Epidemiology Flashcards
National Clinical Trials (NCT) number
- unique ID # given by clinicaltrials.gov once research protocol is submitted prior to study initiation
Positive vs Negative Kurtosis
Positive - more cluster around mean
Negative - less cluster around mean
Type 1 error
NOT ACCEPTING null hypothesis when it was actually true, and you should have accepted it
Type 2 error
Accepting the null hypothesis when it is actually FALSE, and you should not have accepted it
p-value
probability of making a Type 1 error if the Null Hypothesis is rejected
Nominal, 2 groups, Independent
Pearsons Chi squared test
Fishers Exact
Nominal, 2 groups, Related
McNemar
Nominal 3+ groups, Independent
chi squared test of independence
Fishers Exact (less than 5 people)
Nominal 3+ groups Related
Cochran
Nominal Survival
Log Rank
Nominal Correlation
Contingency Coefficient
Nominal Prediction/Association (Regression)
Logistics Regression
Ordinal 2 groups Independent
Mann Whitney
Ordinal 2 groups Related
Wilcoxon Sign Rank
Ordinal 3+ groups Independent
Kruskal Wallis
Ordinal 3+ groups Related
Friedman
Ordinal Post-Hoc tests
Student Newman Keul
Dunnett
Dunn
Nominal Post-Hoc test
Bonferroni Correction
adjusts p value for # of comparisons being made
Ordinal Survival
Cox-Proportional Hazard
Survival Tests
compares proportion of events over time, or time-to events, between groups
Correlation tests
provides a quantitative measure of the strength and direction of a relationship between variables
Regression tests
predict the likelihood of some event
Ordinal Correlation
Spearman Correlation
Ordinal Regression
Multinominal Logistics Regression
Interval 2 groups Independent
student t test
Interval 2 groups Related
paired t test
Interval 3+ groups Independent
ANOVA
confounders = ANCOVA
Interval 3+ groups Related
Repeated Measures ANOVA
confounders = RP ANCOVA
Interval Survival
Kaplan Meier
Interval Correlation
Pearson Correlation
Interval Regression
Linear Regression
Student Newman Keul
compares all pairwise comparisons possible
all groups equal in size
Dunnett
compares all pairwise comparisons against single control
all groups must be equal in size
Dunn
compares all pairwise comparisons possible
all groups are NOT equal in size
Tukey and Scheffe tests
compares all pairwise comparisons possible
Tukey MORE conservative than Student NK
Scheffe - most conservative
Interventional vs Observational Studys
Interventional = forced allocation
Observational = NO forced allocation
- most observational study designs not able to prove causation
Simple Random Sampling
assign random numbers, then take randomly selected numbers
Systemic Random Sampling
assign random numbers, randomly sort, select highest/lowest number, then take every Nth number to get sample size
Stratified Simple Random Sampling
stratify sampling by characteristic (gender) then use Simple random sampling
Stratified Disproportionate Random Sampling
disproportionately use stratified random sampling when baseline population is not at the desired proportional percentages to the referent population
Cluster Multi-Stage Random sampling
same as multi stage random sampling but ALL elements clustered together are selected for inclusion
Multi-Stage Random sampling
uses simple random sampling at multiple-stages towards patient selection
Convenience Sampling
decided on what fraction of population is to be sampled and how they will be sampled
there is some known or unknown order to the sample generated by the selected scheme which may introduce SELECTION BIAS
4 key principles of bioethics (ABJN)
Autonomy - self rule/self determination
Beneficence - do good for patient, not society
Justice - equal/fair treatment of patient
Nonmaleficence - do no harm
IRB Review Lvls: Full Board, Expedited, Exempt
Full - used for ALL interventional trials w/more than minimal risk to patients
Expedited - minimal risk and/or no patient identifiers
Exempt - no patient identifiers, environmental studies, use of existing data/specimens (de-identified)
Increasing Strength of Evidence (lowest to highest)
test tube –> animal –> case report –> case series –> ecological –> cross sectional –> case-control –> cohort –> interventional –> systemic reviews
Interventional Phase 0
assess drug-target actions, healthy volunteers, very small N (<20), very short duration (single dose to few days)
Interventional Phase 1
assess safety/tolerance and pharmacokinetics of doses, healthy/disease volunteers, small N (20-80), short duration (few weeks)
Interventional Phase 2
assess effectiveness, diseased volunteers (narrow inclusion criteria), larger N (100-300), short-medium duration (weeks to months)
Interventional Phase 3
assess effectiveness, diseased volunteers (expanded inclusion criteria), larger N (500-3000), longer duration (few months to years)
LAST PHASE BEFORE FDA Approval
Interventional Phase 4
assess long-term safety, effectiveness, optimal use, diseased volunteers, population N, wide range of durations
Simple, blocked, Stratified Randomization
Simple - equal prob for allocation into one of groups
Blocked - ensure balance within each interventional group
Stratified - ensures balance with known confounding variables
Epidemiology Definition
public health basic science which studies the DISTRIBUTION and DETERMINANTS of health-related states or events in specific populations to control disease and illness and promote health
Epidemic, Outbreak, Endemic, Pandemic`
Epidemic - occurrence of disease in excess of normal compared to baseline
Outbreak - epidemic limited to localized increase (Cluster)
Endemic - constant presence of disease within an area above normal (constant epidemic)
Pandemic - world-wide epidemic
Incidence (Risk or Attack Rate)
of new cases of illness / # of people at risk of illness
subtract out number of people that already have the disease
Incidence Rate
of new cases of illness / person-time at risk for disease
Prevalence
of existing cases of disease / # of persons in population
Crude Morbidity Rate
persons w/disease / total population
Crude Mortality Rate
of deaths (all) / total population
Cause-Specific Morbidity Rate
persons w/specific illness / total population
Cause-Specific Mortality Rate
deaths by specific illness / total population
Case-Fatality Rate
deaths by specific illness / total # of cases
Cause-Specific Survival Rate
of cause-specific cases alive / # cases of disease
Proportional Mortality Rate (PMR)
of illness specific deaths / total # deaths in population
Maternal Mortality Rate
of female deaths related to pregnancy / 100,000 live births
Risk Ratio
Risk of Outcome (Exposed) / Risk of Outcome (Non-Exposed)
A/(A+B)) / (C/(C+D)
Odds Ratio
Risk of Exposure (Diseased) / Risk of Exposure (Non-Diseased)
(A/C) / (B/C)
If there is a 15% difference between crude and adjusted measures of association…
a confounder is present
Crude = OR/RR Adjusted = RR/OR
Three Requirements for Confounders
- independently associate with exposure
- independently associate with outcome
- not directly in causal-pathway
What is a confounder?
a 3rd variable that distorts an association between the Exposure and the Outcome
Confounding Control: Randomization, Restriction, Matching, Stratification, Mutivariate
Random - allocate equal # of known confounders between groups
Restriction - use only subjects without predetermined confounder
Matching - selected in matched pairs
Stratification - evaluate association between within various strata
Multivariate - mathematically factor out effects of confounder(s)
Effect Modification
a 3rd variable that, when present, modifies the magnitude of effect of a TRUE association by varying it within different strata
Hawthorne Effect
individuals alter/modify their behavior because they are part of a study and know they are under observation
Hill’s Criteria
Strength - size of the measure of association (bigger = better)
Consistency - repeated observations in different populations under different circumstances in different studies
Temporality - cause precede the effect/outcome in time (closeness is better)
Biologic Gradient - presence of gradient of risk associated with the degree of Exposure (more good = inc. health)
Plausibility - presence of biological feasibility to the association, which can be understood and explained
Belmont Report (3 guiding principles)
- Respect for persons - research should be voluntary, subjects autonomous
- Beneficence - research risks are justified by potential benefits
- Justice - risk and benefits of research are equally distributed
When is cohort design useful?
When studying a rare exposure
- typically retrospective
Sensitivity
how well a test can detect presence of disease which in fact disease is present
- proportion of time that a test is positive in a patient that DOES have the disease
Sensitivity Equation
True Positive (A) / (True Positive + False Negative (C)) x 100%
Specificity
how well a test can detect the absence of disease when in fact the disease is absent
- proportion of time that a test is negative in a patient that does not have the disease
Specificity Equation
True Negative (D) / (True Negative + False Positive (B)) x 100%
Positive Predictive Value
how accurately a positive test predicts the presence of disease
- proportion of True Positive’s in patients with a positive test
Positive Predictive Value Equation
True Positive / (True Positive + False Positive (B)) x 100%
Negative Predictive Value
how accurately a test predicts the absence of disease
- proportion of True Negatives in patients with a negative test
Negative Predictive Value Equation
True Negative / (True Negative + False Negative) x 100%
Reasons to pick Case Control (4)
- unable to force group allocation
- limited resources
- disease is rare
- exposure data is difficult/expensive to obtain an/or very time inappropriate
Case Control Strengths
- good at assessing multiple exposures of ONE outcome
- used when diseases are rare
- determine association (NOT CAUSATION)
- less expensive
Case-Control Study
- observational studies allowing passive observation of natural events occurring in individuals with disease of interest compared to people without disease of interest
- groups assigned based on DISEASE STATUS
Cohort Study
- observational studies allowing passive observation of a natural events occurring in naturally-exposed and unexposed groups
- group allocation based on EXPOSURE-status or GROUP MEMBERSHIP
Reasons to pick Cohort (4)
- Unable to force group allocation
- limited resources
- exposure of interest is RARE
- more interested in incidence rates or risks for outcome of interest
Prospective, Retrospective, Ambidirectional Cohort Studies
Prospective - exposure group selected on post or current exposure, both groups followed into future
Retrospective - exposure and outcome have already occurred, groups allocated on past history of exposure
Ambidirectional - use retrospective design to asses past differences but also adds future data collected on additional outcomes prospectively from start of study