Statistics, Clinical trials, Epi Flashcards
What is incidence?
Incidence: # of newly diagnosed cases divided by the total population over a specific period of time
What is prevalence?
Prevalence: # of existing cases divided by the total population over a specific period of time
o Prevalence = incidence x average duration of disease
Which bias is this?
The appearance of longer survival that is a result of diagnosis at an earlier time during the course of the disease rather than an improved outcome
Lead-time bias
Which bias is this?
screening tests selectively identify patients with slowly progressive disease; preferentially selecting for slowly progressing tumors will give the false appearance of improved survival
Length-time bias
Which bias occurs if gastric cancer is more common on the SE of Asia compared to eastern Africa?
I believe the answer is the following, but I am not sure:
- Prevalence-incidence bias (AKA= length or survival bias): survival of cases is related to exposure and exposed cases may be over- or underrepresented in sample (cross-sectional studies, case-control studies of rapidly fatal cancers)
* *Geographic variations in cancer incidence can be a result of differences in prevalence of the underlying cause including environmental and ethnic (ie. genetic) differences, or to differences in diagnostic criteria.
Another possible answer could be this bias instead:
2. Detection bias: detection of cases is related to exposure to potential disease-related factor, with cases in exposed group over- or underrepresented (case-control studies)
- What phase of a clinical trial would the following be describing?
All patients receive drug. At the end of this period, response assessment is made.
(1) If a response is noted, the subject continues the drug. If progression is noted, they’re removed.
(2) If SD occurs, they are randomized to either receive the drug or discontinue therapy.
(ie. after months of stable disease the patient is changed to another arm with placebo) - What would be the advantage of this type of trial?
- What would be the disadvantage of this type of trial?
- Randomized discontinuation trials: phase II trial designed to evaluate efficacy of drugs that achieve SD
- decrease # of subjects and increase power
- ethics of DCing drug, potential for carry-over effect after drug is DC’ed, failure to detect short duration of activity
Stopping rules can be used for randomized phase II or III trials:
Which is a reason for stopping a trial?
a. Study drug is clearly better than control
b. Study drug is clearly worse
c. Study drug is very unlikely to be better
d. a, b
e. a, c
f. all of the above
Stopping rules are designed to protect subjects from unsafe drugs, accelerate availability of superior drugs, and help ensure transfer of resources and patients to alternative trials
- f- This is a trick question because they all are reasons
- How can you avoid study bias?
2. Which phase is this essential for?
- Once randomization has occurred, subsequent analysis should be reported on an intent-to-treat basis
(i. e, include those who died before treatment was received, or who did not complete treatment course). This will minimize bias based on temporal issues or treatment toxicity. - Essential for phase III trials; some forgiveness for phase II
The significance of study is 0.06. What do you do to increase this?
Increase # of patients enrolled?
What type of study is being described below?
– follows group of people or population
o Outcome generally incidence rate
o Assumes homogenous group; results may not apply to the individual
Ecological
What type of study is being described below?
– follows group of people to determine outcome of exposed vs. unexposed within the group; can be prospective or retrospective
o Prospective: begins with healthy population to assess risk of disease of interest; allows measurement of exposure before the onset of disease
Can directly calculate relative risk -> determines incidence
Cons: expensive, high rate of drop outs, requires large n, slow to perform
o Retrospective: confounding arises from heterogeneity in treatment
Cohort
What type of study is being described below?
– retrospectively evaluates newly diagnosed cases and compares these to controls to determine effect of exposure
o Cannot calculate RR; must use odds ratio instead
o Reduces cost and requires fewer cases than cohort studies; useful if long latent period
o Only able to evaluate 1 outcome variable; simply used to generate a hypothesis for later work
o Subject to selection bias + recall bias
Case-control
What type of study is being described below?
– sample individuals at a specific time point to compare effects of exposure
o Cannot calculate RR; must use odds ratio instead
o Includes both new and previously diagnosed cases (survival bias)
o Useful when studying prevalence
Cross-sectional
Define each of these and state the formula:
- Sensitivity
- Specifcity
- PPV
- NPV
.1 Sensitivity (proportion with disease who test positive) = A / (A+C)
- Specificity (proportion without disease who test negative) = D / (B+D)
- PPV (probability that animal with positive test actually has disease) = A / (A+B)
o Related to specificity - NPV (probability that animal with negative test is healthy) = D / (C+D)
o Related to sensitivity
Kaplan Meir curves:
- Look for # of censored individuals and realize what?
- When you see curves that cross, make sure to look for or understand what?
3 If curves are different, ask what two questions?
- Realize there are too many censored cases
- Short-term survival is better in one group, but long-term survival is better in the other
- Is difference clinically important?
Is the difference statistically significant? Depends on magnitude of difference, variability of data, trial size, etc.
