Prep for USMLE 
Forum  Resources  New Posts  Register  Login  » 

Author  5 Posts  
mn2_200
Forum Newbie Topics: 25 Posts: 234 
A casecontrol study is conducted to assess the risk for intussusception in infants under the age of 1 year who receive the rotavirus vaccine. The medical records of all those who received the vaccine and those who did not receive the vaccine over a 6month period are reviewed. Results show 125 cases per 100,000 infantyears for infants who received the vaccine compared to 45 cases per 100,000 infantyears for infants who did not receive the vaccine. The investigators conclude that the relative risk for intussusception is 1.9 times greater in infants who receive the rotavirus vaccine (95% confidence interval of 0.5–7.7 and p=0.39). Which of the following is the most accurate interpretation of these results? A ) The results do not show an association between rotavirus vaccine and intussusception, but they may be related B ) The results show sufficient statistical power to identify an association between rotavirus vaccine and intussusception C ) Rotavirus vaccine is associated with a 39% risk for intussusception D ) Rotavirus vaccine causes intussusception in 1.9% of infants E ) Rotavirus vaccine prevents 80 cases of intussusception per 100,000 infantyears  
Prep4USMLE.com
Advertisement 


xenopus
Forum Junior Topics: 15 Posts: 646 
I'll go with A) The results do not show an association between rotavirus vaccine and intussusception, but they may be related. Reading the stem indicates that there's an excess of 80 cases of intussusception in the vaccinated kids using the number of people in the denominator. That might clue you into blaming the Rotavirus vaccine for the excess of cases. But this is a casecontrol study which usually is retrospective, so another variable might still be responsible for that association (the study hasn't been randomized). But the strength of that possible association needs statistical confirmation, and that comes from the RR (Relative Risk) of 1.9. Notice that its Confidence interval at 95% ranges all the way from 0.5 to 7.7. In other words, it goes from a protective range (0.5; rotavirus protects from intussusception) to harming range (7.7; rotavirus goes with intussusception). If the RR is 1.0 means no risk. If <1.0 is protective; if >1.0 harming but in order to be one of them its corresponding CI95% has to remain either in the protective range (say 0.70.9) or in the harming range (say 3.37.7). When the RR spans all the way from <1.0 to >1.0 (protective to harming) it suggest that crosses 1.0 or no risk. This is further strenghtened by the statistical significance given here, which is larger than 5% (p>0.05). In this case, the p value was 0.39 or 39% which means that RANDOMNESS explains the 39% of the differences between the number of cases of Intussusception vs noIntussusception. Too high level of randomness!!!. Maximum tolerable is 5% (sometimes less if the sample is too huge, in that case Bonferroni correction needs to be applied). So what's the Confidence Interval at 95%? In this case they're talking about the CI for RR. It means that if you repeat the experiment 100 times; in 95 of those times the RR would range from 0.5 up to 7.7. In other words, from protective passing through 'no risk =1.0" and going up to a lot of risk (7.7). If it's going to protect or is going to harm it has to stay on only one side. You can not be good or bad at the same time. Hopefully I got it right  
mn2_200
Forum Newbie Topics: 25 Posts: 234 
good explanation but can u explain why b is wrong? i remember if u increase the sample size u increase the power ?  
xenopus
Forum Junior Topics: 15 Posts: 646 
mn2_200 wrote: good explanation but can u explain why b is wrong? i remember if u increase the sample size u increase the power ? The power of the study is OK but the study's design is flawed. It's a case control study and there's so much you can conclude from that. A better study would be a prospective cohort study and the best of course would be a clinical trial randomized and placebo controled, double blind.  
Misrati
Forum Junior Topics: 43 Posts: 610 
Agree with Xeno (A)! im not very good in stat but I remember those few things; if CI contains #1 means NO statistical difference which makes a contradiction with the study results.  
This thread is closed, so you cannot post a reply.
Similar forum topics case control study about casecontrol study Case control/Cohort?  Related resources Kaplan USMLE Step 1 Home Study Program Case Files Peds Premier Review Home Study Kit 
Advertise  Support  Premium  Contact 