HEDU 3720: Introduction to Epidemiology
Problem Set: Randomized Trials
Point values are in parenthesis.
1. Draw and label the design of a randomized trial (2).
2. A double-blinded, placebo-controlled, randomized clinical trial is considered the gold standard for experimental research. Explain what each of the three italicized terms means (3).
3. Define selection bias (1).
4. What is meant by sample size (1)?
5. You are starting up a randomized clinical trial to determine whether a new drug is more effective than the current treatment for a disease called phrenobia. You are particularly concerned that two characteristics (or variables) besides drugs might influence how well persons with phrenobia do: whether or not they are bald and whether or not they are bow-legged. You decide to use stratified randomization as a study design to increase the chances that your new and current treatment groups have about the same number of bald, bow-legged subjects.
You enroll 2000 patients; 1400 have hair and 600 are bald. Of the 1400 with hair, 1100 have straight legs and 300 are bow-legged. Of the 600 who are bald, 400 have straight legs and 200 are bow-legged. Draw a chart showing the stratification of your study population by hair and leg attributes and, after you’ve randomized, the final composition of your new and current treatment groups. Assume that randomization, results in equal assignment of all four strata to treatment groups. (Your chart should be similar in format to that in Figure 10-4 on p.206 of your text.)(2)
6. You are going to conduct a clinical trial to determine whether a new drug relieves arthritis pain any better than aspirin. What are the 4 possible outcomes of your trial (4)?
7. You decide to use the table of random numbers on p. 202 of your text to assign subjects to either the new drug or aspirin in your clinical trial on arthritis. You begin the assignment of subjects at the intersection of Row 5 and Column 10 and move through the table horizontally and to the right in assigning subjects. If odd numbers are assigned the new drug and even numbers are assigned to aspirin, what is the assignment of the first 10 subjects (10)?
8. At the end of your clinical trial on arthritis you conclude that the new drug is not any better than aspirin. Later trials on the same drug prove that it really is better. (a) What kind of error have you made? (b) What do you call the probability of making this kind of error? (c) If that probability were 0.32, what would be the power of your study (3)?
9. If, in a given clinical trial, the probability of making a type I error (α) is 0.01, then what would be the p-value for that study (2)?
10. Use the tables on p. 221 to answer these questions. Medication X, the current treatment for phrenobia, has a cure rate of 35%. Your start-up drug company hopes your new medication Y will increase the cure rate to 60%. You want to be able to detect that improvement with α =0.05 and β=0.20.
(a) How many patients will you need in each treatment group if you want to be able to detect a difference that is either better or worse than Medication X?
(b) How many will you need if you are only concerned with detecting an outcome that is better than Medication X (2)?