All parenteral delivery systems for medication administration must ultimately interface with the target delivery tissue during usage. During the delivery process, multiple factors including device design, human factors and ergonomics, and tissue biomechanics may impact delivery success and consistency. For the treatment of diabetes, self-administered insulin therapy via daily injections using pen injection systems is established and widely practiced.
1 Pen needles are an essential component of these systems providing the injection site interface and ultimate conduit for insulin delivery into the target subcutaneous (SC) tissue. A body of literature is available on skin and SC tissue thickness at common anatomical insulin injection site locations (abdomen, thigh, arm, buttocks) for both adult and pediatric populations.
2-6 This information has been utilized to develop shorter and finer gauge pen needles to facilitate SC delivery while minimizing intramuscular (IM) and intradermal (ID) injections, leakage, pain, and injection anxiety.
7-11 In addition, published injection technique guidelines
12-14 identify best practices for diabetes therapy management and provide recommendations for patient education, injection site selection and rotation, and needle length selection to improve therapy adherence and efficacy.
Pen injection devices themselves have continually advanced, incorporating design elements that improve device function and usability.
15-17 Similarly, a multitude of experimental and simulation-based investigations examining needle-tissue interactions have influenced needle dimensions/bevel design and procedures for use.
18-24 However, there is limited published data investigating the impact of pen needle (PN) device hub designs, user injection technique factors and their interaction, on needle penetration depth. In a prior study, highly variable intra- and intersubject applied skin force and injection techniques were observed during simulated injection procedures, including significant differences between injection sites and dose delivery volumes.
25 This technique variability may originate from various learned behaviors due to human factors including injection site accessibility or visibility, restricted or reduced hand dexterity, individual preference, or perceived comfort, regardless of best-practice instructional training.
A PN design that would decrease the injection variability resulting from these inter- and intrapatient differences in technique and applied force could help reduce the variation often seen in insulin absorption.
26 The purpose of this in vivo preclinical study was to examine the influence of PN hub design across a range of clinically observed skin application forces on needle penetration depth (NPD), deposition consistency, and consequent IM injection risk.
Methods
Force Measurement
The Insertion Force Measurement Tool (IFMT) and data acquisition system were custom designed to measure applied force exerted against the skin during injection procedures. The IFMT housing contains threaded features for connection of a single pen injector system (ClikSTAR®, Sanofi-Aventis, Frankfurt,Germany) and various PN types via the existing thread connectors on each component. A transfer needle within the housing provides a fluid path conduit between the cartridge and the PN to enable dose administration. A donut load cell (Futek, Irvine, CA, USA) resides within the housing to acquire and transmit force data during the PN injection. Acquired force data was observed in real-time using a custom LabVIEW™ graphical user interface.
Study Devices
Four 32 G × 4 mm PN devices were evaluated in this study: the BD Nano™ 2nd Gen, also known as BD Nano™ PRO in some markets (Becton Dickinson, Franklin Lakes, NJ, USA: Investigational Device) and three commercially available PNs with the conventional posted-hub design (Device A = BD Nano™ Ultra-Fine™, Becton Dickinson, Franklin Lakes, NJ, USA; Device B = Clickfine®, Ypsomed AG, Burgdorf, Switzerland; Device C = Unifine® Pentips®, Owen Mumford, Woodstock, Oxfordshire, UK;
Figure 1). Posted-hub PN designs utilize a small diameter cylindrical feature extending 3-4 mm axially from the hub base to support the needle. The patient-end (PE) length (ie, the absolute length between PN polymer hub and needle tip) for each PN tested in the study was measured under 60x magnification, uniquely numbered and recorded to allow subsequent comparison between PE length and NPD. PE length dimensions for all test devices met ISO 11608-2: 2012 requirements; 4 mm ±1.25 mm.
Animal Model
To date, no in vitro mechanical model exists that can simulate the complexities of an intact in vivo integumentary system. Swine are the preferred models in wound healing and plastic surgery research due to the similarities of their skin to human skin tissue
27 and are an ideal species for evaluation of insulin infusion/injection products apart from clinical evaluation. Swine are tight-skinned animals like humans, the presence of a dermal collagenous network makes their skin elasticity more like humans than other mammal models.
28,29 The model possesses structural and compositional dermal and SC tissue characteristics (pigmentation, hair paucity, skin thickness, dermal to epidermal ratio, adipocyte size and layering, compliance, drainage, absorption, and resistance)
29-33 and biological variables (breathing, movement, tissue variability, similar vasculature patterns)
29,31 similar to humans that affect devices during use. In addition swine provide sufficient surface area to allow multiple devices to be tested in parallel without significant discomfort or physical impairment to the animal, at dosing rates and volumes applicable to humans. A difference between swine and human SC structure is the presence of the panniculus carnosus
34 in swine, a layer(s) of striated muscle in the SC tissue that contributes to skin movement (twitch reflex). In this study injections and in vivo imaging were administered to the flank of 35-40 kg Yorkshire swine. The flank injection site provides a diminished panniculus carnosus and SC thickness (~8 mm) similar to humans.
Animal Care
All animal care and preclinical procedures were conducted in accordance with National Research Council standards of care and ethics at an AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care) accredited facility under an experimental protocol approved by the internal Institutional Animal Care and Use Committee (IACUC).
Study Design and Procedures
The study was conducted over a multiweek period consisting of two procedures per week for each animal. After inducing anesthesia the animal was placed under the fluoroscope (Glenbrook Technologies LabScope™) and positioned to enable device insertion perpendicular (90°) to the beam path. Prior to administration an array of 20 injection sites were identified and numbered from anterior to posterior, on both right and left sides of the animal. For each procedure 20 µl (2 IU, U-100 volume equivalent) injections of contrast agent (Omnipaque™ 350 mg Iodine/ml, GE Healthcare, Oslo,Norway) were administered to the flank of the anesthetized swine, with injection order randomized by PN test device.
Force levels selected for evaluation were based upon prior simulated use study results where experienced insulin pen users of both genders applied noncannulated prototype PN designs to the skin surface and simulated injections to the abdomen and thigh. A frequency distribution for these measured application forces pooled across gender, prototype PN device, and injection site was generated (
Figure 2). Applied force levels for this study were centered on the following values: 0.25 lbf [1.1 N], 0.75 lbf [3.3 N], 1.25 lbf [5.6 N], and 2.00 lbf [8.9 N]. The force level range encompasses greater than 80% of observed application forces from the simulated use study and force targets were anticipated to provide adequate separation between levels based on the operator ability to apply target forces within ~±0.25 lbf [±1.1 N]. For each study device, n = 75 injections per force level were administered.
The operator targeted a specific application force level for each injection during delivery of contrast agent. Force data was acquired and displayed in real-time by the IFMT system during each insertion and injection for the target force levels (
Figure 3). Each force data file included unique identifiers (date, animal number, injection site side/number) to enable cross-reference to device NPD, PE length, and erythema observations.
A fluoroscopic image was collected immediately after injection to visualize depot location and obtain depth measurements. The injection volume was limited to 20 µl, as this created a reliably detectable marker for measuring needle total penetration depth without convolution of depot spreading and diffusion from larger volumes that would reduce the accuracy of penetration depth measurement (
Figure 4). NPD was measured from 2D fluoroscopic images and erythema Draize scale ratings scored and recorded immediately after injection. Absolute NPD depth measurements were calculated against a radiopaque mm scale (0.1 mm graduations) placed in the image FOV during image capture. A parallax correction factor was applied to all NPD measurements to accommodate the minor magnification difference between the relative position of the imaged tissue and calibration standard to the X-ray source. The correction factor was consistent for all measurements and applied to the data prior to statistical analysis. Identical study methods were utilized for subsequent procedures.
Peak forces were identified on each profile utilizing the
findpeaks function (MATLAB R2017b), followed by manual operator review to confirm programmatic peak selection. In the event multiple peaks were recorded on the force tracing due to animal respiratory motion the operator manually selected the peak used for analysis. Peak forces were correlated against NPD and erythema scores for analysis. The primary endpoints for this study were device NPD, with corresponding measured applied force, and injection site erythema scores (
Table 1). IM risk was calculated using preclinical NPD distributions obtained from the study and published human skin thickness data; method described below in the Statistical Analysis section.
Statistical Analysis
NPD endpoints were analyzed using a linear mixed effect model, and erythema scores using cumulative logit model, with device and force category factors as main effects including interactions. Multiple comparisons were performed per device and per device per force category using the contrast method, Wilcoxon or variance tests adjusted for multiple comparisons using Bonferroni’s adjustment method. A statistical model was developed to calculate ID and IM risk using device NPD measurements and published skin thickness data
3,8 at the four common insulin injection sites for adults, pooled and stratified by factor (gender, BMI, injection site). Tissue ID thickness, tissue IM depth, and device NPD values were generated using a random number generator function from the respective distributions. Monte Carlo simulations were run at a sample size n = 500 for each combination of tissue thickness and NPD. For each pair of (1) ID depth and NPD and (2) IM depth and NPD, the ID and IM risk was calculated as the proportion of tests where either NPD < ID depth or NPD > IM depth was TRUE, respectively.
Results
In total, 1200 injections (30 procedures, 40 injections/procedure) were completed over a multiweek period on Yorkshire swine (n = 7). 1188 injections were available for analysis; n = 8 injections were excluded due to bent needles during administration which would affect depot location, and n = 4 were excluded due to force acquisition error. Measured applied forces were compared to the target predefined force levels; any measurements outside the targeted force range (<7% occurrence for any category) were reassigned to the appropriate force category.
Needle Penetration Depth
The mean NPD pooled across force levels demonstrate deeper insertion depth for posted-hub PN designs (
Figure 5). The comparative analysis for NPD and variance is presented in
Table 2. Insertion depth (mean ± SD) for BD Nano
™ 2nd Gen (4.25 ± 0.56 mm) was significantly less than Device A (4.97 ± 0.68 mm), Device B (5.85 ± 0.72 mm) and Device C (5.64 ± 0.73 mm) 32 G × 4 mm PN devices using a superiority margin of 0.5 mm. Reference methods section for device key. Amongst posted-hub design NPD, Device A was significantly less than Device B and Device C; no significant difference was observed between Device B and Device C. The nonposted BD Nano
™ 2nd Gen NPD variance was significantly less than comparator commercial devices; no significant differences in variance were observed among the posted-hub PN designs.
Mean NPD of the BD Nano
™ 2nd Gen was closer to the target 4 mm PE needle length across the range of applied forces compared to posted-hub PN devices, which displayed ~0.5 to 1.75 mm deeper NPD with differences proportional to increased applied forces (
Figure 6,
Table 3). To ascertain conformity of NPD to the stated 4 mm PE length the proportion of injections falling within a 4-4.5 mm injection depth range was evaluated for each device, pooled across injection forces. BD Nano
™ 2nd Gen demonstrated a statistically significant higher percentage of injections within this range (41.1%), versus Device A (17.9%), Device B (3.3%), and Device C (4.7%) posted-hub devices (
Figure 7,
Table 4). Device A was statistically significantly different from Device B and Device C; no significance difference between Device B and Device C. Measure of the linear relationship between device NPD and labeled needle PE length using Pearson’s rho correlation was completed across and within force levels; weak negative correlation (0 < |
r| < .3) between these variables was observed indicating labeled PE length alone is not a reliable predictor of penetration depth.
Injection Site Erythema
Assessment of injection site tissue effects was completed using erythema Draize scoring immediately after insertion and dose delivery. Majority of sites had no observable erythema reported for BD Nano
™ 2nd Gen (80.5%) compared to the other test articles (Device A: 45.9%; Device B: 35.8%; Device C: 36.1%). Erythema associated with use of BD Nano
™ 2nd Gen was significantly less than the commercially available, posted-hub PN devices both pooled across (
P < .001) (
Table 5) and within applied force levels greater than 0.5 lbf [2.2 N] (
Table 6). Erythema scores were observed to be proportional to the applied force for posted-hub PN devices, presenting higher scores with increasing forces (
Figure 8).
Estimated ID and IM Risk
ID and IM risk was calculated from paired NPD/tissue thickness measurements utilizing the preclinical NPD distribution, pooled across forces for each device, related to human skin thickness distributions obtained from literature
3,8 for each injection site (
Figure 9). For each device, calculated IM risk at injection sites increased with decreasing BMI and was more pronounced in males versus females within BMI categories. The thigh site had the highest calculated IM injection risk when pooled across gender and BMI, followed by arm, abdomen, and buttock. Calculated IM risk for BD Nano
TM 2nd Gen was decreased for each gender at all injection sites and BMI levels, and displayed ~2-8x IM risk reduction versus posted-hub designs across injection sites when pooled across gender and BMI (
Table 7). Average calculated ID risk was 0.2% for BD Nano
TM 2nd Gen and 0.0% for posted-hub device irrespective of group (gender, BMI, and injection site). The largest ID risk observed from subgroup analysis was 1.8% for the BD Nano
TM 2nd Gen device in males: BMI ≥ 30, buttock injection site.
Discussion
This preclinical study is the first to examine the relationship between forces applied during self-injection procedures and PN hub design on in vivo NPD, and is the first to demonstrate meaningful differences in NPD and calculated IM injection risk based on PN hub design. The primary objective was to quantify device NPD and calculated ID and IM needle insertion potential based on these measurements. Fluoroscopic imaging historically has been utilized to guide needle and catheter insertions to targeted anatomy in medical procedures. We adapted this methodology to enable direct observation of phenomenon such as tissue compression during needle insertion, including in situ visualization and characterization of injectate depositions.
35,36 A small 20 µl (2 IU, U-100 volume equivalent) administration of contrast agent was utilized to represent in vivo needle tip position to mitigate the influence of the injection process on NPD measurements, such as compressive forces placed on tissue at the insertion site, and prevent masking of NPD measurement due to inhomogeneous depot dispersion and spreading seen with larger volume deliveries. Measurements were completed immediately post device removal after tissue relaxation from the compression during delivery.
A statistically significantly smaller mean NPD and depth variance for BD NanoTM 2nd Gen was demonstrated compared to commercially available 4 mm posted-hub PN designs across a range of clinically relevant skin application injection forces. Although increasing application force produced deeper mean NPD for all PN hub designs tested, the force dependent relationship was less pronounced for the BD NanoTM 2nd Gen PN design. Across posted-PN designs, mean NPD was greater than 4.6 mm at the lowest applied force level (0.0-0.5 lbf [0.0-2.2 N]), and on average less than 10% of injections were measured within a 4-4.5 mm depth range with the majority of measurements greater than 5 mm. In contrast, BD NanoTM 2nd Gen displayed limited mean NPD changes across the applied force range (4.14 mm @ [0-0.5 lbf] to 4.45 mm @ [1.75-3.0 lbf]) more closely targeting the 4 mm PE length, with ~41% of injections within a 4-4.5 mm depth range. The influence of PE length on resulting NPD was considered and examined through linear correlation of the respective measured length and depth for each device; the weak relationship observed both across and within applied force levels reinforces the finding that PE length alone is not a reliable predictor of penetration depth.
Although differences in individual injection technique are known to physicians and diabetes HCPs, a means to quantify inter- and intrapatient variability has not been previously identified, and consequently the potential impact on delivery consistency has not been investigated. In our prior study, quantifying forces applied to the skin during PN injection and observation of diverse injection techniques revealed a previously unknown and highly variable factor during user injection technique; applied force.
25 In addition to needle length, the physical act of injection and the skin-interfacing PN design influence NPD. PN designs with a single post feature surrounding the needle base have a small surface area and create an area of concentrated pressure on and below the skin surface during device application to the skin. Increasing the applied force will further compresses underlying tissue resulting in deeper needle seating and injection. Conversely, PN designs that do not sufficiently engage the skin during insertion may limit needle penetration producing dose leakage or overly shallow delivery. An optimized PN interface designed to ensure full needle insertion, while subsequently increasing skin surface contact area to distribute the applied skin pressure, will limit needle penetration to the target PE needle length regardless of applied forces (
Figure 10). Mean NPD also differed between the tested posted-hub PN devices. The observed disparity may reflect slight differences between PN device geometry at the skin interface, however these were not subsequently investigated.
The standard route for insulin delivery is injection or infusion into the SC adipose tissue, from where it is absorbed into the bloodstream. ID and IM administration is generally to be avoided due to resulting differences in insulin kinetics and the potential impact on therapy predictability. IM insulin injection uptake variability has been associated with muscle activity level, and subsequently less predictable glycemic control.
12,26,37-42 More rapid insulin uptake and increased bioavailability have been previously demonstrated in both preclinical animal models and clinical studies using ID delivery of insulin and other protein therapeutics.
43-54 Conversely, injection site leakage due to shallow SC delivery could result in reduced insulin delivery, elevating blood glucose levels.
NPD results indicated that the BD Nano
TM 2nd Gen PN hub design may reduce the impact of unintended user applied injection force differences and subsequently inadvertent IM injections compared to posted-hub PN devices of equivalent lengths. Analyses were conducted to investigate ID and IM risk potential of the different hub design using skin and SC tissue thickness measurements at four common insulin injection sites (abdomen, thigh, arm, buttock) from diverse adult population with diabetes reported previously.
3,8 Based upon this dataset, men have a higher risk of IM injection compared to women at all injection sites, with an inversely proportional relationship to BMI for both genders. As IM risk is directly related to SC thickness thigh presents the highest risk, followed by arm, abdomen, and buttock injection sites. A considerable reduction in calculated IM risk between the BD Nano
TM 2nd Gen PN hub design and commercial posted-hub designs was observed; ~2-5-fold decrease for thigh, abdomen, arm and ~2-8-fold decrease for buttock. No appreciable difference in ID risk was observed between device designs. Limitations of this study include use of a preclinical animal model in place of self-injection by human subjects for NPD measurements. This was a necessary consideration due to the risk of required repeated subject exposure to ionizing radiation for deposition imaging via fluoroscopy. In addition, not all commercial PN hub designs were evaluated in the study.
The 2014-2015 ITQ survey
55 reported the abdomen as the most frequently used injection site (90.9%) amongst respondents, followed by thigh (43%), arm (31.9%) and buttock (13.8%). A majority of respondents (70%) reported using a < 8 mm length needle, representing a ~25% increase from the 2008-2009 ITQ survey. PNs have progressively become shorter and thinner to decrease IM risk, increase deposition into SC tissue with less pain, minimize dose leakage from injection site, and enhance patient therapy adherence. IM risk is related to both needle length and injection site, however differences between equivalent PE length devices have not been previously demonstrated. The statistically significant differences between posted-hub PN designs and the BD Nano
TM 2nd Gen indicates that the hub design, coupled with injection technique, is a key factor significantly influencing needle depth accuracy and consistency.
Conclusions
We have used noninvasive imaging methods to observe and characterize factors influencing device performance for SC injection therapy. SC delivery devices must reliably administer medication to the target tissue space considering both the mechanical limitations of the device system and physical limitations and behaviors of the human operator. This study is the first to evaluate PN design impact on in vivo needle depth between equivalent labeled PE needle length devices. The results indicate hub design, in addition to needle length and injection technique, is a primary factor governing resulting NPD. Reducing the impact of variable user applied forces, which cannot be otherwise controlled, may improve SC delivery consistency during self-injection procedures.
Acknowledgments
We thank the numerous extended team members from BD who contributed to the successful planning, organization, execution, and analysis of this study.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: CR, BCR, DM, RK, BS, and RJP are employees and shareholders of BD, which sponsored this work.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Becton Dickinson.