crabsallover highlights, key points, comments / links.
Figure 1: In eight eligible trials (25570 patients, 674 cancer deaths): 327 deaths due to cancer occurred with patients treated with aspirin out of 14035 in the 8 trials. 347 deaths due to cancer occurred with control patients out of 11535 in the 8 trials.
With Aspirin: 327/14035 = 2.33% died of cancer
With Control: 347/11535 = 3.01% died of cancer
Over about 6 years 0.68% fewer died of cancer because they took a daily Aspirin.
Reclassification of cause of death in a small number of cases in UK-TIA and TPT had little effect on the pooled estimate (686 (684 in figures below) deaths; OR 0·80, 0·69—0·93, p=0·004, webappendix p 3).
Web-Figure 1: In eight eligible trials 335 deaths due to cancer occurred with patients treated with aspirin out of 14035 in the 8 trials. 351 deaths due to cancer occurred with control patients out of 11535 in the 8 trials.
With Aspirin: 333/14035 = 2.37% died of cancer
With Control: 351/11535 = 3.04% died of cancer
So over about 6 years 0.67% fewer died of cancer because they took a daily Aspirin. The number needed to treat with Aspirin over 6 years to prevent one death is 149 (100%/0.67%).
A relative 22% reduction of death with Aspirin ((3.04-2.37)/3.04)
In 3 of the trials (TPT, UK-TIA, BDAT) 1634 (in the 3 trials 740, 287, 607) died of cancer over 20 years follow-up out of 11845 (2545/2540, 806/815, 3429/1710)
13.79% died of cancer in 3 trials over 20 years (1634 died out of 11845 trialists). Assuming a 22% relative reduction of 13.79%, 10.75% taking aspirin would die over 20 years, a difference of 3.04%
Using Spinning the Disk at UnderstandingCertainty.org http://understandinguncertainty.org/spinning input deaths.
or more optimistically
The Number needed to treat: 33 (100%/3.04%)
Aspirin reduces cancer but it also reduces risk of heart attacks and strokes. However Aspirin increases risk of death from fatal bleeds (brain haemorrhages, fatal stomach ulcers etc).
The all-cause mortality, which includes fatal bleeds is stated by Peter Rothwell:-
'The reduction in cancer deaths on aspirin during the trials resulted in lowered in-trial all-cause mortality (10·2% vs 11·1%, OR 0·92, 0·85—1·00, p=0·047, webappendix p 4)'
Peter Rothwell on 14th personally communicated with crabsallover when I asked'
... Can you let me know how you arrived at the 10% all-cause mortality figure when taking aspirin daily?
Peter Rothwell said
The estimate was based on the observed effect in the 8 RCTs. The OR for all cause mortality in the trials was 0.92 (see paper). Excluding patients who were randomised but stopped treatment before the first follow-up it was 0.88 in those trials with the data available. So "about 10%" seemed reasonable. However, we are in the process of doing a formal health economics paper, which I will send you in due course. The estimate of effect of all-cause mortality that we published (OR=0.92) could well be an underestimate because after exclusion of patients who were non-compliant (so-called "on-treatment analysis") the effect was larger (OR=0.88) (unpublished data).
All Cause Mortality:
1432/14035 (aspirin) =10.2% v 1275 /11535 =11.1% (control)
|Aspirin reduces your chance of experiencing All-Cause mortality by 8%|
|All-cause mortality is reduced with Aspirin|