Epidemiology/Biostats Flashcards by Meghan May

Equation for and definition of Absoulte risk reduction (ARR)

ARR = percentage indiating actual difference in event rate between control and treatment groups
ARR= Control Rate - Treatment Rate

How well did you know this?

Not at all

Perfectly

Equation and definition of Relative Risk Reduction RRR

RRR = ARR/control rate

%indicating relative reduction in the treatment event rate compared to the control group

How well did you know this?

Not at all

Perfectly

Relative risk RR definition and equation

Ratio of the probability of an event occurring in the Tx group compared to control group

How well did you know this?

Not at all

Perfectly

Number needed to treat or NNT definition and equation

NNT is number of individuals that need to be tx to prevent a negative outcome in 1 patient
NNT = 1/ARR

How well did you know this?

Not at all

Perfectly

A confidence interval that crosses ____ or a P-value > erro cuttoff value _____ denotes a result is NOT statistically signification

crosses the null value (often solid line)

P value >0.05

How well did you know this?

Not at all

Perfectly

Study that is RETROspective; looks for odds of a previous exposure on the development of a rare dx manifestation.
Person has rare disease…lock backward at other groups that are otherwise matched, assess risk of exposure
What is a bias of this group?

Case Control Study

Recall bias

How well did you know this?

Not at all

Perfectly

Odds ratio are used for Case Control
What is equation?
How is it used?

Start with those that have a disease; then look for chance of having an exposure
Odds ratio = Cases exposed/Cases not exposed
Controls expose/ controls not exposed

How well did you know this?

Not at all

Perfectly

Study that is a ‘snapshot’ in time to determine whether or not they have a certain risk factor and PREVELANCE of certain disease of interest.

Cross sectional study

How well did you know this?

Not at all

Perfectly

Categorizes patients with and w/o specific disease and compares the risk factor frequency between the two gruops

Case Control Study

How well did you know this?

Not at all

Perfectly

Group of case reports regarding individual patients with similiar clinical manifestations or received similar tx; each case described in detail and may be used in rare disease or novel tx

Case Series

How well did you know this?

Not at all

Perfectly

Study that follows individuals who do NOT have the dx/outcome of interests and who may or may not have a certain risk factor OVER TIME… study to assess disease development

Prospective cohort study

How well did you know this?

Not at all

Perfectly

Occurs when an extraneous variable CHANGES THE DIRECTION OR STRENGTH of the effect the independent variable (exposure or treatment) has on the dependent variable(outcome)

Effect modification

How well did you know this?

Not at all

Perfectly

An extraneous variable associated with BOTH the exposure and outcome obscures the association btwn exposure and outcome

Confounding bias

How well did you know this?

Not at all

Perfectly

How can you tell the difference btwn effect modification and confounding bias?

Confounding bias has NO change in the strength or direction of the effect is seen with stratification; with randomization you mix up confounding variable

How well did you know this?

Not at all

Perfectly

occurs when 2 disease interventions are compared and on intervention diagnoses the disease earlier than the other leading to erroneous presumption that there is an effect on the outcome

Lead time bias

How well did you know this?

Not at all

Perfectly

Occurs when trials w/ significant positive results are published but trials with negative/null results are not

Publication bias

How well did you know this?

Not at all

Perfectly

What does it mean when the standard error of measurement bars overlap btwn two samples?

Means there is a non-statistically significant difference

How well did you know this?

Not at all

Perfectly

The applicability of a study’s results beyond the group that was initially assessed; ‘How generalizable are the results of the study to other populations?”

External Validity

How well did you know this?

Not at all

Perfectly

Relates to conclusions regarding cause and effect in a study; answers question ‘are we observing/measuring what we think we are measuring/observing?”; threat to this is counfounding

Internal validity

How well did you know this?

Not at all

Perfectly

The ability to detect an effect if that effect exsists. Depends on sample size; bigger is better

Power

the Effect Size = standard deviation

How well did you know this?

Not at all

Perfectly

Non overlapping confidence intervals always imply statistically significant DIFFEReNCE BUT; the ones that do overlap:

0.39 % (95% -0.58, -0.07)
0.51% (95% -0.71, -0.14)

may or may not have statistically significant difference.

How well did you know this?

Not at all

Perfectly

Increase for ___when multiple simultaneous hypotheses are tested at set p values
Ie testing multiple secondary endpoints increases probability of erroneously finding a statistically sig result

Type I errors (false positives) or alpha; probability of rejecting a null hypothesis when it is in fact true

How well did you know this?

Not at all

Perfectly

Measurement of inter-rater RELIABILITY or extent to which inter-rater agreement is an improvement on change agreement alone

Kappa stastic

How well did you know this?

Not at all

Perfectly

Intention to treat analysis is that participants in trials should be analyzed in the groups they were randomized, regardless of whether they received or adhered to allocated intervention and regardless if they withdrew, this helps prevnt..

Crossover and dropout and overall preserves randomization

How well did you know this?

Not at all

Perfectly

Represents failure to reject a null hypothesis when it is false. dependent on the power of a study which depends on sample size

Type II error

Phenomenon when pts symptoms are alleviated by otherwise ineffective tx likely due to individual expecting the tx to work

placebo affect

Correlation coefficient (r) ranges from values -1 to +1 which indicates a positive or negative direction of association between 2 variables 0 is _____ ____ is a strong association

0 is null value | -1 to +1 is stronger association; -1 is negative correlation while +1 is positive correlation

What does SnNout stand for

Test with high Sensitivity, a Negative results will help rule OUT a diagnosis

What does SpPin stand for

Given a test with high Specificty a Positive result would help to rule IN a dx

An epidemiological parameter that doesn't vary with disease prevalence, provides clinically useful information for individual patients. = a probability of a given test result occurring in a pt with a disorder compared to the probability of the same result occurring in a patient without the disorder

Likelihood Ratio

How to calculate likelihood ratio from sensitivity and specificity

Postive test result = Positive LR = sensitivity / (1-specificity) Neg Test result: Negative LR = (1-sensitivity)/ specificity

when a study uses gold standard testing selectively in order to confirm a positive (or negative) result of preliminary testing. This can overestimate or underestimates sensitivity or specificity

Verification bias

when control group unintentionally receives the tx or intervention, thus reducing the difference in outcomes btwn control and tx group.

Contamination bias

What method of analysis is good in clinical trials with high drop out or non-compliance?

Intention to treat; you include all subjects as initially allocated after randomization, regardless of what happens during study period.

The goal of this trial is to prove that a new drug or intervention is not unacceptably worse than an alternate drug or intervention.

Noniferiority trials

On data from subjects who completed the intervention originally allocated at randomization are analyzed

Per protocol analysis; you end up losing benefit of randomization and will overestimate the real effect of the intervention

The likelihood of an event occuring in a treatment group relative to the control group

Hazard ratio

The null value for hazard ratio (HR) = | A HR that is _____ indicates event is less likely to occur in tx group than control group

The null value for hazard ratio (HR) = 1.0 | A HR that is < 1.0 indicates event is less likely to occur in tx group than control group

In order for confidence intervals to be statistically significant...

They need to NOT include the null value (1.0) Significant ~ CI 0.32-0.69 NOT sigfniciant~ CI 0.72- 1.2

Proportion of people with a particular condition who end up dying from that condition

Case fatality ratio

Proportion of people in whom an illness develops out of the total population at risk for the disease

Attack rate

burden of disease measure that evaluates the impact of specific diagnoses/treatment son indivdual or economic impact on health intervention on population. Years of life lost and lived with diability used for..

Quality adjusted life years or QALY | Use 'Time Trade Off' of TTO; so 5 years in disable condition = 1 year of normal health

burden of disease measure that evaluates the impact of specific diagnoses/treatment son individual or economic impact on health intervention on population. These reflect the difference between the current situation and the ideal situation; ie living in perfect health up to standard of life expectancy; commonly used in global disease burdens

Disability adjusted life years (DALY) | Use years of life lost and years lived with disability used for this calculation

In a case control study, we commonly use ____ as a measure of association

Odds ratio

Type of study design that is experimental and utilizes >2 interventions and all combinations of these interventions

Factorial

type of study that generally is observational in which a specific population or group is studied at one specific point in time, thus gives a picture of the whole group at that particular point in time

Cross sectional

Form of retrospective observational study in which subsets of controls are matched to cases and analyzed for variables of interests

Nested study

Seeks to determine whether an intervention works in real life conditions

Pragmatic study

evaluates the association between a quantitative dependent variable and independent variables of interest while controlling for the effects of the other factors

Multiple linear regression

results when an external variable has a positive or negative impact on the observed effect of a risk factor on disease status

Effect modification Stratification can help detect effect modification you should report separate measures of outcome

A univariate analysis does not adjust for ____ while a multiple regression analysis does

confounding factors

In a patient population with low disease prevelance the positive predictive value is likely to be low due to ...

higher number of false positives relative to true positives

Inciting pathologic events or exposures to risk sometimes occurs years before clinical manifestations become apparent... thus we monitor over time. Phenomonon...

concept of latent period

In highly skewed distribution, the best measure of central tendancy is..

mean

In highly skewed distribution the best measure of Central Location is

median

refers to repeating primary analysis calculations after modifying criteria or variable ranges; goal is determine whether modifications significantly affect intial results

Sensitivity analysis

Occurs when a test diagnoses a disease earlier than another does resulting in time from diagnosis until death appears to be prolongued but its not

lead time bias

systemic differences btwn groups in term of treatment response or prognosis

selection bias

Test to study difference in mean values among several different groups

Analysis of Variance or ANOVA | *must have homoscedasticty or homogeniety of variance for the test to be valid

What does teh F statistic mean in ANOVA test

based on variation w/in and btwn different groups; used to obtain p value

Can test difference between two paired means; patietns serve as their own control; ie mean BP before and after tx in same subject

Paired T-test

___% patients lie w/in 1 STD of mean ___% patients lie w/in 2 STD of mean ___% patients lie w/in 3 STD of mean

68% 95% 99.7%

A cross sectional study will look at + vs negative risk factors and compare...

disease PREVALANCE

A Prospective cohort study will look at + vs - risk factors and compare...

Disease INCIDENCE

A ____ will select patient with a particular disease and patietnts without disease and then retrospectively determine exposure status

Case (disease) Control (not diseased)

Action taken before a patient develops a disease and targeted at preventing outcomes of the disease itself at initial stage

Primary prevention

Action that attempts to halt progression of disease at it's initial stage before irreversible changes take place to prevent complications

Secondary prevention

____ in funnel plots suggest publication bias

Asymmetry

Repeating primary analysis calculations by modifying certain criteria or variable ranges to determine if modifications significantly initial results

Sensitivity analysis

Epidemiology/Biostats Flashcards

(69 cards)