Lecture 7 Flashcards
All clinical trials in the US must be registered where
Clinical trials.gov. Maintained by national library of medicine.
Relatively new- first available in 2000
What to balance when choosing a sample size
Limit cofounding variables but keep study generalizable to the entire population
Develop a good ____ before sample size calculations
Research hypothesis. A simple hypothesis contains one predictor and one outcome variable.
Null hypothesis
No difference between groups. Proving the alternative hypothesis is not the goal.
reject the null hypothesis
Means that you reject the negative, results in a positive.
Reject that there is not significant difference. Meaning that there is a significant difference.
Analytic study sample size estimation (4)
- State null and alternative hypothesis. One or two sided?
- Select a statistical test based on predictor and outcome variables. Use Standard dev.
Ex: Chi square, T test, correlation coefficient. - Estimate effect size. Difference between the two groups in your comparison that is meaningful.
- Specify alpha (Type I error) or beta (type 2 error). Amount of error you are willing to accept.
Type I error (alpha). Standard accepted amount?
Probability of a false positive.
a= 0.05 or 5%
Type II error (beta). Standard accepted amount?
Probability of false negative.
b= 0.10-0.20
Power = 1-b
Power usually = 80-90%
How to find your standard effect size
Significant difference/ Standard deviation
Descriptive study sample size estimation (3)
Estimate proportion with dichotomous outcome or standard deviation of continuous outcome
Specify CI
Specify confidence level (95%)
Sample size and power common errors
Sample size estimated too late in process.
Misinterpreting dichotomous variables.
Dichotomous generally results in a ____ sample size than continuous variables
larger.
Most effective way to change prescribing habits of a clinician
Direct to healthcare provider advertising. Drug rep travels to dr offices.
Direct to consumer advertisting
Push and pull method. Push retail displays and pull people into drs offices to ask about drug.
FDA regulated.
What happened in 1997 when FDA relaxed regulations about marketing
Significant increase in direct to consumer advertising and prescribing rate of drugs. Since, things have been further relaxed. Primary argument is to allow consumers to be more involved in healthcare.
Academic Detailing
Good. Goal is to improve patient care and affect prescribing.
Academic detailers have no link to pharmaceutical industry. Independent.
Occurs in academic or non-commerical institutions. Education of prescribers by trained clinicians on EBM/RCTs.
Good in between of academia and drug industries coming to us.
Sunshine act
Transparency. Minimize bribery by drug industries.
How can you apply population EBM data to individual patient care?
- Identify for the individual patient the objective of treatment.
- Cure, prevention or recurrence, reassurance, relief. - Select most appropriate treatment
- Specify treat goals and endpoints.
- Discuss benefits vs risks/alternatives/costs
CONSORT statement evaluates what kind of study
RCT
Surrogate endpoints
A variable that is relatively easily measured and predicts are rare of distant outcome of therapeutic intervention.
Itself is not a direct measure of either harm or clinical benefit.
Biomarker intended to substitute for a clinical endpoint.
Why would you use a surrogate endpoint?
Reduce cost, allow assessment of treatment whose primary outcomes are invasive or unethical.
Downside of surrogate endpoints
May not be a true indicator of success rate of treamtment Indirect.
Represents only a single measure.
May come from animal studies.
Examples of surrogate endpoints
Pharmacokinetics. In vitro measures. Macroscopic appearance of tissues. Changes in levels of biological markers Radiological appearance
Pros and cons of animal studies
Pros:
- Homogenous and inbred populations reduce cofounding variables.
- Can evaluate multiple generations
- Animals kept in strictly controlled environments.
Cons:
- May not reflect same process in humans.
- May need higher doses.
- we are more diverse
Features of ideal surrogate endpoint
- Dose-reponse effect. Reliable, reproducible relationship with dose and response.
- True predictor or disease.
- Sensitive. Positive predictive value. (its good to be sensitive)
- Specific. Negative predictive value (specific= nit picky. negative)
^so precise cut off between normal and abnormal.
Dry eye and MMP-9
Surrogate endpoint for dry eye. Evaluates for dry eye by inflammation.
Example of anecdotal evidence you should ignore when working with drug reps
Ignore “famous” optometrists using the product.
STEP questions when working with drug reps
Safety Tolerability Efficacy Price and apply CONSORT checklist for reporting results.
Two types of endpoints to a clinical trial study
Clinical (true) endpoint
Surrogate endpoint
The manipulation of surrogate endpoints by manufactures to cast their products in the most favorable light in their sales literature was the focus.
Clinical endpoint
A characteristic or variable that reflects how a patient feels, functions, survives.
