Intentional Large Insulin Overdose Captured on a Continuous Glucose Monitor

Continuous glucose monitoring (CGM) technology has emerged as an important diabetes management tool. Use of CGM with real-time glucose data can improve glycemic control with reduction in HbA1c and reduced time spent in hypoglycemia in both children and adults with type 1 diabetes. The magnitude of clinical benefit is related to duration of use.¹

Despite the advantages of CGM, it has not been widely integrated into routine management of type 1 diabetes. T1D exchange data demonstrates that CGM technology is being used by only 6.5% of people with type 1 diabetes in United States and that among individuals who have used CGM, two-thirds stopped using it.² One of the major challenges facing widespread adoption of CGM technology is reduced accuracy and reliability of CGM systems, with a subsequent negative impact on adherence and effectiveness.³

We present a case of deliberate insulin overdose in a person with type 1 diabetes wearing blinded CGM, highlighting the issue of accuracy of existing CGM systems in the hypoglycemic range.

Case

A 25-year-old man with poorly controlled type 1 diabetes which had been diagnosed at the age of 10 years. His complications include laser-treated preproliferative retinopathy, microalbuminuria, and recurrent diabetic ketoacidosis. He has a personality disorder with multiple previous presentations with deliberate self-harm including insulin overdose. A Medtronic iPro2 blinded CGM system using an Enlite sensor (Medtronic, Northridge, CA) was arranged for 7 days to guide intensification of insulin therapy. On day six of monitoring, he presented to the emergency department with a deliberate overdose of 300 units of insulin aspart (Novorapid, Novo Nordisk, Copenhagen, Denmark). His venous blood glucose was 1.6 mmol/L (28 mg/dL) and serum insulin level was 452.3 mu/L (fasting reference range 3-15 mu/L) at admission. He was treated with dextrose infusion, glucagon injection, and 40% glucose gel over 9 hours with in-patient monitoring for 36 hours. At the end of day 7, the sensor was removed and data were downloaded.

Result

Severe hypoglycemia was managed appropriately with parenteral and enteral glucose, and glucagon. Analysis of CGM data (Figure 1) showed failure of the sensor to detect hypoglycemia with the reported sensor glucose ranging between 5 and 6 mmol/L (90-108 mg/dL). While the mean absolute relative difference (MARD%) for the whole 7 days of CGM was 17.7%, it was 52% in the first 9 hours post–insulin overdose.

Figure 1.

Relationship between ISF glucose and that of venous blood post–insulin overdose.

Discussion

Despite clinical benefits of CGM and data suggesting continuous knowledge of glucose is a research priority, uptake is lower than may be expected.² This in part due to reimbursement challenges, discomfort, complexity, the need for calibration and the limited life span of sensors, but also reflects alarm fatigue related to sensor inaccuracy, particularly in the hypoglycemic range. For people who persevere with CGM despite these limitations, an increase in frequency of false positive alarms for hypo- and hyperglycemia may negatively impact on quality of life, concordance and the effectiveness.

Efforts to improve sensor accuracy are ongoing. One of the avenues for enhancing sensor accuracy is modification of sensor components such as the coating membrane³ which serves several important functions. Its primary function is to extend the glucose detection range of the sensor by limiting glucose flux to the electrode surface and enzyme layer. This prevents glucose oxidase saturation at high glucose concentrations and avoids oxygen deficit (in first-generation glucose sensors) by maintaining a balance between oxygen and glucose fluxes. Permselective membranes also prevent electroactive interferents from reaching the electrode and enhance sensor biocompatibility.⁴

Equally important is the method of electron shuttling between the redox center of the glucose oxido-reductase enzyme and electrode surface. In first-generation glucose sensors, oxygen acts as the electron acceptor with production of hydrogen peroxide. One of the challenges to first-generation glucose sensors is the higher potential required for oxidation of hydrogen peroxide (around +700 mV). At this potential, oxidation of other electroactive interstitial fluid (ISF) analytes occurs. The use of redox mediators in second-generation glucose sensors allows for a lower potential and reduces interference. However, the risk of leaching and lack of mediator biocompatibility is a barrier to mediated implantable glucose sensors. The only approved second-generation CGM system utilizes an osmium mediator.⁴

The use of simultaneous glucose sensors allows redundancy for technical failure as a single malfunctioning sensor can be voted out on the basis of divergent results, improving accuracy and precision of CGM. However, redundancy cannot mitigate for inaccuracy from calibration error or sensor delay. This, however, can be addressed by simultaneous use of two sensors utilizing different methods of glucose sensing (eg, electrochemical and optical), in “orthogonal redundancy.”⁵

Development of new calibration algorithms to compensate for the time lag between blood and ISF is another avenue for enhancing sensor accuracy.⁶

In the presented case the MARD for the whole sensor period was 17.7%, comparable to published sensor data. Sensor performance improved once hypoglycemia was corrected, suggesting that calibrations were at appropriate times and emphasizing that it is in the hypoglycemic range that sensor accuracy deteriorates.

Conclusion

This is the first reported case of deliberate insulin overdose captured on CGM, highlighting a limitation of existing CGM technology with reduced sensor accuracy in the hypoglycemic range. Many technologies are being pursued to overcome these challenges.