USMLE forum
Step 1
Step 2 CK
Step 2 CS
Matching & Residency
Step 3
  <<   < *  Step 3   *  >   >>  

* Fred biostatistic question / explanation needed
  ahmedcsck - 05/11/18 01:01
  A phase 2, multicenter trial was conducted to determine the efficacy of a new vaccine for prevention of HIV infection. The study enrolled 4000 subjects, aged 20 to 65 years. Of these subjects, 2100 were men and 1900 were women; 2500 were white, 1000 were African American, 300 were Hispanic, and 200 were Asian/Pacific Islanders. Results of the trial showed no overall benefit of the vaccine. However, post hoc analysis disclosed a small but statistically significant vaccine protection among African American subjects. Which of the following is the most accurate rationale for questioning the validity of efficacy analysis of the HIV vaccine among the African American study subjects?

(A) Allocation bias favored African American subjects
(B) HIV infection is more prevalent among African American populations
(C) The study was not blinded
(D) There was a Type II error
(E) Vaccine response among African American subjects was not the primary outcome measure

The answer should be E
Report Abuse


* Re:Fred biostatistic question / explanation needed
  rpl - 05/14/18 14:15
  Ans should be D. there is a type 11 error due to confounding bias.
Report Abuse

* Re:Fred biostatistic question / explanation needed
  ahmedcsck - 05/18/18 22:45
  No Buddy, this is one of the Fred questions for step3 and their answer is E.  
Report Abuse

* Re:Fred biostatistic question / explanation needed
  lamonti - 05/19/18 10:05
  True. types of outcome measure could be primary, secondary, or post hoc. The results among AA was a post hoc, not a primary outcome measure.  
Report Abuse

* Re:Fred biostatistic question / explanation needed
  rossy17 - 07/27/18 06:12
  Presumably the randomization was done with the primary outcome in mind. And in a large enough trial, the randomization probably achieved its purpose of equalizing the distribution of unknowns across treatment groups. A study might be randomized for sex, for example, so there is a roughly equal distribution of males and females selected for each treatment group.
But a Secondary, Post-Hoc analysis may be looking at entirely different outcomes or relationships that were not thought of during the planning of the trial. If the original study did not randomize on factors that could influence these secondary outcomes, then the results could potentially be biased.

For example, if subjects were originally randomized on sex but not on age, then a secondary analysis looking at, for example, muscle mass, could be strongly influenced by the age of the subjects in each treatment arm. And just by chance, there could be more older people in one arm of the study than the other arm and they would likely have less muscle mass just because of their age.
Report Abuse

          Page 1 of 1          

[<<First]   [<Prev]  ... Message ...  [Next >]   [Last >>]

Logon to post a new Message/Reply




Step 1 Step 2 CK Step 2 CS Matching & Residency Step 3 Classifieds
LoginUSMLE LinksHome