Correlation between measures of hypoglycemia and glycemic improvement in sulfonylurea treated patients with type 2 diabetes in India: Results from the OBSTACLE hypoglycemia studyS Kalra1, MC Deepak2, P Narang3, V Singh3, A Maheshwari4
1 Department of Endocrinology, Karnal Bharti Hospital, Karnal, Haryana, India
2 Department of Diabetes, Madhav Diabetes Center, Chennai, Tamil Nadu, India
3 Department of Medical Affairs, MSD Pharmaceuticals Pvt Ltd, Lucknow, Uttar Pradesh, India
4 Department of Diabetes, Sri Ramkamal Diabetes and Heart Clinic, Lucknow, Uttar Pradesh, India
Correspondence Address: Source of Support: The study was sponsored by MSD Pharmaceuticals Private Limited,, Conflict of Interest: Deepak MC, Kalra S, and Maheshwari A are clinicians based in India and were investigators for this study. They have received payments for their services for the current study and have also previously received payments against other scientific services provided to MSD Pharmaceuticals Pvt. Ltd., India. P Narang, and V Singh are full-time employees of MSD Pharmaceuticals Pvt. Ltd., India. DOI: 10.4103/0022-3859.132322 Clinical trial registration NCT00907881
Source of Support: The study was sponsored by MSD Pharmaceuticals Private Limited,, Conflict of Interest: Deepak MC, Kalra S, and Maheshwari A are clinicians based in India and were investigators for this study. They have received payments for their services for the current study and have also previously received payments against other scientific services provided to MSD Pharmaceuticals Pvt. Ltd., India. P Narang, and V Singh are full-time employees of MSD Pharmaceuticals Pvt. Ltd., India.
Clinical trial registration NCT00907881
Background: This study aimed to assess correlation between measures of hypoglycemia and glycemic control in patients with type 2 diabetes mellitus (T2DM) treated with sulfonylureas. Materials and Methods: T2DM patients being initiated on a sulfonylurea (SU) on background of a failing oral antihyperglycemic regimen were followed up for 12 weeks. (HbA1c) was measured at baseline and end of follow-up. Hypoglycemia was assessed using Stanford Hypoglycemia Questionnaire at week 12. Results: Of the total 1069 patients enrolled, 950 were considered evaluable. A weak negative correlation was observed between end of follow-up HbA1c values and hypoglycemia score, using both linear regression analysis (correlation coefficient -0.12; P = 0.0002) and negative binomial regression (β slope -0.09; P = 0.0010). A similar correlation was also observed between change in HbA1c from baseline and hypoglycemia score (β slope -0.07; P = 0.0048). Mean HbA1c reduction was lowest (0.65 ± 2.27%) in patients not reporting any hypoglycemia and highest (1.28 ± 2.40%) in patients with hypoglycemia score greater than median of 2 (P = 0.0031). There was no correlation between hypoglycemia frequency and end of follow-up HbA1c values (P = 0.4111). Conclusion: With addition of SU on a background of a failing oral anti-hyperglycemic regimen, the extent of glycemic control correlates directly with measures of patient reported hypoglycemia.
Keywords: HbA1c, hypoglycemia, Stanford hypoglycemia questionnaire, sulfonylurea, type 2 diabetes
Intensive glycemic control, with target glycosylated hemoglobin (HbA1c) of <7%, is a critical component of type 2 diabetes mellitus (T2DM) management and minimizes long-term complications. ,, Recent Indian studies show that only 20.4% of diabetes care physicians set HbA1c of <7% as target for their patients  and only 19.7% of patients achieve this value.  This clinical inertia has often been attributed to presence or fear of hypoglycemia, both among physicians and patients.  It can thus limit the aggressiveness of drug therapy, decrease adherence to diet and reduce patients' willingness to take medications as directed.  Hypoglycemic symptoms may also lead to increased calorie intake and worsening of glycemic control. The primary objective of the present study was thus to assess correlation between glycemic control (as measured by HbA1c) and hypoglycemia symptom score (as measured by Stanford Hypoglycemia Questionnaire) in patients of T2DM, 12 weeks after addition of a SU to an ongoing regime of oral hypoglycemic agent(s).
The study protocol, informed consent document (ICD) and other study documents were approved by a central Institutional Ethics Committee (IEC) for all study centers and written, informed consent was obtained from all participants.
An observational Study to Assess Correlation between Glycemic Control and Hypoglycemia in Patients with Type 2 Diabetes Treated with Sulfonylureas (SUs) (OBSTACLE hypoglycemia study) was designed to assess association between HbA1c and patient reported hypoglycemia as measured using Stanford Hypoglycemia Questionnaire, 12 weeks after addition of a SU to an ongoing regime of oral hypoglycemic agent(s). This was an open-label, multi-centric, prospective observational study conducted across 33 diabetes care clinics from 18 cities. The cities were of varying sizes, spread across 13 states and represented all zones of India. Patients with T2DM of either gender were enrolled if they were aged 18-75 years and were being considered for addition of SU on a background of at least one other oral hypoglycemic agent, as per the clinical judgment of the participating physician. Patients with type 1 diabetes mellitus (Type 1 DM) or gestational diabetes, or those having received insulin or insulin secretagogues (SU, meglitinide, DPP-4 inhibitors) at the time of enrolment or 6 months prior were excluded.
Consecutive patients meeting selection criteria and consenting to participate were enrolled by the treating physicians (study investigators) in this study. HbA1c levels were assessed both at enrolment (baseline, Visit 1) and end of 12-week follow-up (Visit 2). Hypoglycemia was assessed using Stanford Hypoglycemia Questionnaire. This tool was used as a diary card to be filled-in by patients in case if they experience any such symptom anytime during the follow-up period and was also administered by physician at the follow-up visit. The Stanford Hypoglycemia Questionnaire, originally developed and validated in Spanish Language, is a patient friendly tool that helps score hypoglycemia on basis of presence or absence of seven common symptoms within a recall period of one week.  The symptoms include morning headaches, nightmares, night sweats, light-headedness, shakiness or weakness, intense hunger, and loss of consciousness. Scores on this scale can range from 0 to 7, with higher score indicating more hypoglycemia symptoms. For the purpose of this study, the questionnaire was made available in 11 languages (Hindi, English, Urdu, Punjabi, Oriya, Tamil, Telugu, Malayalam, Kannada, Marathi, and Gujarati). Patients reporting at least one of the seven symptoms were considered to have experienced hypoglycemia. These episodes were further classified as mild (little or no interruption of activities and no assistance needed to manage symptoms), moderate (some interruption of activities and no assistance needed to manage symptoms) or severe (needed the assistance of others to manage symptoms or required medical assistance).
The primary endpoint was coefficient of correlation as measured using linear regression analysis for association between the two variables. Other end-points were coefficient of correlation between change in HbA1c from baseline and hypoglycemia score and coefficient of correlation between HbA1c at week 12 and frequency of hypoglycemia. (Author - something is wrong here- co efficient of correlation is a statistic not an end point- please reword)
Correlation analysis between HbA1c and hypoglycemic scores was performed based on Karl Pearson Coefficient of Correlation. Correlation analysis was performed for multiple endpoints. All statistical hypothesis tests were performed at 5% level of significance (two-sided test). Statistical analysis was performed for all enrolled patients. All statistical analyses were performed using SAS ® version 9.2 for Windows (SAS Institute Inc., USA).
Samples size calculation
Keeping 0.70 as the anticipated sample correlation, and 0.74-0.66 as the population correlation, at 5% level of significance and 80% power for a two-sided hypothesis, a sample size of 1138 patients was considered planned
Of the 1069 patients that were enrolled in the study, 950 patients, having the values for HbA1c at week 12 and hypoglycemia score without any major protocol deviation, were considered evaluable for primary analysis. Glimepiride was most commonly prescribed SU (75.3%). Most patients (89.06%) were on single background oral hypoglycemic agent; metformin was the most commonly prescribed background oral hypoglycemic agent (90.27%). Over the mean follow-up period of 14.34 ± 2.80 weeks, mean HbA1c decreased from 8.94 ± 2.27% to 8.01 ± 1.90%. Mean hypoglycemia score at end of follow-up period was 0.98 ± 1.41. The majority (n = 557, 58.63%) of patients had hypoglycemia score of 0, followed by 129 patients (13.58%) with a hypoglycemia score of 1 and 95 (10.00%) patients each with a hypoglycemia score of 3 and 4. Only five patients (0.53%) had the highest observed hypoglycemia score of 6. Baseline characteristics of study patients are described in [Table 1].
Using linear regression analysis, a weak negative correlation was observed between hypoglycemia scores and HbA1c values at the end of study (correlation coefficient -0.12; 95% CI -0.18 to -0.06). This correlation was statistically significant. (P = 0.0002). The data was skewed as the majority of the patients had the hypoglycemia score as '0'. Hence, a negative binomial regression model was used to reassess the correlation and weak negative correlation was confirmed with β slope = -0.09 (95% CI; -0.1468 to 0.0374) and P = 0.0010 [Figure 1]. Subgroup analysis for primary end-point was performed for patient-groups formed on the bases of gender, age, body mass index (BMI), and duration of diabetes. Though a weak negative correlation was consistently observed in each of these sub-groups, statistical significance could be demonstrated only in a few of them, perhaps because of smaller number of patients in many of the sub-groups [Table 2]. Using a negative binomial regression model, we also observed a weak negative, statistically significant correlation between change in HbA1c from baseline to end of follow-up (reduction measured as negative change) and hypoglycemia score with β slope of -0.07 and P = 0.0048 [Figure 2].
We did not observe any correlation between hypoglycemia frequency (measured as hypoglycemia events per 100 patient-years) and end of follow-up HbA1c values (β slope -0.05; P = 0.4111). When grouped on the basis of reported hypoglycemia score, the mean reduction in HbA1c was lowest among patients not experiencing any hypoglycemia (0.65 ± 2.27%), intermediate among patients with hypoglycemia score lower than median score of 2 (1.06 ± 2.15%) and highest among patients with hypoglycemia score greater than median of 2 (1.28 ± 2.40%). The difference between the two extreme groups was statistically significant (P = 0.0031).
A binomial logistic regression was built to assess relation of hypoglycemia score and presence or absence of glycemic control (defined as end of follow-up HbA1c of <7%). We observed an increased probability of adequate glycemic control (HbA1c <7%) when hypoglycemia scores were high (OR 1.19; P = 0.0002) and a decreased probability when hypoglycemia scores were low (OR 0.84; P = 0.0002).
As per the Indian Council of Medical Research guidelines for diabetes treatment, SUs, alongside metformin, remain the mainstay of treatment for T2DM.  As per results of various drug utilization studies conducted in India, SUs are used extensively, usually as first-line or as second-line agents, to manage patients with T2DM. ,,, The published incidence of patient reported hypoglycemia associated with SUs usage ranges widely from 16% to 63%. ,,,, However, as per a recent meta-analysis of randomized controlled studies involving SUs, incidence of hypoglycemia episodes with confirmed blood glucose value of ≤55.8 mg/dl (3.1 mmol/L) was reported to be 10.1%. 
Association between frequency of hypoglycemia and glycemic control has been consistently reported in the past. Hypoglycemia episodes were observed more frequently in intensive therapy arms of large studies like UKPDS, ACCORD, ADVANCE, and VADT. ,,, In a pooled analysis of studies comparing insulin lispro mixtures and insulin glargine in patients with T2DM on background metformin therapy, an inverse relationship was observed between all confirmed hypoglycemia episodes and HbA1c at endpoint.  Further, there was an increase in frequency of hypoglycemia by 1.4 episodes/patient/year for every 1% reduction in HbA1c and patients with confirmed hypoglycemia had lower HbA1c than patients without hypoglycemia.  In a meta-analysis of 11 randomized trials comparing insulin glargine and Insulin Neutral Protamine Hagedorn NPH in patients with type-1 and type-2 diabetes, the frequency of hypoglycemia was found to have a curvilinear relationship with HbA1c, increasing at lower end-point HbA1c values, irrespective of the type of diabetes.  In another study of patients treated on an out-patient basis with a wide range of combinations involving oral agents and insulin, lower HbA1c level was reported as independent predictor of hypoglycemia. 
OBSTACLE Hypoglycemia is the first prospective study specifically designed to assess correlation between glycemic control and patient reported hypoglycemia associated with SUs in T2DM patients. In this study, self-reported hypoglycemia was observed in nearly half of the patients. Such high prevalence of hypoglycemia in our study could be due to the fact that initiation of SU was the trigger for enrolment in this study and that the experience of hypoglycemia is likely to be the highest during initial weeks of SU therapy. We observed statistically significant association between patient-reported hypoglycemia symptom score and various measures of glycemic control, including follow-up HbA1c, change in HbA1c from baseline, likelihood of having follow-up HbA1c of <7%. These results are consistent with studies described above and confirm that aggressive glycemic management with SUs is associated with increased risk for hypoglycemia as HbA1c level approaches the American Diabetes Association goal of less than 7.0%. 
Unlike in insulin-treated patients with type 2 diabetes, , we did not observe any correlation between frequency of patient-reported hypoglycemia episodes and follow-up HbA1c. This could be because risk and frequency of hypoglycemia associated with insulin therapy is substantially higher than that associated with SUs.
Self-reported hypoglycemia is an important determinant of quality of life, fear of hypoglycemia, and self-treatment modification practices among patients of diabetes. ,, Results of this study hence explains, at least partially, deep-rooted clinical inertia among diabetes care practitioners in India and low prevalence of adequate glycemic control in real life clinical practice. It highlights need for major changes in diabetes treatment paradigm in India, including increasing emphasis on physician and patient education about hypoglycemia, routine use of patient-friendly tools (like Stanford Hypoglycemia Questionnaire) to monitor undesirable effects of diabetes medications, collaborative decision-making process between patients and physician involving due consideration of various treatment options and agreed-upon glycemic goals, and increasing acceptance of newer pharmacological modalities.  With such a multi-pronged approach, it is possible to minimize risk of hypoglycemia and achieve good glycemic control in higher proportion of patients.
Although this study was large, nationwide, and multi-centric and conducted in real-life practice, some of its potential limitations that should be considered while interpreting its results are its observational, uncontrolled design and the fact that hypoglycemia episodes were not confirmed with blood glucose measurements.
In T2DM, with addition of a SU on background of a failing oral antihyperglycemic regimen, there is an inverse correlation between follow-up HbA1c and patient-reported hypoglycemia. There is a progressive increase in patient-reported hypoglycemia as HbA1c value in such patients approaches ADA target of HbA1c <7%.
This study was sponsored by an unrestricted grant from MSD Pharmaceuticals Pvt. Ltd., India. The authors would like to thank Sunita Nair and Javed Shaikh, Capita India for editorial support in preparing the manuscript. The authors would also like to thank Clinigene International Ltd. for clinical data management and analysis.
[Figure 1], [Figure 2]
[Table 1], [Table 2]