Potential Insulin Underdelivery from Prefilled and Reusable Insulin Pens in Cases of Premature Needle Withdrawal: A Laboratory Evaluation

Introduction

The gold standard therapy for patients with type 1 diabetes (T1D) is intensive insulin therapy with continuous subcutaneous insulin infusion or multiple daily injections. ¹ For multiple daily injection treatment, insulin pens (prefilled or reusable) have emerged as the reference at the expense of syringes and vials. ² Although scarcely explored in the literature, the clinical experience of many physicians is that the injection techniques of patients with T1D are often inconsistent with manufacturers' recommendations, especially regarding the delay before withdrawing the needle after the injection. Clinical practice also demonstrates that glycemic control of T1D patients remains challenging with high intra- and interday variability, particularly in younger patients, even when intensive insulin therapy is used. Misuse of insulin pens could contribute to this issue. In the present study, our aim was to evaluate potential insulin underdelivery (PIUD) in cases of premature needle withdrawal after an injection with a prefilled or reusable insulin pen.

Materials and Methods

This laboratory evaluation assessed prolonged leakage of five prefilled insulin pens (lispro Kwikpen^® [lisK] [Eli Lilly, Indianapolis, IN], aspart Flexpen^® [aspF] and detemir Flexpen [detF] [Novo Nordisk, Bagsvaerd, Denmark], glulisine Solostar^® [gluS] and glargine Solostar [glaS] [Sanofi, Paris, France]) and three reusable insulin pens (Humapen^® Luxura HD with lispro cartridge [lisH] [Eli Lilly], Novopen^® Echo with aspart cartridge [aspN] and detemir cartridge [detN] [Novo Nordisk], and JuniorSTAR^® with glulisine cartridge [gluJ] and glargine cartridge [glaJ] (Sanofi)]. These three reusable pens were chosen because they all offered a half-unit injection, which is an important feature for patients with low insulin needs (e.g., children).

All pens and cartridges were provided by the hospital pharmacy of the investigation center. BD Micro-Fine Ultra^® 4-mm needles (Becton, Dickinson and Co., Franklin Lakes, NJ) were applied to the prefilled insulin pens as recommended by the manufacturer. All of the experiments were performed in a single laboratory at ambient temperature by the same investigator to improve reproducibility. Insulin pens were operated as recommended. A digital chronometer was used to ensure time compliance during the assessments.

After priming with 4 units (U), all of the pens were assessed with the same method. In brief, the 5-U setting was selected, and the injection button was pushed with the maximum thumb strength of the operator to express insulin into a discard beaker for 2 s. Then, while the injection button remained depressed for an additional 8 s (the total time was 10 s), the remaining drops were ejected into another beaker. This amount of insulin collected was called PIUD and was weighed on an analytical balance that had been previously calibrated and tared with an accuracy of 0.01 mg (Sartorius Pesage Industrie, Les Ulis, France). This experiment was repeated 10 times using two different prefilled pens from each brand. For reusable pens, the assessment was also repeated 10 times using two different cartridges for each type of insulin.

The same procedure was also applied for 10 U of insulin. For this dose, insulin was discarded for 3 s, but the total time during which the injection button remained depressed was 10 s (thus, the time for collecting PIUD was 7 s).

Discard times of 2 and 3 s for the doses of 5 and 10 U, respectively, were chosen because they corresponded to the time of an observed insulin stream before dribbling. Overall, 20 PIUD measurements were performed for each pen with each insulin type at each dose. To convert insulin mass into units, we used a proportional relationship after weighing 10 U of each insulin type eight times.

Quantitative data are expressed as mean±SD values. Student's t tests were used for intra- and intergroup comparisons. Statistical analysis was performed using SAS version 9.4 software (SAS Institute Inc., Cary, NC). A P value of <0.05 denoted statistical significance.

Results

Prefilled pens

For a 5-U injection, mean PIUD was 0.53±0.35 U for rapid-acting analogs (ranging from 0.28±0.09 U [aspF] to 1.01±0.07 U [gluS]; P<0.001) and 0.27±0.07 U for basal analogs (ranging from 0.22±0.04 U [detF] to 0.31±0.05 U [glaS]; P<0.001). Expressed as a percentage of the selected dose, PIUD represented up to 20.2%/10.7% and 6.3%/2.9% for 5/10 U of glulisine and glargine with Solostar, respectively (Table 1 and Fig. 1).

OPEN IN VIEWER

FIG. 1.

Potential insulin underdelivery (PIUD) expressed as a percentage of the selected insulin dose for 5- and 10-unit (U) doses when simulating premature needle withdrawal after injection with (upper panel) five prefilled insulin pens (lispro Kwikpen, aspart Flexpen, glulisine Solostar, detemir Flexpen, and glargine Solostar) and (lower panel) three reusable insulin pens (lispro cartridge in Humapen Luxura HD, aspart cartridge in Novopen Echo, glulisine cartridge in JuniorSTAR, detemir cartridge in Novopen Echo, and glargine cartridge in JuniorSTAR). *P<0.001.

OPEN IN VIEWER

Table 1.

Potential Insulin Underdelivery for Five Prefilled Insulin Pens and Three Reusable Insulin Pens

	PIUD after an injection of
	5 units	10 units	Comparisons
Prefilled pens
Lispro Kwikpen	0.30±0.07 (6.0±1.3)	0.31±0.05 (3.1±0.5)
Aspart Flexpen	0.29±0.08 (5.6±1.8)	0.28±0.08 (2.8±0.8)
Glulisine Solostar	1.01±0.07 (20.2±1.5)	1.07±0.08 (10.7±0.8)	P<0.001 versus lispro Kwikpen and aspart Flexpen
Detemir Flexpen	0.22±0.04 (4.4±0.9)	0.26±0.05 (2.6±0.5)
Glargine Solostar	0.31±0.05 (6.3±1.1)	0.29±0.05 (2.9±0.5)	P<0.01 versus detemir Flexpen (only for 5 units)
Reusable pens
Lispro Humapen Luxura	0.15±0.03 (3.0±0.5)	0.18±0.03 (1.8±0.3)
Aspart Novopen Echo	0.43±0.06 (8.6±1.2)	0.43±0.05 (4.3±0.5)	P<0.001 versus lispro Humapen Luxura and glulisine JuniorSTAR
Glulisine JuniorSTAR	0.30±0.05 (5.9±1.0)	0.29±0.03 (2.9±0.3)	P<0.001 versus lispro Humapen Luxura
Detemir Novopen Echo	0.41±0.08 (8.3±1.7)	0.41±0.09 (4.1±0.9)	P<0.001 versus glargine JuniorSTAR
Glargine JuniorSTAR	0.15±0.03 (3.1±0.6)	0.16±0.04 (1.6±0.4)

Data are mean±SD values, expressed in units (% of selected dose). The five prefilled insulin pens were the lispro Kwikpen, aspart Flexpen, glulisine Solostar, detemir Flexpen, and glargine Solostar. The three reusable insulin pens were the lispro cartridge in Humapen Luxura HD, aspart cartridge in Novopen Echo, glulisine cartridge in JuniorSTAR, detemir cartridge in Novopen Echo, and glargine cartridge in JuniorSTAR.

PIUD, potential insulin underdelivery.

Discussion

We have demonstrated in this laboratory study that during an insulin injection with prefilled or reusable insulin pens, prolonged insulin leakage after the initial stream represents a non-negligible amount of insulin, especially at small doses. Thus, in patients who do not properly use insulin pens and prematurely withdraw the needle, prolonged leakage implies that some insulin remains undelivered. This phenomenon may be critical for subjects with low insulin needs (such as children) because the percentage of potentially undelivered insulin is negatively correlated with the selected dose. For example, in one prefilled pen, the PIUD was 20.2%, 10.7%, and 6.0% at the 5-, 10-, and 20-U doses, respectively (data not shown).

In this particular situation of premature needle withdrawal, the reusable pens Humapen Luxura HD and JuniorSTAR should be preferred to their prefilled counterparts (Kwikpen and Solostar) because they were more efficient and had a significantly lower PIUD.

These discrepancies among pen brands should be mostly explained by structural differences. First, the softness of the rubber used in the plunger might play an important role. Indeed, when the injection stroke starts, the first thing that happens is the screw pushes on the rubber plunger in the cartridge. The plunger, being made of rubber, will compress a small amount before it starts moving forward. A softer rubber will compress more at the start of the injection stroke and requires more time to expand back to its original shape after the injection is complete. Conversely, a harder rubber compresses less at the start and expands back to its original length more quickly. Additionally, the device's internal frictional properties might contribute to the injection time observed with insulin pens: the injection force applied on the injection button is indirectly transferred through all of the plastic elements of the pen to the insulin column and is also used for rotating the dose indicator, which must return to zero for the next injection. These steps result in frictional forces that probably occupy an incompressible amount of time, which is longer than that observed with a syringe where the injection force exerted on the piston is directly transferred to insulin, without any lag time. These structural characteristics also explain that the injection force required and the length of dial extension may be different between pens. ³

Furthermore, the physical characteristics of insulin, such as its viscosity, are likely involved in the duration of insulin stroke because we observed significant differences in PIUD by insulin type in the same pen (PIUD is significantly lower for glargine compared with glulisine with Solostar and JuniorSTAR).

Finally, inner needle diameter should also be involved; however, in our study, we used the same 4-mm needles, ruling out this alternative hypothesis. Our choice to use these 4-mm needles exclusively was guided by recent data demonstrating that these small needles result in the same glycemic control as 5- and 8-mm needles with less pain in adults and children. ^4,5

However, insulin underdelivery can easily be avoided by following the manufacturer's recommendation on the package insert of each insulin pen. To ensure that the entire dose is delivered, the needle should remain under the skin for 5 s (Kwikpen and Humapen Luxura HD), 6 s (Flexpen and Novopen Echo), or 10 s (Solostar and JuniorSTAR) after injection (per the official device manuals). For the Solostar pen, the recommended time before withdrawing the needle is longer, which is consistent with our results of a higher PIUD with the misuse of this pen.

Beyond the manufacturer's recommendations for the proper use of all insulin pens, an expert panel advised counting slowly to 10 before withdrawing the needle to deliver the full dose and prevent medication leakage. ⁶ This recommendation relies on an in vivo study that explored the leakage from six different insulin pens after 1–7 s of needle withdrawal; the amount of insulin lost was variable depending on the pen used and the timing of needle withdrawal. ⁷ This study was conducted over 10 years ago, and most of the pens tested are no longer used. More recent studies have assessed the accuracy of most of the prefilled and reusable insulin pens evaluated in our study (Kwikpen, Flexpen, Solostar, Novopen Echo, and Humapen Luxura HD). All of the tested insulin pens met the requirements for accuracy according the International Organization for Standardization (ISO) guidelines for insulin pen injectors (ISO 11608-1). ^{8 –11} However, our laboratory evaluation provides additional data because none of these previous studies has explored prolonged insulin leakage from the needle tip after premature needle withdrawal.

The loss of insulin at the needle tip should be distinguished from another related and previously described situation, which is insulin leakage from the puncture. ¹² Stewart and Darlow ¹² explored the amount of insulin leakage from the skin shortly after giving an injection using a filter paper technique. They demonstrated that 23% of insulin injections were followed by insulin leakage of more than 5% of the injected dose, resulting in irregular insulin delivery that may increase glycemic variability. ¹³ We demonstrate in our study that insulin leakage from the needle tip could also contribute to irregular insulin delivery and impaired glucose control in case of premature needle withdrawal.

Our study has several limitations. Only one needle size (4 mm) was explored. Using an automated test system to depress the injection button of the pen would have been more reproducible, but we preferred testing our hypothesis with manual handling as it more closely mirrored reality and allowed for global exploration of pen ergonomics (length, weight, pen diameter, injection button access, and stroke smoothness). ¹⁴ We acknowledge that pen handling in children's hands might have revealed different results compared with our evaluation by an adult operator (especially concerning the thumb length and strength required to push the injection button). Finally, this laboratory evaluation cannot preclude what might happen during a real injection through the skin; therefore, the potential consequences on patient glycemic control are only speculative.

However, our results should promote clinician awareness of the importance of carefully teaching the insulin injection technique, especially for patients requiring small doses. Patient education should be reinforced, highlighting the importance of waiting before withdrawing the needle after an insulin pen injection to avoid insulin underdelivery, which can represent up to 20% of the selected dose.