Behavioral Sciences Flashcards
Relative Risk
Equation
When is it used?
What is it?
[a/(a+b)]/[c/(c+d)]
Cohort Studies
(Risk of developing disease in exposed group)/(risk in unexposed group)
Odds Ratio
Equation
When is it used?
What is it?
ad/bc
Case-Control Studies
Odds that the group with the disease was exposed to risk factor (a/c)/(Odds that the group without disease was exposed (b/d)
Observer Bias
Investigator is affected by prior knowledge
Confounding
Exposure-disease relationship can be explained by another variable
Lead-Time Bias
Apparent prolongation of survival because of earlier diagnosis
Recall Bias
Inaccurate pt recall
Selection Bias
Subjects selected biasedly or biased from selective loss of follow up
Case Control Study Mode of investigation Timeframe Design Basic Question Measurement
Observational Retrospective Compare groups of people w/ and w/o disease and look for prior exposure or risk factors "What Happened?" Odds Ratio
Cohort Study Mode of investigation Timeframe Design Basic Question Measurement
Observational Prospective or Retrospective Compare a group with a given exposure or risk factor to a group w/o it and look if exposure ↑ likelihood Who will or Who has developed disease Relative Risk
Cross Sectional Study Mode of investigation Timeframe Design Basic Question Measurement
Observational
Right Now
Collect data from a group of people to assess frequency of disease and relative risk at a particular point in time
“What is happening?”
Prevalence, Relative Risk, Can show Risk Factor Association but not causality
Twin Concordance Study
Design
Measurements
Compares frequency with which monozygotic or dizygotic twins develop the same disease
Measures Heritability
Adoption Study
Design
Measurements
Compares siblings raised by biological vs adoptive parents
Heritability and influence of environmental factors
Clinical Trial
Population involved
Design
What improves study
Experimental study involving humans
Compares therapeutic benefits of 2 or more treatments or treatment and placebo
Randomized, Controlled, and Double Blinded studies are better
Double vs Triple Blind
Double: Patients and Doctors do not know whose getting treatment
Triple: Researchers analyzing data do not know who got treatment
Phase I Clinical Trial
Sample
Purpose
Small # of health volunteers
Assesses safety, toxicity, and pharmacokinetics
Phase II Clinical Trial
Sample
Purpose
Small # of patients w/ disease
Assesses treatment efficacy, optimal dosing, and adverse effects
Phase III Clinical Trial
Sample
Purpose
Large # of patients comparing 2 drugs or placebo
Compares new treatment to current treatment
Phase IV Clinical Trial
Sample
Purpose
Postmarketing surveillance trial of patients after approval
Detects rare or long term adverse effects
Sensitivity Rate of what? Formula (2) Purpose Use
"SNOUT - SeNsitivity rules OUT" "PID - Positive In Disease" True Positive Rate a/(a+c) = 1-(false neg rate) Rules out disease Used for screening in diseases w/ low prevalence
Specificity Rate of what? Formula (2) Purpose Use
"SPIN - SPecificity rules IN" "NIH - Negative In Health" True Negative Rate d/(d+b) = 1-(false pos rate) Rules in disease Confirmatory test after positive screening test
Positive Predictive Value
What does it show?
Formula
What does it vary with?
Proportion of + tests that are true positives
a/(a+b)
Varies directly with prevalence or pretest probability: high pretest probability –> high PPV
Negative Predictive Value
What does it show?
Formula
What does it vary with?
Proportion of negative test results that are true negative
d/(c+d)
Varies inversely with prevalence or pretest probability: High pretest probability –> low NPV
Incidents
(New cases over specified period)/(Population at risk)
Prevalence Equations (2)
(Existing cases)/(Population at risk)
Incidence x Average disease duration
Prevalence vs Incidence in chronic disease
Prevalence > Incidence
When are RR and OR equal?
When prevalence is low
Attributable Risk
Equation
What is it?
[a/(a+b)]-[c/(c+d)]
Risk in exposed group - Risk in unexposed group
Absolute Risk Reduction (ARR)
What is it?
c/(c+d) - a/(a+b)
Event rate in placebo minus treatment groups
Number Needed to Treat
What is it?
Equation
Number of pts who need to be treated for 1 patient to benefit
1/(absolute risk)
Number Needed to Harm
What is it?
Equation
Number of pts who need to exposed to a risk factor for 1 pt to be harmed
1/(attributable risk)
Precision
What is it?
What reduced precision?
When precision is increased, what happens?
Consistency and reproducibility of a test. The absence of random variation
Random error ↓ precision
↑ precision –> ↓ SD
Accuracy
What is it?
What kind of error does it measure?
What reduces accuracy?
Trueness of test measurements (validity)
Absence of systematic error or bias in the test
Systemic error ↓ accuracy
Sampling bias
Subjects not representative of the general population
A form of selection bias
Late-Look Bias
Information gathering at an inappropriate time (i.e. survey to study a fatal disease - only those still alive will be able to answer survey)
Procedure Bias
Subjects in different groups are not treated the same way
Hawthorne effect
Group being studied changes behavior because they know they are being studied
In a normal distribution, how do measurements of central tendency relate?
Mean = Median = Mode
Standard Deviation
What roman numeral?
Percentages?
σ
(+1σ) and (-1σ) account for 68% of n
(-2σ) to (+2σ) account for 95% of n
(-3σ) to (+3σ) account for 99.7% of n
SEM
What is it?
How does it vary?
σ/√n
SEM will ↓ as n ↑
Positive Skew
How do measurements of central tendency relate?
Which is least affected?
How does graph look?
Mean > Median > Mode
Mode is least affected by outliers in the sample
Asymmetrical distribution with long tail on right
Negative Skew
How do measurements of central tendency relate?
How does graph look?
Mean < Median < Mode
Asymmetry with longer tail on left
Statistical Hypotheses
Null Hypothesis
Alternative Hypothesis
There is no association between the disease and the risk factors (H0)
There is some association between the disease and the risk factor (H1)
Type I error Symbol What is it? AKA What is used to calculate it?
α
Error of stating there is a difference when there is not. Accepting H1 and rejecting H0 when H0 is true.
False-Positive Error
Used to calculate p