Biostats + Epidemiology Flashcards
Case Control Study - Type + Design + Examples
Type - Observational + Retrospective
Design - Compares a group of people with disease to those without - Looks for prior exposure/risk factor – What happened?
Examples - Odds Ratio (OR) + Pt. with COPD had a higher chance of hx of history of smoking than those without COPD
Cross-Sectional Study - Type + Design + Examples
Type - Observational
Design - Compares a group of people with disease to a group without disease (At a specific time aka Cross-Section!) - What’s Happening?
Examples - Disease Prevalence + Can shows risk factors but does not establish causality
Cohort Study - Type + Design + Examples
Type - Observational - Can be prospective or retrospective
Design - Compares a group with an exposure to one without - determines if exposure increased the likelihood for disease
Examples
Prospective - Who will develop disease (exposed of not)
Retrospective - Who developed the disease (exposed or not)
Relative Risk - Smokers had a higher risk of developing COPD than not smokers
Case Control Study vs. Cohort Study
Case Control - With or without disease - Look at Exposure
Cohort - With or without exposure - Look at Disease
Twin Concordance Study - Design + Examples
Design - Compares the frequency with which both monozygotic twins vs. dizygotic twins develop disease
Examples - Measures heritability + environmental factors (Nature vs. nurture)
Adoption Study - Design + Examples
Design - Siblings raised by biologic vs. adopted parents
Examples - Nature vs. Nurture
Clinical Trials - Number of Phase + Ways to Increase Quality
Number of Phases - 4
Increase Quality - Randomized + Controlled + Double or Triple Blinded
Phase 1 Clinical Trial - Sample + Purpose
Sample - Small group of healthy volunteers
Purpose - Is it safe (safety + toxicity + pharmacokinetics)
Phase 2 Clinical Trial - Sample + Purpose
Sample - Small Number of Patients with the Disease
Purpose - Does it work - Compares treatment effectiveness + optimal dosing vs. adverse effects
Phase 3 Clinical Trial - Sample + Purpose
Sample - Large number of patients randomly assigned to treatment or gold standard (control)
Purpose - Is it good or better than the current standard of care
Phase 4 Clinical Trial - Sample + Purpose
Sample - Post marketing surveillance
Purpose - Monitor for long term side-effects and rare impacts
Drawing the Diagnostic Testing Table
Top = Disease (+/-) Left = Test (+/-) 2x2 TP FP FN TN
Sensitivity - Definition
Proportion of all people with disease who test positive - Probability that a test actually detects the disease when it is present (True positive rate)
Rules out disease and indicates a low rate of false negatives
Sensitivity - Calculations Mnemonic
SN-N-OUT - High Sensitivity, when negative - rules the disease out (very unlikely that a negative is a false negative)
= TP/ (TP + FN) = 1/ (1 + 3)
= 1 - False Negative Rate
If sensitivity = 100% then TP/(TP + FN) = 1 and FN = 0
Specificity - Definition
Proportion of all people without disease who test negative - probability that a test indicates non-disease when the disease is in fact absence - True negative rate) - Low false positive rate so high values + positive tests help rule in a disease
Specificity - Calculation + Mnemonic
SP-P-IN - Specificity, when positive, rules the disease in - If you have a positive test you have the disease
= TN / (TN + FP)
= 1 - False Positive Rate
= 4 / (4 + 2) - On Table)
If specificity is 100% than TN/(TN + FP) = 1 and FP = 0 - All positives are true positives
Positive Predictive Value (PPV) - Definition + Calculation
Definition - Proportion of positive tests that are true positive - Probability that a person actually has the disease given a positive test result - Varies directly with the prevalence or pretest probability
Calculation - PPV = TP / (TP + FP) = 1 / (1 + 2)
Negative Predictive Value (NPV) - Definition + Calculation
Definition - Proportion of negative tests that are true negative - probability that you are disease free given a negative result
Calculation - NPV = TN / (TN + FN) = 4 / (3 + 4)
Incidence vs. Prevelance
Incidence looks at new incidents
PrevALLence looks at all current cases
Prevalence = Incidence Rate X Average Disease Duration
Chronic Disease = Prevalence > Incidence
Common Cold/Fast Disease = Prevalence = Incidence (Approx.)
Prevalence - Define + Calculate
Prevalence - Likelihood of the disease being present in the population
Prevalence = # of Existing Cases / Population Size (At Risk)
Incidence Rate - Definition + Calculation
Incidence is the number of new cases (incidents) in a given period)
Incidence Rate = # of New Cases in a period / Population at risk during the same period
Odds Ratio (OR) - Definition + Use + Calculation
Definition - Odds that a the group with the disease was exposed to a risk factor divided by the odds that the group without the disease was exposed
Use - Case Control Studies
Calculation - OR = (a/c) / (b/d) = ad/bc
Relative Risk (RR) - Definition + Use + Calculation
Definition - Risk of developing disease in the exposed group divided by the risk in the unexposed group - If prevalence is low that RR is roughly equal to OR Use - Cohort Studies Calculation = [A / (A+B)] / [C / (C+D)] A/(A+B) C/(C+D)
Relative Risk Reduction - Definition + Use + Calculation
Definition - Proportion of risk reduction attributable to the interventions as compared to a control
Use -
Calculation - RRR = 1 - RR —- E.g. if 2% of Patients who receive the flu shot develop flow and 8% of patients who do not receive it develop flu —> RR = 2/8 –> RRR = 1 - .25 = 75% Reduction
Attributable Risk (AR) - Definition + Use + Calculation
Definition - Difference in risk between exposed and unexposed groups - Proportion of disease occurrences that are attributable to the exposure
Use - How significant is the exposure in causing disease
Calculation - AR = A/(A+B) - C/(C+D)
Absolute Risk Reduction - Definition - Use - Calculation
Definition - Difference in risk (not proportion) attributable to the intervention compared to a control
Use - Determine treatment impact on Risk
Calculation - 8% who get placebo get disease vs. 2% of treatment group - ARR = 8% - 2% = 6% Risk Reduction
Number Needed to Treat or Harm - Definitions + Calculation
Needed to Treat - Number of patients who must be treated for 1 patient to benefit = 1/ARR
Needed to Harm - Number of patients who must be exposed to a risk factor for 1 to be harmed = 1/AR
Precision - Definition + Application
Definition - Constancy and reproducibility of a test (absence of random variation) - Random error reduces precision (Higher Precision reduces SD)
Application - Likelihood of hitting the same target it (where it is correct or not)
Accuracy - Definition and Application
Definition - Trueness of a test measure (validity) - Absence of systematic error or bias - Systematic error reduces accuracy
Application - Getting the correct result
Major Bias Categories (3) + Types of Bias (8)
1) Recruiting Based Bias (Selection Bias) - Loss of Follow-Up + Healthy Worker + Volunteer Bias
2) Performing Bias - Recall + Measurement + Procedure + Observer-Expectancy Bias
3) Interpreting Bias - Confounding Bias + Lead-Time Bias
Selection Bias - Definition + Examples (2) + Reduction Strategy
Definition - Nonrandom assignment to participate in the study resulting in a non-representative sample population
Examples - Loss of Follow-Up (If the disease has early mortality you don’t know why they are leaving) + Volunteer Bias (healthier than the general population)
Reduction Strategy - Randomization + Comparison Group Selection
Recall Bias - Definition + Example + Reduction Strategy
Definition - Performing Bias - Awareness of disorder alters recall by subjects (common in retrospective studies)
Example - Patients with disease recall exposure after learning about similar exposures
Reduction - Decrease time from exposure to follow-up
Measurement Bias - Definition + Example + Reduction Strategy
Definition - Information is gathered in a way that distorts it
Example - Hawthorne effect - Groups who know they are being studied/receiving treatment behave differently then they normally would
Reduction - Placebo control group + blinding
Procedure Bias - Definition + Example + Reduction Strategy
Definition - Subjects in different groups are not treated the same
Example - Patients in the treatment groups spend extra time in highly specialized hospital units
Reduction - Double blinding - physicians treat both groups the same
Observer-Expectancy Bias - Definition + Example + Reduction Strategy
Definition - Researchers belief in the efficacy of the treatment changes the outcomes of the treatment (self-fulfilling prophecy)
Example - If the observer expects the treatment group to recover he is more likely to document positive results
Reduction - Double Blinding
Confounding Bias - Definition + Example + Reduction Strategy
Definition - Interpreting Bias - When a actual is associated with both the exposure and outcome but is not causal - Distorts the outcome and confuses the causal effect
Example - Pulmonary disease is more common in coal workers than the population - but coal workers are also more likely to smoke
Reduction - Multiple/repeated studies + crossover studies + matching studies
Lead-Time Bias - Definition + Example + Reduction Strategy
Definition - Interpreting Bias - Early detection is confused with increased survival (seen with advanced screening techniques)
Example - Early detection makes it seem like we are surviving longer but the disease course has not been slowed, we simple know about it longer
Reduction - Measure back end survival - Adjust based on disease severity at the time of diagnosis
Measures of Central Tendency (3)
Mean - Average
Median - Middle Number
Mode - Most Common Number
Measures of Dispersion (2) - Definition + Calculation
Standard Deviation = How much variability exists from teh mean in a set of values = SD
Standard Error of the Mean - An Estimation of how much variability exists between the sample mean and true population mean = SEM = SD * sqrt(sample size) —- SEM decreases as sample size increases
Normal Disturbution
Gaussian - Bell Shaped - Mean = Median = Mode
1 SD = 34% - 68% of all samples
2 SD = 42.5% - 95% of all samples
3 SD = 49.35% - 99.7%of all samples
Non-normal Distributions (3)
Bimodal - Suggests 2 Different Populations (E.g. Suicide Age by Rate)
Positive Skew - Typically Mean > median > Mode - Asymmetry with a right side tail
Negative Skew - Typically Mean < Median < Mode - Asymmetry with a left side tail
Null Hypothesis - Definition + 2x2 Table
Definition - Hypothesis of no difference (no associate between a risk factor and a disease) = Ho
2x2 Table - Top = Reality (1 = Alternative Hi vs. 2 = Null Ho) ……. Left = Study Results
1 = Alternative in Study + Reality = Power = 1-Beta
2 = Alternative in Study + Reality is Null = Alpha = Type 1 Error
3 = Null in Study + Reality is Alternative = Beta = Type 2 Error
4 = Null in Study + Reality
Alternative Hypothesis - Defintion
Hypothesis of some difference (there is some associate between a disease and risk factor) = Hi
Outcomes of Statistical Hypothesis Testing (4)
Correct for Hi —> Test power with 1-Beta
Correct for Ho —>
Type 1 Error
Type 2 Error
Type 1 Error - Definition + Meaning
Definition - Stating that there is an effect or difference when none exists (Accepting the Hi when the correct result is Ho) - False Positive Error
Alpha = Probability of a false positive - Likelihood you saw a correlation when none existed (Convicted an innocent man) - p is judged against a preset significance - E.g p <0.05 there is a less than 5% chance the result occurred by chance alone
Type II Error - Definition + Meaning
Definition - Stating that there is not an effect or difference when one exists (Claiming Ho when Hi was correct) - False Negative Error
Meaning - Beta is the probability of making a Type II Error - B=Blind to something that did exist (setting a guilty man free)
Power - Defintion
Probability of of rejecting the null when it is false (E.g. Finding Hi when HI is correct) = 1 - Beta
Confidence Interval - Definition + Interpretation
Definition - Range of values in which a specified probability of the means of repeated samples would be expected to fall
Interpretation
If the CIs of 2 groups do not overlap - Significant difference
If the CIs of 2 groups do overlap - No significant difference
T-Test - Definition + Example
Definition Compares differences between the means of 2 groups
Example - BP in Men vs. Women
ANOVA - Definition + Example
Definition - Compares the differences of 3 or more means
Example - BPs of 3 Ethnic Groups
Chi-Square - Definition + Example
Definition - Checks differences between 2 or more percentages of categorical outcomes (NOT MEANS)
Example - Comparing the percentage of people with HTN across 3 ethnic groups
Pearson Correlation Coefficient (r) - Definition
Definition - Always between -1 and +1 –> Close to +1 = stronger positive correlation between 2 variables
Disease Prevention Stages (3) - Define + Examples
PST - Primary (Prevent) Secondary (Screen) + Tertiary (Treat)
Primary - Prevent disease occurrence - Vaccines
Secondary - Screen (for early disease) - Pap Smears
Tertiary - Treat (reduce disability from disease) - Chemotherapy