The anti-diabetic drug phenformin was banned in many countries because of the high risk of lactic acidosis (LA), mostly arising in elderly with predisposing conditions [1]. To our knowledge, phenformin is still legal in Italy, Brazil, Uruguay, China, Poland, Greece and Portugal, and cases of LA are continuously reported worldwide [2–5]. Phenformin-induced LA is an unpredictable complication that may occur even in people without predisposing diseases, causing death in 50% of cases [1]. In addition, LA is often misdiagnosed, and many cases may go unrecognized, particularly in older persons affected by various chronically symptomatic illnesses. As recently emphasized, the reported cases of phenformin-induced LA may thus represent only ‘‘the tip of an iceberg’’ [3]. Phenformin can always be replaced by metformin, a safe and useful drug. A recent meta-analysis showed that metformin does not cause LA more frequently than diabetes per se [6]. Age is an absolute contraindication to phenformin, because older patients have increased prevalence of medical conditions favouring LA [1]. These persons, however, may just represent the subgroup with the highest prescription rate, since they are less likely to attend diabetes outpatients clinics where phenformin prescription is avoided [7]. We aimed to test this hypothesis in 1,273 elderly diabetic patients from the Gruppo Italiano di Farmacovigilanza nell’ Anziano Study, consecutively hospitalized in 1997–1998 in 81 participating wards of Internal or Geriatric Medicine located throughout Italy [8]. Patients were grouped according to whether or not they were taking phenformin before hospitalization. Comparison of users versus non users of phenformin before admission was made by ANOVA one way for continuous variables and chi-square test for categorical variables. Variables significantly distinguishing groups in the univariate analysis were entered in a multivariable logistic regression model to identify independent correlates of phenformin use before hospitalization (dependent variable). Odds ratios (OR) and 95% confidence intervals (CI) were calculated. All statistical analyses were performed using SPSS software (version 10.0; SPSS Inc. Chicago, IL). Phenformin was used by 151 patients (11.9%). Mean age was significantly higher in users compared to non users (74.9 ± 10.3 vs. 72.4 ± 11.7, P = 0.015). The clinical characteristics of users and non-users, and the results of univariate analysis are described in Table 1. The multivariable logistic regression analysis demonstrated that age C80 years (OR 1.90, 95% CI 1.09–3.31) and a combined use of insulin and oral hypoglycaemic agents (OR 3.05, 1.56–5.96) were positive correlates of phenformin use, whereas dementia (OR 0.20, 0.06–0.66), decompensated F. L. Fimognari (&) R. Pastorelli Unita Operativa di Medicina Interna, ASL Roma G, Ospedale Leopoldo Parodi-Delfino, Colleferro (Rome), Via F. Verdinois 30, 00159 Rome, Italy e-mail: filippo.fimognari@virgilio.it