Reasons for Metformin Non-Use in Type 2 Diabetes Mellitus in a Hospital: A Retrospective Observational Study

Objective: To investigate and analyze the reasons for metformin non-use in a hospital. Methods: Research-related non-probability physicians and patients lled questionnaire. Results: Physicians’ main inuencing factors were severe liver and kidney dysfunction; gastrointestinal adverse reactions and 11 other causes. Secondary factors.included the appearance of hypoglycemia and other adverse reactions (e.g., skin rash) . Patients’ main reasons included: worry about drug’s inuence on liver and kidney function, gastrointestinal adverse reactions, hypoglycemia, and further weight loss, etc. Statistical analysis showed metformin has certain effects on the mean blood glucose and the mean glycated hemoglobin levels. Conclusion: Based on the research results, we can design more targeted medication education programs.


Introduction
Diabetes affects 2% of the population, consumes 5% of the health service budget, creates appreciable interference with aspects of patients' lives, such as employment, and thereby, poses a signi cant burden to individuals and society. In 2017, the annual cost incurred by patients diagnosed with diabetes was estimated at $327 billion, including $237 billion in direct medical costs and $90 billion in reduced productivity. After adjusting for in ation, the economic costs of diabetes increased by 26% from 2012 to 2017 [1], and this increase is attributed to the increased prevalence of diabetes and the increased cost per person with diabetes [2]. The prevalence of diabetes in China, a country with a vast population, increased from 0.67% in 1980 to 10.4% in 2013 [3].
Scienti c and technological developments have improved the understanding and diagnosis of diabetes. Guidelines for diabetes prevention and treatment recommend adherence to the principle of combining prevention and treatment strategies, based on evidence-based medicine, and focusing on the applicability and practical value of clinical interventions. These guidelines play an important role in standardizing clinical medical practice and improving the current situation of diabetes prevention and control in China. The practice guideline updates are crucial to understand the emerging international trends in the prevention, diagnosis, monitoring, and treatment of diabetes and its chronic complications as well as the new evidences for practicing evidence-based medicine. Therefore, it is important that health services, especially primary care institutions, follow the latest guidelines.
The 2017 edition of the type 2 diabetes prevention guidelines of the Chinese Diabetes Society and the 2018 edition of guidelines by the American Diabetes Association recommended changes in the indications for glucose-lowering drug selection with regard to metformin should be updated throughout the treatment course unless there are contraindications; thus, the 2017 treatment guidelines expanded the range of application of metformin [1,3,4,5]. Many large clinical studies, including the UK Prospective Diabetes Study (UKPDS), have revealed that metformin is a multi-mechanism glucose-lowering agent with adequate e cacy [6,7,8,9] that is currently a rst-line treatment for overweight or obese patients with type 2 diabetes. Metformin can not only inhibit the decomposition of liver glycogen and improve insulin resistance safely and effectively in lowering blood sugar [10], but also has other effects, such as controlling body weight [11], improving polycystic ovary syndrome [12], inhibiting tumor growth [13,14,15], regulating the lipid pro le and blood pressure [16,17], reducing hypercoagulability, improving vascular reactivity, and reducing the incidence of and mortality from cardiovascular diseases [18,19].
This research was conducted for the screening of eligible physicians and patients and used a research questionnaire for ascertaining real-world metformin usage in primary care hospitals. The study aimed to evaluate the causes for metformin non-use through an analysis of the causes identi ed with Pareto law analysis and to reveal whether the use of metformin strictly complies with the updated latest guidelines and instructions. We aimed to provide supporting data for clinical pharmacists to focus on the development of medication education programs for the relevant physicians and patients with regard to metformin use. Simultaneously, the study was intended to eliminate relevant misinterpretations of this drug usage by physicians and patients to promote the rational and standardized use of this drug, improve medication compliance, and maximize the clinical effect of this drug.

Objectives
In this non-probability sampling survey, we used a basic-level hospital HIS (Hospital Information System) to obtain diabetes-related prescription information for all patients with type 2 diabetes who were treated from January 1, 2018 to December 31, 2018. Metformin usage was retrospectively analyzed, and physicians whose prescriptions did not adhere to the guidelines for investigations and patients with diabetes who had unused metformin prescriptions were evaluated to ascertain the reason for the deviation from the guideline.

Study Subjects
This retrospective study was approved by the Ethics Committee of Beijing Changping District Hospital of Integrated Traditional Chinese and Western Medicine and included clinicians who dealt with type 2 diabetes and all patients with type 2 diabetes who were treated at the study center during the study period. The study inclusion and exclusion criteria are described in Box 1. The stripping and shedding criteria were as follows: adverse events, lack of e cacy, loss of follow-up, automatic withdrawal of subjects, and incomplete data that could affect the e cacy evaluation. 2. Information on metformin use: Whether metformin was used and the dose and prescribing frequency.
3. Patient's health status: Blood glucose control, complications, and liver and kidney function.
Compared the glycemic control of metformin users and non-users and the metformin use and non-use among patients with type 2 diabetes.

Statement
All methods were performed in accordance with the relevant guidelines and regulations. All participating patients in this study signed the informed consent form and lled out the questionnaire.

Questionnaire development
Data from a mixed-methods systematic review was used to investigate the reasons why physicians and patients were not using metformin. The physicians' questionnaire included four sections: 1. Demographic information, including the participants' age, sex, number of years quali ed, and the specialty, 2. Participants' description of situations wherein metformin was not used for treating type 2 diabetes. 3. Participants' knowledge of and willingness to understand the latest version of the guidelines and instructions. 4. Participants' analysis of why they did not receive the latest version of guidelines and instructional knowledge update as well as preferences for how they wished to obtain relevant information. Sections 1 and 3 used closed questions to collect quantitative data. Sections 2 and 4 contained two open-ended questions to allow the participants to answer without restriction. The in-built survey logic ensured that participants were shown pertinent questions based on their previous answers. The patient questionnaire was divided into two parts: demographic information, including age, sex, years since being diagnosed with type 2 diabetes, clinical diagnosis, test value,of liver and kidney function and blood sugar control and subjective factors underlying the reasons for the patients' failure to use metformin. Prior to the study completion, participants were encouraged to share any additional information that they deemed relevant in order to capture useful insights that were unaddressed elsewhere in the questionnaire. The nal versions of the questionnaires can be found in the Online Supplementary File 1 and File 2.

Data storage
All electronic data were stored in password-protected computer les that were only accessible to study investigators. Participants who disclosed personal details were additionally protected by coding in data les, and these password-protected les will be retained for 10 years, in conformance with ethical policies.

Data analysis
Pareto's law was applied in cause analysis to identify the main, secondary, and general in uencing factors. The speci c method involved calculating the number and percentage of cause distributions and then sorting them from high to low. A cumulative percentage ≤80% indicated the main reason, which is represented by "A", the secondary reason, with a cumulative percentage of 80%-90% represented by "B", and the general reason with a cumulative percentage >90% represented by "C" [20].
Demographic data was tabulated, and a primary descriptive analysis of the data was performed using IBM SPSS Statistics, V.26.0. Count data was presented as percentages, the measurement data was expressed as average value, and multivariate logistic analysis was used for regression analysis. P < 0.05 was considered statistically signi cant. Cronbach's coe cient was used to evaluate the internal consistency of the questionnaire. Factor analysis was used to evaluate structure validity. [21].

Questionnaire results and analysis of physician surveys
A total of 111 clinician participants completed the questionnaire survey, most of them are young and middle-aged physicians. 52.25% of the physicians surveyed had read the latest guidelines, and 63.06% understood the clinical status of metformin has been further improved. Endocrinologists had a better understanding of this drug and their adherence to the latest version of the guidelines was better than that of physicians in other specialties. Correlation analysis, regression analysis and F test show that there is no signi cant correlation between sexes age division, physician level, years of experience and the reasons why doctors did not choose metformins. Table 1 shows that, physicians consider severe liver and kidney dysfunction, gastrointestinal adverse reactions, severe infection and trauma, etc. 11 factors as the main factors to consider stopping metformin. In addition, the appearance of hypoglycemia and other adverse reactions were secondary factors. Alcoholism, vitamin B12, and uncorrected folate de ciency are the general factors.
A few physicians considered older age, and stable chronic heart failure, with regular cardiac and renal function examinations, in their decision making.

Results and analysis of patient questionnaire
A total of 10,508 patients participated in the survey, and 3730 of these patients completed the questionnaires.
From Table 2, we can infer that the main reasons for refusal to use metformin included concerns about the in uence of the drug on liver and kidney function, gastrointestinal adverse reactions, hypoglycemia, etc. Secondary factors were forgetting to take medication, over-interpreting medicine speci cations, and obsessing that new drugs are better than old ones. The general in uencing factors included the fear that drugs may affect planned conception and di culty in purchasing the drug locally, etc.  An independent sample t-test shows the mean blood glucose level showed no signi cant difference (P>0.05) between subgroup but the hospitalized samples showed a signi cant difference in the mean HbA1c (P<0.05). Speci c analysis showed that the signi cance of hospitalization for the mean HbA1c was 0.01 (T =3.928, P =0.001), and the speci c comparison showed that the mean HbA1c value of outpatients (8.53) was signi cantly higher than that of inpatients (8.38).
The in uence of hospitalization and metformin on mean blood glucose level and mean HbA1c were studied using a two-way analysis of variance. Hospitalization did not show a signi cant difference (F=1.408, P = 0.235) in mean blood glucose level but metformin showed signi cant effects (F=1441.405, P = 0.001), indicating that the drug will affect the difference of mean blood glucose. Hospitalization was signi cant (F = 15.994, P = 0.001) in mean HbA1c, indicating that whether hospitalization will cause a difference in the mean HbA1c. Metformin showed signi cant effects (F = 15.936, P = 0.001), indicating that the main effect exists and that metformin will cause a difference in the mean HbA1c. Speci c differences were analyzed by single-factor ANOVA.

Discussion
The physicians participating in the survey are mainly young and middle-aged and the attending doctors with 5-14 years of experience. The ones with the above characteristics will be the main targets for clinical pharmacists to deliver the latest relevant guidelines. Related analysis and regression analysis revealed that departments, sexs, age divisions, physician level, and years of employment did not have an impact on the reasons why doctors did not choose metformins.. Therefore, it was not possible to speci cally analyze the impact of independent variables on dependent variables. With regard to the patient samples, the main reasons for refusing to use metformin included the following: concern about the in uence of the drug on liver and kidney function, gastrointestinal adverse reactions, hypoglycemia, and further weight loss, etc. As we all know, metformin has certain requirements for liver and kidney function but does not cause further damage and reasonable weight loss can improve insulin resistance to a certain extent. The results show that the patients lacked reliable access to scienti c knowledge of medicine. Therefore, clinical pharmacists should guide and educate patients with type 2 diabetes according to the above-speci ed medication errors in order to promote the patients' deep scienti c understanding and reasonable use of this drug.
Patients with gastrointestinal adverse reactions can be made to develop tolerance for the adverse reactions by initiating low metformin doses and gradually increasing the drug dose or by using extended-release formulations [22,23].
The independent samples t-test revealed that there is no signi cant difference in blood glucose levels but a signi cant difference in the mean HbA1c between inpatient and outpatient treatment.
A two-way analysis of variance indicates that metformin has certain effects on the mean blood glucose and the mean HbA1c.

Strengths and Limitations
This is the rst study to investigate the perceptions of physicians with regard to the reasons for metformin not being used among patients with type 2 diabetes. Moreover, alongside the data from physicians presented in this article, this study provides an overview of the patients' reasons for refusal to use metformin. In the context of the changes to the guidelines and speci cations (the expanded scope for the application of metformin and its improved clinical status), it is particularly interesting to have conducted this study to see how district hospitals are implementing the latest version of the guidelines and speci cations.
It was not possible to contact all physicians and patients due to the privacy policy. Physicians who resigned from the hospital before starting the survey would have been unaware of the questionnaire. Furthermore, some patients who came to the hospital infrequently and left no contact information would have been di cult to contact to obtain valid questionnaire information, and some patients, for various reasons, were unwilling to cooperate to complete the questionnaire. It is impossible to determine the accuracy of information, such as the course of diabetes, in the case of some psychiatric patients who cannot communicate normally, and their medical history can only be narrated by their family members, therefore, the risk of bias remains unknown and should be considered when interpreting the results.

Conclusion
There is room for improvement for physicians, especially those who are not endocrinologists, to better align themselves with the latest guidelines for the management of patients with type 2 diabetes. Accordingly, we advise that health administration departments and hospital leaders should increase the number of more relevant knowledge training workshops to promote more standardized knowledge and to increase the input of information construction with literature download as the main body to update diagnosis and treatment behavior. From this survey, it can be clearly seen that clinical pharmacists, who constitute a bridge for information transmission, should not only serve the endocrinology department, but also promote endocrine-related pharmaceutical care to non-endocrine departments with patients who have diabetes. In terms of drug understanding of patients with type 2 diabetes, due to the lack of medical literacy, there is still some misunderstanding about metformin; thus, there is still a long way to go for physicians, and clinical pharmacists need to do to further correct and guide patients to correct the understanding bias to use drugs more scienti cally and rationally.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementarymaterial.doc