Home
> Academic OB/GYN Podcast, Green Journal, Grey Journal, Journal Articles > Academic OB/GYN Podcast Episode 18 – Journals for April 2010
Academic OB/GYN Podcast Episode 18 – Journals for April 2010
We discuss articles from the April 2010 Journals: Ovarian mass malignancy prediction, an new analysis of the labor curve, the ACOG Practice Bulletin on Thrombophilias and its sudden withdrawal, and a few other bits.
Categories: Academic OB/GYN Podcast, Green Journal, Grey Journal, Journal Articles
Hi Nicholas
First of all thanks for all the hard work, I look forward to the podcast and even though I am a radiologist, the podcast helps me do a better job.
I wanted to comment on the Ovarian mass article. Quite resently I have become aware of the statistical value of Likelihood Ratio values in predicting risk based on multiple factors. In general the process begins with baseline risk assessment using demographics and familiy history. The baseline percentage risk will often be calculated using validated tools such as Gail model and BRCAPRO for breast tumor, Framingham score for coronary disease, or FRAX score for fracture risk in osteoporosis. One can then simply multiple the baseline risk by the likelihood ratios calculated for subsequent testing. For ovarian CA this might include the ultrasound characteristics and CA-125. The product represents a valid approxmation of absolute risk.
Now here is the trick, although most articles do not calculate likelihood ratios (and their 95% confidence intervals), you can calulate the LR youself whenever you have the 2X2 table, or the sensativity/specificity values using the following simple formula
Formulas
Positive Likelihood Ratio= (a/(a+c)) / (b/(b+d))
Negative Likelihood Ratio= (c/(a+c)) / (d/(b+d))
– or –
Positive Likelihood Ratio= (sensitivity)/(1-specificity) = TP/FP
Negative Likelihood Ratio= (1-sensitivity)/(specificity) = FN/TN
——————————————————————————–
You can also use a host of convient web based calculators. Once you have the LRs for a really good article you can use them to estimate actual risk as a guide for you and your patient. In addition, you can campare LR for different studies of the same process, if the LRs are comparable from different groups they are probably useful, if they are widely different, maybe not so reliable. Anyway the calculation of OR has helped me interpret findings in everything from heart failure to Down’s syndrome. Recommended highly.
Thanks again
John Terry MD
** Editor’s Note – I edited this note to fix some typos and confusion between Odds Ratio and Likelihood ratio **
LikeLike
Thanks for the great comment. I completely agree with what you are saying. Likelihood ratios would be a good topic for a post, as they are very useful in interpreting the results of sensitivity and specificity values.
LikeLike
Thank you for all your work on the latest podcast. I really appreciated the coverage of the labor curve article. It seems like all the controversy over the Friedman curve would never actually lead to change in practice until there was a competing curve that health care practitioners could use instead. Hopefully this article is, at the very least, a start.
(For demographics tracking, I’m a graduate nursing student with an eye toward pursuing midwifery.)
LikeLike