Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
86 Cards in this Set
- Front
- Back
|
What is epidemiology?
|
The study of factors that determine occurrence & distribution of disease in a human population
|
|
|
What is a case-controlled study?
Asks what? Measures what? |
Compares a group of people with disease to a group without (study of past)
Asks "what happened" - observational & retrospective Measures odds ratio (OR) |
|
|
What is an Epidemic?
Endemic? Pandemic? |
Epi - increase in the occurence of the disease
End - disease or condition constantly present in a human population Pandemic - widespread epidemic with large geographical distribution |
|
|
Primary prevention of disease?
Secondary prevention? Tertiary Prevention? |
P - reduction in occurence through immunization, sanitation, education
S - early detection & treatment of disease T - reduction in complications of disease through palliation & rehabilitation |
|
|
Statistic?
Example? Type of data? |
estimate of a population parameter based on a random sample from a population
Examples: mean * Descriptive Statistic - for quality assessment, draw preliminary conclusions |
|
|
What is an OR?
What does it approximate? When? |
measures the strength of relationship between two variables (Ex: you have a higher odds ratio to get lung cancer if you smoke) - basically compares two groups of people and the likelihood of an event taking place (exposed/unexposed)
Approximates relative risk if prevalence of disease not too high |
|
|
Test Statistic?
Example? Type of Data? |
calculated value from the sample relating the sample to a probability distribution in significance testing
Examples: chi square, t-test * Inferential statistic - looks at differences between averages, proportions, etc. - determines if this was statistically significant or the null hypothesis can be accepted (<95% confidence) |
|
|
Common Source Epidemic?
Point Source Epidemic? Propagative Epidemic? |
CSE - all individuals exposed to a causative agent from common source (Ex: water supply) - indirect transmission (NOT host-host)
PSE - all individuals exposed to causative agent at one point in time (food poisoning from wedding cake) PE - transmission is person-to-person or through vector (AIDS) |
|
|
What is a cohort study?
Asks what? Measures what? |
Compares a group with a given risk factor to a group without it to assess whether the risk factor increases prevalence of the disease (study of future outcomes) - observational & prospective
Measures Relative Risk |
|
|
Qualitative Data?
Type of Measurement? Quantitative Data? Type of Measurement? |
Qualitative - presence or absence of sx, physical sign, etc. - CATEGORICAL
Quantitative - absolute or relative amount of phenomenon - CONTINUOUS |
|
|
Morbidity?
Mortality? |
- rate of occurence of disease or condition in a particular population over a specific time period
- rate of death due to specific disease or condition in a particular population over a specific time period |
|
|
Discrete scales of measurement?
Frequency vs Counts? Continuous scale? Nominal Scale? Ordinal Scale? |
Discrete - whole units that are subdividable
* Frequency - # of patient with sx (independent) * Count - # of episodes (no independent) Continuous (Nondiscrete) - whole numbers that are not divisible Nominal - DISCRETE - assessment to categories, units not subdivisible Ex: ABO blood typing Ordinal Scale - DISCRETE -categories ranked in ascending order (numbers arbitrary) Ex: pain scale 1-5 |
|
|
What is RR?
|
Percentage of people that will develop a disease based on a specific risk factor
RR = people with risk factor/total people divided by people without risk factor/total people |
|
|
What is a cross-sectional study?
Asks what? Can show what? |
Measures the frequency of a disease (and the relative risk factors) at a particular point in time)
Asks "What is happening" Can show risk factor associated with disease but does not establish causality (cause & effect relationship) |
|
|
What is Attributable Risk?
|
Difference in a rate of condition between an exposed population & an unexposed population - proportion of disease attributable to exposure
|
|
|
What is Absolute Risk?
What is Absolute Risk Reduction? |
Reduction in risk associated with treatment compared to placebo (untreated) - Calculate by taking # of Diseased/Total Group
For Reduction, calculate risks of two groups and subtract one from the other |
|
|
What is a Twin Concordance Study?
Measures what? |
Compares frequency of both twins (monozygotic or dizygotic) developing a disease
Measures heritability of disease |
|
|
What is an adoption study?
Measures what? |
Compares siblings raised by biological vs adoptive parents
Measures heritabilty & genetics of a disease |
|
|
What is incidence?
What is prevalence? How are they related? |
Number of NEW people with a disease in a given time period
Number of TOTAL people with disease in a given time period Prevalence = Incidence x Disease Duration * So a chronic disease will have a much higher prevalence than incidence |
|
|
Point vs Period Prevalence?
What can alter rates of disease over time? |
Point - number of cases in a population at a specific point in time
Period - number of cases in a population in a specific time period - Change in ability to recognize disease (new testing), change in effort to recognize disease (more publicity), change in definition of disease |
|
|
How do you calculate prevalence rate?
How do you calculate incidence rate? Who is included in the population at risk? |
# of total cases in a population at a specific time/ # of people at risk for the disease
# of NEW cases in a specific time/ # of people at risk for the disease * People at risk are only those WITHOUT the disease (those that have the disease DON'T count when calculating incidence) * Use whole population (diseased & none) for prevalence |
|
|
What is an observational study?
What is an experimental study? |
Experimenter doesn't control environment or manipulation
Experimenter controls the independent variables and observes effects |
|
|
What is odds?
What do the OR values mean? If >1.0? <1.0? = 1.0? |
probability of an event occurring divided by the probability of it not occurring (P/1 - P)
>1.0 = increased risk <1.0 = decreased risk =1.0 = no change in risk |
|
|
Prospective Cohort Study vs. Retrospective?
|
Prospective - healthy people tracked forward to see incidence of disease (people not selected for risk factors, incidental risk factors)
Retrospective - healthy people tracked forward to determine incidence of disease (people identified by past records to choose people with certain risk factors) |
|
|
Determining causality?
Best to worst experiment designs? Strength of association? Consistancy? Temporal Relationship? Level of Exposure? Dose dependance? Plausibility? Specificity? Analogy? |
Design (Best to Worst) - experimental, prospective cohort, historical (retrospective) cohort, case control, cross-sectional area
Strength - OR, RR Consistency - others with different designs saw same relationship Temporal - time between exposure and disease Level - acute vs chronic exposure Dose - does it matter? Plausibility - is it consistent with current science? Specificity - does it produce more than one effect? Analogy - similar agents cause same effect |
|
|
NNT? Equation?
NNH? Equation? Absolute Risk Reduction? |
NNT - number needed to treat (number of persons who need to take treatment for one person to benefit from treatment) = 1/Absolute Risk Reduction
NNH - Number needed to harm (number of people who need to be exposed to a risk factor for one person to be harmed) = 1/Attributable Risk ARR = 1 - Absolute Risk |
|
|
What is a meta-analysis?
Limited by what? |
pools data from several studies to come to an overall conclusion, highest echelon of clinical evidence
Limited by quality of individual studies used in analysis |
|
|
What is a clinical trial?
Best when? |
Experimental study involving humans that compares therapeutic benefits of 2 or more treatments or treatment & placebo
Manipulation (independent variable) is controlled by investigator, study is prospective (forward in time) and measures selected outcomes (dependent variable) Best when randomized, controlled, and double-blinded |
|
|
Phase I Clinical Trial?
Purpose? |
Uses small number of patients, usually healthy
Assesses safety, toxicity, and pharmacokinetics (maximum dose tolerated before toxicity) |
|
|
Phase II Clinical Trial?
Purpose? |
Uses small number of patients with disease of interest
Assesses treatment efficacy, optimal dosing, adverse effects of drugs to treat |
|
|
Phase III Clinical Trail?
Purpose? Best when? Phase IV Clinical Trial? Purpose? |
Uses large number of patients randomly assigned either to the treatment under investigation or placebo/treatment commonly used
Compares new treatment options to older ones Better if double-blinded Post-approval surveillance of drugs - looks for larger scale medical usage & effectiveness |
|
|
How do you determine quantitative test cutoffs?
Use what graphs/values? |
Pick a normal distribution, cutoffs are outside of this (Ex: 2 SD above mean), use percentiles (top 5%), therapeutic (minimum level that requires treatement), risk factors (when do they start to develop)
Use specificity, sensitivity, PPV, NVP and plot a ROC Curve (plots these values) to determine optimum cutoff |
|
|
Qualitative vs. Quantitative tests?
Examples? |
Qualitative has one of two outcomes (positive or negative - Example is a urine pregnancy test), Quantitative has continuous outcomes - patient classified based on cut off values (Example: Diabetics have fasting blood sugar >126 at least twice)
|
|
|
ROC Curve? What's on x-axis? What's on y-axis?
|
x - FPR & Specificity - stay to left to maximize
y - FNR & Sensitivity - stay to right to maximize |
|
|
Diagnostic Accuracy?
|
(TP + TN)/Total Tests (proportion of all tests that were correct
|
|
|
General definition of Bias?
Selection Bias? Sampling Bias? Recall Bias? Late Look Bias? |
One outcome is favored over another
Selection - subjects AREN'T randomly assigned Sampling - subjects AREN'T representative of population at risk Recall - subjects response is altered by subject's memory and knowledge of disease Late Look - information is gathered at an inappropriate time |
|
|
Procedure Bias?
Confounding Bias? Lead time Bias? Pygmalion Bias? Hawthorne Bias? |
Confounding - occurs with closely related factors (effect of one factor can distort another)
Lead time - early detection thought to mean increased survival time Pygmalion - researcher's belief of a treatment's efficacy will alter outcome Hawthorne - group being studied changes its behavior since it knows its being studied |
|
|
Reduce bias how?
Blind Studies? Placebo Response? Randomization? Cross-over Studies? |
Blind - (often double blind) - patient and doctor not aware
Placebo Response - at least one third of patients on placebo have response Randomization - patients randomly assigned to groups Cross over Studies - when halfway through the placebo & drug group switch (you act as you own control) |
|
|
What is reliability?
Interrater reliability? Test-Retest reliability? |
Reproducibility of results
Results similar with different testers Results similar when patient retested |
|
|
What is validity?
Includes what? |
Measures whether test assesses what it's supposed to
Includes specificity & sensitivity |
|
|
What is a predictive value?
|
Measure of percentage of test results that match the actual diagnosis
|
|
|
PPV?
Measures what? Equation? |
Positive Predictive Value = portion of positive test results that are true positives
Probability a patient has a disease based on a positive test value PPV = TP/(TP + FP) |
|
|
NPV?
Measures what? Equation? |
Portion of negative test results that are true negatives
Probability a patient has a disease based on a negative test value NPV = TN/(TN + FN) |
|
|
Sensitivity?
Measures what? Equation? FN of zero? |
Portion of all people with disease that test positive
Measures that ability of a test to detect a patient with the disease Sensitivity = TP/(TP + FN) * When FN is near zero, sensitivity is near one meaning it is especially able to detect people with a disease (TP) |
|
|
Specificity?
Measures what? Equation? |
Portion of all people without disease that test negative
Measures the ability of a test to detect a patient without a disease Specificity = TN/(TN + FP) * When FP near zero, specificity is near one meaning it is especially able to detect people without disease |
|
|
Likelihood Ratio Positive?
|
Ratio of true positives to false positives
LRN = Sensitivity/(1 - Specificity) |
|
|
Likelihood Ratio Negative?
|
Ratio of true negatives to false negatives
|
|
|
What is a Research Hypothesis?
Types of comparisons in studies? |
Clinical Question to be answered by a study
Difference/Superiority (determines better/worse) Equivalence (are things equal) Non-Inferiority (is this no worse than...?) |
|
|
What are examples of Retrospective Studies?
What are examples of Prospective Studies? |
Retro - Cohort, Case Controlled, Cross-Sectional
Pro - Observational Cohort, Clinical Trials (controlled) |
|
|
Determine what when evaluating treatment groups?
- Demographics? - Definition of what? - Control? Sample Size and Power? |
Use a mix of demographics (age, sex, race), use appropriate control groups (positive or negative (placebo), adequately define inclusion & exclusion criteria
Inadequate subjects to detect difference (under-powered) or too many subjects so that differences become insignificant (over-powered) |
|
|
What is statistical significance in a study?
|
(p value of <0.05 meaning the results would be repeatable in a subsequent study and were not left up to chance)
|
|
|
What perameters should you check in a study?
Subgroup Analysis? Surrogate Endpoints? Early reporting? Sponsorship? |
Underpowered trails (too few subjects)
Inappropriate Tests Data Manipulation Study Limitations Subgroup Analysis (keep testing until you find something) Surrogate endpoints (short-term endpoints to estimate long term effects) Early reporting of results (before all information in) Sponsorship - who paid for this study! |
|
|
What is the FDA? OHRP? ORI? IRB? DSMB?
|
FDA - Food & Drug Administration
OHRP - Office Human Research Protection (NIH) ORI - Officer Research Integrity (Science Misconduct) IRB - Institutional Review Board (Individual Institutions) DSMB - Data Safety Monitoring Board (Primary Investigator Research Committee) |
|
|
What is a protocol?
What parameters does it include? About research? who's in it? protocols? variables? |
formal document providing complete description of clinical trial - states research hypothesis, population being studied, experimental design, use of control groups, methods of blinding & assigning participants, description of dependent & independent variables (and how they're measured), statistical analysis & data management
|
|
|
What is a research hypothesis?
|
states a presumed effect (relationship) between variables, usually one dependent and one independent - Goal is to prove or disprove it
|
|
|
Difference vs Equivalence Trial?
|
Difference proves superiority of one thing over another, Equivalence shows no difference between the two
|
|
|
Simple Random Sample? Systemic Sample? Stratified Random Sample?
|
Simple - just pick random people based on no other factors
Systematic - pick every ninth name on a list (some system dictates choice Stratified Random - balance males and females or ages (so not completely random) |
|
|
What is a non-probability sample? Convenience sample? Quota sample? Purposeful sample?
|
Can be biased...
Convenience - availability (any subject that will volunteer) Quota - minimal number in sample based on requirements Purposeful - very specific criteria (limits people eligible) |
|
|
Randomized Block Design? Replicated Design? Factorial Design?
|
Randomized - block nuisance factors that may alter results (use same same person to administer test for a group of people, group all women together and administer the test simultaneously, ect)
Replicated - repeat entire experiment with different subjects, pool data Factorial - analyze all factors and interactions |
|
|
Dependent Design? Latin Square Design? Repeated Measures?
|
Dependent - Each subject is measured more than once ... Pre-test, Post-test, or cross-over designs
Latin - use latin square to assess all variables Repated - measure each subject repeatedly across various levels of a factor |
|
|
Blinding? Open Label? Single? Double? Triple?
|
Open - no blinding
Single - patient unaware Double - patient & investigator unaware Triple - patient, investigator, clinical sponsor unaware |
|
|
Ordinal data? Nominal? Interval? Ratio?
|
O - ordered sequential category (pain is mild, moderate, severe)
N - ordered non-sequential category (pain is crushing, burning, aching) I - continuous data, no true zero (survival) R - continuous data, true zero (blood levels) |
|
|
Level I EBM Study?
Level II? 1? 2? 3? Level III? |
Level I - Evidence from at least one properly designed, double-blind, randomized controlled trial
Level II-1 - Evidence from well-designed controlled trials without randomization Level II-2 Cohort or case-control analytical studies, usually from more than one research venter Level II-3 - evidence obtained from multiple time series with or without the intervention Level III- opinions of respected authorities, based on clinical experience, descriptive studies, or reports of expert committees |
|
|
What is a variable?
What is a dependent & independent variable? |
quantity that can change under different circumstances
Independent - what is being manipulated by researchers (like giving everyone one medication a day) Dependent - result dependent on independent variable (side effects from drug) |
|
|
What is a statistical hypothesis?
Null vs Alternative Hypothesis? |
Null Hypothesis - hypothesis that might be falsified during an experiment - researcher collect datas, and calculates how probable that data was assuming the null hypothesis were true. If the data appears very improbable (usually defined as a type of data that should be observed less than 5% of the time) then the experimenter concludes that the null hypothesis is false. If the data looks reasonable under the null hypothesis, then no conclusion is made. The null hypothesis could be true, or it could still be false; the data gives insufficent evidence to make any conclusion.
Alternative Hypothesis - negation of the null hypothesis |
|
|
What is a standard deviation?
Equation? What is variance? Equation? |
SD = average distance of observations from their mean
square of = ((value 1 - mean) + (value 2 - mean) ...)/ n - 1 V = deviation from mean squared |
|
|
What is a z-score?
Equation? |
difference between individual values and population mean in terms of standard deviations
z = (score in distribution - mean of distribution) /standard deviation |
|
|
Standard error of mean?
Equation? |
standard deviations of sample in frequency distribtion
SEM = standard deviation/ square root of n |
|
|
Normal Distribution?
Within 1 SD? 2 SD? 3SD? Bimodal? Positive & Negative skews? |
Bell Shaped (Gaussian)
68% = -1.0 - 1.0 95% = -2.0 - 2.0 99.7% = -3.0 - 3.0 B = two humps P = right tail - mean > median > mode N = left tail - mean < median < mode |
|
|
Distribution of data?
Gaussian? Kurtosis? Skew? |
Gaussian - normal (bell-shaped)
Kurtosis - peaked Skew - in one direction or other |
|
|
Mean? Median? Mode?
|
average, middle value, most represented value
|
|
|
Statistical Tests?
|
used to analyze data from epidemiology studies
|
|
|
Parametric Statistical Test?
Nonparametric? |
evaluate statistically significant differences between groups when distribution scores are normal (bell-shaped) and sample size is large - MOST powerful
distribution is not normal or sample size is small |
|
|
What is a met-analysis?
Uses what? Provides what? |
combining results from many small studies into a single measure of efficacy & safety
* Use chi square analysis - provides odds ratio from each study and overall OR |
|
|
Goal in Research Hypothesis?
|
Disprove null hypothesis by rejecting with some confidence (this drug will have no effect on disease), and accept the alternative hypothesis
|
|
|
What is power?
Depends on what? Equation? |
probability of rejecting null hypothesis when it is infact false, or accepting alternative hypothesis when it is infact true
- Size of expected effect - Difference in compliance of groups - Total number of endpoints experienced by population P = 1 - Beta (type II error) |
|
|
What is a p value?
|
Probability of observing the test result by chance
|
|
|
What is a t-test? Dependent vs Independent?
What is ANOVA? X-squared? |
T-test - checks the difference between the means of two groups - allows you to reject null hypothesis if need be
t = mean1 - mean2/ SE(mean1 - mean2) SE = standard error * can be independent (nonpaired) meaning its from two different groups at the same time or dependent (paired) meaning its from the same group at different times ANOVA - analysis of variance of 3 or more variables (so checks difference between mean of three or more groups) X-squared - compares percentages or proportions |
|
|
Confidence Interval?
Z score for 95% CI? Reject Null Hypothesis when? Accept? No statistical difference when? |
range of values in which a specified probability of the means of repeated samples would be expected to fall - assumes normal distribution (parametric value)
CI = mean +/- value (smaller CI means better data) For a 95% CI (p=0.05), z = 1.96 (95% of data is within 2 SD of mean) Reject - when zero is included in interval (no statistically significance) Accept - when one is included in interval (at least one SD difference between values) If the CI overlap, no statistical difference |
|
|
What is a correlation coefficient?
Range? Ideal? Coefficient of determination? |
r = how well correlated data is
* between -1 and 1 - the closer to 1, the more correlation between the two variables Coefficient of determination = r squared (reported) |
|
|
Chi square test?
|
x squared = sum of (observed - expected)2/expected
* measures difference among frequencies in a sample * categorical test - compare proportions |
|
|
Survival Analysis?
Looks at what? Predicts what? |
method to analyze and understand time-to-event data
* Measures occurrence of event with the time it takes place (dependent variable) * Predicts multiple things - time to death, time to recurrence, time to progression, etc. |
|
|
Characteristics of time-to-date data?
Censoring? Left vs. Right? Interval? Most common type? |
Share common starting event (exposure to risk factor, infection, etc.), can have single or multiple occurrences
Left - event occurred before observation period Right - event hasn't occurred yet (MOST COMMON) Interval - exact time of event unknown |
|
|
Survival Function?
Probability Density Function? Hazard Function? Relationship between them? |
Probability individual survives longer than time t = S(t)
Probability that failure will occur at time t = f(t) Conditional Failure Rate - probability of failure at time t given that the pt has survived until time t = h(t) h(t) = f(t)/ S(t) |
|
|
Kaplan Meier Method?
|
estimates survival functions from life-time data
S(t) = pi(1-(d/Y)) * pi = product of all estimates * d = number of events at t * Y = number at risk at t |
|
|
Logrank test?
|
compares two or more groups to determine if there is a survival difference
* Null Hypothesis = S(t)1 = S(t)2 |
