Unit 2 Flashcards by Danica Wilson

What are the two types of study designs?

descriptive, explanatory

How well did you know this?

Not at all

Perfectly

Intended to describe a disease condition - signs, lesions, outcomes, occurrence of microbes, etc.

Descriptive study

How well did you know this?

Not at all

Perfectly

Seeks to identify causes. Has a hypothesis, has controls. Used to investigate a treatment, intervention, or risk factor, in hopes of providing causal evidence.

Explanatory (or causal) study

How well did you know this?

Not at all

Perfectly

List the two main types of explanatory studies:

experimental, observational (epidemiologic)

How well did you know this?

Not at all

Perfectly

List the three types of observational studies:

cohort
case control
cross sectional

How well did you know this?

Not at all

Perfectly

Planned comparison between 2 groups - one receiving one treatment, another receiving a different tx for a naturally disease - researcher has some control

Experimental - Clinical trials

How well did you know this?

Not at all

Perfectly

Researcher has the greatest control and involves the use of experimental animals. This type of study is the best design to prove cause or demonstrate efficacy

Experimental - laboratory

How well did you know this?

Not at all

Perfectly

Sometimes called “natural studies” as they occur freely in nature.

Observational studies

How well did you know this?

Not at all

Perfectly

Why are observational studies under the category of “explanatory type” studies?

the goal is to assess cause

How well did you know this?

Not at all

Perfectly

What is the difference between observational and experimental studies?

Which animals go to treatment groups is not under control of the researcher for observational

How well did you know this?

Not at all

Perfectly

Observational studies contribute to the body of evidence implicating a ______ as a cause. They often do not provide sufficient ________, in one study, to establish a cause.

factor; evidence

How well did you know this?

Not at all

Perfectly

a group sharing a defining characteristic

cohort

How well did you know this?

Not at all

Perfectly

Type of observational study that is prospective in time:

cohort

How well did you know this?

Not at all

Perfectly

What are the two types of cohorts featured in a study?

one exposed to a factor

2. one NOT exposed to a factor

How well did you know this?

Not at all

Perfectly

Subjects are followed in time, and incidence of one or more diseases are compared between the two groups:

cohort

How well did you know this?

Not at all

Perfectly

For a cohort study, what does a relative risk >1 indicate?

an increased risk in exposed, compared to unexposed

How well did you know this?

Not at all

Perfectly

For a cohort study, what does a relative risk =1 indicate?

that the risk in exposed is the same as the risk in the unexposed

How well did you know this?

Not at all

Perfectly

For a cohort study, what does a relative risk <1 indicate?

that the disease in the exposed is less than the unexposed

How well did you know this?

Not at all

Perfectly

A relative risk < 1 for a cohort study indicates that exposure has a “sparing effect”. What does this mean?

a reduction in risk associated with exposure

How well did you know this?

Not at all

Perfectly

What kind of exposures might give a sparing effect?

vaccinations

How well did you know this?

Not at all

Perfectly

Measure of the strength of association between a factor and a disease:

relative risk

How well did you know this?

Not at all

Perfectly

Relative risk can ONLY be calculated in what 2 studies?

cohort and cross-sectional

How well did you know this?

Not at all

Perfectly

Cannot be estimated in a case-control study:

relative risk

How well did you know this?

Not at all

Perfectly

What is the equation for relative risk?

(proportion with disease in exposed)/(proportion with disease in unexposed)

How well did you know this?

Not at all

Perfectly

Well suited for studying disease and exposures that occur relatively commonly:

cohort

Well suited to study the effect of multiple outcomes following a single exposure:

cohort

What are two benefits of cohort studies?

1. researcher has control over data quality (recorded in real time) 2. time sequence of "cause" and disease is clear

Factor A occurs before disease X

temporal relationship

Factor A is present very often in cases, and not in controls:

strength of association

The more factor A you have, the greater the disease chance

dose-response relationship

Based on what is known, Factor A could cause disease X

biological plausibility

The relationship between Factor A and disease X is seen repeatedly, time and again

Consistency of multiple studies

Other causes are not likely or impossible

rule out other possible causes

Removal of Factor A results in diminished disease

reversible association

A selected group within the population is sampled once and exposure and disease are simultaneously measured:

cross sectional (prevalence) study

How are cross sectional studies not like cohort studies?

animals are NOT followed in time to establish disease incidence

What are the advantages of cross-sectional studies?

- short, fast, inexpensive - can gather data on multiple diseases/exposures - provides preliminary evidence for further study

Cohort studies are usually conducted in a:

prospective manner

Animals with a disease

"cases"

Animals without disease:

"comparands"

Study in which animals with a disease are compared with one more controls:

case-control

Can be used to study rare (infrequent) disease:

case-control

Why are case-controls considered retrospective studies?

look into the past and compare the frequency of occurrence of risk factors for cases vs. controls

Odds of exposure in cases compared to odds of exposure in non-cases

odds ratio

What two measurements assess the strength of association between exposure and disease (but in slightly different ways)?

1. relative risk (cohort) | 2. odds ratio (case-control)

The probability of an event occurring to the probability of the event not occurring:

odds

Which measure of association can be used in case-control AND cohort studies?

odds ratio (RR is easier for cohort though)

When are OR and RR pretty close?

when the disease is rare (<10%)

How to calculate OR =

(ad)/(bc)

An odds ratio > 1 indicates:

an increased risk

An odds ratio = 1 indicates:

there is no increased risk

An odds ratio < 1 indicates:

a "sparing" effect

estimates risk of exposure in cases, compared to exposure in controls:

OR (case-control)

estimates risk of disease in exposed, compared to risk of disease in non-exposed:

RR (cohort)

A population of animals is sampled at a specific point in time:

cross-sectional

How is each animal classified in a cross sectional study?

according to status of outcome and risk factor, simultaneously, at the time of the snapshot

Can be used to assess absolute risks in the population

cross-sectional

Typically refers to the assessment of serum antibody concentration:

serology

The point at which we have diluted a serum sample and measured until there is no longer a reaction:

"end point titer"

What is the idea behind testing serology?

you can determine the last dilution at which an Ag:Ab reaction occurs (end point titer)

What does seropositivity depend on for each titer?

cut-off point

You must be over the critical titer to be classified as:

positive

A 2 fold increase (or 1 dilution) between week 1 and week 4 is considered:

laboratory error

A four fold increase (or 2 diluations) between week 1 and week 4 is considered:

to be greater than lab error, reflecting active Ig production

Why do we often use the Log2 transformation when talking titers?

much easier to work with when they assume a normal distribution

Ability to correctly classify (detect) disease animals. Expressed as a proportion:

sensitivity

The test that is used to determine if a disease is truly present or not. Other tests are compared to it to determine their "accuracy"

gold standard

Of those truly D+, the proportion correctly classified by Test Z

Sensitivity

Ability of a test to correctly detect (classify) non-diseased animals.

Epidemiologic specificity

The ability to measure the correct substance (i.e. not measuring particles or molecules other than the target):

analytical specificity

how close a test result is to the truth

accuracy

What may be different if comparing populations in early vs. late stages of disease?

sensitivity

Properties of the test that allow for comparison between tests (i.e. 2 test for detection of FeLV):

sensitivity, specificity

Higher sensitivity improves _____ and helps identify the disease. It rules ____ disease.

NPV; OUT (SnOUT)

How does lower specificity alter PPV?

decreases it

Lower specificity =

more false positives

A more sensitive test reduces _______ ________ and increases NPV.

false negatives

SnOUT -

helps rule out disease

A more specific test reduces _____ __________ and improves PPV.

false positives

SpIN -

rules in disease

By altering the cutpoint, you alter:

sensitivity and specificity

Increased sensitivity results in:

lower specificity, fewer false negatives, more false positives

When is it advantageous to alter the cut-off point in a test?

if you want to rule in/out a disease

Two or more tests are conducted sequentially based on the results of a previous test:

serial testing

In serial testing, only animals that test ________ to the first test are tested again.

positive

In serial testing, only animals that are positive on __________ are considered positive.

all tests

Used for diagnosis when time is not crucial or if diagnosing a patient when disease positive has a grave outcome.

Serial testing

Conduction two or more tests on a patient at the same time:

parallel testing

If any one test is positive, the animal is categorized as sick:

parallel testing

When is parallel testing used?

rapid assessment/medical emergencies

Increased sensitivity -->

fewer false negatives

decreased specificity -->

more false positives

Because sensitivity is reduced, you will allow more diseased animals to remain in the herd:

serial testing

Relatively small number of positive tests. The pool of positive tests is nearly all “Sick”. But not all sick are in the test + pool.

serial testing

Positive test identifies animal at high probability of Illness, Confidence in Positives SpIN

serial testing

More positive tests. Positive pool captures almost all ill, but includes many non-ill too.

parallel testing

The negative test pool ID's "well" animals. Confidence in Negatives SnOUT

parallel testing

single numbers (points) derived from a sample

point estimate

represents our best estimate of the true value of the population parameter:

point estimate

a range of values that the point estimate could reasonably take

confidence intervals

The probability of obtaining the observed value (or more extreme value) when the null hypothesis is TRUE

P value

Typically states that there is no difference between the group being compared:

H0 (null hypothesis)