Flashcards in Exam 1 - DI 726 Deck (30)
What studies provide what level of evidence?
Systematic reviews of RCTs (Highest)
Case Reports/Case Series (Lowest)
Retrospective, where comparison groups are designated by outcome.
Randomized Control Trial
Study Design where comparison groups are randomly assigned by investigator
Case Report/ Case Series
No comparison group, just a showcase of a individual(s) in their disease state
Retrospective observational Cohort Studies
Study design where comparison groups are not designated by outcome, and data is analyzed after event takes place.
When would cohort studies be the best study to use?
- In investigations with multiple endpoints (e.g. risk of death from various causes)
- When estimating incidence in exposed populations
- In investigation of exposure to rare environmental factors
- Investigation of cofactors that vary over time
When would case-control studies be best used?
- Investigation of rare diseases
- Investigation of exposure to multiple agents
Meaning of Measurement
Internal validity vs external validity
Internal Validity - does study use good methods/prove hypothesis
External Validity - Is study applicable/does it represent target population/ Replicable
Reliability vs Clinical Importance
Reliability - Statistically significant (precision); result not due to chance (p-value/CI); consistent to other studies
Clinical - Significant (relevance); outcome important to patients; treatment effect meaningful
Control with Cross Over Studies
Use self control by exposing experimental subjects and controls to periods of exposure and control.
Study uses results from previous study to compare to subjects from current study.
Placebo Control Trial
- Gold Standard in controls
FDA requires 2 positive Placebo control trials for new drug approval.
Controls for factors other than intervention over time
* Treatment Standards
* Disease Waxing/Waning
- Patient expectations produce positive effect, most commonly seen in symptoms/disease w/ stronger psychological component (Pain, depression, Nausea)
Ethics of PCTs
- Unethical to use placebo in place of standard care
- Must have informed consent and patient must not be harmed
What are the three analytical designs used in active control?
- Shows efficacy is better than standard
- Have strong evidence for effectiveness
- Can establish generic equivalency
- Do not establish effectiveness by itself
- Assumes standard is effective
- Establish not inferior to standard therapy
- Does not establish efficacy by itself
- assumes standard is effective
What is association?
The statistical measure of dependence/relationship between two events, characteristics or variables.
- measured in samples
* use statistical tests and modeling to make inferences about population
What is causal inference?
Possible explanations for observed data/association
Inferring causal relationship based upon body of research and knowledge.
- no set rules, guides for weighing evidence in support of causation.
What systems for determining Causal relationship exist?
- Koch's Postulates
- Bradford Hill's criteria for Causation
What is the only required Bradford Hills Criterion for causality?
Temporality of exposure occurring before disease.
What are the components of causal inference and how are they tested?
1. True/Causal Effect
2. Chance (Statistical Tests)
(Study Design/Statistics/Subjective Assessment)
4. Effect Modification
5. Bias (study design/Subjective assement)
What is Risk?
A term for measures of probability of a given outcome (incidence)
- Risk Measures quantify probability of a given outcome in a group of people
What are measures of association?
Measures of association compare risks between groups of people.
In case series studies:
1. What are we studing?
2. How do we measure?
1. Study of "case" patients only
2. Describe exposures or outcomes
3. Convenience sample, or hypothesis generation
4. Cannot measure association