Bridging the diabetes care gap in elderly: Lessons from automated insulin delivery systems

SUMMARY

Kudva et al, in NEJM Evidence, examine the efficacy and safety of automated insulin delivery (AID) systems in older adults with type 1 diabetes.¹ This randomized, multicentre, crossover trial compares three insulin delivery methods: Hybrid Closed-Loop (HCL) systems, Predictive Low-Glucose Suspend (PLGS) systems, and Sensor-Augmented Pump (SAP) therapy. Eighty-two participants aged 65 to 86 years were enrolled in the study, all of whom had prior experience with insulin pumps and continuous glucose monitoring systems (CGMs). An additional inclusion criterion was that at least 1.5% of glucose readings were <70 mg/dl during the run-in period, indicating a sufficient baseline level of hypoglycaemia risk to evaluate the benefit of the AID system.

Individuals with severe cognitive impairments or major comorbid conditions were excluded. The trial followed a crossover design, with each participant using the three systems for 12 weeks in randomized order: HCL (automatically adjusts basal insulin delivery based on CGM readings and user-input meal boluses), PLGS (suspends insulin delivery when low glucose levels are predicted), and SAP (provides CGM data but requires users to adjust insulin doses manually). The primary outcome was the percentage of time with glucose levels below 70 mg/dl (hypoglycaemia). Secondary outcomes included time-in-range (TIR; glucose 70–180 mg/dl), glycated haemoglobin (HbA1c), and adverse events such as severe hypoglycaemia and diabetic ketoacidosis (DKA). The mean (SD) baseline HbA1c level was 7.2 (0.9). The HCL system significantly reduced the time spent in hypoglycaemia (glucose <70 mg/dl; 1.58 [0.95]) compared to SAP therapy and PLGS (2.57 [1.54] and 1.67 [0.96], respectively). Both HCL and PLGS systems demonstrated marked reductions in night time hypoglycaemia, a critical safety benefit for older adults. The HCL system improved TIR by 8.9 percentage points (95% CI, 7.4 to 10.4 percentage points) compared to SAP therapy. HbA1c levels decreased modestly by 0.2% (95% CI –0.3 to –0.1 percentage points) with HCL, reflecting enhanced glycaemic management. PLGS provided smaller, yet clinically meaningful, improvements. Adverse events were minimal, with severe hypoglycaemia occurring in fewer than 4% of participants across all methods. Two cases of DKA occurred during the HCL phase, attributed to user errors and system interruptions.

COMMENT

The prevalence of type 1 diabetes in older adults is rising due to improvements in life expectancy and advancements in diabetes care.² However, older adults with type 1 diabetes face unique challenges in managing the condition, including increased risk of hypoglycaemia, hypoglycaemia unawareness, difficulties with self-management due to cognitive decline, physical frailty, and potential for comorbid conditions.^3,4 These challenges necessitate innovative approaches to insulin delivery that reduce the burden of diabetes management while ensuring safety and efficacy.⁵ AID systems, which integrate CGMs, insulin pumps, and advanced algorithms, have emerged as a promising technology to optimise glycaemic control. These systems automatically adjust insulin delivery based on real-time glucose readings, thereby reducing the need for manual adjustments.^6,7

The findings from this study have important clinical implications.¹ It is essential to note that the findings of this study are not applicable to elderly patients with type 2 diabetes, as they were excluded from the study cohort. AID systems, particularly the HCL system, represent an effective tool for improving diabetes management in older adults. Their ability to reduce hypoglycaemia risk, which is particularly relevant in a population with impaired ability for self-management and heightened hypoglycaemia unawareness, makes them a major addition to the therapeutic armamentarium for this population.

However, the relatively small sample size and short study duration may not capture the long-term benefits and challenges associated with AID systems, especially for older adults who may face changing health issues. The exclusion of individuals with severe cognitive impairments or major comorbid conditions limits the applicability of this study to frailer populations. These individuals might benefit most from simplified and automated systems due to their difficulty in managing complex diabetes regimens. Furthermore, the lack of wash-out periods in the multiple cross-over design raises questions about the influence of carry-over effects on observed outcomes. Clarification on how these effects were accounted for in statistical analyses would strengthen the reliability of the results.

Lack of diversity in the study population regarding ethnicity and socioeconomic status might limit generalizability. Participants included in the study were required to have prior experience with insulin pumps, potentially excluding those with limited technological proficiency. The study does not detail how cognitive or physical impairments might affect the usability of the systems in general, nor their differential impact on the various types of AID systems evaluated. Patient-reported outcomes, such as quality of life, satisfaction with the technology, or perceived ease of use, were not reported. These factors are crucial in determining the real-world adoption and sustainability of AID systems in older populations, who may face technology literacy barriers. The study does not address the financial implications of AID systems. This is particularly relevant for older adults,⁸ who often live on fixed and meagre incomes, especially in our subcontinent. Accessibility issues, including the need for comprehensive training and ongoing technical support, are also overlooked. These factors are critical for ensuring equitable adoption of AID systems. While the study reports two cases of diabetic ketoacidosis (DKA) in a short duration of time, it does not provide sufficient detail on the circumstances or contributing factors, such as user errors or system malfunctions. This detail is important given the potential safety concerns associated with AID systems. The rates of severe hypoglycaemia were low; however, the study did not investigate whether individual characteristics or device usability challenges influenced these events. Differences between the HCL and PLGS systems were relatively small, making it unclear if the added complexity of HCL justifies its use for all older adults.

While this study provides preliminary insights into the utility of different AID systems in a population where they have not been previously extensively evaluated, future studies of longer duration, involving a larger sample size and diverse populations, will be required to consider the broader context of real-world implementation and accessibility. They should investigate the durability and long-term effects of AID systems, including their impact on diabetes-related complications, healthcare utilisation, and mortality rates. Incorporating patient-reported outcomes, caregiver perspectives, and usability assessments would provide a more comprehensive understanding of the usage of AID systems in the lives of older adults. Research should prioritise inclusion of individuals with cognitive impairments, physical disabilities, and lower socioeconomic status to better reflect the diversity of the older adult population with type 1 diabetes mellitus. Evaluating the cost-effectiveness of AID systems could facilitate broader adoption in healthcare systems.