A study to assess the unmet medical needs associated with the use of basal insulin in patients with type 2 diabetes

Abstract Aim To describe in a real‐world setting, the proportion of patients with a symptomatic hypoglycaemic event and the proportion of individuals with type 2 diabetes, who newly or recently initiated with basal insulin, achieving individual or general HbA1c target. Materials and Method DINAS‐AR was a national prospective observational study to assess the unmet needs in patients with type 2 diabetes treated with basal insulin with or without oral antihyperglycaemic drugs and/or GLP‐1 receptor agonist. The study was conducted at 19 hospitals. Results A total of 385 uncontrolled patients (≥18 years) who recently initiated basal insulin or who initiated treatment within a year prior to study enrolment entered the study. Outcomes were follow‐up incidence of hypoglycaemic events, change of HbA1C and achievement of HBA1c <7% or individual target. A total of 44 patients (11.9%) reported the occurrence of ≥1 symptomatic hypoglycaemia event(s). HbA1c reductions were greater in patients who had recently initiated treatment with basal insulin (between 15 and 90 days prior to study entry) vs patients who initiated treatment within 1 year. A total of 80 patients (31.6%) achieved individual HbA1c target (or target <7.0%) at Week 24. Furthermore, the proportion of patients achieving this target without symptomatic hypoglycaemia was 26.1% (n = 66). A lower percentage of glycemia target achievement was observed in patients reporting hypoglycaemia (n = 14), 20.6% of all patients reporting hypoglycaemia event(s) vs (n = 66) 35.7% of all patients without hypoglycaemia event reported. Conclusion In this real‐world study, although the hypoglycaemia rate was not high in adults with type 2 diabetes treated with insulin, there was a lower percentage of patients that achieved glycemic target among those reporting hypoglycaemia events vs patients who did not report them.


| INTRODUC TI ON
Type 2 diabetes is a chronic condition, and its frequency is increasing mainly due to the impact of obesity growth. 1 In Argentina, the diabetes prevalence reported in the last national risks factors survey, carried out every 4 years and evidenced an increase of around 30% in the last 10 years, showing a prevalence of 12.7% in 2018. 2 A new international study using Markov model shows that by 2060, the number of US adults with diagnosed diabetes is projected to nearly triple, and the per cent prevalence rate will double. 3 The burden of the disease is significant both to individuals and to the society as a whole.
Its related costs are primarily triggered by late complications. 4 Type 2 diabetes disease is characterized by progressive β-cell dysfunction, and it requires the stepwise addition of several therapeutic strategies in order to achieve an adequate metabolic control.
These interventions typically start with changes in lifestyle implemented right upon diagnosis and are followed by the initiation of oral antihyperglycaemic drugs and subsequently injectable therapies, including complete replacement of severely reduced endogenous insulin secretion. 5 Evidence from previous interventional studies in diabetes has clearly shown that an adequate long-term glycemic control plays an important role in reducing the risk of developing late complications. 6 This is well established in the case of microvascular complications.
The relationship between poor glycemic control and macrovascular complications is still a controversial issue to be largely discussed, although epidemiological studies have consistently shown that the poorer the glycemic control, the worse the cardiovascular outcomes. 7 Therefore, the ultimate objective of the therapies administered is to attain adequate glycemic control aimed to delay or prevent late complications. 5 In spite of the overwhelming evidence showing the crucial role of adequate glycemic control in the management of type 2 diabetes and despite the increasing number of antidiabetic drugs available for prescription, a considerable percentage of patients are still unable to reach the 7.0% HbA1c general target, which places them at a higher risk of complications. 8 One of the reasons why achieving the goal is limited in the diabetic population is the potential impact of hypoglycaemia.
As the improvement in metabolic control is typically associated with a higher frequency of hypoglycaemia, both physician and patient are reluctant to achieve an adequate glycemic control. Physicians are not willing to expose their patients to an increased risk of hypoglycaemia while patients would also like to avoid the feared experience of an hypoglycaemic episode.
Different therapeutic agents are associated with different risk levels of hypoglycaemia. 5 Undoubtedly, the antihyperglycaemic therapeutic approach with the highest risk of hypoglycaemia is insulin therapy. Therefore, it is not surprising that reluctance is high to initiate therapy and to optimize the therapeutic regimen used in the case of insulin therapy. 9 Even in the simplest approach, in the case of basal insulin therapy, which is the most popular scheme to initiate insulin treatment for type 2 diabetes, the high level of clinical inertia results in late insulin treatment initiation and suboptimal dosing, both aspects being clearly associated with suboptimal general glycemic control in many type 2 diabetes patients. 8,9 Although there seems to be clinical consensus on the fact that treatment-associated hypoglycaemia is a key factor leading to insulin late initiation and reluctance to optimize the dose (titration period) administered to reach the HbA1c target, evidence available to prove the relationship between treatment-associated hypoglycaemia and failure to reach the glycemic target is surprisingly scarce. 10 As the frequency of severe hypoglycaemia is relatively low in patients with type 2 diabetes on insulin therapy, it is extremely challenging to establish the relationship between severe hypoglycaemic events and the failure to reach the HbA1c target. On the other hand, nonsevere events, which might not be captured in some databases, are more common and can be terrifying and undesirable in real life, and therefore, such events could have a negative impact on target achievement. 11 So, the question is, then, how to reliably capture nonsevere events in order to confirm those relationships in real clinical practice.
A possible way to record these nonsevere events is to monitor patients in clinical practice and ask them to record any hypoglycaemic event they undergo in real time.
The main purpose of this study is to reliably detect hypoglycaemic events in uncontrolled type 2 diabetes patients who recently initiated treatment with basal insulin or who initiated treatment within a year prior to study enrolment, and to establish the relationship between hypoglycaemic episodes occurring during the 24-week observational period and the achievement of glycemic target. In addition, the study focused on describing the proportion of patients reaching their individual HbA1c target and/or general 7.0% glycemic target.
Establishing the link between nonsevere hypoglycaemic events and the failure to reach the glycemic target is important not only to confirm the assumption that both outcomes are related (hypoglycaemia is considered a safety outcome while reaching the HbA1c target is a surrogate outcome for late diabetes complications) but also to emphasize the importance of nonsevere hypoglycaemic events to health decision makers.
Nowadays, payers attribute cost implications only to severe hypoglycaemic events and associated nonsevere events with no direct healthcare costs. A confirmation that nonsevere hypoglycaemic events have a harmful effect on glycemic control could change this point of view and could help prescribing physicians to substantiate the need for new and better antihyperglycaemic drugs associated with a lower risk of hypoglycaemia, as those drugs, including new basal insulin preparations, are typically associated with the clinical benefit of reducing the frequency/rate of nonsevere hypoglycaemic events.
As hypoglycaemia could be associated with other undesired clinical and health economics outcomes, 12 the study includes several secondary objectives to describe the potential impact of hypoglycaemia, with special focus on reliably collected nonsevere hypoglycaemic events, on factors such as weight gain, fear of hypoglycaemia, treatment adherence and discontinuation, and use of healthcare resources.

| Study design and participants
This study was a multicenter, prospective follow-up at week 24, singlecohort and noninterventional study, conducted at 19 hospital sites in Argentina. After informed consent signature, clinical and laboratory data (baseline) were captured for eligible patients and clinical visits were conducted according to local practice. Data were collected at study entry and at Weeks 12 and 24.
Written informed consent was obtained from each patient participating in the study by means an informed consent form (ICF) approved by each site EC/IRB before starting the study. All relevant aspects of the study were explained to the patient before obtaining informed consent and prior to carrying out any activity that is not part of routine care.
The study aimed to enrol 400 participants with T2DM ≥18 years of age who had recently initiated treatment with BI, human or analogue, at least 15 days before enrolment or were on treatment with BI for <12 months with or without oral antihyperglycaemic drugs and/or GLP-1 receptor agonist.
Participants were required to have HbA1c between 7.5% and 11.0% (≥58 to ≤97 mmol/mol) for newly initiated BI users and between 7.5% and 10.0% (≥58 to ≤86 mmol/mol) for previously initiated BI users, and to be willing to perform self-monitoring of blood glucose (SMBG) and to complete a patient diary. Exclusion criteria included treatment with rapid-acting o premix insulin within the next 3 months. Participants were also excluded if they were more likely to have type 1 diabetes (<40 years old and had initiated insulin within 1 year of diabetes diagnosis), or if they were, or planning to become,

pregnant.
To help eliminate bias, investigators were advised to include consecutive patients suitable for the study.

| HbA1C targets
At baseline, an individualized long-term HbA1c target was set for each participant by their physician. In the case of those for whom an individualized target was not set, a general HbA1c target less than 7.0% [<53 mmol/mol] was defined, based on current guidelines (Table 1). A separate 12-week objective was set by physicians, based on the HbA1c level they anticipated patients would be able to reach by Week 12; however, the results of this objective are not the focus of this report.

| Objectives and end-points
The main purpose of this study is to reliably detect hypoglycaemic events in uncontrolled type 2 diabetes in patients who recently initiated treatment with basal insulin or who initiated treatment within a year prior to study enrolment, and to establish the relationship between hypoglycaemic episodes occurring during the 24-week observational period and the achievement of glycemic target. In addition, the study could focus on describing the proportion of patients reaching their individual HbA1c target and/or general 7.0% glycemic target.

To describe the proportion of patients reaching the "24-week
HbA1c" target (defined as the HbA1c level expected to be reached by the patient at Week 24, as judged by the physician).

To describe the proportion of patients reaching the "24-week
HbA1c" target (defined as the HbA1c level expected to be reached by the patient at Week 24, as judged by the physician) without symptomatic hypoglycaemia.

| Data analysis and statistics
The sample size was determinated with an estimation based on a Missing data or unknown responses were not counted for percentages. For hypoglycaemia, patients with a missing value confirmed during the study were not analysed, and therefore, they were not counted as "without hypoglycaemia" in all analyses.
It is not planned to test any statistical hypothesis in a confirmatory sense. All estimates were described in their entirety and evaluated descriptively. Confidence intervals and p-values were interpreted in the perspective of the exploratory nature of the study.
Missing data were not imputed for this study and were handled by eliminating from analysis of respective parameter including the affected variable in all patients.
All extreme values (outliers) were excluded from analysis.
These were related to data entry errors from investigators not properly corrected (eg, entry of an HbA1c value in a SMBG data field). However, the quantity of outliers was low (outliers concerning to laboratory results were identified in less than 5% of study sample).
A multivariable analysis was conducted to explore study entry (baseline) predictive factors associated with the occurrence of symptomatic hypoglycaemia up to Week 12 in the study population.
Four multivariable logistic regression models were used to analyse the impact of covariates (such as HbA1c, body weight, insulin dose, whether the patient has been recently initiated, diabetes years of evolution, renal or hepatic failure, and concomitant antihyperglycaemic medication) in the presence of hypoglycaemia.  Table 2).

| Baseline characteristics
The analysis evidenced no major clinical differences between newly  (Table 1a).
Most of the patients (n = 343, 89.1%) reported the concomitant use of, at least, one oral antidiabetic medication. Comparison among medication classes showed that metformin was the most common medication used (n = 326, 84.7%), at an overall level and in both groups, followed by DPP-IV inhibitors (n = 110, 28.6%).
Comparative analysis by group revealed that sulfonylureas were received in higher proportion by patients who recently initiated basal insulin (P = .047), whereas no statistically significant differences between groups were found in the use of the rest of oral antidiabetic medications (Table 1b).

| Primary end-point
Analysis from study data revealed that a total of 44 patients (11.9%;

| Basal insulin therapy at Week 24
Overall, the average interval between the initiation of current basal insulin therapy and study entry was of 111 days (SD: 107).
In about half of patients (n = 189, 49.1%), a long-acting basal analogue was the type of insulin prescribed, being the most common insulin type used in both groups.

| Change in HbA1c, fasting glucose (measure by SMPG) and FPG (measure by laboratory) at
Week 24 At Week 24, HbA1c had reduced from baseline by 1.55% in newly initiated participants and by 0.8% in previously initiated participants without significance from Week 12 to Week 24 (P = .065; Table 2, Figure 1).

TA B L E 2 Change in mean and by group HbA1c
Group

| Achievement of HbA1c targets at week 24
A total of 68 patients (26.9%) achieved both general HbA1c target <7.0% and HbA1c level defined for the patient, as judged by the physician. Among these patients, a slightly higher proportion in target achievement was observed in the group of patients who recently initiated basal insulin (Table 5). Furthermore, the proportion of patients achieving the target defined for the patient without symptomatic hypoglycaemia was of 22.9% (n = 58).
A total of 80 patients (31.6%) achieved individual HbA1c target (or general target <7.0% if individual target was not defined) at Week 24. Among these, a slightly higher proportion was observed in the group of patients who had a recent initiation of basal insulin (Table 5). Furthermore, the proportion of patients achieving this target without symptomatic hypoglycaemia was of 26.1% (n = 66).
Lack of adherence to lifestyle recommendations (n = 80, 43.2% of patients assessed at Week 24) was reported as the main reason why HbA1c target was not achieved, followed by the lack of adherence to titration (n = 36, 19.5%) Comparative analysis according to the occurrence of hypoglycaemic event(s) evidenced a lower proportion of glycemic target achievement in the group of patients who reported hypoglycaemia (

| Hypoglycaemic events
There were no severe hypoglycaemic events reported during the study.

24-baseline Baseline
Week 12 Week 24 Overall 176. 18   Week 24, with an overall cumulative incidence of 0.012. No statistically significant difference was found between groups in the proportion of patients developing nonsevere hypoglycaemic event(s) up to Week 24 (P = .396), and no symptomatic hypoglycaemic events leading to hospitalization were reported during the study.

Previous initiation of basal insulin (n = 149, %)
Overall (N = 385, %, %)  Abbreviation: n/a, not applicable. a Comprises medications received at any moment during the study period.

| D ISCUSS I ON
The DINAS-AR study was an observational, national, prospective were lower than predicted (20%-45%) based on conservative estimates from previous randomized clinical trials. 14,15 The limited increase in insulin dose in this study may have further contributed to the observed low incidence and rates of hypoglycaemia, as in the DUNE study.
The DINAS-AR study may have been limited by several factors.
Firstly, most of the medical investigators who treated these patients work in centres of excellence and are specialists; therefore, it could be assumed that the rate of patients in HbA1c target would be lower.
Furthermore, hypoglycaemia data were collected by physicians based on patient diaries, which may be subject to recall bias. 19 Additionally, the short observational period may also reduce the generalizability of the results, and the association between hypoglycaemia and target achievement may not necessarily have per-

| CON CLUS IONS
In this study, we quantified hypoglycaemia events in a sample of patients with type 2 diabetes who recently initiated treatment with basal insulin or who initiated treatment within a year prior to study entry, who were followed up during a 24-week period.
Hypoglycaemic event(s) were reported in 11.9% (95% CI: 8.9%-15.6%) of patients assessed at Week 12 and 18.7% of patients assessed at Week 24 while keeping similar proportions between both groups. Neither severe nor events requiring hospitalization were reported. Body weight reduction and basal insulin dose increase were identified as predictive factors for hypoglycaemia in a multivariable analysis. We also explored the relationship between the occurrence of hypoglycaemia and the achievement of glycemic (HbA1c) target at Week 24. In this regard, statistically significant drops in the HbA1c Min-max 0.00-4.00 0.00-4.00 0.00-4.00 0.00-4.00

ACK N OWLED G EM ENTS
The authors thank the study participants, trial staff and investigators for their participation (participating physicians are listed in Appendix Table 1).

CO N FLI C T O F I NTE R E S T
GF has served on advisory panels for Novo Nordisk, Sanofi and AstraZeneca. CF has served on advisory panels for Sanofi, Praxis Pharmaceutical, AstraZeneca, Novo Nordisk and MSD, and has received speaker fee from Eli Lilly, Sanofi, Novartis and Novo Nordisk.