EPIB final

Bias is a

systematic error

• Can mask an association when there really is one

• Can create spurious association when there’s none

Two types of bias are…

selection and observational/information

What is Selection bias?

process of selecting participant affects exposure and outcome

• changes target population

What is self-selective bias?

• selection bias

individuals decide for themselves whether to participate, they may share characteristics that differ systematically from those who do not volunteer

What is non-response Bias?

• selection bias

individuals might not respond or participate

What is Loss to Follow-up (Attrition Bias)?

• selection bias

Participants drop out of study

what is Hospital-based Selection (Berkson's Bias)?

• selection bias

In studies conducted within a hospital setting, the relationship between an exposure and a disease may be overestimated

What is Survival Bias?

• selection bias

researchers focus on those who did survive, not those who didn’t

What is Healthy Worker Bias?

• selection bias

healthy people → more likely to work

choosing people in occupational are more healthy

What can you do to fix selection bias?

• use the same criteria

• large sample size

• obtain all relevant subject records

What is information bias?

Occurs after subjects have entered the study

What is recall bias?

• information bias

people might not recall their events well

What is recall bias bad?

Can result in over/under-estimate of measure of association

What is Interviewer Bias?

• information bias

interviewer scopes for much information from specific participants (info is sought out)

What is Misclassification bias?

Occurs when the subject’s exposure or disease status is

erroneously classified

→ cause groups to be more similar than they actually are

What is Confounding?

The relationship between an exposure and a disease outcome is influenced by a 3rd

variable

Prevention strategies attempt to control confounding through the study design itself are?

• Randomization

• Restriction

• Matching

Analysis strategies seek to adjust measures of effect to remove confounding distortions are?

• Stratification

• Multivariate techniques

Two pros of randomization?

Convenient, inexpensive

Two cons of randomization

• need large sample size

• need control over exposure and ability to assign

Pros of restricting?

• Provides complete control of known confounders

Cons on restricting

• cannot control for unknown confounders

two pros of matching

• Fewer subjects are required

• May enhance the validity

Cons of matching

• costly

Two pros of stratification?

• Logical strategy with straightforward computation

• Minimal assumptions must be

two cons of stratification?

• small data in some strata

• loss of data due to categorizing

Two pros of multivariate techniques?

• Can use continuous variables

• Allow for control of multiple variables in a single analysis

Two cons of multivariate techniques?

• Requires statistical software -

• Potential for misuse

ODDS RATIO is used for?

case-control studies

Interpretation of ODDS RATIO?

The [CASES] have [OR] times higher/lower odds of [EXPOSURE] compared to the [CONTROLS]

How to calculate attributable risk?

AR = incidence *exposed - incidence *unexposed

What is attributable risk?

Additional risk of disease in the exposed that can be attributed to the specific exposure

What is Attributable risk % (etiologic fraction)?

Proportion of incident cases in the exposed group that are due to the exposure

How do you calculate AR%?

AR% = AR / incidence *exposed x 100%

How do you calculate relative risk?

= (a/[a+b]) / (c/[c+d])

RR is used for?

cohort studies

Interpretation for RR?

The [EXPOSED] have [RR] times higher risk of [OUTCOME] compared to the [UNEXPOSED]

What is random error?

Error due to chance

How can random error occur?

• poor precision

• variability in measurement

• sampling error

p-value

the probability that the results observed are due to chance.

(less than) <0.05 p value means

typically considered statistically significant

>0.05 p value means

not statistically significant

systematic error is

Error due to flaws in the study design or conduct

systemic error is

not due to ERROR

● Does not decrease with larger sample size

Confounders must be…

1. associated with outcome/exposure

2. Not an intermediate step on the causal pathway

What is screening?

recognition of disease through examinations

Does screening provide a definite diagnosis?

NO, provides info for diagnostic exam

What is an example of screening?

Newborn screening → stick-heel test with newborn’s blood is done within 48 hrs of birth. Any records should be sent to the pediatrician.

What is Non-Invasive Prenatal Testing (NIPT)?

NIPT - testing for chromosomal disorders

what are the types of screening?

Multiphase

Mass

Selective

What is Multiphase screening

the use of two or more tests together among large groups of people

What is mass screening?

screen large proportions of the population for a given disease regardless of risk status

What is selective screening?

screening only those who have higher risk of disease

What is opportunistic screening

detection of conditions unrelated to the patient’s chief complaint

(BP rate)

Why is selective screening good?

• more cost effective

• likely to yield more cases

Explain the iceberg concept of disease. Why does this matter?

• The tip of the iceberg, corresponds to active clinical or apparent disease 

• Accounts for only a small proportion of host infections and exposures to disease agents

Reliability (precision)

produces stable and consistent results when administered repeatedly.

Validity (accuracy)

Test can accurately distinguish between those who have disease and those who do not

Sensitivity is..

The proportion of people with disease who are correctly identified by the test as positive for the disease. (True Positive Fraction)

How do you calc. for sensitivity?

a/ a+c x 100

interpretation of sensitivity?

Out of all cases/those who truly have the disease, X% of them correctly screened positive.

What is specificity?

The proportion of healthy people who are correctly identified by the test as negative for the disease. (true negative fraction)

How do you calc. for specificity?

d / d+b x 100

What is the interpretation of specificity?

Out of those who don’t actually have the disease, X% of them correctly screened negative.

What is predictive value?

Describes the ability of a test to accurately predict a positive or negative result.

What is positive predictive value?

The ability of a test to identify true positives

How to calc. for positive predictive value?

a / a+b x 100

Interpretation of PPV:

Out of those who screened positive, X% of them truly have the disease/are true cases.

What is Negative Predictive Value?

The ability of a test to identify true negatives

How do you calculate for NPV?

d/ c+d x 100

Interpretation of NPV?

Out of those who screened negative, X% of them truly don’t have the disease.

It is possible for a measure to highly reliable but…

invalid

It is NOT possible for a measure to be valid and…

highly reliable

What is Colonization?

agents multiply on the surface of the body without invoking tissue or immune response

What is Infestation?

the presence of a living infectious agent on the bodys exterior surface, upon which a local reaction may be invoked

What is incubation time?

Time interval from exposure to agent to first signs of illness/symptoms

What is infectivity?

The capacity of an agent to enter and multiply in a susceptible host and produce illness

e.g. Polio and measles have high infectivity