Wang · Journal of diabetes and its complications 2017 · Retrospective cohort study · n=41,204

Differential effects of metformin on age related comorbidities in older men with type 2 diabetes.

Cited 138 times in the scientific literature.

Level 3 - non-randomized controlled study

Retrospective non-randomized cohort study

PubMed 28190681 · doi:10.1016/j.jdiacomp.2017.01.013 · record verified 2026-08-29

What was done

A retrospective cohort of 41,204 US male veterans aged ≥65 years with type 2 diabetes and free of age-related comorbidities (cardiovascular disease, cancer, depression, dementia, and frailty-related diseases) during 2002–2003 was followed from 2004 to 2012. Latent class modeling adjusted for confounders was used to evaluate the association between metformin use and incidence of age-related comorbidities.

What was found

The cohort had a mean age of 74.6 ± 5.8 years, baseline HbA1c 6.5 ± 0.97%, and 8,393 (20.4%) metformin users. Four comorbidity classes were identified: Healthy Class (53.6% of cohort), where metformin reduced likelihoods of all comorbidities from 0.14% in dementia to 6.1% in cardiovascular disease; High Cancer Risk Class (11.6%), with likelihood reductions of 13.3% for cardiovascular disease, 45.5% for cancer, 5.0% for depression, and 13.7% for frailty-related diseases; High CVD Risk Class (17.4%), with likelihood reductions of 48.6% for cardiovascular disease, 3.2% for cancer, 2.8% for depression, and 6.3% for frailty-related diseases; and High Frailty Risk Class (17.2%), with likelihood reductions of 18.8% for cardiovascular disease, 3.9% for cancer, 3.8% for dementia, 15.6% for depression, and 23.8% for frailty-related diseases.

Why it matters

It demonstrates that the potential disease-slowing associations of metformin differ substantially across distinct comorbidity risk profiles in older diabetic patients.

Limits

The cohort was restricted to older male veterans, limiting generalizability to females and younger populations. As an observational study, it is subject to residual confounding and indication bias. The abstract does not report confidence intervals, p-values, or the mortality outcomes mentioned in the objective.

Cited by