Automated Insulin Delivery Systems for the Treatment of Elderly People With Type 1 Diabetes: A Case Series and Short Review

The treatment of type 1 diabetes in elderly patients poses several clinical dilemmas, including multiple comorbidities, established complications, impaired cognitive function, frailty, and the increased risk of severe hypoglycemia, which may lead to the pursuit of less stringent glycemic goals (1,2). Additionally, technological advances such as continuous glucose monitoring (CGM) systems, sensor-augmented pump (SAP) therapy, and, more recently, hybrid closed-loop automated insulin delivery (AID) systems have changed the landscape of modern type 1 diabetes treatment (3). However, data on the efficacy and safety of AID in older adults (>70 years of age) with type 1 diabetes are extremely limited.

We present here three cases of elderly patients with type 1 diabetes who started treatment with an AID system (Medtronic MiniMed 780G with Medtronic Guardian 4 CGM sensors and the CareLink online data platform). The patients set up an account with the online data platform, which collects and reports information directly from the AID system. These reports include ambulatory glucose profile (AGP) reports, as well as data including time the CGM sensor was used; time spent in the AID system’s automated mode; units and percentages of insulin administered as basal, bolus, and automatic corrections; grams of carbohydrate consumed per day; and number of manual boluses administered by the user.

Case Presentations

Case 1

A 76-year-old man with type 1 diabetes for 46 years and an A1C of 8.0% was being treated with multiple daily injection (MDI) insulin therapy with the analog insulins degludec 22 units once daily and lispro 3 units at breakfast and 5 units at lunch and dinner in fixed doses without carbohydrate counting. He had been using an intermittently scanned CGM system (Abbott FreeStyle Libre 2) for 1 year. His medical history included coronary artery bypass surgery, a surgically treated peptic ulcer, macular edema treated with intraocular anti- vascular endothelial growth factor injections, peripheral neuropathy, and frequent episodes of hypoglycemia with impaired awareness of hypoglycemia and at least three episodes of severe hypoglycemia in the past 5 years. He was fully independent and had good mobility, no cognitive deficits, and no severe visual impairment. He lived in an apartment in the same neighborhood as his son, who also has type 1 diabetes and had been using an AID system for 1 year.

While analyzing his CGM AGP data, we discovered an increased time in level 1 (55–69 mg/dL) and level 2 (<54 mg/dL) hypoglycemia (time below range [TBR] 4% and 2%, respectively) (Figure 1). At his request and with the agreement of his son, who accompanied him and agreed to participate in the education and initial support of his father, it was decided to switch him from MDI to AID insulin therapy. After two sessions of individual at-home training by a qualified trainer in the placement of insulin catheters and glucose sensors and the handling of the AID pump, he came to the diabetes clinic.

Figure 1.

Case 1: AGP report of patient on insulin analog injections.

During the visit, the system’s operating settings were programmed and the pump was initiated. An account was set up on the AID system’s online data platform, through which the son could upload data from the AID system. After 1 week of running the pump in open-loop (nonautomated) mode and analyzing the data, the final settings were programmed for operation of the system in automated mode and a glucose value of 110 mg/dL was selected as the therapy target. Three months later, that patient’s AGP report showed that his TBR had been reduced to 0%. All other parameters had also improved significantly; in particular, his degree of glucose variation had decreased from 42.6 to 28% (Figure 2). Also, his need for basal insulin decreased from 22 to 15 units within 24 hours, while his total insulin dose remained unchanged at 35 units. His A1C value was 7.1%. Furthermore, his adherence to AID was excellent (94% sensor use and 96% of time in automated mode), with no reported adverse effects and a high level of patient satisfaction.

Figure 2.

Case 1: CareLink data report 12 weeks after starting AID.

Case 2

A 75-year-old man with type 1 diabetes for 45 years and an A1C of 6.7% had been treated for 20 years with a traditional insulin pump (AccuChek, Roche) and an intermittently scanned CGM system (initially FreeStyle Libre 1 and later FreeStyle Libre 2). His medical history included coronary artery bypass surgery, ischemic stroke, arterial hypertension, and proliferative retinopathy, for which he underwent photocoagulation. He lived with his wife, was fully independent, and had good mobility, no cognitive deficits, and no severe visual impairment.

Review of his AGP report data revealed excellent glycemic control (Figure 3). The patient’s pump had technical faults because of its age, and at his request, it was decided to replace it with an AID system. After individual at-home training by a specialized trainer highlighting the specificities of the AID system, the system was remotely initiated in nonautomated mode using the settings from his previous pump. A data platform account was set up under the supervision and guidance of one of the patient’s children.

Figure 3.

Case 2: AGP report of patient on conventional insulin pump.

After running the pump in nonautomated mode for 1 week and assessing the AGP data remotely via the data platform, the final settings for closed-loop operation were made, selecting a therapeutic target glucose level of 110 mg/dL. Three months later, the AGP report showed further improvement in glycemic time in range (TIR; 70–180 mg/dL), with no change in TBR. The patient’s basal insulin requirement decreased from 26 to 21 units per 24 hours, and his total daily dose of insulin likewise decreased from 63 to 60 units (Figure 4). His A1C value was 6.8%, and his adherence to therapy was excellent (96% sensor use and 99% in automated mode). No adverse effects were reported, and the patient was very satisfied.

Figure 4.

Case 2: CareLink data report 12 weeks after starting AID.

Case 3

A 74-year-old woman with type 1 diabetes for 48 years and an A1C of 7.5% had been treated with an SAP (Medtronic MiniMed 640G) with predictive low glucose suspend functionality for 6 years. Her medical history included chronic atrial fibrillation, hypothyroidism, surgical cataracts in both eyes, and nonproliferative diabetic retinopathy. She lived with her husband and was visited daily by her daughter, who has helped her manage her diabetes over the past few years and helped her set up an account with the pump’s online data platform. She can look after herself, is mobile, and has no severe visual impairment. She has experienced mild cognitive impairment for about 1 year without a significant impact on her daily life.

Her AGP report showed that, despite using her pump’s predictive low glucose suspend feature, her TBR in level 1 and level 2 hypoglycemia was 2% and 1%, respectively (Figure 5). Because of her pump’s age and technical defects, it was decided to replace it with an AID system. After two individual at-home training sessions by a qualified trainer, the new pump was put into operation remotely, initially retaining the settings from her previous pump.

Figure 5.

Case 3: CareLink data report of patient using an insulin pump with the predictive low glucose suspend feature activated.

After 1 week, her data were analyzed remotely via the online platform, and the final settings were programmed for her pump to operate in automated mode, with a therapeutic target glucose level of 110 mg/dL. Three months later, her TBR in level 2 hypoglycemia was 0%, and the TBR in level 1 hypoglycemia had increased by 1 percentage point (Figure 6). Her TIR increased from 60 to 77%, and her total daily insulin dose increased by 4.8 units because of an increase in prandial insulin of 4.4 units. Her A1C was 7.2%. Her adherence to therapy was excellent (98% sensor use and 95% in automated mode), no adverse effects were reported, and the patient was very satisfied.

Figure 6.

Case 3: CareLink data report 12 weeks after starting AID.

Questions

1. Is the use of AID treatment in older patients with type 1 diabetes safe and effective?

2. What are the potential barriers to the implementation of AID treatment in elderly patients with type 1 diabetes?

Commentary

We have presented three cases of older adult patients with type 1 diabetes in whom the AID therapy was used safely and effectively. In fact, in the first case, by switching from an MDI insulin regimen to an AID system, the patient was able to achieve the overall treatment goals of AGP for elderly patients (i.e., TIR >50%, level 1 hypoglycemia <1%, level 2 hypoglycemia 0% time above range 1 [TAR1; 181–250 mg/dL] <50%, and time above range 2 [TAR2; >250 mg/dL] <10%) (4). Of particular note is the almost complete elimination of hypoglycemia with the use of AID in a patient with a history of multiple hypoglycemic episodes and impaired perception of hypoglycemia, both of which are common complications in elderly patients with type 1 diabetes (2). In the other two cases, in which the patients had been long-time insulin pump users, the AID system increased their TIR without prolonging their already low TBR. In the third patient, who was initially treated with SAP therapy, using the AID system led to a significant increase in TIR of 17 percentage points with a simultaneous elimination of level 2 hypoglycemia, while level 1 hypoglycemia increased from 2 to 3%.

Data from clinical trials on the efficacy and safety of AID in patients with type 1 diabetes >65 years of age are extremely limited. A recent clinical trial investigated the efficacy and safety of closed-loop insulin administration compared with SAP therapy in patients with type 1 diabetes >60 years of age with a disease duration of >10 years who were already treated with an insulin pump (5). Thirty participants with a mean age of 67 years and a mean disease duration of 38 years were randomized into two 4-month periods of treatment. The first was with an AID system in automated mode. The second was with the same pump in open-loop (nonautomated) mode with a glucose sensor (i.e., SAP) but with the features of suspend in prediction of hypoglycemia and suspend in hypoglycemia both disabled. The mean TIR was 75.2% with the AID automated mode and 69% with SAP, for a mean difference of 6.2 percentage points (P <0.0001). In addition, the time spent in hypoglycemia <70 mg/dL was significantly reduced by 0.5 percentage points for all 24 hours and by 0.8 percentage points overnight with AID. The frequency of severe hypoglycemia did not differ significantly between the AID and SAP groups (three and two episodes, respectively). Only one episode of diabetic ketoacidosis occurred with the SAP, whereas no serious adverse events were recorded with the AID system.

In a retrospective observational study, Toschi et al. (6) collected data on 48 patients with a mean age of 70 years, a mean diabetes duration of 42 years, and a mean A1C of 7.3% who were prescribed an AID system at a tertiary diabetes center. Of note, nine of the patients did not eventually start the new treatment because they had problems adjusting to the operational requirements of the system, vision problems, or problems with hand movement (carpal tunnel syndrome). Of the patients who eventually received the new treatment, all but one had already been treated with an insulin pump, and all but two were already CGM users. With the use of the AID system, TIR increased from 62 to 76% (P <0.001), time in level 1 hypoglycemia decreased from 2 to 1% (P = 0.03), glycemic variability decreased from 35 to 27% (P = 0.001), and the glucose management indicator (a CGM-derived estimate of A1C) decreased from 7.0 to 6.8% (P = 0.002). The time in level 2 hypoglycemia and the total daily insulin dose remained unchanged.

Finally, Choudhary et al. (7), who analyzed data from >100,000 AID system users, reported that, in the >13,000 users >56 years of age, TIR achieved was 76.8% and was comparatively higher than in all users and in users <15 years of age.

All three of our patients were already familiar with the use of the technology to some degree, as they had already been using a CGM system, and two of them had also used an insulin pump. Thus, the benefit of switching to the AID system was added to the advantage of using CGM with or without an insulin pump over traditional self-monitoring of blood glucose with a glucose meter for patients >60 years of age (8–10). Of note, after adequate training, the adherence of our patients in using their glucose sensor (>94% of time per week) and using the AID system in automated mode (>95% of time per week) correlate with optimal AID system efficacy in all age-groups of users with type 1 diabetes (11,12).

Despite having comorbidities and complications of diabetes, our patients remained independent, with good mobility and no significant impairment in cognitive function. They also had family members who were able to help with the immediate resolution of practical problems related to the use of technology. Often, severe impairments in cognitive function, visual acuity, and mobility are serious barriers to the use of technology to treat type 1 diabetes (13). There are no data from clinical trials on the safety and efficacy of AID use in individuals experiencing such issues, who are usually excluded from clinical trials. Therefore, assessment of cognitive function, vision, and physical abilities in elderly patients with type 1 diabetes is essential before technology-based therapeutic interventions are implemented (13,14).

All three of our patients were satisfied with the use of the AID system and felt that it helped to improve their daily management of diabetes. In a pilot study, Bisio et al. (15) found a similar level of satisfaction with various psychosocial outcomes and sleep quality with the use of AIDs in elderly patients with type 1 diabetes.

In summary, we presented three cases in which AID therapy was successfully implemented in elderly patients with type 1 diabetes who did not have clinically significant cognitive impairment and did have good mobility and visual acuity, the ability to manage self-care, and a supportive family environment for the use of technology. Further data from targeted clinical trials are needed to demonstrate the safety and efficacy of AID systems in a broader population of older patients with type 1 diabetes.

Clinical Pearls

• AID therapy seems to be safe and effective in elderly patients with type 1 diabetes without clinically significant cognitive impairment.

• Reducing hypoglycemia is the most important anticipated benefit from the use of AID in elderly people with type 1 diabetes.

• Sufficient residual mobility and visual acuity, as well as supportive family environment, may be important prerequisites for the successful implementation of AID treatment in these patients.