Multifactorial causation

Question 1

Which of the following statements on multi-factorial disease causation is incorrect?

A. Multi-factorial causation comprises several factors, or components, that result in sufficient cause.
B. It is necessary to identify all of the components before prevention can be successful.
C. Multi-factorial have many components, none of which are strictly necessary or sufficient on their own to cause disease
D. In multi-factorial causation, every cause also has multiple causes.
E. In multi-factorial causation, causes of disease are interacting.

Answer 1

B

Question 2

Which of the following statements on disease causation is correct?

A. A necessary cause will always produce disease regardless of other factors.
B. A necessary cause is a combination of causes that will inevitably cause disease
C. A component cause is sufficient to cause disease on it’s own.
D. A cause of a disease increases a person’s risk of developing the disease.
E. Disease is always the result of a single event or exposure

Answer 2

D

Question 3

An outbreak of Salmonella occurs at a cafe. What type of causal factor is Salpmonella typhimurium?

A. Predisposing

B. Enabling

C. Precipitating

D. Reinforcing

Answer 3

C

Precipitating= specific disease agents, e.g. microorganisms

Question 4

In a study of Aboriginal and Torres Strait Islander women, it was found that having a partner who smoked was associated with an increased risk of smoking during pregnancy. This is an example of which type of causal factor?

A. Predisposing

B. Enabling

C. Precipitating

D. Reinforcing

Answer 4

D

Reinforcing factors include repeated exposure to factors or environmental conditions that may aggravate an established condition

Question 5

List some measures that can be taken to coorrect for confounding

Answer 5

• Multivariate analusos
• Restriction
• Matching
• Stratification

n.b. multivariate analysis: confoudning variables must be identfied and measured (and measured properly)

Question 6

All of the following are important criteria when making causal inferences except for

A. Dose-response relationship

B. Consistency with existing knowledge

C. Strength of association

D. Predictive value

E. Temporal relationship between exposure and disease

Answer 6

D

Question 7

In a large case-control study of pancreatic cancer, 17% of cases were found to have type 2 diabetes at the time of cancer diagnosis, compared to 4% of a well-matched control group (matched by age, sex and several other characteristics) that were examined for diabetes at the same time as the cases were diagnosed. Can it be concluded that type 2 diabetes causes pancreatic cancer?

Yes/No and why?

Answer 7

No, because the study failed to demonstrate the time sequence between onset of diabetes and diagnosis of pancreatic cancer

Information on exposure was collected at time of cancer diagnosis

Question 8

B1 Revision

What are the four types of systematic error or bias?

Answer 8

• Recall bias
• Observer bias
• Measurement bias (or information bias)
• Selection bias

Question 9

A busy, inner city GP practice conducted a survey to evaluate patient satisfaction with waiting times for GP appointments. Questionnaires were placed in the waiting room on one day of the week, and 300 patients attended the practice for appointments on that day. By the end of the day, 45 questionnaires had been completed (response rate 15%). What is the main limitation in the design of this survey?

Answer 9

Selection bias - due to very low response rate.
Note:
Completion of questionnaire may be related to outcome i.e. waiting times: thsoe waiting longer may be more likely to complete questionnaire and express dissatisfaction

Question 10

What is a cause?

Answer 10

An event, condition, or characteristic [or a combination of these factors] that plays an essential role in producing the occurrence of disease. (Webb & Bain p239)

In other words, a cause of a disease increases a person’s risk of developing the disease.

Question 11

True or false: a hazard ration is essentially synonymous with risk ratio

Answer 11

True*

Question 12

B1 revision

List the types of causes and provide examples

Answer 12

A necessary cause - any agent that is required for the development of a given disease (e.g. a specific infectious agent). An outcome cannot develop in its absence. e.g. presence of A necessary cause - any agent that is required for the development of a given disease (e.g. a specific infectious agent). An outcome cannot develop in its absence. e.g. P.carinii and pneumonia (but not all who have P.carinii develop pneumonia…), HIV infection for AIDS

A component cause - a factor that contributes towards disease causation but is not sufficient to cause disease on its own e.g. smoking and lung cancer

A sufficient cause - a factor (or usually a combination of factors) that will inevitably produce disease.
OR
A minimum set of conditions and events that are sufficient for the outcome to occur.
This is a feature of multi-factorial causation e.g. smoking, and asbestos, for lung cancer.

n.b. Some can be both necessary and sufficient e.g. Tay-Sachs mutation

Equally, some causes may be neither necessary not sufficient e.g. for complex chronic diseases e.g. cancer

e.g.s from CDC textbook, and link in footnote

N.B. Public health action does not depend on identifying every component cause but diseae prevention focuses instead on blocking any single component of a sufficient cause (through that pathway) e.g. elimination of smoking– which would prevent lung cancer where smoking greatly contributes

https://www.verywellhealth.com/understanding-causality-necessary-and-sufficient-3133021

Question 13

Describe multi-factoral causation

Answer 13

Multi-factorial causation therefore comprises several factors, or components, that result in “sufficient” cause.

It is not necessary to identify all of the components of a sufficient cause before prevention can be successful.

The removal of one component may interfere with the others and therefore prevent the development of the outcome.

Chronic diseases are multi-factorial - there are many components, none of which are strictly necessary or sufficient on their own to cause disease. See the smoking and lung cancer example below.

Question 14

List and describe the factors involved in causation of disease

Answer 14

• Predisposing factors: age, sex, genetics, previous illness.
• Enabling factors: factors that favour the development of disease, such as poor nutrition, low income, inadequate medical care, or assist in the recovery from disease. Social determinants of health are important enabling factors.
• Precipitating factors: exposure to a specific disease agent.
• Reinforcing factors: repeated exposure to factors or environmental conditions that may aggravate an established condition.

Question 15

B1 revision

Describe the hierarchy of study designs

Answer 15

he investigations that explore a possible cause of a disease often arise initially from clinical observation.

Following this, exploratory studies using routinely available data (e.g. from vital statistics, hospitalisation records, disease registry or surveillance data) may shed some light on the question.

New studies, such as case-control and cohort studies, are then carried out that are designed to determine if there is an association between an exposure and a disease, and whether a causal relationship exists.

Ultimately, a randomised controlled trial would be the next step as it is the strongest study design to test a causal relationship. However, RCT’s are usually used to test the effects of potential beneficial agents and not the effects of harmful exposures.

The greatest weight is given to systematic reviews and meta-analyses of randomised controlled trials (we will cover these when we examine evidence synthesis later in the course).

Question 16

What are the questions to ask when determining if association is causal?

Answer 16

1. Consider alternative non-causal explanations.

Could the observed association be an artefact due to chance, bias or confounding?

Is the association due to a chance occurrence?
Is it due to a flaw in the methodology (bias)?
Is it due to some other factor linked to both the exposure and the outcome (confounding)?

2. A formal evaluation of whether an observed association is causal

If we are confident from our critical appraisal of a study that the association is not due to chance, bias or confounding we can consider additional criteria to assess whether the association is causal.

Question 17

List and describe Hill criteria for causation

Answer 17

Temporality

The exposure precedes the occurrence of disease (essential). Is easier to establish in a prospective cohort study than a case-control study. Also important to understand the length of the interval between exposure and disease.

Plausibility

The causal explanation makes biological sense (essential)

Replication & consistency

Similar results from other studies in a variety of different situations and in different populations that are unlikely to share the same bias.

Strength

What is the strength of association between the exposure and disease? A relative risk (or odds ratio) of 3 or 4 is less likely to be due to bias than one of 2 or less.

Dose-response relationship

Is increased exposure to the possible cause associated with increased effect?

Reversibility (cessation of exposure)

Does the removal or reduction of a possible cause lead to a reduction in the disease risk?

Specificity

Are the findings specific? Is the association specific to one disease?

Many diseases have several causes and many exposures cause several diseases (e.g. tobacco). However, may be useful for some exposures, for example where an association may be limited to a particular group with a particular environmental exposure.

Support from experimental data

Including animal or in vitro studies and intervention studies in humans

Question 18

Which of Hill’s criteria as essential to establishing causality?

Answer 18

It is not an absolute requirement that all criteria are met in order for a exposure to be considered causal. However, a temporal association and biological plausibility are ESSENTIAL criteria.

Question 19

List the two broad measures that can be taken to control confounding

Answer 19

• Study design
• Analysis

Question 20

Describe residual confounding

Answer 20

We can not adjust for confounding factors if they have not been measured. Also, any confounders considered also need to be adequately measured. When assessing an association to be causal you should always consider whether all important potential confounding factors have been considered in the study, and that they have been measured sufficiently. Otherwise, residual confounding may exist.

Question 21

What are the criteria for a confounding variable?

Answer 21

For something to be a confounder it must:

• be a risk factor for the disease in it’s own right, and
• be associated with the risk factor of interest
• must not be an intermediate factor between the exposure and outcomeNote:
  A confounder may:
  Account for all or part of an apparent association
  May cause an overestimate or an underestimate of a true association.

Question 22

Describe the consequences of random and systematic error on results

Answer 22

a) Little systematic or random error: an experienced shooter using a gun with the sights properly aligned - shots will cluster closely around the bulls-eye. The estimate is precise and accurate.

b) Systematic error but little random error: an experienced shooter using a gun with sights not aligned correctly - shots will cluster closely but will not be accurate as consistently falling to the right. The estimate is precise but not accurate.

c) Random error but little systematic error: a less experienced shooter using the first gun - will be accurate but the shots would be more spread out. The result will not precise but will be accurate.

d) Random and systematic error: the less experienced shooter using the second gun. The result will not be precise nor accurate.

Question 23

What is random error and what are the types of random error?

Answer 23

Each sample will include slightly different people - their characteristics will be slightly different from those in other samples - just by chance.

This is known as random sampling error. It is one of the three main sources of random error in addition to biological variation and measurement error.

There are two related ways to evaluate whether an association is due to chance.

1. Statistical significance (hypothesis) testing
2. Confidence intervals

Question 24

What are some ways we can determine clinical significance of results?

Answer 24

Unfortunately, there is no clear answer to how big an effect should be for it to be meaningful.

An relative risk of 2.0 is generally considered to be strong and therefore practically significant.

A RR of less than 2 should not be immediately dismissed without considering the risk (i.e the incidence) in the unexposed group or the total population under consideration.

A small association for a relatively common disease can result in a large number of additional cases; conversely a moderate or strong association for a rare disease may only result in a few additional cases.

balancing statistical and practical or clinical significance for an exposure-disease association can be challenging. Decisions relating to behavioural risk factors, such as coffee consumption, often become a trade off between risks and benefits, plus other complex decision making factors.

Also, a risk factor may be related to several diseases but in different ways. Alcohol consumption is a good example of this - while known to be associated with an increased risk of several cancers, moderate consumption is consistently found to be protective for cardiovascular disease.

In other examples, the funding of new expensive drugs that may only show small gains in survival compared to the cheaper alternative raises the question of what represents a clinically significant improvement and who should pay for it?