Epi MT 4-6 Flashcards
Epidemiology
- describes the amount and distribution of disease within a population, regardless of causality
- IDs who, when, where
- observational, not experimental
Descriptive
-e.g. prevalence of color vision deficiency in boys by ethnicity
Epidemiology
- concerned with causes and effects of diseases within populations
- asks why, how
- association bw exposures and outcomes
Analytic
-e.g. diet and AMD: the Melbourne… study - concluded diet high in some things seem to be assoc with advanced AMD prevalence
4 types of descriptive studies in increasing complexity
Case report
Case series
Cross sectional
Ecological
(Note alphabetical)
Descriptive studies
- detailed description of a single case
- often a unique case
- cannot generalize, but can initiate studies
Case reports
Descriptive studies
- subjects of common characteristics of a disease
- no healthy comparison/control group
Case series
Descriptive studies
- examines relationship bw diease and other variables
- a snapshot of the study population
- can be done in a relatively short period of time with large populations
Cross sectional (aka prevalence study) -e.g. ocualar-visual defect and visual neglect in stroke patients
Descriptive studies
- units of analysis are populations or groups (NOT individuals)
- aggregate/overall population risk is determined
- may be done when group, but not individual data is known
Ecological studies
-e.g. correlation bw dietary fat intake and breast cancer by country (key = by country, not individual)
Descriptive studies
-describe ecological fallacy
Findings from ecological studies may not necessarily be applied to an individual
Analytic studies
- investigator manipulates 1+ risk factors and analyzes the effeccts
- can control external factors
- more expensive and difficult
Experimental
-e.g. randomized clinical trial
Analytic studies
-experimental: randomized clinical trial
—used to evaluate __
—generally most __
Eval intervention (preventative or therapeutic procedure) -evals efficacy and potential harm of intervention
Most scientifically rigorous method
Analytic studies
- people are observed to see whether there’s a relationship bw risk factor and health status
- most common design in epidemiology
- no intervention given
Observational
Analytic studies: observational
- group of ppl with given characteristics followed over time (longitudinal study)
- may be most important epidemiological study
- used for diseases that are common
- good way to eval relationship bw development of disease and risk factor
- not useful for diseases that take a while to develop & expensive
Cohort
Analytic studies: observational
-cohort
—none of the individuals have the disease in the beginning
—follow into the future to observe presence/absence of disease
—compare risk factors
—DIRECT MEASURE OF INCIDENCE AND RISK
Prospective
-e.g. Framingham heart study
Analytic studies: observational
-cohort
—IDs individuals with and without the diease in the present
—go back to a time when all of them were free of disease
—compare groups with regard to risk factors
—incidence measured
Retrospective
-e.g. statin use and cataract surgery (published 2013, data from 1998-2009)
Analytic studies: observational
- cases and controls are compared with regard to suspected risk factors for the disease
- usually small groups, rare conditions
- inexpensive, no need to follow over time
- tough to generalize, no new cases/incidence not measured
Case-control
Risk quantification
- probability of an adverse event taking place in a population within a specified time
- measure of incidence
Absolute risk
Risk quantification
-how to measure excess risk
Relative risk
Risk quantification
-equation for relative risk (risk ratio)
= (risk in people exposed)/(risk in people not exposed)
Risk quantification
-ex: new cls wearer asks what his risk of getting an ulcer is - however, even without CLS, he could get an ulcer, so what he is really asking is ___
Excess risk
Risk quantification
-ex: smokers’ relative risk of AMD is 3.29. What does this mean?
The risk of developing AMD among smokers is 3.29x greater than that of the population of non-smokers (NOT 3.29x greater than overall pop)
Risk quantification
-regarding CL wear and corneal ulcers, what does a risk ratio of 3.0 mean?
The relative risk of developing (a corneal ulcer) is 3x greater in (CL wearers than non-CL wearers)
Risk quantification
-regarding current smokers and AMD, what does a risk ratio/relative risk of 1.26 mean?
Current smokers are 1.26xs more likely to develop AMD than non-smokers
I.e. 26% more likely
Ratios
-used in cohort, not case-control studies
Risk ratio
Ratios
-used in case-control, cross-sectional, and cohort studies
Odds ratio
Ratios
-a comparison of frequency of exposure among cases and controls
Odds ratio
Odds ratio
= (odds in favor of exposure among cases)/(odds in favor of exposure among controls)
-asks if a patient with a disease is more likely to have a risk factor than a patient without the disease
Exposure odds ratio
-asks about the risk factor
Odds ratio
= (odds in favor of disease among exposed)/(odds in favor of the disease among unexposed)
-asks if a patient with a risk factor is more likely to have a disease than a patient without
Disease odds ratio
-asks about disease
Confidence interval
- used to address the (2) of the results
- represents the __ of the study
- usual value
- depends on (2)
Repeatability or variability b/w studies
Precision, more narrow = more precise results
95%
Size of sample and variability
Study results and chance
- chance definition
- null hypothesis
Random error, inherent in all studies
-can be minimized, never eliminated
Idea that there’s no real difference and any statistical differences are due to chance alone
-goal of all studies is to reject it
Statistical significance
- define P value
- P values that are statistically significant
The probabilty that the effect is due to chance alone
SS (rejecting null hypothesis) = P values typically starting at/below 0.05
-means there’s a <5% probability results are due to chance
Sampling
- simple
- stratified
Each person has an equal chance of being selected
Population is divided according to characteristics, then randomly selected from those subgroups
Bias
- selection
- information
- confounding
S: airsing from selection of individuals
I: each group does not receive the same info
-e.g. multiple interviewers with different styles
C: due to association of other factors that influence the outcome
-e.g. coffee drinkers are commonly smokers, which is the risk for heart disease
Confounding bias
-attempts to avoid it
Masking - single or double-blind
Validity
- internal
- external
I: extent to which results reflect the relationship bw exposure and outcome
E: generalizability to similar populations
*want good validity, low bias
Number needed to treat (NNT)
-why it’s used
In clinical trials to put results into perspective
-e.g. you would have to treat 20 pts with OHT to prevent 1 cause of glaucoma developing in 5 years
Clinical trial phases (4)
1 - small group, safety, dosage range, SE
2 - larger group, efficacy, safety
3 -efficacy, SE, comparison with commonly used treatments
4 - effect in various populations and SE associated with long-term use