Stats - Association, Causation, Confounding and Bias

Question 1

Two variables are said to be associated when one is found more commonly in the presence of the other.
What are the three types of association?

Answer 1

Spurious
Indirect
Direct

Question 2

What is a spurious association between variables?

Answer 2

Spurious – A spurious association occurs when the relationship between two variables appears to exist, but it is false or misleading due to the influence of a confounding variable or random chance. In this case, there is no true relationship between the two variables. The confounding factor creates an illusion of a link, but the association is not real. For example, the apparent relationship between ice cream sales and drowning deaths is spurious, as the real factor influencing both is hot weather, not a direct or indirect link between the two variables.

Question 3

What is an indirect association between variables?

Answer 3

Indirect – An indirect association occurs when two variables are genuinely related, but the relationship is mediated or explained by a third variable (a confounder). In this case, the association is real, but the relationship is not direct. For example, socioeconomic status might be linked to health outcomes, but the true mechanism might be mediated through access to healthcare or lifestyle factors, making the relationship indirect. Unlike spurious associations, an indirect association reflects a real connection between the variables, just mediated by the confounder

Question 4

What is a direct association between variables?

Answer 4

Direct – A direct association exists when one variable directly affects another without the involvement of any confounding variable. This means that the relationship is causal in nature, where changes in one variable directly bring about changes in the other. For example, smoking directly causes lung cancer without the need for any intermediary factor.

Question 5

Once the association has been established, the next question is whether the association is causal. Not all associations imply causality, so additional evidence is needed to confirm a cause-and-effect relationship.

What criteria is used to assess causality?

Answer 5

To assess causality, the Bradford Hill Causal Criteria are commonly used

Question 6

What are the 5 parts of the Bradford Hill Causal Criteria?

Answer 6

Strength – A stronger association (e.g., higher relative risk) increases the likelihood that the relationship is causal rather than due to chance or bias. For example, the strong link between smoking and lung cancer supports a causal relationship.

Temporality – The cause must precede the effect. This is a fundamental criterion; if the outcome occurs before the exposure, then it cannot be causal.

Specificity – If a cause leads to a specific effect or disease, the association is more likely to be causal. This criterion has limitations, as many causes can lead to multiple effects (e.g., smoking causes several diseases).

Coherence – The association should align with existing biological and epidemiological knowledge. For example, if an observed association fits with known biological mechanisms, this strengthens the argument for causality.

Consistency – If the same association is observed across different studies, populations, and settings, it strengthens the case for causality. For example, the consistent association between asbestos exposure and mesothelioma across various studies is a strong indicator of a causal relationship.

Question 7

What s the name given to the situation in a trial where one outcome is systematically favoured?

Answer 7

Bias

Question 8

Give 6 examples of selection bias:

Answer 8

1) sampling bias where the subjects are not representative of the population. This may be due to volunteer bias.
2) non-responder bias e.g If a survey on dietary habits was sent out in the post to random households it is likely that the people who didn’t respond would have poorer diets than those who did.
3) loss to follow up bias
4) prevalence/incidence bias (Neyman bias): when a study is investigating a condition that is characterised by early fatalities or silent cases. It results from missed cases being omitted from calculations
5) admission bias (Berkson’s bias): cases and controls in a hospital case control study are systematically different from one another because the combination of exposure to risk and occurrence of disease increases the likelihood of being admitted to the hospital
6) healthy worker effect

Question 9

What is Neyman bias?

Answer 9

Prevalence/incidence bias (Neyman bias): when a study is investigating a condition that is characterised by early fatalities or silent cases. It results from missed cases being omitted from calculations

Question 10

What is Berkson’s bias?

Answer 10

Admission bias (Berkson’s bias): cases and controls in a hospital case control study are systematically different from one another because the combination of exposure to risk and occurrence of disease increases the likelihood of being admitted to the hospital

Question 11

What type of bias?

Difference in the accuracy of the recollections retrieved by study participants, possibly due to whether they have disorder or not. E.g. a patient with lung cancer may search their memories more thoroughly for a history of asbestos exposure than someone in the control group. A particular problem in case-control studies.

Answer 11

Recall bias

In retrospective studies where participants are asked to remember their past exposure to risk factors, it is likely that cases will have thought more about what factors in their past may have caused a disease than controls will have. Controls are therefore less likely to remember an exposure because they don’t link it to any disease process, which may skew the results.

Question 12

What type of bias?

Failure to publish results from valid studies, often as they showed a negative or uninteresting result. Important in meta-analyses where studies showing negative results may be excluded.

Answer 12

Publication bias

Question 13

What type of bias?

In studies which compare new diagnostic tests with gold standard tests, work-up bias can be an issue. Sometimes clinicians may be reluctant to order the gold standard test unless the new test is positive, as the gold standard test may be invasive (e.g. tissue biopsy). This approach can seriously distort the results of a study, and alter values such as specificity and sensitivity. Sometimes work-up bias cannot be avoided, in these cases it must be adjusted for by the researchers.

Answer 13

Work-up (Verification) bias

Question 14

What type of bias?

Only a problem in non-blinded trials. Observers may subconsciously measure or report data in a way that favours the expected study outcome.

Answer 14

Pygmalion effect - Expectation bias

Question 15

What type of bias?

Describes a group changing it’s behaviour due to the knowledge that it is being studied

Answer 15

Hawthorne Effect

Question 16

What type of bias?

Gathering information at an inappropriate time e.g. studying a fatal disease many years later when some of the patients may have died already

Answer 16

Late-Look bias

Question 17

What type of bias?

Occurs when subjects in different groups receive different treatment

Answer 17

Procedure bias

Question 18

What type of bias?

Occurs when two tests for a disease are compared, the new test diagnoses the disease earlier, but there is no effect on the outcome of the disease

Answer 18

Lead-time bias

Lead time is the period between early detection of disease and the time of its usual clinical presentation. When evaluating the effectiveness of the early detection and treatment of a condition, the lead time must be subtracted from the overall survival time of screened patients to avoid lead time bias. Otherwise early detection merely increases the duration of the patients’ awareness of their disease without reducing their mortality or morbidity. Numerous cancer screening procedures were thought to improve survival until lead time bias was addressed.

Question 19

What type of bias?

A form of bias that occurs when measurement of information differs among study groups examples include recall bias, reporting bias, diagnostic bias, and Hawthorne effect, errors in measurement

Answer 19

Information bias

Question 20

What type of bias?

Distortion of exposure, disease relation by some other factor

Answer 20

Confounding bias

Question 21

What type of bias?

This can occur when conclusions about individuals are based only on analyses of group data

Answer 21

Ecological Fallacy

Question 22

What type of bias?

This can occur when exposure can influence diagnosis. For example women taking an oral contraceptive will have more frequent cervical smears than women who are not on the pill and so are more likely to have cervical cancer diagnosed (if they actually have it). Thus, in a case-control study that compared women with cervical cancer and a control group, at least part of any higher pill consumption rates amongst the former group may be due to this effect

Answer 22

Detection bias

Question 23

What type of bias?

Articles of high citation are easy to reach and have higher chance to be entered into a given study.

Answer 23

Citation bias

Question 24

What type of bias?

This can occur when a treatment is studied in more severe forms of a disease. Such results may then not apply to mild forms of the disease.

Answer 24

Disease spectrum bias (aka case-mix bias)

Question 25

What type of bias?

Where the subjects are not representative of the population. This may be due to volunteer bias (aka referral bias). An example of volunteer bias would be a study looking at the prevalence of Chlamydia in the student population. Students who are at risk of Chlamydia may be more, or less, likely to participate in the study

Answer 25

Sampling bias

Question 26

What is selection bias vs information bias?

(Two main categories that the other subtypes fall under)

Answer 26

Selection bias - when selected sample is not a representative sample of reference population

Information bias - when gathered information about exposure, outcome or both is not correct and there was an error in measurement

Question 27

What type of bias?

Interviewer or observer knowledge about in-question hypothesis and disease or/and exposure can take effect on collection and registry of data.

Answer 27

Interviewer/ observer bias

Question 28

What name is given to a variable that is associated with both the outcome and the exposure but has no causative role.

Answer 28

Confounding factor

Confounding can be addressed in the design and analysis stage of a study. The main method of controlling confounding in the analysis phase is stratification analysis.

Question 29

What occurs when there is a non random distribution of risk factors in the populations. Age, sex and social class are common causes of confounding.

Answer 29

Confounding

Question 30

What are the three main ways (in the design stage) to address confounding?

Answer 30

1) Matching (e.g. By age and gender) - an active form of control
2) Randomization (which aims to produce an even amount of potential risk factors in two populations)
3) Restriction of participants (e.g. If watching TV is a known confounder then restrict participants to ones who don’t watch TV)

Question 31

What are the two main ways (in the analysis stage) to address confounding?

Answer 31

1) stratification - a statistical technique that allows to control for confounding by creating two or more categories (strata) in which the confounding variable either does not vary or does not vary very much.
2) multivariate models (e.g. logistic regression, linear regression, analysis of covariance (ANCOVA))

Question 32

What are 4 graphical methods used to detect publication bias?
FP
GP
OFP
NQP

Answer 32

Funnel plot
Galbraith plot
Ordered forest plot
Normal Quantile plot

Question 33

What is a funnel plot?

Answer 33

A funnel plot is a graph used to check for publication bias in systematic reviews and meta-analyses. They are a form of scatter graph that offers an easy visual way of making sure that the published literature is evenly weighted (drug companies have a habit of withholding data that doesn’t support the product).

Question 34

What do the x and y axes represent on a funnel plot?

What do the dots represent?

Answer 34

The x-axis represents some measure of effect size (often a risk ratio) and the y-axis represents some measure of the study size (often the standard error). By convention, the y-axis represents a reverse of the standard error (zero at the top and standard error gets larger towards the bottom).

Each dot on the funnel represents a differ trial in a meta-analysis. Larger trials tend to have smaller standard errors and are located towards the top and smaller studies, with larger standard errors, towards the bottom.

Question 35

What should a funnel plot look like if there is no publication bias?

Answer 35

A pyramid, or symmetrical inverted funnel

Question 36

What does an asymmetrical funnel plot indicate?

Answer 36

An asymmetrical funnel indicates a relationship between treatment effect and study size. This indicates either publication bias or a systematic difference between smaller and larger studies (‘small study effects’)