Whether metformin use might affect the risk of non-Hodgkin lymphoma (NHL) remained to be answered.

A total of 610,089 newly diagnosed type 2 diabetes patients with 2 or more times of prescription of antidiabetic drugs during 1999–2009 were enrolled from Taiwan's National Health Insurance database. They were followed up for NHL incidence until December 31, 2011. Both intention-to-treat and per-protocol analyses were conducted. Cox regression incorporated with the inverse probability of treatment-weighting using propensity scores was used to estimate hazard ratios.

There were 414,783 metformin initiators and 195,306 non-metformin initiators within the initial 12-month of prescriptions of antidiabetic drugs. After a median follow-up of 5.07 years in metformin initiators and 6.78 years in non-metformin initiators, 1076 and 755 patients were diagnosed of new-onset NHL, respectively. The respective incidence was 47.74 and 57.68 per 100,000 person–years and the hazard ratio for metformin initiators versus non-metformin initiators was 0.849 (95% confidence interval 0.773–0.932) in the intention-to-treat analysis. In the per-protocol analysis, the hazard ratio was 0.706 (95% confidence interval 0.616–0.808). Sensitivity analyses after excluding patients with irregular follow-up, with an extension of minimal observation periods of 24 or 36 months, with incretin-based therapies, or in patients enrolled during 2 different periods (i.e., 1999–2003 and 2004–2009) consistently showed a lower risk among metformin initiators in both the intention-to-treat and the per-protocol analyses.

Metformin use is associated with a lower risk of NHL compared with non-metformin antidiabetics.