Metformin as a Cause of Late-Onset Chronic Diarrhea

Melissa T. Foss, PharmD, and Kathi D. Clement, MD


Pharmacotherapy. 2001;21(11) 

In This Article


Diarrhea is not uncommon in the diabetic patient. Estimates for the prevalence of diarrhea in patients with diabetes range from 3.7-22%.[5,6,7] These estimates are derived from referral centers and likely are affected by selection bias, the definition of diarrhea used, and the patient population studied.

A subtype of diarrhea in patients with diabetes is called diabetic diarrhea. Diabetic diarrhea generally affects patients with a long history of type 1 diabetes and insulin treatment (> 8 yrs). Peripheral and autonomic neuropathies are common. Stools are described as voluminous, brown, and watery and may be characterized by steatorrhea. Nocturnal bouts are not uncommon and may be associated with bowel incontinence. Other causes of diarrhea in the patient with diabetes include drugs, damaged a2 receptors in small-bowel enterocytes, anorectal dysfunction, bacterial overgrowth in the small bowel, bile acid catharsis, pancreatic insufficiency, internal anal sphincter dysfunction, celiac disease,[5,6] overuse of sugar-free foods containing nonabsorbable sugar substitutes, and lactose intolerance.

Although diarrhea is a common adverse effect of metformin, only two articles[5,7] describe late-onset diarrhea. A MEDLINE search using the following key words was performed: diarrhea, metformin, late-onset, atorvastatin, cerivastatin, drug interactions, and diabetic diarrhea.

In one study,[5] the authors administered a bowel habit questionnaire to 861 patients with type 1 or type 2 diabetes and performed a comprehensive work-up to define the cause in those reporting chronic diarrhea (> 3 mo). In 1.3% of this sample, the authors diagnosed diabetic diarrhea, and in another 2.4% the diagnosis was nondiabetic diarrhea. Diabetic diarrhea was very rare in patients with type 2 diabetes, occurring in 0.4%. Of those patients with nondiabetic diarrhea, the most common cause was treatment with metformin (8 of 21 patients). These authors reported that six of these eight patients experienced diarrhea that appeared long after the start of treatment (in some cases several years). The other two patients experienced diarrhea after an increase in dosage. Two of the patients with metformin-related diarrhea reported fecal incontinence. All eight patients experienced an abrupt cessation of diarrhea when metformin was discontinued. Four patients were rechallenged: in three, diarrhea returned within a few days, and the fourth patient's diarrhea returned after 2 months. These authors describe the diarrhea in these patients as having similar features to that of diabetic diarrhea: watery stools, often explosive, causing fecal incontinence. They recommend that the diagnosis of autonomic diarrhea not be made in any patient receiving metformin until a trial of metformin discontinuation has been conducted.

A second survey of diarrhea in patients with diabetes was reported.[7] These authors administered a questionnaire to 285 patients at a diabetic clinic. They found that diarrhea was reported by 20% of patients taking metformin as monotherapy and by 20% of patients taking metformin in combination with sulfonylureas. This was statistically significantly greater than the percentage of patients reporting diarrhea in the group taking insulin, the group taking sulfonylureas as monotherapy, or the group who were being treated with diet alone (6% in each group). The diarrhea was characterized as watery and often explosive in 9 of 10 patients receiving metformin monotherapy, and causing soiling of clothes in 12 of the 20 patients receiving metformin monotherapy or combination therapy. Most patients in this series had not reported the diarrhea to their physicians. Twelve patients were advised to discontinue metformin, and all reported resolution of diarrhea within 2-5 days. No dose relationship was observed. These authors concluded that metformin was by far the most common cause of diarrhea in their clinic, with diabetic diarrhea being relatively rare. The length of time these patients had been taking metformin and whether the dosage had been changed were not addressed in this report.

Changes in lipid-lowering drugs initially were suspected as the cause of our patient's diarrhea. Owing to the age at which type 2 diabetes commonly manifests itself and its frequent association with other disease states, it is common for these patients to be taking a variety of drugs. Drug interactions and diarrhea caused by other drugs must be considered potential causes.

The mechanism by which metformin causes diarrhea is unknown. Increased intestinal motility and malabsorption are postulated as possible contributing factors.[8]

Using the Naranjo algorithm,[9] we concluded that metformin probably was the cause of diarrhea in this patient. Characteristics of this case that support metformin as causative include the drug's known propensity to cause this adverse effect during dosage titration, cessation of the diarrhea when the drug was stopped and resumption of diarrhea when the drug was restarted, and the lack of other identifiable causes after a rather thorough investigation. Our case reinforces the findings of the two reports described: metformin may cause diarrhea long after the initial days or weeks of therapy. Chronic diarrhea associated with metformin has been described as watery, often explosive, and frequently causing fecal incontinence. These features are similar to those of diabetic diarrhea, except that the drug-induced disorder may be much more likely in the patient with type 2 diabetes. This phenomenon may be underreported to physicians; therefore, routine questioning of patients taking metformin, both early and later in therapy, may be warranted. Metformin should be considered as the cause of chronic diarrhea in patients with type 2 diabetes who are taking a stable dosage. A drug-free interval may prevent an extensive work-up and considerable patient inconvenience and discomfort.