Exam Flashcards
Describe: case-control studies
People w/ & w/out condition are compared on exposures/ interventions they had in the past
- Prevalence
- Controls are selected based on prior design
- Starts w/ outcome & looks back at exposure
Describe: cohort studies
Groups that have an exposure or receive an intervention are compared to others who do not
- Incidence
- Disease vs. no disease, followed over time
- Exposure 1st
Describe: cross-sectional studies
- Prevalence
- Exposure & outcome are measured at same time
- “Snapshot”. Ex = survey
What type of research questions do RCTs address?
Prevention, screening, diagnostic accuracy, therapy
Define: incidence
# of new occurrences in a population over a period of time - Conveys info about the risk of contracting the disease
Define: prevalence
Proportion who have a particular condition at a point in time
- Conveys info about how widespread the disease is
Define: absolute risk
Risk of developing the disease over a time period in a group of people
Define: relative risk (risk ratio)
Compares the absolute risk btwn groups of people
- Ratio = 1 = no difference
- Ratio < 1 = less risky
- Ratio > 1 = more risky
Distinguish btwn observational vs experimental research designs
Both observe people & measure outcome
- Observational: lower level of evidence & the investigator does NOT specify intervention
- Experimental: investigator specifies intervention
What type of study designs are considered to be observational?
- Ecologic
- Cross-sectional
- Case reports
- Case control
- Cohort
Distinguish btwn controlled vs uncontrolled studies
Controlled:
- More than 1 group studied
- At least 1 comparison or control group is compared w/ experimental group
Uncontrolled:
- Single group studied
- No comparison
Define: selection bias
People in comparison group may not be the same as the people in the experimental group
How do you solve selection bias?
Randomization, allocation concealment
Define: attrition bias
Differences in withdrawals or exclusions of people entered into a study, excluding pts from the analysis
What does attrition bias lead to?
Overestimation of benefit & underestimation of toxicity
How do you solve attrition bias?
“Intention to treat” analysis: Includes all study participants
Define: performance bias
Differences in the care provided & exposure to other factors
- If providers know which group a participant is in, they may act differently towards that person
How do you solve performance bias?
“Double” masking
Define: information bias
Differences btwn groups in how data are collected or assessed
How do you solve information bias?
- All groups must be measured in the same way
- “Double” masking
What are the levels of evidence? From best to worst?
- Systematic review of RCTs –> 2. Single RCT –> 3. Systematic review of controlled observation –> 4. Single controlled observation –> 5. Single uncontrolled observation –> 6. Physiologic studies –> 7. Unsystematic clinical observation
Describe single controlled observation
Investigator compares 2 groups
Describe single uncontrolled observation
Gathering many pt experiences
- Can’t demonstrate cause & effect
Describe physiologic studies
Data about a mechanism of action
- May not reflect larger population
- Can’t demonstrate cause & effect
Describe unsystematic clinical observation
Stories we hear (subjective reporting)
- May not represent the broader experience
- Lowest level of evidence
Define: surrogate endpoint
Short term outcomes are measured
Define: equipoise
Uncertainty about whether experimental intervention is better
- Ethical requirement
Define: placebo
Substance (or procedure) that looks, tastes, sounds, feels the same as experimental intervention
- Controls for placebo effect
- Allows for “masking”
- Ethical in only certain settings
Define: double masking
Participants & investigators don’t know who is getting control or experimental intervention
- Avoid bias
Define: allocation concealment
Process used to ensure that the investigator does not know which group the person will be allocated
Define: intent to treat analysis
Includes all study participants as part of the group @ which they were originally randomized whether or not they received or completed the intervention
Describe: Ecologic studies
- Comparisons of groups instead of individuals
- Ex. rates of disease in groups/regions
Give an example of an ecologic study
Relationship btwn laws against smoking & prevalence of smoking under age 16
Describe: case-report/series
Uncontrolled observational studies
- Describes unusual or new occurrences
Give an example of a cross-sectional study
Association btwn amt of time adults report watching TV & their BMI
Which types of studies have “no randomization, bias, or self-selection”?
Case control & cohort
Give an example of a case-control study
Mothers of children w/ & w/out spinal malformations were asked about diet during pregnancy
Define: risk
Probability of suffering harm/loss
- Proportion of people experiencing the event of interest in a given time
Define: harm
Physical or psychological injury/damage
Define: benefit
Promotes well-being
Define: screening
Testing for a disease in a population w/ no sx
Distinguish btwn sensitivity vs specificity
- Sensitivity: How well the test finds the disease when it’s present
- Specificity: How well the test excludes the disease when it’s not present
Define: validity
Credibility or believability of research
- Are the findings genuine?
- External or internal
Define: reliability
Repeatability of findings
- If the study was done a 2nd time, would you get the same results?
Define: critical appraisal
Process of assessing & interpreting evidence by systematically considering its validity, results & relevance
What is an abstract?
A condensed version of the full text of the paper, presenting the most important info
What is included in an abstract?
- Context/background
- Study purposes
- Basic procedures
- Main findings
- Conclusions
- New/important aspects of study
Define: type I error (False positive)
Concluding that a tx works, when it actually does not work
- Also called “alpha”
- Rejecting the null, when it is true
Define type II error (False negative)
Concluding that there is no evidence that a tx works, when it actually does work
- Accepting the null, when it is false
- Also called “beta”
- Risk of type II error decreases as participants in study increases
Define: innumeracy
Inability to deal comfortably w/ notions of # & chance
- Affects informed consent & decision-making
- Interferes w/ appraisal
Define: framing effect
The presentation & context of info given affects how it is received
Define: absolute risk reduction
Change in the risk of an outcome of a given tx in relation to a comparative one
Define: relative risk reduction
The absolute risk difference, divided by the risk of the control
Define: odds
# w/ event ÷ # w/out event - Expressed as decimal
Define: odds ratio
Odds of an event in the exposed group ÷ odds of the event in the unexposed group
- When odds ratio = 1, means no difference or no effect
