BCPS Discussion

kitiria — Mon, 13 Feb 2012 13:33:38 +0000

As far As I know, you have hit it on spot. I am not a Stats expert but with respect to CI and odds ratios. That is correct. I double checked in the study guide for biostat (the 2011 edition). It does state that CI for OR/RR with a 1 in it is insignificant. All other CI with a 0 in it's range is insignificant.

Original post blogged on b2evolution.

cutant — Mon, 13 Feb 2012 09:23:24 +0000

The reasoning I am using to know when significance counts for CI=0 or CI=1 is the following:

In a prospective trial (i.e. comparing 2 means or difference between 2 estimates e.g. cure rates as %) a CI that includes 0 means no difference between variables can be interpreted as statistically significant.

For observational studies or epidemiologic trials (e.g. cohorts or case controls ) i.e. trials that will include odds ratios or relative risks (e.g. reduced MI) a value of 1 for RR means no dofference in risk and when CI includes 1 there is statistical significance.

In short this means for OR or RR, CI=1=insignificant

In absence of OR / RR, CI=0=insignificant

So far this is working for me - I would appreciate if someone check it out.

Original post blogged on b2evolution.

kickbcps2012 — Tue, 31 Jan 2012 00:08:03 +0000

CI can be used for difference or RR/OR. For CI of difference, if 0 is included, no difference; For CI of RR/OR, when 1 is included, no difference.

Please let me know if you need more clarification. my email is xdf2008@gmail.com.

Original post blogged on b2evolution.

fad2333 — Thu, 29 Sep 2011 21:24:29 +0000

nice to join and get prepared for BCPS exam next year.

thanks

Original post blogged on b2evolution.

bdial — Fri, 23 Sep 2011 17:05:41 +0000

I have one specific stats question that is driving me crazy. I have searched and searched, unsuccessfully, for a clear-cut answer on this one. It relates to confidence intervals. I think that I am correct in stating that an inclusion of “1” within the confidence intervals of cohort (relative risk) and case-control (odds ratio) studies renders any such study statistically not significant. However, I have read in certain situations/study types (and seen in certain study question answers) when looking at confidence intervals that “0” within the confidence interval (not “1”) renders the study findings statistically not significant. When is this the case AND how does one differentiate these cases from OR and/or RR (when “1” renders the findings stat. not significant)? I would appreciate ANY advice that you can give on this one.

Original post blogged on b2evolution.

sommejm — Wed, 14 Sep 2011 13:11:59 +0000

There may be more answers discussed within the BioStats chapter, but they were hard to find bc they are embedded in the notes. If you post the page numbers, I can try to find more. Hope this helps.

pg 149 Descriptive Stats Example

Sample Data Set Mean = 60.8 Median = 61 Mode = 65

Std Deviation = no answer given, will not be expected to calculate

Range can be reported as either 87 - 46 or 41

pg 154-155 Rosuvastatin vs Simvastatin

Which is the most appropriate stat test to determine baseline differences in

A) Sex: Chi Squared

B) LDL: Independent t test aka 2 sample

C) % smoker vs. non-smoker: Chi Squared

Which is the most appropriate test to determine:

A) Effect of rosuvastatin on LDL: Wilcoxen Signed Rank Sum

Very confusing to the group. The key here (as explained) is that the samples are Paired. The measurements in the rosuvastatin group at baseline and compared against the same patients again at the 3 month mark.

B) The primary endpoint: 2 Sample t test

pg 158 Regression example

8c. i. Parametric, continuous values

ii. 31% of variability in Factor Xa conc. is explained by the Enox dose

iii. For 2mg ~ 0.5 U/mL; Cannot predict 3.75 because it is outside range of the experiment

Original post blogged on b2evolution.

sommejm — Wed, 14 Sep 2011 13:10:30 +0000

Original post blogged on b2evolution.

kesamai — Sun, 07 Aug 2011 15:02:16 +0000

Hi, I just wanted to know if anyone has the answers to the review questions from the ACCP review book for the Biostats refresher. I have the self-test questions in the back of the chapter, but there is no key for the questions throughout the chapter... I would love to know if I am on the right track....any suggestions or help?

Original post blogged on b2evolution.