Abstract
Background:
The purpose of this study was to identify the reasons why metformin prescribing is suboptimal.
Methods:
Two semi-structured focus groups with attitudinal questionnaires and a brief educational presentation were held in two US cities. Participants included providers (physicians, pharmacists, midlevel practitioners) caring for patients with type 2 diabetes mellitus (T2DM) in an ambulatory setting. Outcome measures included provider attitudes, behaviors and opinions regarding the use of metformin.
Results:
Participants identified three main themes influencing the use of metformin, including the appropriate timing of metformin initiation, known risks associated with metformin, and procedures to manage safety concerns and mitigate adverse effects associated with metformin. Participant prescribing behaviors of metformin were not consistent with the best available evidence in the settings of renal insufficiency, heart failure, hepatic dysfunction, alcohol use, and lactic acidosis. With minimal education, provider prescribing behaviors appeared to change by the end of the focus group to align more closely with the best available evidence.
Conclusions:
Provider attitudes, behaviors and opinions regarding the use of metformin for T2DM reveals the need for further education to improve appropriate use of metformin. Educational interventions should target prescribing behaviors and opinions identified to be inconsistent with the evidence.
Introduction
Metformin is the only recommended first-line medication for type 2 diabetes (T2DM) because it is the only oral diabetes medication that has reproducibly demonstrated reductions in both cardiovascular morbidity and mortality [Turner, 1998]. However, metformin prescribing is suboptimal. Only 65% of newly diagnosed T2DM patients [Desai et al. 2012] and 25% with ongoing T2DM are prescribed metformin [Cohen et al. 2003]. Reasons for suboptimal prescribing are currently unknown, but may be due to fear of precipitating lactic acidosis in certain at-risk patients. Patients with renal insufficiency, heart failure, chronic obstructive pulmonary disease (COPD), hepatic insufficiency, history of lactic acidosis, and those who abuse alcohol or are of older age are perceived to be at risk for lactic acidosis. Another potential reason for suboptimal prescribing of metformin may be fear of legal implications if prescribing contradicts manufacturer labeling. The current product labeling includes contraindications and precautions for metformin use in patients at-risk for lactic acidosis. In the United States (US), the only stated contraindication to metformin is renal function based on serum creatinine thresholds.
It is perceived that metformin decreases lactate metabolism and thus results in lactic acidosis. However, studies have consistently demonstrated no increase in serum lactate concentrations with metformin use [Cryer et al. 2005; Bolen et al. 2007; Salpeter et al. 2010]. Further, epidemiological studies suggest the risk of developing lactic acidosis is 4.3 per 100,000 patient-years in patients with T2DM who are taking metformin and 5.4 per 100,000 patient-years in patients with T2DM who are not taking metformin [Salpeter et al. 2010]. Despite a lack of evidence supporting the precautions or contraindications to metformin use and the overwhelming evidence supporting metformin’s benefits, 85% of providers do not prescribe metformin in the presence of a precaution or contraindication [Ricci et al. 2009].
Understanding the reasons why prescribing of metformin is suboptimal is needed to implement targeted interventions that promote appropriate use of this evidence-based first-line treatment for T2DM. Therefore, the purpose of this study is to identify reasons why metformin prescribing is suboptimal.
Methods
This mixed methods study involved two focus groups that utilized grounded theory and a modified nominal group technique to enhance participation and limit response bias [Van de Ven and Delbecq, 1974; Delbecq and Van de Ven, 1971]. In addition, a brief educational presentation on evidence of risk of harm associated with metformin was provided to assess for change in perceived prescribing behaviors after the presentation [Trinkley et al. 2015]. Each focus group was 90 minutes and was conducted in Arizona and Colorado, US in 2014. Each group included 5–9 participants, which is in accordance with standards of good focus group conduct [Kitzinger, 1995]. Each session included the following 7 components: (i) informed consent; (ii) prefocus group questionnaire; (iii) a formal moderated discussion guided by clinical case scenarios; (iv) educational presentation; (v) postfocus group questionnaire. This study was approved by the University of Colorado’s Institutional Review Board. Informed consent was provided by each participant at the beginning of each focus group.
The focus groups were led by trained facilitators using a moderator guide with an investigator taking field notes. The moderator guide did not change over the course of the study. The facilitators and observers were pharmacists with training in qualitative research and knowledge of the content to be discussed. Sessions were audio-recorded for transcription to identify key issues and comments. All transcriptions were anonymized.
Focus group participants
Focus group participants were a convenience sample of clinicians identified by the researchers based on ongoing professional relationships. Participants were selected to represent broad clinician characteristics including: discipline (physician assistant, nurse practitioner, pharmacist, physician) and practice setting (primary care, endocrinology, community, academic) to ensure adequate representation from settings that commonly prescribe metformin for T2DM. Invitations to participate were sent via email and each participant received a nominal Visa gift card to partially compensate them for their time. Participants were required to be present for the entire focus group session.
Focus group conduct
Prior to the discussion component of the focus group a presurvey was completed by each participant. The pre-focus group questionnaire included 11 demographic and attitudinal questions. The six attitudinal questions asked participants to describe their attitudes towards prescribing metformin for patients with T2DM and also to describe concerns or issues regarding metformin and the risk of lactic acidosis, including two, five-point Likert scale questions (1 = strongly disagree, 5 = strongly agree), two multiple-choice and two open-ended questions.
Focus group moderated discussion
A total of seven clinical case scenarios followed by open-ended questions were used to assess provider behaviors and opinions regarding metformin prescribing, including (i) perceived risk of lactic acidosis with and without the presence of risk factors for lactic acidosis; and (ii) optimal dosing of metformin. The clinical case scenarios were adapted from a survey used to assess provider attitudes for prescribing metformin [Goldberg et al. 2015]. Each provider was asked to independently identify their treatment strategy for the seven clinical cases from a list of options using audience response technology (Turning Technologies, LLC). Following each individual clinical case, open-ended questions were used to facilitate group discussion regarding rationale for their responses. Open-ended questions were also used at the end of the seven clinical case scenarios to address additional topics that did not come up during the case scenarios, such as legal liability. Responses to the open-ended questions were used to further assess participant responses, awareness and degree of concern of lactic acidosis, and knowledge of contraindications to metformin use.
Postfocus group questionnaire
At the conclusion of the moderated discussion, the participants were presented with a brief (10 minute), evidence-based educational review regarding the use of metformin, including the risk of lactic acidosis with or without risk factors for lactic acidosis. The educational presentation included the incidence of lactic acidosis in a large cohort of 109,656 patients with T2DM and estimates of the association between metformin use, lactic acidosis, and risk factors for lactic acidosis [Trinkley et al. 2015]. Reactions to the information were assessed with the seven-item post-focus group questionnaire. The post-focus group questionnaire asked participants to individually answer the same seven clinical case scenarios they answered during the moderated discussion.
Analysis
Accuracy of each focus group transcription was verified by an independent study investigator who reviewed at least 20% of the transcription and validated by comparison with field notes. Transcriptions and questionnaires were categorized using a common data dictionary. Using a thematic approach, descriptive coding, topic coding, analytical coding, and postcoding were used to categorize the data. Descriptive coding was used to categorize participant demographic characteristics that were collected from the pre-focus group questionnaire. Topic coding included a general categorization followed by recoding to incorporate more specific subcategories. General categorization was based on the seven clinical case themes and recoding was performed to identify concepts discussed within each of the seven themes to provide more depth into reasons for treatment decisions. Analytical coding was used to interpret and assess the meaning of responses to better understand reasons for treatment decisions. Postcoding was used to quantify information when appropriate. Throughout the coding, the literature was consulted to see how the information generated from the focus groups compared. Using ATLAS.ti (ATLAS.ti Scientific Software Development GmbH) for the thematic analysis of topic and analytical coding, codes were assigned to all categories of data. Initial responses to the seven clinical cases were compared with the responses in the post-focus group questionnaire. Quality assurance was conducted on at least 20% of the data via recoding and consistency checking by an independent investigator.
Results
Pre-focus group questionnaire
Participant characteristics
There were 14 participants, 5 in the Arizona focus group and 9 in the Colorado focus group. Participants included 5 physicians, 5 midlevel providers and 4 pharmacists, with practices including community, private practice, academia, urban and a federally-qualified health center. Table 1 describes the demographic characteristics of the participants.
Demographic characteristics of participants (N = 14).
BCACP, Board Certified Ambulatory Care Pharmacist; BCPS, Board Certified Pharmacotherapy Specialist; CDE, Certified Diabetes Educator; DO, Doctor of Osteopathy; MD, Medical Doctor.
Attitudes regarding use of metformin
All participants reported they strongly agree (12 of 14 participants) or agree (2 of 14 participants) with the statement that “metformin is the drug of choice for patients with newly diagnosed T2DM.” However, 9 of 14 participants agreed or strongly agreed with the statement that “metformin is used optimally in practice today,” while 1 participant had a neutral response and 4 disagreed. Although most felt metformin was optimally prescribed, 12 of 13 (1 participant did not answer) felt “metformin was under-prescribed,” and only 1 felt it was over-prescribed. When asked to select an option that best estimates the risk of lactic acidosis with metformin use, 7 of 15 selected an incidence of 1 per 1,000,000 patient-years, 6 selected one per 100,000 patient-years, and 1 selected 1 per 1000 patient-years.
Reasons for under-prescribing metformin
When asked for reasons why metformin is under-prescribed, participant responses were categorized as: (i) metformin was not initiated early enough for T2DM or prediabetes; (ii) metformin was stopped prematurely with changes in renal function, including increased serum creatinine and presence of albuminuria; (iii) providers may fear potential adverse effects, including lactic acidosis and gastrointestinal intolerance; and (iv) metformin should be stopped when insulin is started. The one participant who stated metformin was over-prescribed felt that providers may continue metformin despite increases in serum creatinine.
What providers need to know about metformin
Participants felt providers needed additional education on metformin and common misconceptions regarding its use. Common themes identified included: (i) when to start metformin given specific patient characteristics such as hemoglobin A1C measures, obesity and diagnosis of prediabetes; (ii) actual risk of lactic acidosis with metformin use; (iii) when to adjust or stop metformin for renal function, specifically relying more on glomerular filtration rate (GFR) cut offs versus serum creatinine; (iv) dosing principles to avoid and mitigate gastrointestinal (GI) intolerance and achieve effective doses; and (v) the comparative advantage of metformin over other glycemic medications. In general, participants felt there were misconceptions about the actual risk of lactic acidosis and GI intolerance with metformin use.
Focus group moderated discussion
There was a high-level of agreement among participants with nine overarching themes discussed. Interestingly, participant individual responses to clinical case scenarios at times contrasted with verbal responses to the group discussions. See Table 2 for participant’s responses to the clinical case scenarios.
Participant responses to clinical case scenarios: metformin prescribing behaviors of participants (N = 14).
A1C, hemoglobin A1C; ALT, alanine aminotransferase; AST, aspartate transaminase; CKD-EPI eGFR, Chronic Kidney Disease Epidemiology Collaboration estimated glomerular filtration rate; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 second; INR, international normalized ratio; NYHA, New York Heart Association; T2DM, type 2 diabetes mellitus.
The nine overarching themes are described below and include use of metformin in the settings of kidney insufficiency, heart failure, hepatic dysfunction, COPD, alcoholism, current or historical lactic acidosis, and metformin dosing and contraindications.
Kidney insufficiency
There was consensus that renal function was an important consideration when prescribing metformin, with more severe renal impairment being most concerning. Responses to the clinical case scenarios revealed that all participants would adjust metformin therapy based on either serum creatinine or GFR thresholds, with 10 of 13 indicating they would base their decision on GFR versus serum creatinine (Table 2). Most participants stated that they do not calculate GFR manually because it is included in standard laboratory reports, but one individual stated they would double check accuracy. Participants stated they typically evaluate serum creatinine at annual preventative health visits, but that they would not evaluate serum creatinine more frequently in patients prescribed with metformin.
Heart failure
Half of participants felt heart failure was an important consideration when prescribing metformin. A total of 7 of 14 participants indicated they would adjust metformin therapy based on varying severity of heart failure (Table 2) in response to the clinical case scenarios. Half of participants stated they would adjust metformin if heart failure was unstable or was a primary reason for hospitalization, while the others stated they would not alter metformin therapy based solely on heart failure-related factors. However, all participants agreed that they would not change the frequency of assessing heart failure in patients taking metformin.
Hepatic dysfunction
Approximately half of participants stated hepatic function was moderately important when prescribing metformin, with two stating they would only be concerned with advanced or end-stage hepatic dysfunction. The remainder of participants stated during the discussion that hepatic function played little importance in their decision to prescribe metformin. However, during the clinical case scenarios, all but two of the participants indicated they would adjust metformin therapy for varying degrees of hepatic dysfunction (Table 2). Participants indicated they would not assess hepatic function differently for patients based on metformin use.
Chronic obstructive pulmonary disease
Participants unanimously agreed that COPD or forced expiratory volume measurements have no bearing on their decision to prescribe metformin and that they would not adjust or stop metformin based on any clinical indicators of COPD. Only one participant stated they would adjust metformin therapy for advanced COPD in the clinical case scenarios (Table 2).
Alcoholism
Half of participants felt alcohol use was at least somewhat important when considering metformin. Overall, 9 of 14 participants responded to the clinical case scenarios indicating they would not adjust metformin therapy for any degree of alcohol use/abuse (Table 2). All participants stated they would not assess alcohol use more frequently in a patient taking metformin; however, they otherwise already assessed alcohol use frequently as standard clinical practice.
Current or historical lactic acidosis
Participants stated they would not screen patients with T2DM or those prescribed sulfonylureas for history of lactic acidosis, but two indicated they would screen for history of lactic acidosis when considering metformin. For the clinical case scenario of a patient with a history of lactic acidosis, 5 of 14 participants stated they would regularly prescribe metformin, while the others indicated they would either not prescribe metformin or would prescribe it with more conservative dosing or monitoring (Table 2).
For the clinical case of current lactic acidosis, all participants felt it was an important consideration for the use of metformin and would stop metformin in the presence of lactic acidosis. However, 2 of the 14 participants indicated they would not stop metformin, rather they would decrease the dose of metformin or monitor serum lactic acid concentrations more frequently (Table 2). One participant noted liability concerns are influential factors in their decision to stop metformin. Participants all agreed that they would not screen for incident lactic acidosis in someone taking metformin.
Dosing
Unsolicited by the moderator, three participants reported that they attempt to reach a metformin total daily dose of 1500 mg to achieve better glycemic lowering, stating that lower doses likely had some, but minimal, impact on glycemia. When the hemoglobin A1C is <6% for a patient taking 500 mg metformin once daily, participants generally felt it was fine to continue as is, but a few participants suggested they may stop it altogether.
Contraindications
All but three participants felt that contraindicated means ‘do not prescribe’ under any circumstance. These three participants indicated that “nothing is black and white in medicine” and that situations could exist in which there may be a need to use a medication despite a contraindication.
After the discussion revealed that metformin is contraindicated in patients with certain serum creatinine measurements, participants were split with half indicating they did not agree with the contraindication based on the best available evidence and that they do not necessarily adhere to the contraindication in practice. The other half felt that the contraindication was very important and they would adhere to it, stating they would not want to do something contraindicated or “do the wrong thing.”
Post-focus group questionnaire
After presenting participants with the best available evidence surrounding the use of metformin and the risk of lactic acidosis, all participants again individually answered the clinical case scenarios [Trinkley et al. 2015]. Notable changes in responses to the clinical case scenarios included a decrease in participants who would adjust therapy for heart failure (7 of 14 versus 2 of 14) or hepatic dysfunction (12 of 14 versus 5 of 14), and an increase in participants who would start metformin as usual in someone with a history of lactic acidosis (5 of 14 versus 9 of 14). Table 2 displays the results from the clinical case scenario responses before and after the brief education review.
Discussion
This study indicated that providers who routinely prescribe metformin for patients with T2DM have perceptions regarding the use of this first-line medication that are not always consistent with best available evidence. The participants in this study were identified to broadly represent clinicians that provide ongoing clinical care for patients with T2DM. These study participants were mostly in agreement in their approach towards managing metformin in patients with renal insufficiency, hepatic dysfunction, COPD and those abusing alcohol or with current or historical lactic acidosis. However, participants showed the least agreement in approach to managing metformin in patients with concurrent heart failure. The varied perceptions regarding metformin use in the setting of heart failure were surprising, given the evidence consistently supports the positive benefits of metformin in patients with heart failure [Eurich et al. 2005; Andersson et al. 2010; MacDonald et al. 2010; Aguilar et al. 2011]. Although there was mostly agreement among participants, prescribing behaviors and opinions were in contrast to the best available evidence in the settings of renal insufficiency, heart failure, hepatic dysfunction, alcohol use, and lactic acidosis.
Individual participant responses to clinical case scenarios sometimes differed from verbal statements made during the group discussions, which is likely ‘group think’ that can arise in focus groups. However, during the focus group discussion, there was seldom indication that the participant responses or viewpoints had changed. The influence of ‘group think’ and the introduction of new evidence-based knowledge that occurred just prior to reassessment likely also influenced how participants responded to the clinical case scenarios when they were reintroduced in the post-focus group questionnaire. Participant responses to the clinical case scenarios changed notably in three situations: heart failure, hepatic dysfunction, and history of lactic acidosis. In all three of these situations, participants were more likely to prescribe metformin at the end of the focus group session compared with their initial responses. This change in responses corrected misconceptions and was more aligned with evidence.
This study suggests that misconceptions exist regarding metformin risks that influence prescribing of metformin as per our participant responses. Directed education to mitigate misconceptions is needed to optimize the use of metformin in patients with T2DM. Participant responses identified specific areas in which providers needed additional focused and evidence-based education, specifically in the setting of renal insufficiency, heart failure, hepatic dysfunction, alcohol use, and lactic acidosis. Fortunately, the shifts in participant behaviors and opinions regarding metformin use suggests that minimal interventions may result in improved prescribing of metformin. Achieving such a shift in prescribing behaviors and opinions regarding metformin use among a group who already considers metformin to be a valuable treatment suggests that an even stronger shift in opinion may occur among providers who are less aware of metformin’s value.
There are several interventions that may potentially improve the prescribing of metformin. These could include traditional educational sessions regarding the benefits and risks of metformin in special populations (including the actual evidence-based risk of metformin), or implementation of electronic clinical decision support tools at the point of prescribing. Implementing a clinical decision support intervention within electronic medical records may be particularly effective in special populations because the benefit to risk profile of metformin in several special populations is profound [Turner, 1998; Salpeter et al. 2010; Holstein et al. 1999; Emslie-Smith et al. 2001; Aguilar et al. 2011; Andersson et al. 2010; MacDonald et al. 2010; Eurich et al. 2005, 2013]. Clinical decision support tools can summarize and present clinical information to providers in a manner that increases practice efficiency and quality.
A major limitation of this study is that it is not representative of the larger population of providers who prescribe metformin in patients with T2DM. Although our focus groups represented physicians, midlevel providers and pharmacists from varying practice settings in two different states, other providers may have responded differently. Of note, providers practicing in rural settings were not represented in the focus group, limiting generalizability to this population. Another potential limitation is the possibility that some providers reserved responses for fear of scrutiny from their peers and colleagues; however, all participants were actively engaged and contributed to the discussion. Although there was one participant that was more reserved in speaking up, he did so without hesitation when prompted. Otherwise, there was no dominant voice during the focus groups and participants spoke up equally without argumentative interactions.
Clinicians believe that metformin is an important first-line medication for T2DM that is generally under-prescribed. Our findings and noted shifts in participant responses over the course of the focus groups indicate that misconceptions regarding the use of metformin can be improved with minimal focused and evidence-based provider education. Specific areas identified for further education included when to initiate metformin and actual risks and precautions of metformin use. Based on our focus group responses, additional provider education is also needed to improve understanding of the risks of metformin use in the settings of renal insufficiency, heart failure, hepatic dysfunction, alcohol use, and lactic acidosis.
Footnotes
Acknowledgements
All authors meet authorship criteria and no one who qualifies for authorship has been excluded. All authors contributed significantly to meet the following criteria: (i) substantial contributions to research design, or the acquisition, analysis or interpretation of data; (ii) drafting the paper or revising it critically; (iii) approval of the submitted and final versions.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. This study was supported by a grant from the Skaggs Scholars Program at the University of Colorado, Skaggs School of Pharmacy and Pharmaceutical Sciences.
Conflict of interest statement
The authors declare that there is no conflict of interest.