Is the Use of Metformin Associated with Vitamin B12 Deficiency in Hospitalized Patients?
Keywords: Metformin; Cobalamin; Vitamin B; Deficiency; Type 2 diabetes; Neuropathy
After the United Kingdom Prospective Diabetes Study [1], metformin has commonly been used as the first hypoglycemic agent of choice for the treatment of type 2 diabetes as it has been shown to reduce the risk of diabetes related microvascular and macrovascular complications [1]. In addition to decreasing the risk of stroke, myocardial infarction, and death [2-5], metformin has been associated with less weight gain and a decrease in hypoglycemic episodes [1- 5]. Since metformin is extensively used in the diabetic population, its side effects are well documented. Numerous studies report that approximately 10-30% of patients taking metformin experience deficiency and subclinical deficiency among metformin users vitamin B (B malabsorption [2,3,5-14]. The most current and likely 12 12 method of action is through altering the calcium-dependent uptake of B in the distal ileum [2-9,11,14-16] which is supported by the reversal of metformin-induced B deficiency by supplemental calcium [2-4,7- 9,16]. B malabsorption can lead to B deficiency (<148 pmol/L), 12 12 which can have serious complications, including weakness [7,8,17], dementia [7,15,18], confusion [7], difficulty maintaining balance [7], and neurological changes [7,8,15,17-20]. If not adequately treated, B deficiency can result in permanent nerve damage [7]. Symptoms of B deficiency as also seen in those with marginal B status (148- 12 12 220 pmol/L); further, the high folate status commonly seen in folatefortified countries has been shown to further exacerbate symptoms of an impaired B status (<220 pmol/L) [21-23].
The absence of a standardized protocol to assess B status during metformin therapy is problematic as B -related neuropathy may be misread as diabetic neuropathy [15,20] The primary objective of this study was to determine the prevalence of vitamin B deficiency among metformin users in select clinical settings across Southwestern Ontario. The secondary objective was to determine if there is an association between metformin use and B deficiency. Administration of B supplementation among those with a B deficiency was also assessed to determine if these individuals are at risk of developing irreversible neurological damage.
A retrospective chart review was conducted in 3 university-affiliated hospitals, including 2 acute-care hospitals and a rehabilitation and complex care hospital in Southwestern Ontario. Electronic charts of patients aged 18 and older that were admitted to emergency, neurology, critical care and trauma, and spinal cord and stroke rehabilitation units between January 1, 2010 and December 31, 2012 were examined. These settings were chosen, as they are settings where patients with neurological impairments and/or damage are more likely to present. Sample size for categorical data was calculated to ensure 10-20 events per variable (EPV) [24]. Recording data on 20 variables required a minimum of 200-400 charts to be reviewed. Through random sampling, a total of 710 charts were analyzed.
A data abstraction tool was created for this study based on measures of interest. The primary measure was B status as measured by serum B . There is some inconsistency with regards to cut-offs for Vitamin B deficiency, however, more recently accepted cutoffs are 148 pmol/L for deficiency and between 148-220 for marginal or subclinical deficiency status [25,26]. Secondary variables were metformin use, dose, and duration. Other variables of interest included hospitals, units of admission, sex, age, reason for admission, signs of B deficiency (B ) signs of deficiency were confirmed 12 12 through having at least one symptom of either confusion, dementia, paresthesias, bizarre behaviour, fall, ataxia, and/or hallucination), RBC folate, mean corpuscular volume (MCV), hemoglobin, serum iron, albumin, B supplementation, iron supplementation, folic acid supplementation, multivitamin/multimineral supplementation, and pernicious anemia.
Before data collection began, 3 trained research assistants (RAs) pilot tested the data abstraction tool to ensure it was reliable. A reference manual was created to standardize data collection. Interrater reliability was measured as a percentage of agreement between the 3 RAs for 3 different patient charts. An inter-rater reliability score of 92% was achieved, meeting the recommendations of at least 80% for important variables [27]. Next, the RAs recorded the measures of interest from patients’ electronic charts. If a patient had multiple admissions during the specified time frame, the first admission was used to record the reason for admission and lab data. Charts with missing data will be treated by the maximum likelihood strategy. In this strategy, the absence of a “yes” response results in a “no” response (i.e. if metformin use is not listed anywhere in the chart, it will be 001-005.
recorded as “no metformin use”) [27]. Data was recorded on the data abstraction tool, which was stored on an encrypted USB.
Means and standard deviations were calculated for continuous variables to describe the study population (Table 1). Mann-Whitney U test was used to determined significance for these variables, all of which were not normally distributed (as determined by the Shapiro-Wilk test of normality). Cross-tabulation and the ChiSquare or Fisher’s Exact test were used to determine associations for dichotomous variables. Any relationship with a p value <0.05 was deemed statistically significant. Frequency was used to determine the prevalence of B deficiency among metformin users. Serum methylmalonic acid (MMA) is a more sensitive and specific marker than serum B [7,17-20] since it indicates a metabolic change specific to B deficiency [7]; however, MMA labs were not routinely performed in these hospitals. Other studies have demonstrated that some individuals within the marginal or subclinical deficiency range for serum B value are B deficient with the use of MMA [28,29].
12 12 Statistical analyses were performed using SAS v 9.4.
The sample consisted of 106 individuals on metformin and 552 individuals not on metformin. The participant characteristics are shown in (Table 1). Among metformin users, 51% were female, with an average age of 73 years, approximately 54% showed at least 1 symptom of B deficiency, and 37% were supplemented with varying doses of B . Mean serum B was not significantly different between 12 12 metformin users and non-metformin users (413 ± 273 pmol/L vs. 383 ± 204, p=.2725). There were no significant differences in sex, age, signs of B deficiency or in concentrations for RBC folate, MCV, hemoglobin, serum iron, creatinine or albumin. There were no differences between metformin users and non-metformin users for medication and supplement use. However, when looking at differences for medication and supplement use within metformin users, there are significant differences (Table 1). Of those taking Metformin, a significantly higher proportion were also taking PPIs compared to those who were not (66% vs 34%, P=0.0010) but the reverse was true for H2 blockers. The majority of metformin users did not use any form of vitamin or mineral supplement.
There was information on B status for 391 patients. The overall rate of impaired B status (< 220 pmol/L) among the entire patient population with B data was 18.6% (n=391) and 24% among metformin users (n=78, P < 0.0001). For non-metformin users, the overall rate was 17%. The overall prevalence of B deficiency and subclinical deficiency among metformin users versus non-metformin users was 0.5% vs 4% and 4% vs 10% respectively (P=0.1094) (Table 2). Looking at only metformin users, we can see that the majority of users are classified as having good B status (> 220 pmol/L, P=0.0001).
A total of 220 patients were taking some form of B supplement (n= 658, P=<0.0001). Of these 18.6% were either B deficient or sub clinically deficient and only 66% of those who were either deficient or subclinically deficient were taking a B supplement. The information on B supplement use among patients taking metformin is shown in (Table 3). Only 42% of metformin patients were taking a B supplement. Overall, 11 (14%) patients with either B deficiency or subclinical deficiency status were taking a B supplement versus 8 (10%) patients also with compromised status who were not taking a supplement (P=0.1138).
Abbreviation: SD: Standard Deviation; RBC: Red Blood Cell; MCV: Mean Corpuscular Volume; PPI: Proton Pump Inhibitor.
a B cut-offs defined as < 148 = deficient, 148-220= marginal status, >220 = normal range.
Differences in supplement use by hospital location were also assessed for metformin users (Table 4). Both patients who were 001-005.
classified as being B deficient (<148 pmol/L) were both located at 1 hospital and both were taking B supplements. There is a trending towards significance for B supplement use by hospital location among those with subclinical B deficiency status with one hospital more likely to recommend use of supplements than the other two locations (P=0.0618).
It has been well documented that 10-30% of metformin users experience B malabsorption [2,3,5-14]; however, there is minimal research on the prevalence of B deficiency or subclinical deficiency among metformin users. This study found a 24 % prevalence of either B deficiency (≤148 pmol/L) or subclinical deficiency status (<220 pmol/L) among metformin users compared to 17 % for nonmetformin users. This value is similar to a previous study, which reports a prevalence rate of 23% in metformin users [15]. or to a study by Pflipsen, et al. [20] which reports a prevalence of 22% among individuals with type 2 diabetes. Despite the higher prevalence rates of deficiency within the metformin population compared to the general patient population in this study, the results were not statistically significant perhaps due to the overall low rate of metformin users and lack of B data.
supplementation among metformin users by hospital location.
Table 4: Prevalence of B
12 12 Site 1 Site 2 Y N Y < 148 - - 2 148-220 0 3(18) 1 (6) >220 3 (5) 7 (12) 1 (2) Y: Yes, N: No.
Approximately one-third of the entire patient population were taking a B supplement. In metformin users, the two patients who were B deficient (< 148 pmol/L) were taking a B supplement 12 12 (although there is no record of the actual recommended dosage) but it was even chance as to whether or not those with subclinical B deficiency would be taking a B supplement. Evidence suggests 12 12 that larger doses [3] and longer durations [3,4,14] of metformin use are risk factors for B deficiency. The retrospective nature of this study prevented dose and duration of metformin from being adequately assessed. There also seemed to be a slight difference in the recommendations for B supplementation in metformin users in those with subclinical B deficiency by clinic site with the results of this study trending towards significance. This may be related to the lack of a standard cut-off value and different institutions may use a different cut-off. Or this variation in treatment may be due to lack of awareness that neurological impairment and other symptoms of lack of B may occur in those with subclinical B deficiency if MMA tests 12 12 are not used to test for functional status.
If individuals with a B deficiency do not receive adequate treatment in a timely manner, neurological damage may become irreversible. Treatment options include discontinuing metformin and managing type 2 diabetes through an alternative oral drug or administering B supplementation, either orally or intramuscularly [13]. Tissue B deficiency can occur without serum B deficiency [20]; 12 12 therefore clinicians should consider screening for B using a higher cut-off value to identify all patients at risk or requesting functional markers of B status like MMA. In summary, it is recommended that all patients who are taking metformin, be screened for possible vitamin B deficiency including blood work for serum methylmalonic acid and serum B . If the patient is deemed to be B deficient, using 12 12 serum B of <220 pmol/L, B supplementation should be started 12 12 immediately. All patients should also be monitored weekly until levels reach normal values. This protocol should be implemented in all health care facilities.
The retrospective nature of this study design poses a limitation. Data were collected from electronic records, which lacked information on drug and supplement use, dose, and duration before admission. Therefore, metformin dose and duration could not be reliably analyzed as risk factors for B deficiency. However, the primary objective of determining the prevalence of B deficiency in various clinical settings was sufficient with this study design.
Another limitation of this study was the sole use of serum B concentration, without the use of serum methylmalonic acid to confirm deficiency. There is no consensus as to whether serum B concentration accurately reflects storage levels [3,6].
001-005.
Site 3 N Y N - - - 2 2 (12) 8(47) 3 (18) 17 0.0618 4 (7) 18 (30) 26 (44) 59 0.5732
B deficiency among metformin users has been well documented. The high prevalence of deficiency in various clinical settings in this study suggests that a protocol for B screening is absolutely necessary. All individuals on metformin therapy should receive annual screening for B deficiency. Individuals who are at risk should receive supplementation to reduce the possibility of developing serious complications associated with B deficiency. Further research is necessary to determine if metformin dose and duration are significantly associated with B deficiency in various clinical settings.
All authors were involved in aspects of the study design, data collection, analysis and interpretation. JM, BH and Co’C were involved in the writing of the manuscript.
Egg Farmers of Canada Research Committee and Dietitians of Canada Gerontology Network provided funding to complete this study.