'p' value significance
interesting editorial in JNCI : http://jnci.oxfordjournals.org/content/103/1/2.full
members' comments
'p' value significance
Forum
» Evidence and Technology / Clinical and Biological Evidence
» 'p' value significance
It's a good thought process.
overtly relying on the p Value is fraught with dangers. For example, someone who is hard pressed for
time may just skim over the abstracts including the tendency of the journals to publish only "positive
studies". The prime example that comes to my mind is the Canadian Trial for Carcinoma Cervix which was
inconclusive. Although the p value was not significant, it has been assumed that chemoradiation is not "indicated"
for stage III/IV patients and I have seen juvenile debates on this issue.
Admittedly systematic flaws in trials/patient recruitment hampered the outcome of above mentioned trial but mercifully later in meta-analysis chemoradiation became the norm.
I congratulate you to bring this to our notice Ayan. Great write up and hopefully should spur on interesting
debate in the community about how we report our results and how we should adapt.
Reality is merely a persistent illusion
Ayan
Thanks for the link to the thought provoking article. It will take a long time before we move on to Bayesian stats and learn to understand it. But the point is well made.
For starters we need to stop judging a positive study based on the p-value alone, and the judge the clinical significance of the outcomes using some standard method.
Abhishek
Why are u mad at the Canadian trial? Its simply one trial that did not have a positive outcome. Its only normal when people question a new approach and its relevance to the Indian population. I wonder what the Cracx trial from TMH will show, we will know the results soon hopefully. Does anyone have any unofficial knowledge?
I am not mad at it but just wanted to highlight the way patient recruitment was done (too many early stage patients though).
Yes, even I am looking forward for the TMH outcome. Lets see how it works out :)
Reality is merely a persistent illusion
I presume Abhishek is discussing about the Pearcey et al (NCIC) trial that was hotly debated after the NCI clinical alert .
Here is an excellent edit discussing the issue : http://jco.ascopubs.org/content/20/4/891.full.pdf
Please continue the discussion on statistical significance , i enjoyed your comments —but i must confess this incidental aside into cancer cervix is even better !! let us discuss more on that too —
Ayan
Thanks for the editorial link. Two interesting points:
1) They mention that if 6 RCTs were conducted on the same subject, there is a 74% chance that one of them would be negative anyway.
2) They also mention that one of the reasons why the Pearcey trial is negative is that they did not surgically stage patients, while 4 or the 5 positive trials did. The 5th trial ony had stage IB2 patients. Interesting because we don't surgically stage patients and therefore are as likely to miss para-aortic disease as Pearcey was. Our patient population therefore is more like Pearcey's than the other trials. He also gave the same chemo as we give. Therefore, are we actually benefiting patients with our current treatment protocols?
In my mind the stage III question is still open. The standard is chemRT no doubt, but there is no harm in some introspection.
Yes exactly! Though nothing is perfect :) Any more opinions on this?
Reality is merely a persistent illusion
while on staging , i wonder how many of you have heard Perry Grigsby cry himself hoarse at innumerable meetings and also write extensively regarding staging that PET scan should be incorporated into the staging protocol for cervix cancer primarily because of the possibility of under-estimating nodal disease which FIGO tends to under-emphasize anyways :
http://www.siteman.wustl.edu/ContentPage.aspx?id=581
—reading the Edit i wondered if he has been right all along . What do you feel guys ?
Hi ayan,
FIGO doesnot stop you from incorporating PETCT into the staging workup. What it doesnot allow is upstaging based on that. That is really important as it would mean two things instantly:
- If we make that a mandatory part of the staging it would mean that almost 90% of patients with Ca Cervix will start getting treated without appropiate staging workup
- If we dont make it mandatory but upstage nonetheless the outcome of stage II patients chiplun would treat would proabably be very different from that in Grigsby's centre
On an aside note here where a CT is done routinely in all patients in HN Ca an amazing number become stage IVA even after applying the size criteria etc. No wonder these guys have better outcomes than we do - I had seen this in my thesis when I had done CT for all patients of HN too. Before I left PGI CT was being done all patients with Ca Cervix too and treatment plan was modified dependant on findings. FIGO does allow you to encompass the known and suspected disease and treatment modality is not recommended by them.
Another thing would PET stand upto USFPIO MRI in future?
FIGO is a surgeon dominated body; do look at their recommendations that came in last year. They insist on a surgical staging (!!) rather than a non invasive work up which of course I find it odd. But then there can be no easy answers. FIGO also insists that all the while PET can be used but they disregard overwhelming evidence in favor of PET-CT. Do check out the references; I think it was Grigsby's article only that they have quoted. (I don't have their paper with updated staging recommendations at present).
Reality is merely a persistent illusion
Nice arguments put forward by all of you.
Re CTRT in advanced ca cervix, updated data presented at ASTRO 2010 http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6T7X-514GTCR-C6&_user=10&_coverDate=11/01/2010&_alid=1609124670&_rdoc=1&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5070&_sort=r&_st=13&_docanchor=&view=c&_ct=1&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=eb86143a8d957ac2132ec777009021a8&searchtype=a.
PET scan: Canada is conducting a Phase III trial in this area. Hopefully we will get some more answers soon.
Alberta actively does per and post treatment PET for Advanced cervical cancer patients and we did change our treatment decisions based on PET results. Its is hard to ignore the findings on PET.
FIGO is doing a good job by giving us a staging system which is applicable to majority of the centers all across the globe. Yes it is dominated by Gyn-oncologist becoz we never attend these meetings. IGCS meeting is full of gyn-onc's and very few RO's. In-fact we should be attending meetings where there is a mix of all Oncology streams not just RO.
"p" yes many of us just read abstracts and fail to understand the clinical significance of the data. I personally have a lot of respect towards people who honestly publish negative results, rest just try to find out some significance in there data and publish which has no clinical relevance.
Dr Nikhilesh , many thanks for the updates
Very interesting issue raised by Santam -while Alexandra Taylor et al have worked and published extensively on USPIO for lymph node status and even incorporated the same in the consensus guidelines for pelvic lymph node delineation , Grigsby et al have done the same for PET.
Loads of literature , samples below:
http://www.ncbi.nlm.nih.gov/pubmed/18771811
http://www.ncbi.nlm.nih.gov/pubmed/19792965
So , what does the house opine? Any head on comparisons between the two ?
I think it deserves a separate discussion thread for PET/CT in cervical malignancies. However, while it is difficult to compare the newer 'nano particle detection' with PET-CT, I found a write up something similar to what is on here:
http://www.scribd.com/full/46996310?access_key=key-1vz285e1aqipo48homh2
Which of course feel free to download and share.
PET/CT is inherently more sensitive and accurate than MRI as far as the nodal detection is concerned and there should be a strong case to include in routine work ups (whenever possible). There is NO rationale not to give the benefit of a modality that upstages a patient who would otherwise benefit from intensive therapy.
Reality is merely a persistent illusion
agreed —new thread @ http://www.isocentre.org/forum/t-299357/pet-vs-uspio-mri-for-lymph-node-staging-in-cancer-cervix
abhishek , this paper on scribd compares MRI (but not USPIO) Vs PET , do you have any with head on comparisons of the two?please discuss in the new thread
/forum/t-298852/p-value-significance#post-