Week 19 Flashcards
Give an example of a bad experiment in the real world and its consequence?
MR vaccine and development of ‘autism’
Consequences:
- UK: Reduced vaccination rates
- UK: More outbreaks of measles
- Japan: stopped using MMR jabs
- UK, EU: ↑↑ cases and deaths
What are the 2 broad classes of research design?
What do these look at?
Observational/correlational/descriptive
- Looks for patterns/associations
- Correlation; cannot prove causation
Manipulative/experimental
- Manipulates conditions
- Stronger design; tests cause and effects
- Seeks to prove causation
What are the essentials for experimental design?
Clear and precise question
Controls/comparator group(s)
Randomization
Repetition
Observational studies often develop what type of data?
Categorical
Correlation does not mean…
causation
What is a confounding variable?
A third variable that hasn’t been accounted for that affects the variables
(A and B don’t effect eachother, but both are affected by a third variable)
Who is considered to be the first clinical trial to investigate scurvy (1753)?
James Lind 1716-1794
Using manipulative/experimental studies
First controlled trial of scurvy?
12 sailors with scurvy treated on board ship:
2 = half a pint of sea water a day
2 = 2 spoonfuls of vinegar 3x a day
2 = nutmeg-sized mix of garlic, mustard, myrrh, 3x a day
2 = elixir of vitriol 3x a day (sulfuric acid)
2 = quarts (1L) of cider a day
2 = 2 oranges and 1 lemon a day
all were treated for 6 days in the same location and with the same meals
the 2 fruit treated were (almost) fit for duty after 6 days
Research designs and levels of evidence (how strong results are to publish to the public): most to least strong?
Meta-analysis of (all relevant) randomised controlled trials
- eg Effectiveness of drug X examined by combining all RCT’s in a meta-analysis
Randomised Controlled Trial (RCT)
- eg Group of COPD patients randomly assigned to 1 of 2 groups, new drug X v existing treatment, lung function compared
Quasi-RCT and Clinical Controlled Trial
- eg Provision of smoking cessation service for COPD, incidence of smokers compared before and after introduction of service
Case-control study (i.e. condition v healthy)
- eg Under age 50 COPD deaths retrospectively compared to age-matched controls to identify determinants.
Cohort study
- eg Smokers and non-smokers followed for a number of years and incidence of COPD recorded. Causality hypotheses
Cross-sectional study
- eg COPD and smoking
Single Case study
- eg Effective treatment of 1 COPD patient
randomised controlled trials is the ______ ______ study design for evaluating interventions
“gold standard”
Why may variation in a repeated measure design occur?
Chance or unsystematic events e,g, instrument precision
Systematic inconsistency e.g. different experimenters, different situations
Actual change in the underlying event being measured
Measurement validity is…
The degree to which any measurement approach or instrument succeeds in describing or quantifying what it is designed to measure
What is bias?
A systematic error (caused by the investigator or the subjects) that causes an incorrect (over- or under-) estimate of an association.
Selection bias, the most common type, can occur in several ways:
Selection of a comparison group (“controls”) that is not representative of the population that produced the cases in a case-control study. (Control selection bias)
Differential loss to follow up in a cohort study, such that the likelihood of being lost to follow up is related to outcome status and exposure status. (Loss to follow-up bias)
Refusal, non-response, or agreement to participate that is related to the exposure and disease (Self-selection bias)
Using the general population as a comparison group for an occupational cohort study (“Healthy worker” effect)
Differential referral or diagnosis of subjects
what is a double-blind study?
where the treatment vs control status is unknown to the researchers as well as to the subjects.
Here, blinding avoids the possibility of both measurement bias and the placebo effect
What is a single-blinded study?
might be one where the disease/exposure status is unknown to the researcher.
Here, blinding avoids the possibility of measurement bias by the researcher
basic principles for experimental design?
Formulate question/goal in advance
Comparison/control
Replication
Randomization
Measurement validity and reliability
Measure/record possible confounders, statistically correct for them
Type I vs Type II error?
Type I error = false positive so reject null hypothesis
Type II error = false negative so do not reject null hypothesis
what is statistical power?
the likelihood of detecting a significant difference or significant association, by sampling, if it actually exists between the underlying populations
What is precision, what is accuracy?
precision = repeatability
Accuracy = there is a true value, and how good is the measure at measuring it?
characteristics of a good experiment?
Unbiased: Randomization, Blinding
High precision: Uniform material, Replication
Stratification: Simple, Protect against mistakes
Wide range of applicability: Deliberate variation, Factorial designs,
Able to estimate uncertainty: Replication, Randomization