Evaluation of the Accuracy and Reliability of a Tubeless Insulin Infusion System Under Laboratory Conditions

The Omnipod DASH (OP-D) Insulin Management System (Insulet Corporation, Acton, Massachusetts) is a novel tubeless insulin infusion system. In comparison with the established Omnipod (OP) system (Insulet Corporation), the OP-D features additional functionality such as Bluetooth connectivity; however, the insulin delivery mechanism appears to be unchanged. ¹ Although the impact of insulin delivery accuracy on clinical outcomes is unclear, the accuracy and reliability of insulin pumps are important. The aim of this analysis was to investigate the performance of the OP-D using an established procedure. ²

The methodology for accuracy assessment involved a microgravimetric measurement system based on IEC 60601-2-24 that was adapted for tubeless pumps. ² Insulin aspart (NovoRapid; Novo Nordisk A/S, Denmark) was used as test liquid. Bolus delivery accuracy was determined by measuring 25 consecutive boluses of 0.1, 0.2 and 1 U. The basal rate delivery was assessed at infusion rates of 0.1 and 1 U/h over a period of 72 hours. The reliability of the occlusion detection was determined during delivery of the same basal rates by clamping the soft cannula and determining the time until an occlusion alarm was triggered. ³ All experiments were repeated with nine different disposable infusion devices from the same lot and used in conjunction with three different handheld controlling devices.

The results of the accuracy assessment of bolus and basal delivery are displayed in Table 1. Boluses of all sizes could be, on average, delivered accurately with low mean deviations. However, the increased standard deviations and decreased percentage of boluses falling within ±15% of the target indicate variability between individual bolus deliveries. For basal delivery, the total dose over the 72-hour period was found to be accurate with low mean deviations, but variability between non-overlapping, consecutive 1-hour windows was observed. Compared with the results of the OP,^4,5 the OP-D shows similar results, but increased accuracy during bolus and basal delivery of 0.1 U and 0.1 U/h, respectively. Comparing the results of bolus and basal delivery with the accuracy limits specified by the manufacturer, more than half of 1 U boluses were found outside these limits of ±5%. For the smaller boluses of 0.1 and 0.2 U, the manufacturer defines the limits as ±25% and ±50%, respectively, resulting in a large share of boluses within that range. Applying the manufacturer’s specified limits of ±5% to the tested basal rates, the majority of non-overlapping 1-hour windows were found to fall outside these limits.

Table 1.

Accuracy Results for Bolus and Basal Rate Delivery.

	Bolus delivery (n = 225)			Basal delivery (648 h)
	0.1 U	0.2 U	1 U	0.1 U/h	1 U/h
Mean deviation	1.3%	−2.5%	−0.3%	+7.9%	+1.2%
SD of deviation for boluses/1-h windows	12.9%	18.5%	13.6%	21.7%	10.0%
Boluses/1-h windows within ±15%	76.4%	61.3%	71.6%	52.7%	86.2%
Accuracy limits of manufacturer a	±50%	±25%	±5%	±5%	±5%
Boluses/1-h windows within accuracy limits of manufacturer	100.0%	80.4%	40.4%	19.2%	40.7%

The mean and standard deviation (SD) summarize the relative deviation from the target delivery of individual boluses (bolus delivery) and nonoverlapping 1-hour windows (basal rate delivery) over all tested disposables.

a

The accuracy limits given by the manufacturer for bolus delivery are ±5% for bolus volumes ≥1 U and ±0.05 U for volumes <1 U. The accuracy limits for basal rate delivery are ±5% for basal rates ≥0.05 U/h. ⁶

Regarding the detection of occlusion, the median (minimum-maximum) times of insulin delivery in hh:mm until an occlusion alarm was triggered were 27:37 (26:59-28:29) and 03:16 (03:06-04:40) for basal rates of 0.1 U/h and 1 U/h, respectively. These results are comparable to other tubeless pumps ⁵ and indicate a decreased reliability of occlusion detection for smaller basal rates.

In conclusion, the OP-D showed slightly improved performance characteristics compared with the OP, with acceptable mean deviations but moderate dose-to-dose variability during both bolus and basal rate delivery. However, the clinical relevance of this variability is unclear.