Epidemiology Flashcards
What is epidemiology?
The quantitative study of the distribution, determinants and control of disease in populations.
What is the purpose of a study?
To measure the effects of a particular exposure to an outcome.
What are the causes of correlation?
- Causative
- Confounder
- Chance
- Bias
What is a confounding factor?
- Risk factor or predictive of outcome
- Associated with but not direct consequence of exposure
- Irrelevant to the study objective
- Unequally distributed amongst the
What are the 2 main confounding factors?
- Age
- Sex
What is bias?
Systematic deviation from the truth, producing a mistaken estimate of the relationship of an exposure with the risk of disease.
What are Koch’s postulates for causation?
- The agent appears in every case of the disease.
- The agent occurs in no other disease.
- The agent is capable of inducing the disease in healthy subject.
What are the problems with Koch’s postulates?
- There are usually multiple factors influencing a disease. These include:
1. Agent
2. Host (e.g. species, age, genetics…)
3. Environment (Weather, housing, air…)
What are the different natures of studies?
- Therapeutic: Benefits patients, in addition to improving scientific knowledge.
- Scientific: May not benefit patients, but improves scientific knowledge.
What are the types of studies?
- Observational: Observing outcomes in patients with certain exposure.
- Descriptive: Using pre-existing statistics from other studies to draw conclusion (e.g. meta-analysis).
- Experimental/intervention: Actively exposing patients to certain exposure and measuring outcomes.
What are the criteria for causation?
- Strength of association
- Coherence (biological plausibility)
- Time sequence
- Consistency of association
- Dose response
- Reversibility
What are the different types of observational studies?
- Cross-sectional study: Measure the exposure and prevalence of a disease in a population at one moment in time.
- Case control (retrospective) study: Group of people with disease compared to group of people without disease to determine differences in exposure. (Present –> Past)
- Prospective study: Select group of people with disease and monitor the incidence of disease in group. (Present –> Future).
What are the advantages of case control studies?
- Cheap
- Suitable for rare diseases
- Small number of subjects needed
- Results quickly obtained
What are the disadvantages of case control studies?
- Selection problems for cases
- Selection problems for controls
- Recall/observer bias
- Time sequence difficult to determine
- Only produces estimate of relative risk (odds ratio)
What is an odds ratio (relative risk, RR)?
For study (+ is exposure, - is non-exposure) : Case (+) = a Control (+) = b Case (-) = c Control (-) = d
RR = Incidence (+) / Incidence (-) = (a/(a+b))/(c/(c+d))
What are the interpretations of RRs?
RR > 1: Increased risk
RR = 1: No risk
RR
What is excess/absolute risk (E/AR)?
For study (+ is exposure, - is non-exposure) : Case (+) = a Control (+) = b Case (-) = c Control (-) = d
ER = Incidence (+) - Incidence (-) = (a/(a+b))/time - (c/(c+d))/time
What is proportionate ER?
Proportionate ER = ER / Incidence (+)
What us population ER?
Population ER = ER x Proportion of population exposed
What are the advantages of prospective studies?
- No recall/observer bias
- Allows incidence rates and true relative risk to be calculated
- Allows other associations to be deduced
What are the disadvantages of prospective studies?
- Bias in selection
- Large sample required
- Long follow up (expensive)
- High drop-out rate
- Methodology/criteria change in disease identification over time
What information can be estimated from a retrospective study?
For study (+ is exposure, - is non-exposure) : Case (+) = a Control (+) = b Case (-) = c Control (-) = d
RR ≃ ad/bc
Incidence (+) = a/b
Incidence (-) = c/d
How can bias be reduced in retrospective study?
- Blinding subject/observer to hypothesis
- Use objective methods of obtaining exposure
What are 3 key aspects of trials?
- Randomisation
- Control
- Blinding
Should patients who drop out of a trial be counted in final analysis?
Yes, to ensure randomisation is valid.
What are the types of blinding?
- Single blinding (patients):
- Eliminates placebo - Double blinding (patients + observers):
- Eliminates placebo
- Eliminates recording bias - Triple blinding (patients + observers + statisticians):
- Eliminates placebo
- Eliminates recording bias
- Eliminates analysis bias
What are the advantages of trials?
- Minimises bias
- Minimises confounders
- Quantifies risks/benefits
What are the disadvantages of trials?
- Limited feasibility
- Very specific
What is prevalence?
Proportion of defined group with condition at specific point in time.
What is the equation for prevalence?
Prevalence = Number of cases / Total population
What is incidence?
Rate of occurrence of new cases of a disease over a period of time.
What is the relationship between prevalence and incidence?
Prevalence = Incidence x Duration of disease
What does an increase in prevalence and incidence indicate?
Better treatment (greater duration of disease or survival time)
What does an increased incidence and unchanged prevalence indicate?
High mortality
What is prevalence used to measure?
- Chronic diseases
- Slow onset
- Slow recovery/mortality
What is incidence used to measure?
- Acute diseases
- Quick onset
- Quick recovery/mortality
What is mortality?
Death
What is morbidity?
Illness
What is “Number Needed to Treat (NNT)”?
- The number of people that need to be treated in order for one person to prevent the disease in one person.
- Calculated using ER.
What is population excess risk?
Population excess risk = ER x Prevalence
What are the different types of results that can be obtained from a diagnostic test?
True positive (TP): Test was +ve and person had disease. False positive (FP): Test was +ve and person didn't have disease. True negative (TN): Test was -ve and person didn't have disease. False negative (FN): Test was -ve and person did have disease.
What is sensitivity?
Proportion of true positives correctly identified by test.
Sensitivity = TP/(TP + FN)
What is specificity?
Proportion of true negatives correctly identified by the test.
Specificity = TN/(TN +FP)
What is the predictive value of a positive test?
The likelihood that a person with a positive test has the disease.
Predictive value = TP/(TP + FP)
What is the predictive value of a negative test?
The likelihood that a person with a negative test does not have the disease.
Predictive value = TN/(TN + FN)
How are the predictive values dependent on the prevalence?
- If prevalence is high, positive predictive value also high, and vice versa.
- If prevalence is low, negative predictive value also low, and vice versa.
What is the purpose of an asymptomatic test?
- Detects possible condition in asymptomatic individual before symptoms manifest.
- Allows early interventions that may improve outcome.
What are the criteria for the condition to be worthy of testing?
- Important health problem
- Natural history understood
- Detectable at latent stage
- Early treatment is better than later treatment
What are the criteria of the test?
- Sensitive
- Specific
- High predictive values
- Safe
- Low cost
- Chance of benefit should be higher than chance of harm
What is lead time bias?
Increased survival time because disease diagnosed earlier, not because test extended life of participant.