eoph12@castle.ed.ac.uk (I F Gow) (09/18/90)
The study was described in last week's Lancet. Two points which I personally would be unhappy about are: 1. The Bristol group and the control group were not well age-matched 2. If the length of survival is taken from the date of diagnosis of breast cancer in *both* groups, then there is no significant difference between the groups. The significant difference between the groups is apparent only if survival in the control group is taken from date of diagnosis, and the Bristol group taken from date of entry into the Bristol study. According to the article, entry into the Bristol system could take place either months or several years after the date of diagnosis. I was surprised that this point did not seem to be discussed at all, since it meant that depending on how you do the statistics, there was either a 3x increase in the risk of mortality, or no difference. However, I do not pretend that I can follow the statistical prodecures thoroughly, and it may be that their recjection of one type of test is statistically valid. Iain Newsgroups: sci.med Subject: Re: Question about Insulin and Growth Hormone Summary: Expires: References: <1990Sep13.145836.9241@cadre.dsl.pitt.edu> <4013@ursa-major.SPDCC.COM> <90257.091159WFP101@psuvm.psu.edu> Sender: Followup-To: Distribution: Organization: Edinburgh University Computing Service Keywords: Newsgroups: sci.med Subject: Re: Question about Insulin and Growth Hormone Summary: Expires: References: <1990Sep13.145836.9241@cadre.dsl.pitt.edu> <4013@ursa-major.SPDCC.COM> <90257.091159WFP101@psuvm.psu.edu> Sender: Followup-To: Distribution: Organization: Edinburgh University Computing Service Keywords: