The logrank test
BMJ 2004; 328 doi: https://doi.org/10.1136/bmj.328.7447.1073 (Published 29 April 2004) Cite this as: BMJ 2004;328:1073
All rapid responses
Dear Sir,
In the article by Bland and Altman ( BMJ 2004;328:1073 ) it might have
been more helpful if the authors had expressed the research question in
terms of whether there is a difference between two groups of patients,
rather than whether one group did better ( or worse ) than another. This
would have emphasised the point that a two sided test is appropriate in
most situations in medicine. The fact that a two sided test has been used
might even have been indicated when giving the resulting P value. The
usual tables and computer programs used to calculate P values for the chi
square test with one degree of freedom give P values corresponding to a
two sided test.
Competing interests:
None declared
Competing interests: No competing interests
see Professor Bland's correction :-
' better ' not ' worse ' !
Competing interests:
None declared
Competing interests: No competing interests
Dear sir, madam,
I followed with great interest Boland & Altman article's on Logrank
test. I was a bit confused reading the sentence: "Is this sufficient to
conclude that in the population patients with anaplastic astocytoma have
worse survival than patients with giloblastoma?" as it doesn't make sense
with the survival curve shown. Or am I missing something?
Kind regards
Yours faithfully
Habab Omer
Medical Student,
Manchester
Competing interests:
None declared
Competing interests: No competing interests
I thank Helen Seaman for pointing out that there is a typographical
eror in the second sentence, which should end "is this sufficient to
conclude that in the population patients with anaplastic astrocytoma have
better survival than patients with glioblastoma?", rather than "worse
survival", as at present.
As my coauthor will confirm, I am not a good proof reader.
Apologies.
Competing interests:
None declared
Competing interests: No competing interests
Re: Two sided test use should be emphasised in medicine
I entirely agree with Alan Gibbs that the appropriate analysis in the
example was a two-sided test. We did not intend to imply otherwise. We
made the same point in an earlier Statistics Note: One- and two-sided
tests of significance [1].
[1] Bland JM, Altman DG. One- and two-sided tests of significance.
BMJ 1994; 309: 248.
Competing interests:
None declared
Competing interests: No competing interests