The Effect of Transition From Multiple Daily Injection Therapy to Automated Insulin Delivery in People With Type 1 Diabetes and Limited Diabetes Self-Management

We want to highlight the potential effect of offering automated insulin delivery (AID) systems to people with type 1 diabetes (T1D) and limited self-management. The AID is a promising technology that can improve glycemic control and enhance the quality of life for people with T1D. ¹ A central feature of AID is its dynamic correction that can mitigate the glucose excursions caused by insufficient meal announcements or carbohydrate inputs, which is common among people with limited self-management capacity. Despite this, most consensus reports and national treatment guidelines state that “AID systems still require basic diabetes management skills, including carbohydrate counting.” ² Furthermore, clinical trials investigating the effect of AID systems have primarily included participants already using traditional insulin pumps, thereby excluding participants who are not candidates for insulin pumps due to limited diabetes self-management practices, including carbohydrate counting. ³ This case series reports the effect of transferring people with T1D and limited diabetes self-management from multiple daily injections (MDIs) to an AID system.

We offered 12 people with T1D (mean age = 30.3 ± 4.5 years; 42% women; mean diabetes duration = 16.5 ± 9.0 years) and insufficient glycemic control (mean [range] HbA_1c = 92.3 [68-147] mmol/mol [10.6 (8.4-15.6)%]) due to low diabetes self-management resources to shift from MDI to an AID system. Ten (83%) had one or more psychiatric diagnoses, and nine (75%) had diabetic late complications. The AID initiation included a dedicated 2.5-day course in groups of four to six people and several follow-up visits with diabetes specialist nurses. All gave informed consent to collect and publish their diabetes-related data.

After six months of AID use, mean time in range (TIR) was more than doubled with a 32.0% points (95% CI = 21.6% to 42.4%, P < .0001) increment, time below range (<3.9 mmol/L) was reduced by 1.4% points, and time above range (>10 mmol/L) by 30.6% points (Figure 1). Glycemia risk index (GRI), a composite metric for the quality of glycaemia, ⁴ was reduced by 36.4 (95% CI = 19.2 to 53.56, P = .0007). The results were sustained at nine months after AID initiation. After approximately 5 to 24 months of AID use, HbA_1c levels were reduced by 25.6 mmol/mol (95% CI = −39.4 to −11.7, P = .002). There were no increases in admissions due to diabetic ketoacidosis or severe hypoglycemia.

Figure 1.

Time in ranges before and six months after automated insulin delivery. Mean difference (MD) of TIR from baseline to six months was 32.0% point (95% CI = 21.6 to 42.4). MD of TBR level 2 was −0.3% points (95% CI = −0.9 to 0.2). MD of TBR level 1 was −1.1% points (95% CI = −2.1 to −0.1). MD of TAR level 1 was 6.2% points (95% CI = 0.8 to 11.6). MD of TAR level 2 was −36.8% points (95% CI = −49.1 to −24.4). All MDs were estimated using a dependent sample t-test. Dark red = TBR level 2 (<3.0 mmol/L), red = TBR level 1 (3.0-3.9 mmol/L), green = TIR (3.9-10.0 mmol/L), yellow = TAR level 1 (10.0-13.9 mmol/L), and orange = TAR level 2 (>13.9 mmol/L).

Abbreviations: AID, automated insulin delivery; CI, confidence interval; MD, mean difference; TAR, time above range; TBR, time below range; TIR, time in range.

The transition from MDI to AID in these people with T1D and limited diabetes self-management resulted in substantial short-term improvements in glycemia. Presumably, this patient group is at the greatest need for automated glucose correction as delivered by AID to replace their own diabetes management. It was obtained without compromising safety, which has previously been a concern for implementing insulin pump therapy in these people. Our findings suggest that broader implementation of AID systems in people with T1D should be considered, as it could reduce disparities in diabetes care and improve outcomes for people struggling with their glucose control. Although our case series is small, the results are encouraging and highlight the need for future research on the long-term safety and efficacy of AID use in real-world settings among people with T1D and limited diabetes self-management.