Many of us are fixated on the p-value pertaining to a change from baseline following exposure to an intervention. As long as the p-value is significant (
it is indeed good questions i also have the same doubts and hope for better understand of this issue..
Large studies are good for populations and statistical methods are inevitable consequences to demonstrate whether the observations are chances or real. P-value is one common approach but other proponents suggest the Bayesian approach as an alternative for the reasons you have given.
However at clinical level, individual responses are significant, in spite of magnitude, inasmuch as that response removes or keeps one within a range of risk( for an event). For example BP response may remove one from likelihood of AMI or haemorrhagic stroke but not ischemic stroke.
But patients at highest risk of an event benefit from the smallest of alteration to their risk factor, if there are no other overriding factors. And those very small changes of individual risk factors are multiplied if such changes do occur in several risk factors e.g. little changes in BP, LDL, and HbA1c multiply benefits more than pushing a big change in only one.
Its difficult to address scientifically to this question. If a question is about individual patients then it a subjective matter and their own biology to decide the acceptable change as an outcome in terms of values if we are talking. But in large scale we have to dependent on p value as they are scientific measure of change and hence we can not ignore their importance. But in my view the study with different racial distribution should be encouraged and data should be presented in same way and not single strategy is applicable for all. Yes for me basic biology and subjective response is a matter of concern.............and should be focused.
If one is going to treat any condition, one treats that condition to the point of stabilizing or regressing that disease. P-values are meaningless here. Failure to achieve treatment goals simply forstalls the inevitable. To see how it's done, kindly go to my website at www.bowlinggreenstudy.org, which is free and open to all.
Thank you William. I actually visited the website last week. What a useful site it is. Thank you!
I totally agree eith your perspective that in the context I mentioned, P-values are useless. In fact they can be completely misleading, especially when taken from latge studies.
Thank you for your kind response. I should be doing poster exhibitions at the NLA symposium in Miami and EAS Maastricht next May. If you ar attending either symposium, kindly stop by the poster and we can chat.
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis