In Reply In their letter regarding our JAMA Insights Clinical Update article,¹ Dr Caturano and colleagues point out several additional considerations in prescribing metformin. These include comparable or superior glucose-lowering efficacy compared with most alternatives, as well as evidence for safety and efficacy in pediatric and older patients with diabetes.

We agree with most of these points. However, we urge caution in interpreting observational evidence, such as the article they cite reporting broad-ranging health benefits from metformin in older patients. Observational studies have played a crucial role in assessing metformin’s safety, helping to show that no measurable increase in rates of lactic acidosis accompanies metformin use.² However, the benefits associated with metformin use in many observational studies on outcomes ranging from cancer to dementia are susceptible to bias, including immortal time bias and confounding by indication (in which metformin users at baseline tend to be intrinsically healthier than most comparison groups).³ Observational studies should generally be viewed as hypothesis generating and factored only cautiously into clinical recommendations.⁴ To this point, we agree with Caturano and colleagues that there remains a need for more rigorous research, particularly randomized head-to-head trials comparing metformin (and sulfonylureas) to the newer alternatives.