Research Ethics
What is the purpose of the Institutional Review Board?
Protect rights and welfare of human subjects participating in research
Rejecting the null when the null is true
Type 1 error ( designated as alpha)
The primary outcome of a randomized controlled trial is mortality. What statistical test is best to assess mortality differences between 2 groups?
Chi Square (or Fisher Exact test if small sample size with cell frequency less than 5)
Researchers set the power of a study at 80% to detect a difference between groups. What is the likelihood of a type II error?
Beta (likelihood of Type 2 error) = 20%
Power = 1-Beta.
Which of the following sources of evidence, if well-conducted, should carry the most weight? Rank highest to lowest evidence.
Case Control study
Systematic review
Cohort study
Randomized controlled trial
Systematic review
Randomized controlled trial
Cohort study
Case Control study
What are the three ethical principals of the Belmont Report?
Respect for Persons (autonomy); Beneficence and Justice
Failing to reject the null hypothesis when the alternative hypothesis is true
Type II error (designated as Beta)
In a RCT study looking at obesity in subjects receiving drug A or Placebo, to compare the primary outcome of bodyweight in two groups what do you need to know before picking the statistical test?
If data meets assumptions for Parametric:
Normality; Equal Variance; No extreme outliers; Independence (Data in each group randomly and independently sampled)
Which of these are not Continuous data?
Temperature
Blood pressure
weight
Pain score
Pain Score is Categorical Ordinal, not Continuous
Which of these study designs can be used to assess incidence?
Cross-sectional
Case Control
Cohort
Cohort
How are the 3 ethical principles applied?
Autonomy - Informed consent; Beneficence - Minimize risks/maximize benefits; Justice in Selection of Subects
Researchers calculate a sample size of 180 based on 10% incidence with a power of 80% to detect an increase of 15%. What is the beta value for this study?
1-Beta=Power
1-Beta = 0.80
Beta = 0.20
A RCT study of 200 patients compares study drug (4 groups of varying doses).The primary outcome is blood pressure and data is normally distributed with equal variance. What statistical test is appropriate?
ANOVA
Which measure of central tendency is used for nominal data such as Gender?
Mode
In a vaccine study, 5 out of 100 people who had the vaccine later developed disease; 10 out of 100 unvaccinated people later developed disease. What is the Number Needed to Treat?
ARR = 5/100 - 10/100 = 0.05
1/.05=20
A research study finds the study drug group is doing much better than the placebo group. If the researchers give placebo group an opportunity to receive the study drug, which principle of the Belmont Report is being applied?
Beneficence (maximize benefits)
If you increase alpha, what happens to beta and what happens to Power (1-beta)?
If you increase alpha, beta decreases and power increases.
To increase power of a study, you can increase alpha or increase the sample size.
Results from a case control study looking at tea drinking and breast cancer shows an odds ratio of 1.3 (95% CI 0.9-1.4). Based on this result, is the association between tea drinking and breast cancer an increased risk or decreased risk?
Confidence interval crosses 1. Results are not statistically significant. In this study, tea drinking is not associated with an increased or decreased risk of breast cancer.
Which measure of central tendency is not affected by outliers and can be used for both continuous and ordinal data?
Median
In a vaccine study, 5 out of 100 people who had the vaccine later developed disease; 10 out of 100 unvaccinated people later developed disease. What is the Relative Risk?
Relative Risk =(5/100)/(10/100) = 0.5
Relative Risk less than 1 shows protective factor against the disease.
Who has authority to grant a waiver of informed consent?
The Institutional Review Board
(only for minimal risk studies that meet regulatory requirements)
The difference in mortality between study drug A and study drug B was 6% (p=0.01). What was the chance of making a Type I error rate?
p=0.01 means the chance of making a Type I error is 1 in 100.
A cohort study looked at the association between oral contraception and blood pressure. Blood pressure was measured at two time points (baseline, and 1 year). The mean difference in blood pressure between the two times was equal to the median difference. What statistical test is appropriate?
Paired t-test
Investigators set α at 0.05. Results give a P-value of 0.02.How do you interpret?
●Statistically significant
●There is a 2 in 100 chance of a Type I error
In a 2x2 contingency table for a Case Control Study: Disease (yes/no) and Risk factor (yes/no) what is the formula for Odds Ratio?
OR = odds that case was exposed divided by odds that control was exposed
OR = (A/C)/ (B/D)