Incidence of Hypoglycemia and Hospitalization Related to Chronic Diabetes Complications and Its Effect on Quality of Life Among Patients Initiating Second Line Therapy: DISCOVER Study

The management of patients with type 2 diabetes is a complex process that must be individualized and be patient centered. The aim of this study was to assess the metabolic control, the annual incidence and crude prevalence of hypoglycemia, hospitalization, and complications among patients with type 2 diabetes initiating second-line therapy. This study is an observational, longitudinal, prospective study as a part of the multinational DISCOVERing Treatment Reality of Type 2 Diabetes in Real World Settings (DISCOVER) study, that recruited 519 patients with type 2 diabetes who were non-insulin users, aged ≥ 18 years, and switching to second-line therapy. The cohort was clinically evaluated over three years of follow up. Fear of hypoglycemia was assessed using the Hypoglycemia Fear Survey (HFS II), while the quality of life was assessed using SF36v2 questionnaire. Using second-line therapy improved metabolic control but the annual incidence of microangiopathies were at 61/1000 patient-years, 47/1000 patient-years, and 4/1000 patient-years for neuropathy, retinopathy, and nephropathy, respectively. The incidence of hypoglycemia was 57/1000 patient-years, where 50% were recurrent during the three-years period. The HFS II showed a signicant increase in patients’ worries related to hypoglycemia. The incidence of hospitalization was 31/1000 patient-years, out of which 8/1000 patient-years were related to cardiovascular events, mainly myocardial infarction. Moderate metabolic control was associated with lower incidence of macro angiopathy and an increased incidence and fear of hypoglycemia, while it was associated with improved mental component score when assessing the patients’ quality of life. The treating physician’s decision of treatment intensication should be individualized with consideration of bets of good glycemic control versus the risk of hypoglycemia, especially in elderly patients. the follow up period. Statistical analysis. Descriptive analysis was used to describe demographic variables, patient characteristics, changes in HbA1c levels, blood glucose, lipid prole, body weight, body mass index, and blood pressure. The mean values were presented with standard deviations (±SD) and the medians with interquartile ranges (IQR). All parameters were analyzed during the follow-up period using a P value of < 0.05 as a level of signicance. The per 1000 patient-years incidence of complications and hypoglycemic events either related or unrelated to diabetes were calculated. Frequency analyses were used to report clinical and demographic data for all participants and different subcategories. The categorical data are presented as numbers and percentages. Data from each domain of the Hypoglycemic Fear Survey (HFS II) were analyzed for descriptive statistics using the mean (±SD) and the median (IQR). Multiple imputation was used to account for unreported data and missing values. Imputation was carried out by IVEware (University of Michigan). All statistical analyses were performed using the SAS statistical software system (SAS Institute, Inc., Cary, NC).


Introduction
The management of patients with type 2 diabetes is a complex process that has to be individualized and to be patient centered. It usually follows clinical guidelines with consideration of the patient's personal and clinical characteristics. Patient's age, time since diabetes diagnosis, presence of comorbidities, risk of hypoglycemia, and the physician's experience are the most important factors that play a major role in the selection of management pathway. 1 Type 2 diabetes is an established risk factor for cardiovascular disease, especially when associated with hypertension, obesity, and dyslipidemia, and plays an important role in the observed increasing mortality rates. 2,3 The need for early achievement of good glycemic control to reduce the risk of microvascular and macrovascular complications among patients with type 2 diabetes was well established by the United Kingdom Prospective Diabetes Study (UKPDS) . 4 New classes of glucose-lowering agents for treatment of type 2 diabetes are more commercially available globally in the last three decades. Despite new therapeutic classes, there is an increased risk of hospitalization, heart failure, myocardial infarction, and stroke. 5 One explanation for such increased rates of complications is the association of good glycemic control with increased incidence of hypoglycemia which remains the main restricting factor for treatment intensi cation. 6 Fear of hypoglycemia may also limit achieving good glycemic control and would have negative impact on the patients' quality of life. 7 Healthcare providers and insurance companies are looking for data that predict unmet needs for the best approach to reach good control and reduced morbidity and morality that eventually reduce both direct and indirect cost.
Objective As a part of the International multi-centered study (DISCOVER), 8 which looks at patients with type 2 diabetes initiating second-line therapy, the objective of this study using the data collected from different centers from a country facing an epidemic of type 2 diabetes (Saudi Arabia) is to assess physicians, patients and health care system attitude towards patients failing rst line management. Since it is a prospective study, it provides a better forecast for hospitalization related to chronic diabetes complications and hypoglycemic events in addition to the assessment of quality of life during the initiation of a second line therapy.
classi ed into nephropathy (chronic kidney disease and/or albuminuria), retinopathy (non-proliferative retinopathy (NPDR), proliferative retinopathy (PDR) or received retinal laser photocoagulation), and neuropathy (autonomic neuropathy, peripheral neuropathy, and/or erectile dysfunction). Macrovascular complications were coronary artery disease (angina, myocardial infarction, percutaneous coronary intervention, and coronary artery bypass grafting), heart failure, and implantable cardioverter de brillator use. Cerebrovascular disease was reported in the form of stroke, transient ischemic attack (TIA), carotid artery stenting, or carotid endarterectomy. The peripheral artery diseases (PAD) reported included revascularization procedures, diabetic foot, and amputation. The major hypoglycemic events were de ned as those that required an emergency room visit, hospital admission, a visit to a physician or other healthcare professional or external help from a caregiver or family member; and minor hypoglycemic events were de ned as those which did not require external help.
The changes in HbA1c levels, body weight, blood pressure, and lipid pro le were recorded in relation to treatment changes after the initiation of the second-line therapy or other glucose-lowering medications, insulin therapy, switching between glucose-lowering therapies, or dose changes. During the follow-up period, the incidence of microvascular complications, namely, nephropathy, neuropathy, and retinopathy was reported. The incidence of macrovascular complications was reported if the patient manifested heart failure, myocardial infarction, stroke, and diabetes-related revascularization. All patients' quality of life were assessed using the SF-36 questionnaire, and the fear of hypoglycemic events was assessed using the Hypoglycemia Fear Survey (HFS II), which assess the levels of fear related to hypoglycemia and consists of two subscales: Behavior (HFS-B) and Worry (HFS-W). 9 Part of the data collected included incidence and reasons for hospitalizations or emergency room visits during the follow up period.
Statistical analysis. Descriptive analysis was used to describe demographic variables, patient characteristics, changes in HbA1c levels, blood glucose, lipid pro le, body weight, body mass index, and blood pressure. The mean values were presented with standard deviations (±SD) and the medians with interquartile ranges (IQR). All parameters were analyzed during the follow-up period using a P value of < 0.05 as a level of signi cance. The per 1000 patient-years incidence of complications and hypoglycemic events either related or unrelated to diabetes were calculated. Frequency analyses were used to report clinical and demographic data for all participants and different subcategories. The categorical data are presented as numbers and percentages. Data from each domain of the Hypoglycemic Fear Survey (HFS II) were analyzed for descriptive statistics using the mean (±SD) and the median (IQR). Multiple imputation was used to account for unreported data and missing values. Imputation was carried out by IVEware (University of Michigan). All statistical analyses were performed using the SAS statistical software system (SAS Institute, Inc., Cary, NC).

Results
A total of 477 patients completed the 36 months of follow-up visits and had clinical and biochemical assessments. Their anthropometric parameters did not show signi cant changes; even though waist circumference increased signi cantly, it may not carry any weight due to the non-signi cant change of mean body weight from the baseline. There was a signi cant reduction in all glycemic and lipid parameters in addition to diastolic blood pressure indicating better medical care. Serum creatinine and the albumin/creatinine ratio demonstrated a signi cant increase supporting the noted increase in the incidence of albuminuria of 22/1000 patient-years projecting the increase in the incidence of diabetic nephropathy as shown in Table 1. Table 1 Mean ± SD and Median (IQR) for clinical, metabolic, and biochemical markers at baseline and during follow-up period   The incidence of mild hypoglycemia that did not warrant admission was 57/1000 patient-years, while the incidence of major hypoglycemic events was found to be 0.7/1000 patient-years as shown in Table 2.   (21) 24 (17) 13 (9) 5 (3) 6 (4) The incidence of hospitalization was 31/1000 patient-years, mainly as a single admission. Of those, 8/1000 patient-years were related to cardiovascular events, 2/1000 patient-years were related to serious infection, and 1/1000 patient-years were related to cancer. While the incidence of emergency room visits was 81/1000 patient-years mainly single visits, as shown in Table 3. The emergency room visits were mainly related to cardiovascular disease at an incidence of 10/1000 patient-years, where 6/1000 patient-years were due to myocardial infarctions and 4/1000 patient-years were related to heart failure that were mainly class II according to the New York Heart Association classi cation.  The incidence of retinopathy was 47/1000 patient-years, mainly in the form of NPDR. The incidence of neuropathy was higher, at a rate of 61/1000 patientyears, presenting mainly as erectile dysfunction. Hypertension and hyperlipidemia were found at an incidence of 52/1000 patient-years and 80/1000 patientyears, respectively. Thyroid disorder events were found at an incidence rate of 67/1000 patient-years, where 94% of the events were hypothyroidism. The incidence of cancer was found to be 4/1000 patient-years, mainly colorectal and breast cancer. The most-commonly reported infection was urinary tract infection (UTI) with an incidence of 21/1000 patient-years. Depression was reported at an incidence of 8/1000 patient-years, as shown in Table 4 and appendix 1.    The Hypoglycemia Fear Survey Score (HFS II) for the total cohort demonstrated a signi cant increase in both behavior and worries mean and median scores throughout the three years of follow up, whereas the behavior mean score increased from 7.2±11.6 at baseline to 8.2± 12.1 at 36 months follow up with a p value 0f 0.0238, while worries means increased from 6.4 ±11.9 at baseline to 7.9±11.6 at 36 months follow up with a p value of 0.0446 as shown in gure 1. Table 5 demonstrates the physical and mental components scores for the quality of life for the study participants through the follow up period. The mental component scores were signi cantly higher in each follow up visit compared with the baseline (p<0.0001) with the highest mean score in the 24 months visit at 53.8 ±7.1. The score differences from baseline were 4.0, 6.40 and 5.6 respectively. However, the mean physical component score did not change signi cantly during the 36 months follow up period

Discussion
In this study, the second-line management had signi cantly improved the metabolic control of diabetes in the form of reducing HbA1c during the 36 months follow up period. These real-world ndings were consistent with a recent meta-analysis in which all investigated drug classes lowered HbA1c to a similar extent. 10 The second-line treatments chosen for glycemic control by the treating physicians were not associated with risk of weight gain or signi cant severe hypoglycemia as observed in other studies. 10 The physicians' attitude of clinical inertia observed in this study had affected the degree of reduction in glycemic control to reach the targeted level which is commonly observed in similar studies in different societies 11 . Another explanation could be related to the delay in introducing the second -line therapy which was observed in IMPROVE study that was conducted in eight countries and involved more than 50,000 patients which highlighted the concern that the initiation of second line therapy, particularly insulin, is commonly delayed in clinical practice. 12 The improvement in glycemic parameters with the add on of the second-line therapy was re ected on the lipid parameters, with a signi cant decrease in the total cholesterol, LDL, and triglycerides, and an increase in HDL. It was re ected by reduced crude prevalence of cardiovascular disease during the follow up period especially at the rst six months of follow up (2.3%). Such observation was more pronounced in the Western countries but observed at a lower extent in the South-East Asian countries. 13 The reduction in the crude prevalence of cardiovascular disease in this study could be the result of both mild improvement of glycemic control and lower mean age of this cohort. This highlights the importance of age, as a risk for cardiovascular disease which could outmost glycemic control in this community. The same nding was also observed among patients with ischemic stroke in this community. 14 Such observation was noted in patients with PAD manifested by foot ischemia at an incidence rate of 0.2%, that was lower than what has been reported from the same study in Europe at a rate of 1.2%, 13 which could be a re ection of the younger age of our study cohort and in addition to ethnicity effect.
This cohort also demonstrated an increase in the crude prevalence of microvascular complications, namely: neuropathy, retinopathy, and nephropathy. Each year, for every 1000 patients with type 2 diabetes, 61 will present with neuropathy, 47 with retinopathy, and 4 with chronic renal disease. This increase in the crude prevalence could be explained by the mild reduction in HbA1c that did not meet the recommended targes and the legacy effect in the early phases of diabetes management 15 which warrants more therapeutic intensi cation and earlier interventions. 16 The all-cause hospitalization rate was ten times lower than what was reported in the General Practice Research Database linked to the Hospital Episode Statistics data in England, when for diabetes type and mean age. 17 This could be explained by the limited access to in-patients' admission because of long waiting list. The second reason behind such low hospitalization rate could re ect low incidence of sever complications that warrants admission. In addition, with optimization of diabetes management in the study cohort, hypoglycemia was reported to be more limited to minor events with low recurrence rate that did not warrant hospitalization or emergency room visits.
The main reason for admission in this cohort was cardiovascular disease, which is in accordance with what has been reported in the global DISCOVER data. 13 cardiovascular disease was also the most frequent reason for emergency room visit as expected in such patients. 18 This study con rmed the fact that patients with diabetes are more prone to associated disease like thyroid disease, osteoarthritis, urinary and chest infections. Such ndings support the fact that diabetes mellitus is associated with increased direct and indirect cost due to its complications and associated diseases which pressurize health services and economy. 19 The fear of hypoglycemia restricts patients' likelihood of achieving and maintaining glycemic control and has a negative impact on their quality of life. 20 The HFS II worry score was associated with the use of rst-line insulin secretagogues and the glycemic response to second-line agents or insulin use in the studied cohort. The score for behavior, worries, or both signi cantly increased with more interventions or loss of glycemic control, which is the same nding reported from the same study at an international level by Wang et al.,. 21 This nding underlines the importance of assessing patients' fear of hypoglycemia prior to treatment intensi cation, especially if comorbidities are present. On the other hand, the treatment intensi cation, and the add-on of the second line therapy was associated with improved quality of life (QoL) in terms of mental component score. The difference between the mean QOL score at the baseline and at any follow up visit exceeded the minimal clinically important difference for SF-36v2 score which is more than 3 points. 22 There was an improvement in the mental QoL more than the physical QoL with tight glycemic control which is in line with nding of Lau et.al.,. 23 A possible explanation of such nding is that increased regimen complexity required to achieve better glycemic control and increased risk of hypoglycemia may have a negative impact on patients' perception of physical QOL. On the other hand, increased sense of empowerment associated with improved glycemic control positively impacted the mental component of QOL. 24 The improvement of quality of life despite the increased fear of hypoglycemia indicates that the patients' quality of life is affected more with the improvement of their glycemic control compared with the hypoglycemic fear.
Study limitations. This study has the strength of being a prospective longitudinal hospital-based and being part of an international multicenter study with low dropout rate of 8.1%. The study limitations are related to data collection from the medical le that might have missing data like HbA1c, microalbuminuria, neuropathy testing, and fundus examination. Another limitation is the assumption of patients' adherence to their medications.

Conclusions
In conclusion, early initiation of second-line management would improve glycemic control and reduce complications rate, especially if the targeted level for both blood sugar and HbA1c is achieved. Hypoglycemia should not be a restricting factor when initiating second-line therapy, especially with better patients' education and closed home glucose monitoring. Improved diabetes control was associated with improved quality of life in terms of the mental and physical component scores. The treating physicians should individualize their plan for treatment intensi cation and assess the be ts of good glycemic control versus the risk of hypoglycemia, especially in elderly patients.
Despite the low incidence rate of diabetes complications, this rate is still critical for a country facing type 2 diabetes epidemic, especially when no active prevention program has been adopted or lunched. The annual incidence of these complications could affect the health care. The study ndings are baseline for health planners and insurance companies to improve health care and reduce the disease nancial impact.. Figure 1 Hypoglycemic Fear Survey scores at baseline and during follow-up period

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Appendix1.docx