Upgrade to remove ads
Only $2.99/month

Chapters 14-16

Terms in this set (64)

Exposure-Odds Ratio
Used in case-control studies

Ratio of odds in favor of exposure among the cases to the odds in favor of exposure among non-cases.
Disease-Odds Ratio
Used in cohort or cross sectional studies

Ratio of odds in favor of disease among the exposed to the odds in favor of disease among unexposed.
Prevalence Odds Ratio
Refers to an odds ratio derived cross-sectionally

The odds ratio is derived from studies of prevalent (rather than incident) cases.
Be very familiar with tables chapter 13
Review slides 6-11
Approaches for studying the etiology of disease
Animal models
In-vitro systems (cell culture or organ culture)
Observations in human populations
If an association is observed, the first question we must always ask is:
"Is it real?"
If the observed association is real, is it causal?
A casual pathway can be either....
Direct or Indirect
Direct Causation
Direct: Factor directly causes a disease without any intermediate step
Indirect Causation
Indirect: A factor causes a disease but only through an intermediate step or steps
Note- In human biology, intermediate steps are virtually always present in any causal process
causal relationship #1: necessary and sufficient
A factor is both necessary and sufficient for producing the disease.Without that factor, that disease never develops.
In the presence of that factor, the disease always develops.
This situation rarely, if ever, occurs.
causal relationship #2: necessary, but not sufficient
Each factor is necessary, but not, in itself, sufficient to cause the disease.
Multiple factors are required and often in a temporal sequence

Example: Carcinogenesis is multistage process involving both initiation and promotion.
causal relationship #3: sufficient, but not necessary
Factor alone can produce the disease, but so can other factors that are acting alone

Example: Either radiation exposure or benzene exposure can produce leukemia without the presence of the other. Cancer does not develop in everyone who has experienced radiation or benzene exposure.
causal relationship #4: neither sufficient nor necessary
A factor, by itself, is neither sufficient nor necessary to produce disease.

More complex model, which probably most accurately represents causal relationships that operate in most chronic diseases.
Guidelines for judging whether an observed association is causal 1-8
1. Temporal Relationship
2. Strength of association
3. Dose response relationship
4. Replication of findings
5. Biologic plausibility
6. Consideration of alternate explanations
7. Cessation of exposure
8. Consistency with other knowledge
It is not so much a count of the number of guidelines present that is relevant to causal inference but rather an assessment of the total pattern of evidence observed that may be consistent with one or more of the guidelines.
1. Temporal Relationship
If factor is believed to be cause of disease, exposure to factor must have occurred before the disease developed.

Often easier to establish temporal relationship in prospective cohort study than in case-control study or retrospective cohort study . Note: In case-control and retrospective cohort study, exposure information may need to be obtained or re-created from past records and timing may therefore be imprecise.

Temporal relationship also important for clarifying length of interval between exposure and disease.

Example: Asbestos linked to increased risk of lung cancer, but latent period between exposure and disease is between 15-20 years.
2. Strength of association
Measured by relative risk (or odds ratio).
The stronger the association, the more likely relationship is causal.
3. Dose response relationship
As dose of exposure increases, the risk of disease also increases.

Note: If dose-response relationship is present, relationship may be causal.
However, absence of dose-response relationship does not necessarily rule out causal relationship.
4. Replication of findings
If relationship is causal, we would expect to find it consistently in different studies and different populations

Replication particularly important in epidemiology
5. Biologic plausibility
Refers to coherence with current body of biologic knowledge.
6. Consideration of alternate explanations
Have the investigators taken other possible explanations into account?
Have investigators ruled out such explanations?
7. Cessation of exposure
If factor is cause of disease, we would expect the risk of disease to decline when exposure to factor is reduced or eliminated.
8. Consistency with other knowledge
If relationship is causal, we would expect the findings to be consistent with other data.
Definition of Bias
Any systematic error in the design, [execution] or analysis of a study that results in a mistaken estimate of an exposure's effect on the risk of disease.

Bias is a systematic error in a study and cannot be fixed. It can be acknowledged and discussed.
Main types of Bias
Selection Bias & Information Bias
Selection Bias
If the way in which cases and controls, or exposed and non-exposed individuals, were selected into the study such that an apparent association is observed, this association is the result of selection bias.

In other words . . . selection bias is a method of participant selection that distorts the exposure-outcome relationship from that present in the target population
When does selection bias occur
Occurs when the selection of participants in one group results in a different outcome than the selection for the other group.
Selection bias can result from
Nonresponse: those who do not respond in a study differ from those who don't in regard to demographic, SES, cultural, lifestyle and medical characteristics.
Example of selection bias
A case-control study of alcoholism and pneumonia
Cases and controls selected from hospitalized patients
Alcoholics with pneumonia are more likely to be admitted than non-alcoholics with pneumonia
Risk of pneumonia associated with alcoholism is biased upwards.
Definition of exclusion bias (specific type of selection bias)
Exclusion bias: Results when investigators apply different eligibility criteria to the cases and to the controls in regard to which clinical conditions in the past would permit eligibility in the study and which would serve as the basis for exclusion.
Selection bias in a cross-sectional study
"Selective Survival" - only survivors are included in study

If exposed cases are likely to survive longer than unexposed cases OR

If unexposed cases are likely to survive longer than exposed cases THEN

Conclusions may differ from an appropriate cohort study
Other examples of selection bias #1
-Select volunteers as exposed group and non-volunteers as non-exposed group in a study of screening effectiveness
-Volunteers could be more health conscious than non-volunteers, thus resulting in less disease
-Volunteers could also be at higher risk, such as having a family history of illness, thus resulting in more disease
Other examples of selection bias #2
-Study health of workers in a workplace exposed to some occupational exposures comparing to health of general population
-Working individuals are likely to be healthier than general population that includes unemployed people
(Called Healthy Worker Effect)
-Use prevalent cases instead of incidence cases
Target population
The population to which the results of the study may be extrapolated out to, even if not all members of this population were eligible for sampling, and is often not clearly defined.
-External validity
Source population
The population from which the sample was taken, and therefore all members of this population should have a chance of being selected for inclusion in the study.
-Internal validity
Controlling selection bias
In a case-control study: define criteria of selection of diseased and non-diseased participants independent of exposures

In a cohort study: define criteria of selection of exposed and non-exposed participants independent of disease outcomes in a cohort study

Use randomized clinical trials
Information bias definition
Occurs when information is collected differently between the two groups, leading to an error in the conclusion of the association
Misclassification bias definition
When information is incorrect there is misclassification:
-Differential misclassification
-Non-differential misclassification
Misclassification example
For example, in a case-control study, some people who have the disease (cases) may be misclassified as controls, and some without the disease (controls) may be misclassified as cases.

Or, we may misclassify a person's exposure status: we may believe the person was exposed when the person was not exposed, or we may believe that the person was not exposed, when, in fact, exposure did occur.
Differential misclassification bias
Occurs when the level of misclassification differs between the two groups

Can lead either to an apparent association even if one does not really exist or to an apparent lack of association when one does in fact exist.
Non-differential misclassification bias
Occurs when the level of misclassification does not differ between the two groups.

Usual effect of non-differential misclassification is that the relative risk or odds ratio tends to be diluted, and it is shifted toward 1.0, or shifted toward the null hypothesis

In other words, we are less likely to detect an association even if one really exists
How can disease status be misclassified?
Incorrect Diagnosis

Subject Self-Report

Records Incorrectly Coded in Data-Base
Incorrect diagnosis examples
Limited knowledge about disease
Diagnostic process complex
Inadequate access to technology
Laboratory error
Disease subclinical
Detection bias (e.g., more thorough exam if exposed)
Subject Self-Report examples
Incorrect recall
Reluctant to be truthful (e.g., due to stigma)
How can exposure be misclassified
Imprecise measurement:
Poorly constructed questionnaire
Faulty measuring device or observation technique

Subject's self-report:
Recall of prior exposure status (more of a problem in case-control than cohort studies)
Reluctance to be truthful if behaviors are socially unacceptable

Interviewer bias:
May probe thoroughly for cases than controls

Incorrect coding of exposure data
Surrogate interview
Type of information bias where family members of case are interviewed.
Surveillance bias
If a population is monitored over a period of time, disease ascertainment may be better in the monitored population than in the general population and may introduce surveillance bias.
Recall bias
Cases may have better recall than controls.
Reporting bias
Subject may be reluctant to report an exposure he is aware of because of attitudes, beliefs and perceptions of interviewer.
Controlling information bias
Have a standardized protocol for data collection

Ensure that sources and methods of data collection are similar for all study groups

Ensure interviewers and study personnel are unaware of exposure/disease status
Controlling Information Bias
Have a standardized protocol for data collection

Ensure that sources and methods of data collection are similar for all study groups

Ensure interviewers and study personnel are unaware of exposure/disease status
Cofounding Occurs When...
The observed result between exposure and disease differs from the truth because of the influence of the third variable

Effect of a factor of interest is mingled with (confounded with) that of another factor

Confounding is a situation in which a measure of the effect of an exposure is distorted because of the association of exposure with other factor(s) that influence the outcome under study

Confounding occurs where an apparent association between a presumed exposure and an outcome is in fact accounted for by a third variable not in the postulated causal pathway; such a variable must be itself associated with both presumed exposure and outcome
A confounder must...
1. Be independently associated with the exposure of
interest
2. Be independently associated with the disease
3. Not be an intermediate link in the causal chain between
exposure and disease
Confounding has both direction and magnitude (strength).
Define Magnitude and Strength
The direction depends on the relationships between exposure, disease and the potential confounder.

The magnitude depends on the strength of associations between the confounder and exposure, and between the confounder and disease.
Positive confounding biases
Estimate away from the null.
The measure of association will over-estimate the true association and lead to a stronger increased risk or a more protective factor.
Negative confounding biases
Biases an estimate toward the null.
The measure of association will under-estimate the true association and lead to a weaker increased risk or a less protective factor.
Discussion of confounding should include its....
1.Presence/absence
2. Direction or expected direction
3. Magnitude
How to assess co-founding
1. Analyze data as though confounding was not present: obtain a "crude" measure of association

2. "Control" for the effect of confounder

3. Re-calculate the measures of association - 1 for each level of the confounder - and compare them to the crude estimate - if they are equal across levels of the confounder but differ from the crude estimate, then confounding is present.
Effect modification (interaction)
The presence or level of a third variable has a different impact on disease in different exposure groups.
-This 3rd variable modifies the effect of exposure on disease.
-The magnitude of the exposure-disease relationship depends on whether the 3rd variable is present or absent.
An effect modifier must be...
1. Independently associated with the disease
2. An intermediate link in a causal chain between exposure and disease
3. Has both direction and magnitude.
Issues with co-founding
May lead to errors in the conclusion of a study, but, when confounding variables are known, the effect may be fixed

Confounding can overestimate, underestimate, or change the direction of the observed effect.
Discussion of effect modification should include its:
1. Presence/absence
2. Direction or expected direction
3. The extent to which it differently modifies the effect of exposure in different exposure groups
-Magnitude alone is not enough

We do not want to "remove" the effects of effect modification. Instead, the goal is to evaluate the impact of the effect modifier then account for it in the interpretation
How to asses effect modification
1. Analyze data as though effect modification was not present: obtain a "crude" measure of association

2. "Control" for the effect of effect modifier

3. Re-calculate the measures of association - 1 for each level of the effect modifier - and compare them to the crude estimate - if they are different at different levels of the effect modifier and differ from the crude estimate, then effect modification is present.
NOTE: STRATIFIED ESTIMATES MUST DIFFER FROM EACH OTHER
Synergism
Increase disease risk beyond expected; also known as positive
-person with exposure more susceptible to another exposure
Antagonism
decrease disease risk beyond expected ; also known as negative interaction
-person with exposure less susceptible to another
THIS SET IS OFTEN IN FOLDERS WITH...