s Guidelines for Causation—2

2. Consistency ► Has this association been observed in epidemiologic studies in different places,

populations, and times? ► If the association is causal, it would be expected that it could be replicated in other

populations

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

11

Consistency

► Social Relationships and Mortality Risk: A Meta-analytic Review (Julianne Holt-Lunstad, Timothy B. Smith, and J. Bradley Layton)

► “Social relationships, or the relative lack thereof, constitute a major risk factor for health—rivaling the effect of well established health risk factors such as cigarette smoking, blood pressure, blood lipids, obesity and physical activity” — House, Landis, and Umberson; Science 1988

Source: Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). PLoS Med, 7(7), e1000316. https://doi.org/10.1371/journal.pmed.1000316

12

Forest Plot of Functional Measures

► Functional measures ► Received

social support

► Perceptions of social support

► Loneliness (inversed)

► Outcome: mortality

Source: Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). PLoS Med, 7(7), e1000316. https://doi.org/10.1371/journal.pmed.1000316

13

Conclusions: Social Relationships and Mortality Risk

► “What Do These Findings Mean? These findings indicate that the influence of social relationships on the risk of death are comparable with well-established risk factors for mortality such as smoking and alcohol consumption and exceed the influence of other risk factors such as physical inactivity and obesity. Furthermore, the overall effect of social relationships on mortality reported in this meta-analysis might be an underestimate, because many of the studies used simple single-item measures of social isolation rather than a complex measurement. Although further research is needed to determine exactly how social relationships can be used to reduce mortality risk, physicians, health professionals, educators, and the media should now acknowledge that social relationships influence the health outcomes of adults and should take social relationships as seriously as other risk factors that affect mortality, the researchers conclude.”

Source: Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). PLoS Med, 7(7), e1000316. https://doi.org/10.1371/journal.pmed.1000316

14

Hill’s Guidelines for Causation—3

3. Specificity ► Is the exposure associated with only one disease and is the disease associated with

only one exposure? ► Specificity holds for infectious disease, but is considered one of the weaker causal

guidelines when evaluating cause for a chronic disease

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

15

Specificity

► Examples ► Mycobacterium tuberculosis causes tuberculosis ► Bacillus anthracis causes anthrax

► Doesn’t work as well for non-infectious agents, for example… ► Cigarette smoking causes lung and other cancers, cardiovascular disease, low birth

weight, and many other poor health states ► Lung cancer is caused by cigarette smoking, but also radon, asbestos, indoor cooking

fuels

16

Hill’s Guidelines for Causation—4

4. Temporality ► Does the exposure precede the outcome? ► Temporality is the one causal guideline that must be met in order for a factor to cause a

disease

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

17

Temporality

► Easiest to establish in randomized trials, prospective cohort studies

► More difficult to establish with other study designs

18

Hill’s Guidelines for Causation—5

5. Biological gradient ► Does risk of disease increase as the exposure to the factor increases? ► Also referred to as “dose-response”

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

19

Biological Gradient—1

Dementia incidence in 639 adults followed for >10 years in the Baltimore Longitudinal Study on Aging (BLSA)

Source: Lin et al. (2011). Arch Neuro.

20

Biological Gradient—2

Risk of incident all-cause dementia (compared to normal hearing)*

Severity HR 95% CI p

Mild 1.89 1.00 – 3.58 .05

Moderate 3.00 1.43 – 6.30 .004

Severe 4.94 1.09 – 22.4 .04

*Adjusted for age, sex, race, education, DM, smoking, and hypertension Source: Lin et al. (2011). Arch Neuro.

21

Hill’s Guidelines for Causation—6

6. Biological plausibility ► Is current biological knowledge consistent with a causal association? ► Was an association between the exposure and the disease anticipated a priori based

on current biological knowledge?

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

22

Hill’s Guidelines for Causation—7

7. Coherence ► Is a causal association consistent with what is known from scientific investigations from

other disciplines? ► An association should not conflict with what is known about the natural history and

biology of the disease

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

23

Hill’s Guidelines for Causation—8

8. Experiment ► If preventive actions regarding an exposure are taken, is the disease, in fact,

prevented? ► Can be an unplanned, or “natural” experiment

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

24

Hill’s Guidelines for Causation—9

9. Analogy ► Is there a similar (analogous) association that is known to be causal? ► For example, a drug is known to cause birth defects; if a similar drug is associated with

birth defects, that association may also be causal

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

25

Hill on the Application of These Guidelines

► “Here then are nine different viewpoints from all of which we should study association before we cry causation. What I do not believe—and this has been suggested—is that we can usefully lay down some hard-and-fast rules of evidence that must be obeyed before we accept cause and effect. None of my nine viewpoints can bring indisputable evidence for or against the cause-and-effect hypothesis and none can be required as a sine qua non. What they can do, with greater or less strength, is to help us to make up our minds on the fundamental question—is there any other way of explaining the set of facts before us, is there any other answer equally, or more, likely than cause and effect?”

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

26

Hill’s Causal Guidelines

1. Temporal relationship (the only guideline that must be met) ► If met, it is less likely that an observed association is due to

confounding or bias (or chance): 2. Strength of the association 3. Consistency 4. Specificity 5. Biological gradient 6. Biological plausibility 7. Coherence 8. Experiment 9. Analogy

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300.

27

Other Guidelines to Consider

► Consideration of alternate explanations ► Did the authors consider how bias and confounding may

have impacted their measure of association?

► Cessation of exposure ► Does risk of the outcome decline when exposure to the

factor is reduced?

Source: Gordis, L. (2014). Epidemiology (5th ed.).

28

Implications of a Causal Conclusion—1

► “The judgment that smoking causes a particular disease has immediate implications for prevention of the disease. Having reached a causal conclusion, one of the immediate and appropriate next steps is to estimate the burden of disease that might be avoided through prevention and cessation of smoking...”

Source: US Surgeon General’s Report 2004.

29

Implications of a Causal Conclusion—2

► “…There are also implications of not reaching a causal conclusion… [T]he evidence review may indicate needed areas of research to address remaining gaps and uncertainties that have precluded a causal designation”

Source: US Surgeon General’s Report 2004.

30

Beyond Cause—1

► Effect modification (interaction) ► Does the association between exposure and disease

differ by level of a third factor?

► Generalizability (external validity) ► Does the relationship observed in the study population

hold in other target populations?

Source: US Surgeon General’s Report 2004.

31

From Evidence to Action—1

32

From Evidence to Action—2

33

Beyond Cause—2

► Translation to policy ► Can the exposure be modified? ► Is it cost-effective to intervene? ► Are there mechanisms in place to support intervention? At

what level?

34

Hill on Public Health Action

► “All scientific work is incomplete—whether it be observational or experimental. All scientific work is liable to be upset or modified by advancing knowledge. That does not confer upon us a freedom to ignore the knowledge we already have, or to postpone the action that it appears to demand at a given time.”

Source: Hill, A. B. (1965). Proc R Soc Med, 58, 295-300. Photo by unknown. Creative Commons BY 4.0. Accessed January 31, 2019, at Wikipedia.

35

Lessons Learned

► Defined “risk factor” and “cause”

► Identified properties of a cause ► Direct vs. Indirect ► Necessary vs. Sufficient

► Described the process of causal inference

► Defined guidelines for judging causality