Jump to content

Metformin lowers all cause mortality of diabetics compared to normal people!

Shwet Shyamal

Recommended Posts

This retrospective study was done on a very large number of metformin treated diabetics with pair matched non-diabetics. Diabetics had higher BMI, but were also more often on treatment for high lipid and hypertension. Unadjusted all cause mortality of diabetics was equal to controls (an achievement by itself), where as adjusted mortality was lower!


Link to full text paper:


Link to comment
Share on other sites

Yeah, not so much, unfortunately. The following critique was authored by a Professor of Applied Statistics at the Open University in the UK ("Professor" is a much more senior position in the UK than the USA, denoting something like a department head), and seems cogent and consistent with a slightly cryptic phrase in the abstract, in which case teh study is likely not worth paying much attention to:

Expert reaction to study looking at type 2 diabetes, metformin and lifespan


The title of this paper itself is not helpful in that anyone reading it might get the wrong idea – this study cannot actually answer the question it poses (“Can people with type 2 diabetes live longer than those without?”) for reasons discussed below, and it sounds almost as if there are grounds to advise people without diabetes to take metformin. But in fact the study isn’t saying that at all.

In the press release, Craig Currie says “People lose on average around eight years from their life expectancy after developing diabetes” and goes on to explain why. So if the life expectancy of people with type 2 diabetes is so much shorter, how on earth can they “live longer than people without the disease”, as the title of the release and the paper both say?

The answer is that the comparison in the paper runs only over the time period when the patients with diabetes were on first-line treatment with metformin, on its own (and there’s a similar comparison involving patients whose first-line treatment is with sulphonylureas). At some point after this first-line treatment starts, many of the patients with diabetes would be switched from metformin alone onto a second-line treatment, and this switch is (or should be) necessary because the diabetes or its effects have got worse. But at that point the comparison in this study simply stops.

So the quote in the press release about an eight year reduction in life expectancy, in people who develop type 2 diabetes, is talking about the entire rest of a person’s life after the diagnosis, including the time when they might be on a more aggressive second-line treatment. But the comparison in the paper is looking only at the time before the treatment changes. ...

[MR: This is evidently what the abstract means by using a "censored followup:" they ONLY looked at deaths occurring WHILE the person was on metformin or sulfonylureas: if your disease progressed, and they added on a secnd drug to bring your glucose back under control, you were simply "censored" out of further followup. This would obviously greatly bias the resulting mortality rates (and note: it is for sure from the abstract that they did not actually look at life expectancy, despite the press release AND the title of the abstract: explicitly , they ONLY looked at mortality rates of people while they were only on one drug (or, during the matched number of years of the nondiabetic controls). This means, by definition, that people who were put on the drug most usually used for the mildest diabetes (metformin) and who remained stable and healthy on it, were being compared to more severe patients at outset (on sulfonylureas), and as soon as they got sick they vanished from the analysis! Comparing even average nondiabetic people to unusually successful diabetics is, from the get-go, comparing elite diabetics to merely average aging people. Returning to the critique:]

The researchers did match the controls with patients with diabetes in certain ways, and in their statistical analysis they try to allow statistically for other differences between the people with diabetes and the controls. But the paper itself points out some issues. The researchers could not take into account certain possible confounders (other variables that might affect the comparison) because they did not have data on them for enough of the controls. Even without that important issue, statistical adjustment for confounders is never perfect. The difference in survival between people with diabetes on metformin, and controls without diabetes, was statistically significant but in fact rather small, and probably within the range where it could be explained by residual confounding ...

Further, the paper itself also points out that people with diabetes are more likely be monitored for, and receive interventions for, problems with the heart and circulation. This extra intervention and monitoring, and the possibility of residual confounding, between them cast huge doubt on the possibility that the better survival in the patients taking metformin, compared to controls without diabetes, was simply because they were taking metformin. ...


And, as a reminder: metformin has been tested now at high and low doses in normal, healthy mice, and also in a somewhat flawed study in rats, and in no case has metformin increased maximum lifespan; there was a very mild increase in average LS in the mouse studies, which might well be due to residual effects of a diet of lab chow and no exercise.


1. C. A. Bannister, S. E. Holden, S. Jenkins-Jones, C. Ll. Morgan, J. P. Halcox, G. Schernthaner, J. Mukherjee and C. J. Currie.
Can people with type 2 diabetes live longer than those without? A comparison of mortality in people initiated with metformin or sulphonylurea monotherapy and matched, non-diabetic controls
Diabetes Obes Metab. 2014 Nov;16(11):1165-73. doi: 10.1111/dom.12354. Epub 2014 Jul 31.
PMID: 25041462

Link to comment
Share on other sites

  • 2 weeks later...
  • 2 years later...

A further comment making the same central point has subsequently been published in PubMed Commons:


Andrea V Margulis,  Manel Pladevall,  Nuria Riera-Guardia,  John Seeger,  Elisabetta Patorno, Cristina Varas-Lorenzo

To identify the effect of metformin or sulfonylurea monotherapy, follow-up of treated patients with diabetes was censored after any modification to the initial treatment. Thus, as patients with diabetes progressed to the point they needed treatment intensification, they left the cohort. [That is: as the diabetes in patients on metformin got worse enough to require additional medication(s) — and are therefore the ones who are at greatest risk of mortality — they "disappeared" from the study without being reported to have died -MR]. In contrast, the patients without diabetes ... cannot leave the cohort until he or she dies. ...

Furthermore, the requirement of patients with diabetes to stay on glucose-lowering therapy for 180 days to be eligible ... may have created an additional opportunity to selectively retain healthier individuals ... Good adherence by itself has been shown to reduce mortality [and diabetic patients not sticking to their meds got booted out -MR] ...while comparators without diabetes are only required to survive for the 180 days)
Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...