View Full Version : Please Help Again Data Analysis
QUESTION- THE DRUG AZT WAS THE 1ST EFFECTIVE TREATMET FOR AID. AN IMPORTANT MEDICAL EXPERIMENT DEMONSTRATED THAT REGULAR DOSES OF AZT DELAY THE ONSET OF SYMPTOMS IN PEOPLE WHOL HIV IS PRESENT . THE RESEARCHERS WHO CARRIED OUT THIS EXPERIMENT WANTED TO KNOW THE FOLLOWING.
**DOES TAKING EITHER 500MG OF AZT OR 1500MG OF AZT PER DAY DELAY THE DEVELOPMENT OF AIDS?
****IS THERE ANY DIFFERENCE BETWEEN THE EFFECTS OF TAKING THESE TWO DOSES?
THE SUBJECTS WERE 1200 VOLIMTEERS ALREAD INFECTED WITH HIV BUT NO SYMPTOMS OF AIDS WHEN THE STUDY STARTED.
A. Outline the design of the experiement
MY Answer 500mg per day pd AZT patients with HIV NO SYMPTOMS OF AIDS
1500mg per day of AZT PATIENTS WITH HIV NO SYMPTOMS OF AIDS
B DESCRIBE BRIEFLY HOW YOU WOULD USE A TABLE OF RANDOM DIGITS TO DO RANDOMIZATION REQUIRED BY DESIGN? THEN USE TABLE B AT LINE 110 TO CHOOSE THE FIVE SUBJECTS FOR ONE OF THE GROUPS.
MY ANSWER I WOULD USE THE TABLE IN ORDER TO ENSURE THE ACCURACY OF THE EXPERIMENT. NUMBERS FROM TABLE B 60,230,345,600,780,892
01-18-2006, 12:25 PM
A. you need to have a control group as well, so you need to randomly assign the 1200 patients to the following groups (400 per group):
(1) Control Group - take a placebo
(2) 500mg AZT
(3) 1500mg AZT
then measure the onset of symptoms for each group.
Without a control group, you have no way of knowing whether or not the doses are truly effective, compared to patients who take nothing.
Also, this study should be "double blind" - patients should not know which group they have been assigned to, and doctors should not know which group a patient has been assigned to.
B. You use the table in order to make a random assignment and avoid bias.
I am unsure about the control group? why not just the two treatment plans?
Also when I am using the table to look for my answer will any number I use work ? or should it be a pattern i am looking for?
01-18-2006, 12:56 PM
OK, if you consider the 500mg group as the control or baseline, then you don't need a placebo group, and you would split the 1200 into two groups of 600.
With random number tables, start anywhere - you definitely don't want a pattern.
Thanks :o) Now I understand the control group. We just never dicussed that in my class or it went over my head :o( Thanks again :o)
01-19-2006, 09:41 AM
as i read the question, you definately need a control group.
the first question is "will taking either 500 or 1500 ..." which means you need to know the mean time to onset in the untreated population.
01-19-2006, 10:30 AM
But the question does mention that AZT already demonstrated a delayed onset.
"AN IMPORTANT MEDICAL EXPERIMENT DEMONSTRATED THAT REGULAR DOSES OF AZT DELAY THE ONSET OF SYMPTOMS IN PEOPLE WHOL HIV IS PRESENT ."
This 500mg / 1500mg comparison study doesn't need to have a placebo group, since the AZT effectiveness has already been established - they're now just trying to compare the efectiveness of two dose levels.
01-19-2006, 02:25 PM
if you feel that one previous study lays sufficient foundation then you are correct.
another perspective might be to consider that if we acceept the previous experiment's result then why ask the first research question ("does taking either...") since you allready know the answer? if the goal is to verify the result of the first study and also test the new question of dosage level then you need a control. perhaps the real issues here is: what is the question really asking me to determine?
to me it boils down to wether you wish to rely on the previous work or not. some experimenters would others, like myself, would not unless the previous work(s) was(were) exhaustive enough to be more or less definative. since there appears to only be one such study i would include one just in case the former study was in error.
remember that the statistician always leaves room for their conclusions to be wrong, a 95% confidence level leaves 5% room for error ;)
01-19-2006, 06:54 PM
Yes, I agree that a repeated study can be powerful evidence, but in this case, I don't see a lot of value in repeating a control group. A treatment group, yes, but not a control group.
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.