Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus

Shelley R Salpeter; Elizabeth Greyber; Gary A Pasternak; Edwin E Salpeter

doi:10.1002/14651858.CD002967.pub4

. 2010 Apr 14;2010(4):CD002967. doi: 10.1002/14651858.CD002967.pub4

Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus

Shelley R Salpeter ^1,^✉, Elizabeth Greyber ², Gary A Pasternak ², Edwin E Salpeter ³

Editor: Cochrane Metabolic and Endocrine Disorders Group

PMCID: PMC7138050 PMID: 20393934

Abstract

Background

Metformin is an oral anti‐hyperglycemic agent that has been shown to reduce total mortality compared to other anti‐hyperglycemic agents, in the treatment of type 2 diabetes mellitus. Metformin, however, is thought to increase the risk of lactic acidosis, and has been considered to be contraindicated in many chronic hypoxemic conditions that may be associated with lactic acidosis, such as cardiovascular, renal, hepatic and pulmonary disease, and advancing age.

Objectives

To assess the incidence of fatal and nonfatal lactic acidosis, and to evaluate blood lactate levels, for those on metformin treatment compared to placebo or non‐metformin therapies.

Search methods

A comprehensive search was performed of electronic databases to identify studies of metformin treatment. The search was augmented by scanning references of identified articles, and by contacting principal investigators.

Selection criteria

Prospective trials and observational cohort studies in patients with type 2 diabetes of least one month duration were included if they evaluated metformin, alone or in combination with other treatments, compared to placebo or any other glucose‐lowering therapy.

Data collection and analysis

The incidence of fatal and nonfatal lactic acidosis was recorded as cases per patient‐years, for metformin treatment and for non‐metformin treatments. The upper limit for the true incidence of cases was calculated using Poisson statistics. In a second analysis lactate levels were measured as a net change from baseline or as mean treatment values (basal and stimulated by food or exercise) for treatment and comparison groups. The pooled results were recorded as a weighted mean difference (WMD) in mmol/L, using the fixed‐effect model for continuous data.

Main results

Pooled data from 347 comparative trials and cohort studies revealed no cases of fatal or nonfatal lactic acidosis in 70,490 patient‐years of metformin use or in 55,451 patients‐years in the non‐metformin group. Using Poisson statistics the upper limit for the true incidence of lactic acidosis per 100,000 patient‐years was 4.3 cases in the metformin group and 5.4 cases in the non‐metformin group. There was no difference in lactate levels, either as mean treatment levels or as a net change from baseline, for metformin compared to non‐metformin therapies.

Authors' conclusions

There is no evidence from prospective comparative trials or from observational cohort studies that metformin is associated with an increased risk of lactic acidosis, or with increased levels of lactate, compared to other anti‐hyperglycemic treatments.

Plain language summary

Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus

Metformin, a medication used to lower glucose levels in patients with diabetes mellitus, has long been thought to increase the risk for a metabolic disorder known as lactic acidosis. This review summarised data from all known comparative and observational studies lasting at least one month, and found no cases of fatal or nonfatal lactic acidosis in 70,490 patient‐years of metformin use, or in 55,451 patient‐years for those not on metformin. Average lactate levels measured during metformin treatment were no different than for placebo or for other medications used to treat diabetes. In summary, there is no evidence at present that metformin is associated with an increased risk for lactic acidosis when prescribed under the study conditions.

Background

Description of the condition

Diabetes mellitus is a metabolic disorder resulting from a defect in insulin secretion, insulin action, or both (DeFronzo 1999). A consequence of this is chronic hyperglycemia with disturbances of carbohydrate, fat and protein metabolism. Long‐term complications of diabetes mellitus include retinopathy, nephropathy, neuropathy, and an increased risk of cardiovascular disease. For a detailed overview of diabetes mellitus, please see under 'Additional information' in the Metabolic and Endocrine Disorders Group in The Cochrane Library (see 'About the Cochrane Collaboration', 'Collaborative Review Groups'). For an explanation of methodological terms, see the main Glossary in The Cochrane Library.

Description of the intervention

Metformin is an anti‐hyperglycemic agent that has been used with increasing frequency over the past several years, especially in obese or overweight patients with type 2 diabetes whose blood glucose levels cannot be controlled non‐pharmacologically. There are three main questions that are being addressed. First, how does the drug affect total mortality and the development of long‐term diabetes‐related complications, and are these effects similar in all patient groups with type 2 diabetes? Secondly, what is the effect on cardiovascular risk factors such as obesity, dyslipidemia, and hypertension, and is this effect associated with changes in cardiovascular morbidity and mortality? The third question addresses concerns about the safety of the drug; specifically, what is the risk of fatal and nonfatal lactic acidosis associated with metformin use? These three questions are addressed in three separate reviews. The present review evaluates the risk of lactic acidosis attributed to metformin use, in comparison to placebo and other agents used for glycemic control in patients with type 2 diabetes. The reviews will be continually updated to include relevant new studies.

Adverse effects of the intervention

Adverse effects, principally gastrointestinal, are reported to occur in 20% to 30% of patients receiving metformin therapy and require discontinuation of the drug in less than 5% of patients (DeFronzo 1999). Diarrhea, nausea, vomiting, abdominal bloating, abdominal cramping or pain, flatulence, and anorexia are the most common gastrointestinal symptoms associated with metformin therapy. Other adverse effects reported are headache, agitation, dizziness, and tiredness.

Lactic acidosis is a rare, potentially fatal metabolic condition that can occur whenever substantial tissue hypoperfusion and hypoxia exist (Kreisberg 1980; Olivia 1970). Lactic acidosis is characterised by elevated blood lactate concentration (exceeding 45 mg/dl or 5.0 mEq/L), decreased blood pH (less than 7.35), and electrolyte disturbances with an increased anion gap. The mortality in reported cases have ranged from 8% to 50% (Bailey 1996; Laulau 2001; Misbin 1998). Biguanides are believed to decrease gluconeogenesis from alanine, pyruvate and lactate, and levels of lactic acid could accumulate under certain circumstances (Stang 1999). An earlier biguanide, phenformin, was withdrawn from the market because it was associated with a reported rate of lactic acidosis of 40 to 64 cases per 100,000 patient‐years (DeFronzo 1999; Stang 1999). Metformin differs from phenformin in molecular structure and pharmacokinetics (Sulkin 1997). Metformin, unlike phenformin, is thought to enhance glucose oxidation without significantly affecting fasting lactate production in peripheral tissues (Cusi 1996).

The true incidence of lactic acidosis associated with metformin is not known. The Food and Drug Administration has estimated the rate of fatal or nonfatal lactic acidosis to be five cases per 100,000 persons treated over the course of one year (Misbin 1998). Population‐based studies have estimated a rate of two to nine cases of lactic acidosis in metformin users per 100,000 person‐years (Bodmer 2008; Campbell 1985; Stang 1999; Wilholm 1993). However, most of the reported cases have occurred in patients with severe acute conditions, such as renal failure, that could in themselves have caused the lactic acidosis (Brown 1998; Misbin 1998). In order to estimate the risk specifically attributable to metformin, the background rate of lactic acidosis in patients with type 2 diabetes who are not treated with metformin must be assessed. To this end, a database was used to measure incidence rates in patients with type 2 diabetes in the United States before metformin was introduced, and found a rate of nine cases per 100,000 person‐years (Brown 1998). In addition, other population‐based studies have found similar incidence rates for users of metformin as for other agents such as insulin or sulfonylureas (Aguilar 1992b; Bodmer 2008). This raises the question of whether patients with type 2 diabetes have an increased risk for developing lactic acidosis with metformin use compared to other glucose‐lowering treatments.

How the intervention might work

Metformin hydrochloride is a biguanide that has been in clinical use for over 50 years (DeFronzo 1999; Sterne 1959). Unlike the sulfonylureas, biguanides do not have a hypoglycemic effect in healthy people and do not stimulate insulin release (Cusi 1996). Through its anti‐hyperglycemic effect, metformin lowers both fasting and postprandial blood glucose concentrations in patients with type 2 diabetes. Although the precise mechanism of this effect has not been fully established, evidence suggests that the drug improves both peripheral and liver sensitivity to insulin, reduces basal liver glucose production and increases insulin‐stimulated uptake and utilisation of glucose by peripheral tissues (AHFS 1999). Metformin, even in excessive dosage, normally does not lower glucose concentrations below euglycemia. Metformin accumulates in the wall of the intestine but does not appear to have clinically important effects on glucose absorption. In contrast, studies and systematic reviews have consistently shown that other diabetes treatments, including sulfonylureas and insulin, are associated with a substantial risk for clinically significant hypoglycemia (Bolen 2007; Hamnvik 2009).

Apart from its influence on carbohydrate metabolism, metformin is thought to have other positive effects related to type 2 diabetes and its long‐term prognosis. There may be modest improvements in serum lipids, in particular reductions of fasting serum triglycerides as well as total and LDL‐cholesterol concentrations. Additionally, therapy with metformin may be associated with weight loss or a stabilisation in weight gain. Suggested mechanisms for this effect include the absence of a hyperinsulinemic effect (which if present may increase appetite or lipogenesis) and decreased dietary intake caused by adverse gastrointestinal effects of metformin. There is inconclusive evidence at present on the effect of metformin on the fibrinolytic system and platelet aggregation, that play a role in the development of coronary artery thrombosis (Palumbo 1998).

Studies of metformin

Several trials using metformin alone or in combination with other drugs in patients with type 2 diabetes mellitus have been published. The UK Prospective Diabetes Study (UKPDS) was the first big trial to assess long‐term clinical outcomes related to metformin therapy in persons with type 2 diabetes. The study included overweight patients with newly diagnosed type 2 diabetes, mean age 53 years, who had no coronary artery disease or contraindication to treatment. The results indicated that metformin monotherapy led to a reduction in diabetes‐related endpoints and also in diabetes‐related mortality and total mortality, as compared to insulin, sulfonylurea therapy or diet alone (UKPDS‐34 1998). There were no cases of lactic acidosis in any group.

More recently, a long‐term trial evaluated the addition of metformin or placebo to insulin therapy and found that metformin was associated with a statistically significant 40% reduction in macrovascular morbidity and mortality compared with placebo (Kooy 2009a). The absolute risk reduction for macrovascular events for metformin compared with placebo was 6.1%, resulting in a number needed to treat of 16 to prevent one macrovascular endpoint, over the mean trial duration of 4.3 years. No cases of lactic acidosis were reported during the trial.

Several meta‐analyses have been published evaluating the effect of metformin on glucose regulation, weight, diabetes‐related outcomes, and mortality (Bolen 2007; Campbell 1995; Eurich 2007; Guthrie 1997; Johansen 1999; Saenz 2005). In the meta‐analyses by Campbell, Johansen, Guthrie and Bolen, metformin and other anti‐hyperglycemic treatments lowered blood glucose and glycosylated hemoglobin significantly compared with placebo, but body weight was substantially lower in the metformin group compared to other agents, including insulin and sulfonylureas. In addition, metformin was associated with significantly less clinically significant hypoglycemia than other agents. More recently, the Cochrane review by Saenz found that obese patients treated with metformin showed a significantly greater benefit than sulfonylureas or insulin for any diabetes‐related outcome as well as for total mortality. Finally, the systematic review by Eurich evaluated patients with diabetes and congestive heart failure and a found that metformin was associated with a greater reduction in mortality and hospital admission compared with any other diabetes treatment.

Why it is important to do this review

The available data indicate that metformin use in patients with type 2 diabetes mellitus is associated with a reduction in cardiovascular morbidity and mortality, compared to insulin, sulfonylureas or diet alone (Eurich 2007; Kooy 2009a; Saenz 2005; UKPDS‐34 1998). However, at present metformin use is considered to be contraindicated in many chronic conditions that may increase the risk of tissue anoxia (lack of oxygen) and the development of lactic acidosis, such as cardiovascular, renal, pulmonary and liver disease. These restrictions significantly reduce the number of patients who could benefit from metformin treatment. The present review assesses the risk of fatal and nonfatal lactic acidosis associated with metformin. Other adverse effects associated with metformin use are evaluated in another review.

Objectives

To assess the risk of fatal and nonfatal lactic acidosis associated with metformin use in persons with type 2 diabetes mellitus, compared to placebo or other glucose‐lowering therapies. A secondary objective was to evaluate levels of blood lactate, measured at baseline and during treatment, for metformin compared to placebo or other hypoglycemic therapies.

Methods

Criteria for considering studies for this review

Types of studies

Prospective clinical trials in patients with type 2 diabetes mellitus were included if they evaluated metformin, alone or in combination with other treatments, compared to placebo or to any other glucose‐lowering therapy, and lasted at least one month. Clinical trials were included even if they were not randomised or blinded. In addition, all observational cohort studies evaluating at least one month of metformin use were included in the analysis, as long as they provided the number of patients and the duration of treatment. The excluded trials lasting less than one month were evaluated separately to see if there were any cases of lactic acidosis.

Types of participants

Participants studied were adults with type 2 diabetes mellitus. To be consistent with changes in classification and diagnostic criteria of type 2 diabetes mellitus through the years, the diagnosis should have been established using the standard criteria valid at the time of the trial.

Types of interventions

Metformin, alone or in combination with other treatments, versus placebo or one of the following interventions used with the intention of lowering blood glucose levels:

sulfonylurea (for example, glibenclamide);
thiazolidinedione (for example, rosiglitazone);
meglitinide (for example, repaglinide);
alpha‐glucosidase inhibitor (for example, acarbose, miglitol);
dipeptidyl peptidase‐4 inhibitor (for example, sitagliptin, vildagliptin);
glucagon‐like peptide‐1 agonist (for example, exenatide, liraglutide);
sodium‐glucose cotransport inhibitor (for example, dapagliflozin)
insulin;
non‐pharmacological intervention (for example, diet);
any combination of the above.

Data on participants treated with phenformin were not included in the analysis for lactic acidosis, but were included in measurements of lactate levels.

Types of outcome measures

Primary outcomes

death described as due to lactic acidosis;
reported cases of nonfatal lactic acidosis, as defined by the investigator.

Secondary outcomes

blood lactate levels for metformin compared to placebo or other non‐biguanide therapies, and compared to phenformin.

Covariates, effect modifiers and confounders

Reported cases of renal failure or change in any hypoxic co‐condition (e.g. pulmonary disease). If cases of lactic acidosis were to be identified, their association with concurrent illness would be assessed.

Search methods for identification of studies

Electronic searches

Two investigators (SS, EG) jointly developed search strategies with the help of an information service librarian and the Cochrane Metabolic and Endocrine Disorders Group Trials Search Coordinator.

A comprehensive search of the following databases was performed to identify relevant human clinical trials or meta‐analyses:

The Cochrane Library (issue 3, 2009);
MEDLINE including OLDMEDLINE (until 10/2009);
REACTIONS (until 10/2009);
EMBASE (until 10/2009).

The described search strategy (see for a detailed search strategy Appendix 1) was used for MEDLINE. For use with EMBASE, The Cochrane Library and the other databases this strategy was slightly adapted.

Studies published in any language were included. No additional key words of relevance were identified during any of the electronic or other searches. If, in future searches, additional key works are found, electronic search strategies will be modified to incorporate these terms.

Searching other resources

In addition, attempts were made to contact authors of identified studies in order to obtain additional references, unpublished trials, ongoing trials or missing data not reported in the original trials. Similarly, the metformin manufacturer Bristol‐Myers Squibb Company was contacted in order to retrieve information on metformin trials, published and unpublished.

The search was further augmented by scanning references of identified articles or reviews, and of abstracts at a clinical symposium, reported in the journal Diabetologia, Volume 43, Supplement 1, 2000. The Cumulated Index Medicus was used to search relevant articles from 1959 to 1965.

Data collection and analysis

Selection of studies

Two authors (GP, SS) independently reviewed the records found in the search. Full articles were retrieved for further assessment if the information given suggested that the study evaluated metformin use in patients with diabetes mellitus. If there was any doubt regarding these criteria from the information given in the title and abstract, the full article was retrieved for clarification. In addition, any potentially relevant clinical trials found from scanning references of identified articles or reviews were retrieved.

Two investigators (SS, EG) independently evaluated retrieved studies for inclusion, and consensus was reached in cases of dispute. For publications with additional information on participants included in another publication, the publication with the most information was chosen as the included article, the companion publications are provided in the reference list.

Data extraction and management

Data concerning details of study population, intervention and outcomes were extracted independently by two authors (SS, EG) using a standard data extraction form. The data extraction form included the following items:

general information: published/unpublished, title, authors, contact address, language of publication, year of publication, duplicate publications, sponsoring, setting.
trial characteristics: design, duration, randomisation (and method), blinding (single‐, double‐ triple‐blind), method and check of blinding.
intervention(s): placebo included, interventions (dose, route, timing), comparison interventions (dose, route, timing), co‐medications (dose, route, timing).
patients: inclusion and exclusion criteria, total number and number in comparison groups, sex, age, selected baseline characteristics, diagnostic criteria, similarity of groups at baseline (including any co‐morbidity), withdrawals or losses to follow‐up (description), subgroups.
outcomes: deaths thought due to lactic acidosis, nonfatal lactic acidosis, lactate levels, renal failure, worsening of hypoxemic co‐conditions, length of follow‐up, quality of reporting of outcomes.

Differences in data extraction were resolved by consensus, referring back to the original article. Cases of lactic acidosis were to be tabulated according to the investigator's report. In addition, information was sought from the authors of the primary studies.

Assessment of risk of bias in included studies

The methodological quality of each study was evaluated based on the quality criteria modified from Schulz, Jadad and Stroup (Jadad 1996; Schulz 1995; Stroup 2000). Studies were divided into five categories.

For randomised controlled trials, the following factors were studied:   1. Minimisation of selection bias ‐ a) was the randomisation procedure adequate? b) was the allocation concealment adequate?   2. Minimisation of performance bias ‐ were the patients and people administering the treatment blind to the intervention?   2. Minimisation of attrition bias ‐ a) were withdrawals and drop‐outs completely described? b) was analysis by intention‐to‐treat?   4. Minimisation of detection bias ‐ were outcome assessors blind to the intervention?

Based on these criteria, trials were broadly subdivided into the following three categories:   A ‐ all quality criteria met: low risk of bias.   B ‐ one or more of the quality criteria only partly met: moderate risk of bias.   C ‐ one or more criteria not met: high risk of bias.

For non‐randomised trials, the following criteria were used:   D ‐ Open‐label non‐randomised controlled trials   E ‐ Observational cohort studies

Each trial was assessed independently by two authors (SS, EG), and consensus was reached in cases of disagreement. However, as no events were found in the results, sensitivity analyses using the quality assessments were not done.

Assessment of heterogeneity

Interstudy heterogeneity was to be tested for using the chi‐squared statistic for the assumption of homogeneity, with the statistical significance set at P < 0.1. Possible sources of heterogeneity were to be assessed by subgroup and sensitivity analyses as described below. As no cases of lactic acidosis were found, this was not performed. Small study bias was tested for using funnel plots.

Data synthesis

The treatment effect for fatal and nonfatal lactic acidosis was expressed as a risk difference, by taking the incidence of events on metformin, alone or in combination with other treatments, and then subtracting the incidence of events on placebo or alternative treatments. If there were non‐fatal events found, the first event would be considered for any one patient. We had planned to pool the results, using the fixed‐effect model for dichotomous data. The risk difference could then be converted to the number needed to harm (NNH). In addition, the results could be expressed as the relative risk of lactic acidosis associated with metformin use, compared to placebo or non‐metformin therapy. However, when no cases of lactic acidosis were found in either treatment group, the upper limit for the true incidence of lactic acidosis in the metformin group and the non‐metformin group were calculated separately using Poisson statistics.

Once pooled results revealed no cases of lactic acidosis, it was decided to report on trials that measured blood lactate levels for metformin, compared to placebo or non‐biguanide treatments, and also compared to phenformin. Three outcomes were analysed for the metformin group compared to the comparison groups: (1) the change in lactate levels from baseline to treatment, (2) the mean lactate levels recorded during treatment, and (3) the change in treatment lactate levels from a basal state to peak stimulation, either with food or exercise. The results were recorded as the weighted mean difference (WMD), in mmol/L, and pooled using the fixed‐effect model for continuous data.

Subgroup analysis and investigation of heterogeneity

We planned to perform subgroup analyses in order to explore the association of lactic acidosis with the following factors:

patients with hypoxemic co‐conditions, e.g. chronic renal insufficiency (creatinine >1.5 mg/dl) or renal failure, congestive heart failure, liver disease, pulmonary diseases, and peripheral artery disease;
age greater than 65 years;
metformin use, given as monotherapy or in combination with other medications;
different comparison interventions.

These analyses were not done as there were zero cases to analyse. Instead, information was obtained on how many patients were over the age of 65 or thought to have concomitant hypoxemic conditions.

Sensitivity analysis

We were to perform sensitivity analyses in order to explore the influence of the following factors on effect size:

repeating the analysis excluding unpublished studies, non‐randomised trials, and unblinded trials;
repeating the analysis taking account of study quality, as specified above;
repeating the analysis excluding any very long or large studies to establish how much they dominate the results;
repeating the analysis excluding studies funded by industry sponsors.

The robustness of the results was also to be tested by repeating the analysis using different measures of effects size (risk difference, relative risk, etc.) and different statistical models (fixed‐effect and random‐effects models). As no cases of lactic acidosis were found, sensitivity analyses were not performed.

Results

Description of studies

Results of the search

The electronic database search identified approximately 7000 articles, and of these 386 were potentially relevant studies on metformin use in patients with type 2 diabetes. After reviewing articles and bibliographies, the Cumulated Index Medicus, and abstracts at a clinical symposium, an additional 70 studies were identified. Of these 456 studies, 346 met inclusion criteria. No further articles were found by corresponding with authors, but information from one additional unpublished trial was received from Dr. Evertine Abbink, for a total of 347 included studies.

Missing data

Attempts were made to contact 102 of the authors for the comparative trials using the listed correspondence address, and 30 responses were received. All 30 of the respondents stated that they knew of no cases of lactic acidosis in any of their 34 trials. In addition, the metformin manufacturer Bristol‐Myers Squibb Company responded, stating that they had no unpublished trials to report. They provided a list of trials involving metformin, but none had been overlooked by the search.

Exclusion criteria of the studies

Of the 334 prospective studies, renal insufficiency (defined as a creatinine level of greater than 1.5 mg/dL) was listed as an exclusion criterion in 191 (57%), cardiovascular disease in 154 (46%), liver disease in 179 (54%), pulmonary disease in 46 (13%), and age greater than 65 years in 40 (12%).

Included studies

Studies and participants

Of the 347 studies analysed, 209 were prospective comparative trials, 125 were prospective cohort studies, and 13 were retrospective cohort studies. A total of 96,295 participants were followed for 125,941 patient‐years, with 69,642 participants (70,490 patient‐years) in the metformin group and 26,653 participants (55,451 patient‐years) in the non‐metformin group. The mean age of the participants in the metformin group was 57.1 (SD 8.8) years, with 61% men. In the non‐metformin group, the mean age was 57.2 (SD 9.0) years, with 61% men. From the available data it was estimated that 26% of the participants were over the age of 65 years, who were followed for approximately 32,745 patient‐years. The mean trial duration was 1.3 years, with a range from 0.1 to 10.7 years. The mean study size in the metformin group was 201 participants with a range of 6 to 11,014. The mean study size in the non‐metformin group was 128 participants with a range of 8 to 2,897. The drop‐out rate was estimated to be 9.2%.

Interventions

Metformin was given in daily doses of 1 to 3 grams, with the dosage titrated clinically. Comparison treatments included placebo, diet, insulin, glyburide (glibenclamide), gliclazide, glipizide, glimepiride, chlorpropamide, tolbutamide, acarbose, nateglinide, repaglinide, miglitol, troglitazone, rosiglitazone, pioglitazone, vildagliptin, sitagliptin, saxagliptin, dapagliflozin, and guar gum.

Outcome measures

Outcomes measured included glycemic control (blood and urinary glucose, HbA1, HbA1c, insulin, and C‐peptide levels), insulin sensitivity using a glucose clamp, weight, energy consumption, lipids, lipoproteins, fructosamine, free fatty acids, fibrinogen, plasminogen activator inhibitor, C‐reactive protein, heart rate, blood pressure, lactate levels, bicarbonate, ketones, microalbuminuria, renal and liver function tests, crude mortality rate, time‐related mortality, drug‐related adverse events, death from hyper‐ or hypoglycemia, renal failure, diabetic eye disease, and cardiovascular endpoints (sudden death, myocardial infarction, stroke).

Only 19 trials were specifically designed to assess the incidence of lactic acidosis, but side effects or adverse events were described in almost all the trials. Attempts were made to reach the authors of the trials and 30 investigators replied, all confirming that there were no known cases of fatal or nonfatal acidosis in any of their 34 trials. Serum bicarbonate or lactate levels were measured in 123 of the included studies (45%). Of the comparative trials, 25 measured lactate levels while on metformin and non‐metformin treatment (Bjorntorp 1978; Botha 1977; Campbell 1994; Cavallo‐Perin 1989; Cryer 2005; Cusi 1996; Damsbo 1998; De Silva 1979; DeFronzo 1995; Fritsche 2000; Gregorio 1989; Gregorio 1990; Hother‐Nielsen 1989; Inzucchi 1998; Jackson 1987; Josephkutty 1990; Klein 1991; McAlpine 1988; Nattrass 1977; Pedersen 1989; Rachmani 2002; Raptis 1996; Teupe 1991; Velussi 1992; Wu 1990).

Excluded studies

Studies were excluded for the following reasons: Ten were retrospective cohort studies that did not give information on the number of patients or the length of treatment (Charlton 2008; Cook 2005; Debry 1966a; Debry 1966b; Evans 2006; Masoudi 2005; Monami 2008a; Ong 2006; Sharabashi 2006; Simpson 2006), 33 were prospective cohort studies that did not give information on the number of patients or length of treatment (Bernard 1965; Berhanu 2007; Carpentier 1975; Chow 1995; Clauson 1996; Comaschi 2007; Comaschi 2008; Debry 1964; Derosa 2009; Eurich 2005b; Farah 2008; Forti 2008; Harris 2008; Hermansen 2007; Home 2009; Javaid 2007; Kamber 2008; Kim 2008; Komajda 2008; Lapina 2008; Lin 2008; Messens 1965; Messens 1966; Monami 2006; Muntoni 1965; Nauck 1993; Nauck 2009; Papa 2008; Rambert 1961; Seufert 2008; Sugawara 1962; Teitelbaum 1963; Yegnanarayan 2008), 41 prospective comparative trials were of less than one month duration (Bonfigli 1999; Bruneder 1978; Cacciapuoti 1991; Cunha 2008; English 2007; Faure 2008; Fery 1997; Galuska 1994; Gibson 1995; Gin 1982; Gin 1985; Gin 1989; Giugliano 1979; Gontier 2008; He 2009; Herman 2006; Herman 2006; Hong 2008; Irsigler 1978; Ismail 1978; Isnard 1996; Jansson 1996; Leslie 1987; Lim 1970; Magalhaes 2006; Orlikowska 1966; Panahloo 1995; Perriello 1994; Pilger 1978; Prager 1983; Rigas 1968; Rizkalla 1986; Sambol 1996; Scarpello 1998; Schaffalitzky 1979; Signore 1996; Slama 1984; Sum 1992; Trischitta 1983; Turner 1995; Zapecka‐Dubno 1999), and 21 were retrospective analyses or reviews (Aguilar 1992b; Belsey 2008; Bodmer 2008; Chan 2009; Connolly 1996; Daniel 1997; Eurich 2005a; Guthrie 1997; Hirsch 2009; Johansen 1999; Lalau 1994; Lalau 1995; Mellbin 2008; Monami 2008; Nauck 1997; O'Connor 1998; Rao 2008; Runge 2008; Selby 1999; Tomioka 2007; Zhang 2009).

Risk of bias in included studies

The methodological quality evaluation of the studies was done using the criteria described above. Only information published in the trials was used to determine bias categories. An A low‐risk of bias category was given to 32 trials, a B category to 63 trials, a C category to 84 trials, a D category for 30 trials, and an E category for 138 studies. Of the studies analysed, 94 were double‐blind, randomised controlled trials (32 described the method of randomisation and allocation concealment). Another 115 were single‐blind or open‐label comparative trials (85 randomised and 30 non‐randomised). The 138 cohort studies were all open‐label and observational. The average drop‐out rate was approximately 9%.

Effects of interventions

Incidence of lactic acidosis

When combining the data from cohort studies with the prospective comparative trials, there were no cases of fatal or nonfatal lactic acidosis reported in the metformin group, totaling 70,490 patient‐years, and no cases in the non‐metformin group, representing 55,451 patient‐years. Using Poisson statistics with 95% confidence, the upper limit for the true incidence of metformin‐associated lactic acidosis was 4.3 cases per 100,000 patient‐years, and the upper limit for the incidence of lactic acidosis in the non‐metformin group was 5.4 cases per 100,000 patient‐years. When combining data from metformin and non‐metformin groups together the upper limit for the true incidence of lactic acidosis in all patients with type 2 diabetes was 2.4 cases per 100,000 patient‐years.

Association with hypoxemic co‐conditions

Another outcome to be assessed was the number of participants with worsening of their hypoxemic co‐conditions during the trial. An accurate assessment of the incidence of renal failure or worsening of other conditions could not be made because two of the large trials did not provide adequate data (Fisman 2001; UKPDS‐34 1998). On correspondence with the authors of these trials, this information could not be provided.

There was insufficient information to estimate the number of participants studied with hypoxemic co‐conditions such as renal insufficiency, cardiovascular diseases, liver diseases, or pulmonary disease. Instead, each of the trials included in this analysis was characterized as to whether any of these conditions were listed as exclusion criteria. If the patients were listed as healthy or that standard contraindications were used, it was assumed that all of these conditions were excluded. Renal insufficiency was usually defined as a creatinine level of greater than 1.5 mg/dl. Of the 334 prospective studies, 143 (53%) allowed for the inclusion of renal insufficiency, following 37,360 patient‐years of metformin use, and 324 (97%) allowed for the inclusion of at least one of the contraindications listed above. It was estimated from the available data that 26% of the participants in the studies were over the age of 65 years, who were followed for approximately 18,327 patient‐years of metformin use.

Blood lactate levels

For those trials that provided the data, the baseline lactate level measured prior to metformin treatment was 1.13 ± 0.25 mmol/L. There was no difference in the net change of lactate levels from baseline for metformin compared to placebo or non‐biguanide therapies, with a weighted mean difference (WMD) of 0.12 mmol/L (95% CI ‐0.01 to 0.25). The mean lactate level during metformin treatment was 1.24 ± 0.31 mmol/L, which was not significantly different from non‐biguanide comparisons (WMD 0.04 mmol/L, 95% CI 0.00 to 0.13, P = 0.07), and was 0.75 mmol/L lower than with phenformin (95% CI ‐0.86 to ‐0.65). Lactate levels during metformin treatment, measured before and after stimulation (by a meal or strenuous exercise), were 2.3 ± 1.7 mmol/L. This was not significantly different for metformin compared to the non‐biguanide group (WMD 0.09 mmol/L, 95% CI ‐0.03 to 0.22) or to phenformin (WMD ‐0.37 mmol/L, 95% CI ‐1.06 to 0.32). Four trials that measured lactate levels did not provide data to be analysed, but reported levels to be normal during metformin and non‐metformin treatment (DeFronzo 1995; Fritsche 2000; Gregorio 1989; Raptis 1996).

Heterogeneity was noted in the three trials that measured lactate levels after stimulation by food or exercise during treatment with metformin or non‐biguanide therapies, probably due to the fact that each was performed under different conditions. The results were not significantly different when the random‐effects model was used (WMD 0.04 mmol/L, 95% CI ‐0.45 to 0.53). In addition, heterogeneity was noted in the three trials measuring mean lactate levels for metformin compared to phenformin treatment. When the random‐effects model was used the difference was no longer statistically significant (‐0.64 mmol/L, 95% CI ‐1.63 to 0.35).

Small study bias

Funnel plots of the effect size versus standard error were evaluated for the included trials in the analysis. The funnel plot used for the incidence of lactic acidosis was unable to provide evidence for or against the possibility of small study bias, since all of the trials found no cases of lactic acidosis. Funnel plots for the measurement of lactate levels showed no convincing evidence for significant small study bias.

Discussion

Summary of main results

In order to evaluate the risk of lactic acidosis attributed to metformin use, pooled data from all known prospective comparative trials and observational cohort studies with durations of at least one month were analysed. No cases were found in 347 trials with 70,490 patient‐years of metformin treatment. In fact, on review of 94 additional trials that were excluded from analysis (those that lasted less than one month or were of unclear duration) no cases of lactic acidosis were found. The risk difference for metformin compared to non‐metformin treatment, calculated using Poisson statistics, was 0.00 per 100,000 patient‐years (95% CI, ‐5.4 to +4.3). This indicates that the upper limit for the true incidence of metformin‐associated lactic acidosis is 4 cases per 100,000 patient‐years, and the upper limit for the incidence with other non‐biguanide treatments is 5 per 100,000 patient‐years. Of the trials that measured blood lactate levels, there was no significant difference for metformin compared to placebo or non‐biguanide treatments, and was lower for metformin than for phenformin.

The mean duration of studies included in this review was 1.3 years, with a wide range from 1 month to 10.7 years. As no cases of lactic acidosis were found in any trial, the association of lactic acidosis with duration of treatment could not be assessed. In addition, excluded trials of less than one month duration were evaluated to see if lactic acidosis occurs shortly after initiation of treatment, and no cases were found.

At present, metformin is often considered to be contraindicated in patients with chronic renal insufficiency, liver function abnormalities, congestive heart failure, peripheral vascular disease, pulmonary disease, or age greater than 65, as these conditions may increase the risk of tissue anoxia and therefore the development of lactic acidosis. In this review, 324 (97%) of the 334 prospective studies allowed for the inclusion of patients with at least one of these contraindications, and 26% of all participants were estimated to be older than 65 years, with no adverse effects observed. However, it is not clear how many participants with each of these conditions were included in the trials, so the safety of metformin in the presence of these standard contraindications cannot be assessed. One trial (Rachmani 2002) questioned the standard contraindications by studying 393 patients, all with at least one contraindication to metformin use, and found no cases of lactic acidosis over four years of the trial duration. All of the patients in this trial had renal insufficiency, with mean plasma creatinine levels of 1.5 to 2.5 mg/dl (mean level 1.8 mg/dl).

Overall completeness and applicability of evidence

Metformin is a biguanide anti‐hyperglycemic medicine that has been in use for over 50 years (Sterne 1959). Metformin treatment in patients with type 2 diabetes has been shown to reduce cardiovascular events and mortality when compared to insulin, sulfonylureas or diet alone (Eurich 2007; Kooy 2009a; Saenz 2005; UKPDS‐34 1998). Studies have consistently shown that treatments other than metformin, such as insulin and sulfonylureas, are associated with significant weight gain and a substantial risk for clinically significant hypoglycemia (Bolen 2007; Campbell 1995; Guthrie 1997; Hamnvik 2009; Johansen 1999).

Concern about the risk of lactic acidosis has led to recommendations that metformin be withheld in persons with chronic conditions that in themselves can cause lactic acidosis. These recommendations, if followed, would reduce the number of patients eligible to receive metformin by approximately one half (Brown 1998). It has been found that in clinical practice these standard contraindications are largely disregarded, with 54% to 73% of patients on metformin having at least one standard contraindication to treatment (Calabrese 2002; Holstein 1999; Sulkin 1997). In two studies, approximately 15% of patients on metformin admitted to a hospital had concurrent renal insufficiency (Calabrese 2002; Holstein 1999). In this meta‐analysis, 97% of the studies allowed for at least one of the standard contraindications.

Metformin has been implicated as a cause of lactic acidosis because a related biguanide, phenformin, had been associated with several cases of lactic acidosis and was removed from the US market in 1977 (Aguilar 1992b). Despite their similarities, phenformin has a chemical structure significantly different from metformin. Unlike metformin, phenformin can impair oxidative phosphorylation in the liver, thereby increasing lactate production by anaerobic pathways (Cavallo‐Perin 1989; Irsigler 1978; Pilger 1978; Sirtori 1994; Velussi 1992). In contrast, metformin inhibits hepatic gluconeogenesis without altering lactate turnover or lactate oxidation (Cusi 1996; Scheen 1996; Stacpoole 1998). In addition to the trials analysed in this review, several other trials have confirmed that metformin treatment does not significantly elevate blood lactate levels, even in the presence of renal impairment or advanced age (Connolly 1996; Debry 1964; Giugliano 1993; Irsigler 1978; Lalau 1990; Menzies 1989; Pagano 1983; Pilger 1978; Trischitta 1983).

At present the only evidence to indicate that metformin use is associated with lactic acidosis comes from reports of approximately 330 cases that have occurred in patients while on metformin treatment (Bergman 1978; Gan 1992; Lalau 1994; Luft 1978). The incidence of lactic acidosis occurring in patients on metformin has been estimated from population studies to be 2 to 9 cases per 100,000 patient‐years (Bodmer 2008; Misbin 1998; Stang 1999; Wilholm 1993). Essentially all of the cases reported were in patients with severe underlying conditions that in themselves could have caused the lactic acidosis.

Lactic acidosis has also been reported in diabetic patients not treated with metformin, typically under conditions in which there is significant tissue hypoperfusion or hypoxia (Aguilar 1992b; Bodmer 2008). To assess the rate of lactic acidosis in diabetic patients on treatment other than metformin, a population study followed patients with type 2 diabetes who were treated in the USA prior to the introduction of metformin and after the withdrawal of phenformin (Brown 1998). This study found the rate of confirmed lactic acidosis to be approximately 10 per 100,000 patient‐years, which is equivalent to that seen with metformin treatment. Another study evaluated all cases of metabolic non‐ketotic acidosis in patients with type 2 diabetes that occurred during 600 emergency admissions to a University hospital (Aguilar 1992b). The rates of non‐ketotic acidosis per 1000 emergency admissions were 29 for sulfonylureas, 48 for insulin, and no cases for those on metformin treatment. All cases of non‐ketotic metabolic acidosis found were associated with severe precipitant disease that could have caused lactic acidosis. More recently, a nested case‐control analysis using the United Kingdom General Practice Research Database found that the crude incidence rate for lactic acidosis in patients with diabetes was 3.3 cases per 100,000 person‐years among users of metformin and 4.8 cases per 100,000 person‐years among users of sulfonylureas (Bodmer 2008). The investigators of these observational studies concluded that it is the underlying systemic dysfunction and not the particular treatment that is the main determinant for the appearance of lactic acidosis. In support of that conclusion, the results of this review reveal that there is no evidence of an increased risk of lactic acidosis associated with metformin use, as compared with other diabetes treatments, when prescribed under the study conditions.

Potential biases in the review process

This review has several limitations. Essentially all the data included in this analysis were from published trials, and this may have produced biased results. A funnel plot of effect size versus standard error was unable to provide convincing evidence for significant small study bias, since no cases were found in any trial. It is interesting to note that many of the comparative trials included in the analysis were sponsored by pharmaceutical companies producing anti‐hyperglycemic medications other then metformin, in which case a bias may be to publish adverse effects for metformin.

Another difficulty is that in order to assess the risk of a rare occurrence such as lactic acidosis, it may be necessary to evaluate more than 70,000 patient‐years of metformin treatment. It is especially difficult to assess the risk of lactic acidosis in the presence of standard contraindications such as renal or hepatic insufficiency because it is unclear exactly how many of the participants had these conditions. For that reason, no conclusions can be made about the safety of metformin use in the presence of these conditions. Despite these limitations, the most important conclusion from this review is that, at present, there is no evidence from prospective comparative trials or observational cohort studies to support the hypothesis that metformin is associated with an increased risk for lactic acidosis.

Authors' conclusions

Implications for practice.

There is no evidence from prospective comparative trials or from observational cohort studies that metformin treatment increases the incidence of lactic acidosis compared with other anti‐hyperglycemic treatments. This review was not able to quantitatively assess the safety of metformin treatment in the presence of each of hypoxic co‐conditions.

Implications for research.

Large prospective, comparative trials are necessary in patients with type 2 diabetes mellitus who have conditions that are presently considered contraindications for its use. For example, a large trial could be performed in patients known to have chronic renal insufficiency. Outcomes to be followed would include the incidence of lactic acidosis as well as diabetes‐related complications and total mortality.

What's new

Date	Event	Description
16 March 2010	New citation required but conclusions have not changed	Erratum author Salpeter EE: 'posthumous' deleted.

Date	Event	Description
12 November 2009	New citation required but conclusions have not changed	The third update as of October 2009 revealed no cases of fatal or nonfatal lactic acidosis in over 70,000 patient‐years of metformin use. Metformin did not significantly affect lactic acid levels.
11 November 2009	New search has been performed	An update search covering the period 2008 to October 8, 2009 identified 137 potentially relevant publications out of 1660 scanned references. From these 73 studies were included as new trials.
30 September 2007	New search has been performed	Second update: No cases of fatal or nonfatal lactic acidosis were found in over 50,000 patient‐years of metformin use. Metformin did not significantly affect lactic acid levels.
31 August 2005	New search has been performed	This is an update of the first version of this review, published in issue 2, 2002.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Net treatment effect, lactate levels (mmol/L)	7	222	Mean Difference (IV, Fixed, 95% CI)	0.12 [‐0.01, 0.25]
1.1 Net treatment effect, lactate levels (metformin minus non‐metformin, mmol/L)	7	222	Mean Difference (IV, Fixed, 95% CI)	0.12 [‐0.01, 0.25]
2 Mean treatment lactate levels (mmol/L)	19	1547	Mean Difference (IV, Fixed, 95% CI)	‐0.09 [‐0.13, ‐0.05]
2.1 Mean treatment lactate levels (metformin minus non‐metformin, mmol/L)	16	1387	Mean Difference (IV, Fixed, 95% CI)	0.04 [‐0.00, 0.09]
2.2 Mean treatment lactate levels (metformin minus phenformin, mmol/L)	3	160	Mean Difference (IV, Fixed, 95% CI)	‐0.75 [‐0.86, ‐0.65]
3 Peak stimulated lactate levels (mmol/L)	4	92	Mean Difference (IV, Fixed, 95% CI)	0.08 [‐0.05, 0.20]
3.1 Peak stimulated lactate levels (metformin minus non‐metformin, mmol/L)	3	72	Mean Difference (IV, Fixed, 95% CI)	0.09 [‐0.03, 0.22]
3.2 Peak stimulated lactate levels (metformin minus phenformin, mmol/L)	1	20	Mean Difference (IV, Fixed, 95% CI)	‐0.37 [‐1.06, 0.32]

Methods	TRIAL DESIGN: Retrospective cohort study   DURATION: at least one year, then 12 weeks.
Participants	COUNTRY: Norway   SETTING: Endocrinology center Treatment N: 28, with 14 on folate and 14 on placebo.   Metformin + placebo AGE: 57+/‐2.8. Metformin + folate AGE: 62+/‐2.5. Metformin + placebo SEX: 79% men. Metformin + folate SEX: 71% men. INCLUSION: patients with type 2 DM, treated with metformin for a minimum of 1 year   EXCLUSIONS: vitamin use that would interfere with the study.
Interventions	TREATMENT: metformin, at least 1g/day. One‐half of patients on folate 0.25 mg/day + iron 60mg/day, and one‐half on iron 60mg/day.   COMPARISON: none.
Outcomes	Fasting homocysteine, cysteine, cysteinylglycine, vitamin B12, and folate.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Randomised controlled trial   DURATION: 3 months
Participants	COUNTRY: United States   SETTING: research laboratory Treatment N: 15   Control N: 8.   Treatment AGE: 53 +/‐3   Control AGE: 51 +/‐4   Treatment SEX: 64% men   Control SEX: 87% males   INCLUSION: Type 2 DM   EXCLUSIONS: abnormal laboratory values, vascular disease
Interventions	TREATMENT: metformin‐blind versus open‐label metformin, dosage adjused clinically. COMPARISON: placebo
Outcomes	Fasting and postprandial glucose, insulin, and free fatty acids.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	High risk	C ‐ Inadequate

Methods	TRIAL DESIGN: Prospective cohort study   DURATION: 6 months
Participants	COUNTRY: United States   SETTING: outpatient and research center   Treatment N: 11   Control N: 0   AGE: not listed   SEX: not listed   INCLUSION: diet‐treated type 2 DM EXCLUSIONS: laboratory abnormalities, diabetic vascular complications, or abnormal electrocardiogram
Interventions	TREATMENT: metformin 1‐2.5 g/day COMPARISON: none
Outcomes	Plasma glucose, insulin, and free fatty acids.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Double‐blind randomised controlled trial ‐ unpublished   DURATION: 2 months
Participants	COUNTRY: Netherlands   SETTING: outpatient   Treatment N: 12   Control N: 12   AGE: unclear   SEX: not listed.   INCLUSION: Type 2 DM   EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin 500 mg TID COMPARISON: Glibenclamide
Outcomes	Glucose, HbA1.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	TRIAL DESIGN: Abstract of a double‐blind randomised controlled trial   DURATION: 2 months
Participants	COUNTRY: Netherlands   SETTING: outpatient   Treatment N: 12   Control N: 60   AGE: unclear   SEX: not listed   INCLUSION: Type 2 DM   EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin, dosage adjusted clinically COMPARISON: glibenclaminde or glimerperide or acarbose
Outcomes	Vasodilator responses to diazoxide.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	TRIAL DESIGN: Prospective observational cohort   DURATION: 6 months
Participants	COUNTRY: Georgia   SETTING: Outpatient   Treatment N: 26   Control N: 0   AGE: 59.7   SEX: 0% men   INCLUSION: Type 2 DM, obese postmenopausal women   EXCLUSIONS: None listed
Interventions	TREATMENT: Metformin, 1700 ‐ 2500 mg/day
Outcomes	Leptin, adiponectin, insulin resistance
Notes

Methods	TRIAL DESIGN: Prospective cohort study of metformin in a randomised trial   DURATION: 12 weeks
Participants	COUNTRY: Italy   SETTING: outpatient   Treatment N: 107   Control N; 0   AGE: 57.7   SEX: 45% men   INCLUSION: type 2 DM   EXCLUSIONS: clinically significant cardiovascular disease, carbohydrate disorders, elevated triglycerides
Interventions	TREATMENT: metformin plus vildgliptin or placebo
Outcomes	Beta‐cell function and insulin sensitivity
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Prospective randomized controlled trial   DURATION: 3 months
Participants	COUNTRY: Mexico   SETTING: outpatient   Treatment N: 26   Control N: 23   AGE: < 65 years   SEX: not listed   INCLUSION: type 2 diabetes and incipient nephropathy   EXCLUSION: hypertension, malignancy, hepatic or cardiovascular disorders
Interventions	TREATMENT: metformin, dosage unclear   COMPARISON: glibenclamide, dosage unclear
Outcomes	Metabolic control, blood pressure, unsulin, lipids
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Open‐label randomized controlled trial   DURATION: 48 weeks
Participants	COUNTRY: China   SETTING: Outpatient   Treatment N: 22   Control N: 60   AGE: Not stated   SEX: Not stated   INCLUSION: Type 2 DM   EXCLUSIONS: None listed
Interventions	TREATMENT: Metformin, 750‐1500 mg/day COMPARISON: Repaglinide or Rosiglitazone
Outcomes	Glycemic and metabolic control
Notes

Methods	TRIAL DESIGN: Double‐blind randomized controlled trial   DURATION: 4 months
Participants	COUNTRY:   SETTING: Outpatient   Treatment N: 15   Control N: 16   AGE: 56   SEX: 55% men   INCLUSION: Type 2 DM   EXCLUSIONS:
Interventions	TREATMENT: Metformin, 1 gm BID COMPARISON: Pioglitzaone, 45 mg/day
Outcomes	Insulin action
Notes

Methods	TRIAL DESIGN: Prospective comparative study   DURATION: 3 months
Participants	COUNTRY: United States   SETTING: outpatient   Treatment N: 14   Control N: 7   Treatment AGE: 49   Control AGE: 54   SEX: not stated   INCLUSION: type 2 DM on oral sulfonylurea   EXCLUSIONS: alcoholism, chronic renal failure, liver disease, cardiopulmonary disease
Interventions	TREATMENT: metformin, dosage unclear   COMPARISON: sulfonylurea, dosage unclear
Outcomes	Vitamin B12 measurements
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Crossover randomised controlled trial   DURATION: 2 months
Participants	COUNTRY: Turkey   SETTING: University clinic Treatment N: 36   Control N: 36   AGE: 30‐63   SEX: 100% men   INCLUSION : Type 2 DM with poor control   EXCLUSIONS: microvascular or macrovascular complIcations, liver function abnormalities.
Interventions	TREATMENT: Metformin 500mg TID   COMPARISON: acarbose
Outcomes	Insulin , c‐peptide, fibrinogen, lipids, HbA1.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	TRIAL DESIGN: Prospective randomised controlled trial   DURATION: 2.5 months
Participants	COUNTRY: Venezuela   SETTING: outpatient   Treatment N: 53   Control N: 9   Treatment AGE: 52   Control AGE: 55.3   SEX: not states   INCLUSION: type 2 DM   EXCLUSIONS: Age > 60 years
Interventions	TREATMENT: metformin, 500 mg TID   COMPARISON: metformin, 500 mg TID plus glimeperide 0.5 mg daily
Outcomes	Glycemic control, insulin, insulin resistance
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN:   Two prospective double‐blind randomised controlled trials   DURATION: 2 years
Participants	COUNTRY: United Kingdom   SETTING: outpatient   Treatment N: 960   Control N: 319   AGE: not stated   SEX: not stated   INCLUSION: type 2 DM   EXLCUSIONS: not listed
Interventions	TREATMENT: study 1: metformin, dosage unclear plus pioglitazone 15‐45 mg daily. Study 2: metformin 850‐2550 mg daily   COMPARISON: study 1: metformin, dosage unclear plus gliclazide 80‐320 mg daily.   study 2: pioglitazone 15‐45 mg daily
Outcomes	Lipids and lipoproteints
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Prospective cohort study in a randomised trial of extended release metformin   DURATION: 12 weeks
Participants	COUNTRY: India   SETTING: outpatient   Treatment N: 40   Control N: 0   AGE: 57.3   SEX: not stated   INCLUSION: type 2 DM   EXCLUSION: renal or hepatic dysfunction, congestive heart failure
Interventions	TREATMENT: metfromin, up to 2 gm daily   COMPARISON: none
Outcomes	Glycemic control
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Blinded randomised controlled trial (unclear if double‐blind)   DURATION: 2 months
Participants	COUNTRY: United Kingdom   SETTING: outpatient   Treatment N: 22   Control N: 22   AGE: unclear   SEX: not listed   INCLUSION: Type 2 DM not controlled on sulfonylureas EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin 1‐2 g/day + chlorpropamide   COMPARISON: placebo + chlorpropamide
Outcomes	Plasma glucose and weight.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	TRIAL DESIGN: Prospective, cross‐over comparative trial; not randomised   DURATION: 8 weeks
Participants	COUNTRY: Sweden   SETTING: outpatient   Treatment N: 21   Control N: 21   AGE: 58   SEX: 52% men   INCLUSION: Type 2 DM on long‐term biguanide treatment EXCLUSIONS: abnormal renal function or liver function
Interventions	TREATMENT: Metformin, 1.5‐3.0 g/day   COMPARISON: phenformin, 50‐100 mg/day (not analysed)
Outcomes	Fasting glucose and fasting lactate levels.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Open‐label cross‐over trial; not randomised   DURATION: 1 month
Participants	COUNTRY: South Africa   SETTING: general practice   Treatment N: 21   Control N: 21   AGE: unclear   SEX: not listed   INCLUSION: Type 2 DM   EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin, dose adjusted clinically   COMPARISON: phenformin, buformin (not analysed), and untreated controls.
Outcomes	Heart rate, blood lactate, and lactate/pyruvate ratios, at baseline and with exercise.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

PERMALINK

Risk of fatal and nonfatal lactic acidosis with metformin use in type 2 diabetes mellitus

Shelley R Salpeter

Elizabeth Greyber

Gary A Pasternak

Edwin E Salpeter

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Description of the condition

Description of the intervention

Adverse effects of the intervention

How the intervention might work

Studies of metformin

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Covariates, effect modifiers and confounders

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Assessment of heterogeneity

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

Missing data

Exclusion criteria of the studies

Included studies

Studies and participants

Interventions

Outcome measures

Excluded studies

Risk of bias in included studies

Effects of interventions

Incidence of lactic acidosis

Association with hypoxemic co‐conditions

Blood lactate levels

Small study bias

Discussion

Summary of main results

Overall completeness and applicability of evidence

Potential biases in the review process

Authors' conclusions

Implications for practice.

Implications for research.

What's new

History

Acknowledgements

Appendices

Appendix 1. Search strategy

Data and analyses

Comparison 1. Fatal/nonfatal lactic acidosis.

1.1. Analysis.

Comparison 2. Blood lactate levels.

2.1. Analysis.

2.2. Analysis.

2.3. Analysis.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Methods	TRIAL DESIGN: Open‐label randomised controlled trial   DURATION: 4 weeks
Participants	COUNTRY: United Kingdom SETTING: outpatient   Treatment N: 39   Control N: 67   AGE: 57   SEX: 21% men   INCLUSION: Type 2 DM   EXCLUSIONS: renal failure, congestive heart failure
Interventions	TREATMENT: Metformin 850 mg BID   COMPARISON: phenformin (not analysed)
Outcomes	Fasting glucose, body weight, and lipids
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Open‐label comparative trial; not randomised   DURATION: 4 months
Participants	COUNTRY: Spain   SETTING: outpatient   Treatment N: 12   Control N: 24   AGE: unclear   SEX: 50% men   INCLUSION: Type 2 DM   EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin 850 mg TID   COMPARISON: insulin or acarbose
Outcomes	Lipids, blood pressure, HbA1, body weight, insulin sensitivity.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Double‐blind crossover randomised controlled trial   DURATION: 6 months
Participants	COUNTRY: Italy   SETTING: outpatient   Treatment N: 10   Control N: 10   AGE: 51+/‐2.1   SEX: 60% men   INCLUSION: Type 2 DM   EXCLUSIONS: liver or kidney disease, heart failure, other drugs, or chronic infection
Interventions	TREATMENT: Metformin 850 mg BID   CONTROL: phenformin 50 mg BID (not analysed)
Outcomes	Weight, glucose, HbA1, and blood lactate levels at different times of day.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	TRIAL DESIGN: 4 prospective double‐blind randomised controlled trials   DURATION: 1 year
Participants	COUNTRY: United States   SETTING: multi‐center   Treatment N: 298   Control N: 541   Treatment AGE: 55.6   Control AGE: 57   Treatment SEX: 80% men   Control SEX: 71% men   INCLUSION: type 2 DM, 35‐75 years, poorly controlled   EXLCUSIONS: heart attack or stroke
Interventions	TREATMENT: metformin, alone or in combination with other medications   COMPARISON: pioglitazone, gliclazide, sulfonylureas
Outcomes	Glycemic control, insulin sensitvity
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Single‐blind crossover comparative trial; not randomised   DURATION: 2 months placebo, 4 months treatment
Participants	COUNTRY: United Kingdom   SETTING: outpatient   Treatment N: 27   Control N: 27   AGE: 56.3   SEX: 95% men   INCLUSION: Type 2 DM with coronary artey disease, claudication   EXCLUSIONS: none listed
Interventions	TREATMENT: metformin 500 mg TID   COMPARISON: placebo
Outcomes	Cholesterol, plasma fibrinogen.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Ongoing prospective observational cohort   DURATION: 3 years
Participants	COUNTRY: United Kingdom   SETTING: Outpatient   Treatment N: 178   Control N: 0   AGE: 54.1   SEX: Not stated   INCLUSION: Type 2 DM, 35‐80 years, poor control   EXCLUSIONS: None listed
Interventions	TREATMENT: Metformin, with glicazide, repaglinide or pioglitazone, varying doses
Outcomes	Deterioration of glycemic control
Notes

Methods	TRIAL DESIGN: Prospective double‐blind randomised controlled trial   DURATION: 20 weeks
Participants	COUNTRY: France   SETTING: outpatient   Treatment N: 222   Control N: 150   Treatment AGE: 56.7   Control AGE: 55.4   Treatment SEX: 60% men   Control SEX: 58% men   INCLUSION: type 2 DM age 35‐70, poorly controlled   EXCLUSIONS: severe chronic disease, morbid obesity, major cardiovascular event
Interventions	TREATMENT: metformin with or without glimepiride   COMPARISON: Glimepiride
Outcomes	Glycemic control
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Prospective comparative trial   DURATION: 8 weeks
Participants	COUNTRY: Yugoslavia   SETTING: outpatient   Treatment N: 23   Control N: 23   AGE: not stated   SEX: not stated   INCLUSION: type 2 DM   EXCLSUIONS: not stated
Interventions	TREATMENT: metformin, dosage unclear   COMPARISON: placebo
Outcomes	Plasma xanthine oxidase, thiobarbituric acid‐reactive substance, lactate and frutosamine
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Open‐label, nonrandomised comparative trial   DURATION: 6 months
Participants	COUNTRY: Italy   SETTING: outpatient   Treatment N: 23   Control N: 57   AGE: 56   SEX: 40% men   INCLUSION: Poorly controlled Type 2 DM   EXCLUSIONS: cardiac, liver or renal disease, contraindication to oral hypoglycemic medications
Interventions	TREATMENT: Metformin, dosage adjusted clinically + glibenclamide   COMPARISON: glibenclamide or diet
Outcomes	Basal glucose, HbA1, renal and liver functions, lipids.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	TRIAL DESIGN: Abstract; randomised controlled trial, placebo‐controlled; unclear if single‐blind
Participants	COUNTRY: United States   SETTING: outpatient   Treatment N: 484   Control N: 161   AGE: not listed   SEX: not listed   INCLUSION: Type 2 DM   EXCLUSIONS: none listed
Interventions	TREATMENT: Metformin, dosage adjusted clinically, versus metformin + glyburide   COMPARISON: glyburide or placebo
Outcomes	HbA1
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear