As they say, don’t trust anybody and then verify.
Mr. Debonneuil quotes (albeit disbelievingly) Mr. Rae who (very confidently) quotes a meta-analysis by Dr Bartolucci et al, (PMID 21481826) who read a bunch of articles and came to the conclusion that all-cause mortality effects of low-dose aspirin are null.
Yours truly is non-plussed, having heard an aspirin a day keeps the doctor away.
A bit later however, Dr Bartolucci publishes an erratum:
“The conclusion concerning “All cause mortality” in the paper was incorrect. There is a statistically significant advantage for aspirin in reducing the risk. In Table 3 the odds ratio should be 0.934 with 95% Confidence Interval (0.874, 0.999) and p-value for treatment favoring aspirin, p = 0.0.045. This information also applies to Figure 3 with z-value = −2.004. The p-value for Heterogeneity in Table 3 is 0.989. The reported result was due to incorrect data input for the AAT and POPADAD studies, In Figure 3 the correct odds ratio for AAT is 0.940 with lower 95% confidence limit = 0.756 and upper limit = 1.169, z-value = −0.556 and p-value = 0.578. Likewise for the POPADAD study the correct odds ratio is 0.919 with lower limit = 0.677 and upper limit = 1.247, z-value = −0.545 and p-value = 0.586.”Gee-whiz, who could have guessed? You write an article whose main claim is aspirin does not save net lives, you even get it published, and, oops, you are forced to eat your words and say the opposite in an erratum. To make it even more painful, a Dr Sutcliffe (PMID 24074752 ), in another similar meta-analysis, shows that, yes, aspirin saves net lives in primary prevention.
It is hard to escape the impression that aspirin is good for you.
RafalOn Tue, Aug 12, 2014 at 2:07 AM, Edouard Debonneuil wrote:>