Closed-Loop Insulin Therapy in Older Adults with Type 1 Diabetes: Real-World Data

Introduction

In the recent years progress of diabetes-related technologies has been able to integrate pump and continuous glucose monitoring (CGM)—hybrid closed-loop system—to improve glycemic control and reduce risk of hypoglycemia. ^{1 –3} Data in older adults are still scant due to the small number of patients in this age group enrolled in clinical trials and limited insurance coverage. The use of hybrid closed-loop systems have the potential to mitigate hypoglycemia, which is the most feared complication of insulin therapy in older adults. However, it is not clear if closed-loop control (CLC) can be used by older adults effectively and if it provides similar benefits as in younger adults.

The CLC (Control-IQ, Tandem Diabetes Care) is the first hybrid closed-loop system covered by Medicare and, therefore, became widely available to older patients with diabetes. In randomized control trials (RCT), this system has been shown to improve time in range and reduce time spent in hypoglycemia in children and adults, and in small group of older adults. ^2,4

The aim of this study was to assess the clinical characteristics of older persons with type 1 diabetes (T1D; age ≥65 years) from a tertiary diabetes center starting CLC and its impact on CGM metrics at the 3-month subsequent follow-up clinical visit.

Methods

Older patients (age ≥65 years) who were prescribed CLC (Control-IQ, Tandem Diabetes Care) in a single tertiary center between January 1 and December 31, 2020 were identified. Clinical information and CGM data were gathered retrospectively from the electronic medical records. The study protocol was approved by the Institutional Review Board.

Data from the clinic visit immediately before initiation of CLC and the subsequent 3-month follow-up visit—as per standard of care—were analyzed.

CGM metrics are reported as per CGM International Consensus report. ⁵

Results

Forty-eight patients were prescribed CLC. The mean age of this cohort was 70 ± 4 years, age of onset of diabetes was 27 ± 14 years, duration of diabetes 42 ± 14 years and baseline HbA1c was 7.3% ± 0.8%. Prior use of pump was seen in 83% (n = 45) and prior use of CGM was seen in 83% (n = 45) of patients. Nine patients did not start CLC, out of which two patients were on multiple daily injections and had reported trouble with dexterity (carpal tunnel) and vision issues. Four patients declined to move onto CLC due to challenges encountered in starting a new system, and three patients did not have a clear reason documented in their medical records. Two patients were excluded from the analysis because they had type 2 diabetes. Of the 37 patients with T1D who started CLC, all but 1 were prior pump users, and all but 2 patients were prior CGM users. CGM metrics, pre-CLC initiation, were not available for the two persons not previously on CGM.

In CLC users, the time in range increased from 62% ± 14% to 76% ± 9%, (P < 0.001) and time in hypoglycemia (glucose ≤70 mg/dL) decreased from 2% (1–3) to 1% (1–2); P = 0.03 compared with pre-CLC. However, the time spent in severe hypoglycemia (glucose ≤54 mg/dL) did not change [0% (0–0.5) to 0% (0–0.5); P = ns]. In CLC users, time spent in hyperglycemia (glucose >180 mg/dL) also decreased from 30% ± 11% to 20% ± 9% (P < 0.0001), but not the time of severe hyperglycemia (glucose of >250 mg/dL) (6 ± 7 to 3.2 ± 3; P = ns) (Table 1).

Coefficient of variation improved from 35% ± 6% to 27% ± 8% (P = 0.001) and glucose management indicator decreased from 7.0% ± 0.5% to 6.8% ± 0.4% (P = 0.002) in CLC users. CLC use averaged 94% ± 9% of the time over 2 weeks. HbA1c at 3 months was missing for many patients due to Covid-19 restrictions for the in-person visits (Table 1).

Total daily insulin did not change after CLC initiation (30 ± 12 U/day vs. 31 ± 14 U/day; P = ns); however, basal insulin decreased from 53% ± 13% to 50% ± 14% (P = ns) and bolus from 46% ± 13% to 38% ± 17% (P = 0.05). In CLC mode autocorrection represented 10% of insulin administered (Table 1).

Discussion

To our knowledge this is the first report on the use of the CLC technology in older population with T1D in real world. Our data show that older adults can use CLC outside the study environment and have an improvement in their CGM metrics; however, it also highlights potential enablers and barriers on starting CLC use. Interestingly, in our cohort, CGM metrics before initiation of CLC were already meeting the targets set by the CGM International Consensus report for the age in regard to time in range (goal >50%) and time above range (goal <50%), whereas time below glucose of 70 mg/dL was above goal (<1%). ⁶ After 3 months of the initiation of CLC, glycemic metrics further improved and time spent in hypoglycemia decreased, however did not reach the recommended target of <1%. These data highlight challenges in meeting hypoglycemia targets for older patients on a complex insulin regimen.

Our findings showing improvement in CGM metrics with the use of CLC in this older cohort are consistent with the RCT of CLC in children and adults with T1D. ² A subanalysis of that study looking at the 15 older participants showed that the use of CLC minimally improved hypoglycemia that was already low at baseline. Compared with the older participants of that study, our cohort of patients from clinic starting CLC were older and spent greater time hypoglycemia before initiating CLC. ⁴

The use of diabetes-related technologies (pump and CGM) in our cohort was high before initiating CLC compared with the overall older T1D population. ⁷ Almost all of the patients started on CLC were already using pump and CGM.

We reviewed the charts for the few patients who were prescribed the CLC but did not start the system. Physical issues (dexterity and visual problem) and potentially cognitive challenges (declined to start a new system for fear of failure to learn) were the most common barriers. Physical and cognitive barriers may need to be assessed by clinicians before prescribing a new diabetes-related technology in older adults. ⁸

We also found that the total amount of insulin delivered before and on CLC were similar; however, the insulin distribution changed, with less bolus insulin administered, replaced by the autocorrection. The delivery of insulin guided by glycemic metrics improved glycemic targets and reduced glycemic variability as previously described. ²

Our study timeline overlapped the COVID-19 lockdown, and so the data on laboratory HbA1c after starting CLC are not available. The lockdown and the limited access to in-person visits may have affected patients' and clinicians' decision to prescribe or change to a new diabetes-related device. However, it is important to note that even with limited in-person contact, most of the older adults who were prescribed CLC were able to use it successfully.

In conclusion, in real world, most of the older adults with T1D who were prescribed CLC were able to use it successfully, improving glycemic control and reducing time in hypoglycemia. However, discussion between clinicians and patients of potential barriers to initiate a new device should be carried out carefully.