The advancement of point-of-care diagnostics and the decentralization of healthcare have created a need for the simple, safe, standardized and painless collection of blood specimens. Here, we describe the design and implementation of a capillary blood-collection device that is more convenient and less painful than a fingerstick and venepuncture, and collects 100 µl of blood. The technology integrates into a compact, self-contained device an array of solid microneedles, a high-velocity insertion mechanism, stored vacuum, and a microfluidic system containing lithium heparin anticoagulant. The use of the device requires minimal training, as blood collection is initiated by the single push of a button. In a clinical study involving 144 participants, haemoglobin A1c measurements from device-collected samples and from venous blood samples were equivalent, and the pain associated with the device was significantly less than that associated with venepuncture. The device, which has received premarket clearance by the US Food and Drug Administration, should help improve access to healthcare, and support healthcare decentralization.
Clinical diagnostics is integral to many fields of medicine, and blood continues to be the preferred sample medium for testing. Yet, blood sample collection technology has not kept pace with the advances in diagnostic assays and instrumentation that have allowed rapid measurements to be performed on small volumes of blood outside of the traditional clinical laboratory setting. Blood samples for diagnostic testing are still routinely obtained by venepuncture or by lancing the finger (‘fingerstick’), both of which have certain disadvantages.
Venepuncture requires substantial training to perform the collection, and success in obtaining a sample is dependent on the skill of the phlebotomist performing the blood draw in addition to the accessibility of the patient’s veins. The use of a hypodermic needle for venepuncture not only poses the risk of needle-stick injury to the phlebotomist, but can also cause pain and anxiety for the patient1,2, which can be a barrier to receiving proper healthcare3,4. In addition, the volume of blood typically collected by venepuncture (≥2 ml) is no longer required for many tests5,6,7,8. For example, many currently commercially available diagnostic instruments can measure panels of analytes from a blood sample of around 100 µl.
Sampling capillary blood by fingerstick requires less training than venepuncture and can be self-administered. Disadvantages to fingerstick sampling are that it is painful, and it can be difficult to reliably obtain a blood sample of sufficient volume and quality for testing. It is often impractical to collect and mix small volumes of blood with anticoagulants or stabilizers, so capillary samples are usually tested immediately at the point-of-care. The practice of lancing ‘alternate sites’ such as the arm, thigh, or palm has been used to alleviate the pain associated with a fingerstick, but the lower capillary densities of these regions make it even more difficult to obtain an adequate sample volume for testing9,10,11. On the basis of these limitations of both venepuncture and fingerstick sampling, there is a need for a simple, safe, standardized, and painless method for blood collection both within and outside traditional healthcare settings12.
Here, we describe the design and implementation of a device for painless and automated capillary blood collection that uses microneedles to pierce the skin and vacuum to withdraw blood. The device requires minimal training, protects the user from exposure to sharps and blood, mixes the sample with anticoagulant throughout the collection process, and stores the sample for subsequent analysis. The device yields 100 µl of heparinized whole blood, which is suitable for many of the most frequently ordered point-of-care tests, including chemistry and wellness screening panels. We discuss the design and optimization of the key technological components of the device and demonstrate the clinical utility of the device in the measurement of haemoglobin A1c (HbA1c) with two point-of-care instruments.
Painless skin puncture using microneedles
Microneedles have been studied for a variety of applications, including transdermal drug delivery13,14,15,16,17, extraction of interstitial fluid (ISF)18 and transdermal biosensing19. Hollow microneedles have also been explored for use in obtaining blood; however, it is unclear if they could reliably be used to collect blood of sufficient quality and quantity for diagnostic applications20,21,22. We chose to use solid microneedles in our device to create punctures in the skin from which blood is then extracted. We also hypothesized that an array of microneedles, as opposed to a single microneedle, would be necessary to ensure a high probability of striking a sufficient number of capillary blood vessels to obtain good blood flow.
The elasticity and tear-resistance of skin make needle penetration difficult, with significant skin compression required before the stratum corneum is breached23. This difficulty is magnified for insertion of microneedle arrays, where slow manual insertion has been shown to produce incomplete or non-uniform insertion24. The fracture toughness and viscoelastic properties of the skin can be overcome by using a high velocity insertion mechanism to give the microneedles sufficient kinetic energy25. The velocity needed to achieve reliable insertion depends on the geometry and mass of the microneedle array, but is commonly on the order of 1 to 10 m s–1 (refs 26,27).
To identify a suitable microneedle insertion method for our device, we examined seven actuation mechanisms: six found in commercially available finger lancing devices and a snap dome. Snap domes are a type of disc spring typically found in membrane switches such as a keypad. Disc springs with certain ratios of thickness to height are bi-stable28, where deflection beyond a certain point will produce a stable inverted configuration. When an inverted snap dome is triggered, it will accelerate rapidly as it returns to its resting state. We recorded the movement of each mechanism using high-speed video, and determined the maximum acceleration and the maximum velocity (Fig. 1a). Based on these measurements, we selected the snap dome for use in the device. The mean maximum velocity of the snap dome was 8.7 m s–1, which is sufficient for reliable microneedle insertion. In addition, the snap dome had a mean maximum acceleration of 86,600 m s–2, which was more than three times greater than the acceleration of the next highest insertion spring. With this acceleration, the snap dome can reach its maximum speed within a very short travel distance (1.8 mm), allowing a device to have a low profile.
To determine the optimal microneedle array dimensions for use in our device, we began by evaluating the effect of microneedle length on blood flow rate and perceived pain over a range of 750 to 1250 µm. This was investigated by inserting microneedle arrays into the skin using the snap dome mechanism and then applying vacuum to the puncture sites to encourage blood flow. It was found that the blood flow rate from the punctures substantially increased with microneedle length up to 1000 µm, then levelled off (Fig. 1b). Pain sensation also increased with increasing microneedle length. From these results, we selected a microneedle length of 1000 µm (Fig. 1c) to efficiently access the capillary beds and maximize blood flow rate while minimizing stimulation of the nerve endings that causes pain29,30.
Capillary density per unit area of skin is reported to be in the range of 14–70 capillaries per mm2, depending on location31,32,33. It is likely that an array of multiple microneedles would be necessary to ensure that at least one capillary is struck during insertion into the skin. To ensure a high probability of capillary blood access, we used Monte Carlo simulation as a convenient method to combine assumptions about the distribution of capillaries in the skin and the configuration of a microneedle array. Our simulations suggested that an array with eight needles of 350 µm width and 50 µm thickness should result in at least one capillary strike during insertion into the skin (p = 0.94689). An array with 30 microneedles would provide an even greater degree of redundancy (5–16 capillary strikes, p = 0.96163). Arrays of similar needles have previously been reported to be sufficiently rigid to provide good insertion with minimal pain30. From these results, a final microneedle array design that has 30 needles with 50 µm thickness, 350 µm width, and 1000 µm length was selected for the device.
Blood extraction using stored vacuum
We observed that when skin is simply punctured with microneedles, no more than trace amounts of blood emerge from the puncture sites. The application of vacuum has been shown to improve the blood volumes that are collected when sampling from alternate sites11,34. We found that applying vacuum to the area surrounding the microneedle puncture sites encourages localized blood flow, stretches the puncture sites open, and leads to substantially increased blood flow rates. To optimize the level of vacuum used for blood collection, we tested devices with −5, −12 and −20 inHg of vacuum, using a vacuum reservoir connected to each device. The blood flow rate from the puncture sites was observed to increase with increasing vacuum up to −12 inHg, beyond which there was no significant increase (Fig. 1d). We observed considerable subject-to-subject variability in flow rate, which we believe is related to differences in capillary density between different people. The final version of the device is evacuated to apply a vacuum in the range of −15 to −16 inHg to the skin.
Selection of the blood-collection site
The density of blood vessels in the skin varies significantly between locations35. We tested devices on the forearm and upper arm of volunteers, and observed higher flow rates for most individuals when sampling from the upper arm with no substantial difference in pain (Table 1). We therefore selected the upper arm as the initial sampling location for the device.
Anticoagulation during blood collection
Blood samples collected by venepuncture are commonly mixed with an anticoagulant to allow the samples to be analysed several hours or days after collection. Blood is collected into an evacuated tube containing anticoagulant in wet or dried form, then the phlebotomist mixes the anticoagulant with the blood by inverting the tube multiple times immediately after collection. Our goal was to collect an anticoagulated blood sample without the need for such manipulation by the user. We therefore designed features that would gradually release anticoagulant into the blood sample as it is collected and ensure thorough mixing as the sample flows through the device. We used dried lithium heparin as the anticoagulant for all work described here. As blood emerges from the microneedle puncture sites, the sample is collected by the device and mixed with anticoagulant through the following steps: (1) the blood first contacts a capillary ring in the immediate vicinity of the puncture sites, which wicks blood from the skin and into the device; (2) dried anticoagulant located in the capillary ring is steadily dissolved by the blood as it flows into the device; (3) a flow channel that connects the capillary ring to the sample collection reservoir contains mixing features that evenly distribute the anticoagulant into the sample during the collection process (Fig. 2).
The combination of very low Reynolds number of blood flow in our device (Re < 1), and very low diffusion coefficient of heparin in blood (D ≈ 1.2 × 10−10 m2 s–1)36, means mixing can only be achieved through convective flow within the channel. We evaluated different static, low-Reynolds-number mixers and decided to implement a staggered herringbone mixer37. This mixer consists of a series of chevrons in alternating directions, which create advection in the direction perpendicular to the main direction of flow. The advection occurs in a manner that stretches fluid filaments and creates a contact surface between the fluid filaments that grows exponentially with the distance travelled along the channel. The dimensions of the channel and chevrons were optimized using computational fluid dynamics (CFD) simulations, to achieve efficient mixing and low back-pressure. We predicted that a channel of 28 mm length, 1.2 mm width, and 0.225 mm height containing 16 chevrons 0.3 mm deep would achieve adequate mixing. Figure 3a shows the result of a simulation of this flow -channel design, in which two streams of liquid enter the channel at the left and exit fully mixed at the right. In vitro experiments using dyed liquid confirmed that this design achieved full mixing in the transit time between the channel entrance and the reservoir (typically 10–30 s) (Fig. 3b).
For effective anticoagulation of the blood sample, we found it necessary to ensure sufficient anticoagulant is available in the capillary ring throughout the duration of blood collection. We slowed the dissolution of anticoagulant by using a hydrophilic polymer membrane layer to partially obstruct blood access to the dried anticoagulant in the capillary ring. The membrane was patterned with small cuts, such that some blood wicks under the membrane and dissolves the anticoagulant, then mixes with the remaining blood sample. The device collects approximately 150 µl of blood when dead volume is accounted for and yields 100 µl in the sample reservoir, therefore, we evaluated whether sufficient anticoagulant was released beyond 150 µl. Table 2 compares the anticoagulant release profile for devices with and without this membrane to control the anticoagulant release. The concentration of anticoagulant in the later fractions was higher and variability in the anticoagulant concentration within each fraction was reduced with the controlled release design.
Final form of the device
An integrated version of the device, commercialized as 'TAP', uses a stainless-steel microneedle array, a one-step actuation mechanism, and a pre-evacuated vacuum chamber, in conjunction with a microfluidic channel and sample reservoir system containing lithium heparin anticoagulant. These components are assembled in a compact (4.7 cm diameter, 3.5 cm height), single-use plastic housing, and all the materials in the device are biocompatible. The basic operation of the TAP device is illustrated in Fig. 4. The skin is first swabbed with alcohol and the device is applied to the skin. A hydrogel adhesive on the bottom of the device forms an air-tight seal between the device and the skin. Blood collection is initiated by pushing the button on the top of the device. Upon actuation, the microneedles are rapidly deployed into the skin and immediately retracted into the device, leaving punctures in the skin surface. The vacuum stored in the device is then applied to the punctured area of skin. The vacuum draws blood from the skin punctures through the flow channel and into the fixed-volume sample reservoir. Once the sample reservoir is filled, a visual indicator turns red to signal that the collection is complete. The device is then removed from the skin and can be transported to the sample testing location. The base of the device has a foil seal, which can be pierced with a pipette tip to access the reservoir and extract the sample for testing. Throughout every step of blood collection with the device, exposure to potentially infectious blood and sharps is minimized. The TAP device typically takes 3 minutes (mean 2:54, range 0:34–7:00, standard deviation 1:21, n = 643) to collect a sample (mean yield of 103.5 ± 19 µl, with 91.4% of devices yielding > 90 µl, n = 643). The collected blood is generally free of clots (a single small clot up to 2 mm in diameter was observed in 4.8% of samples, n = 475) and haemolysis is generally low (mean plasma haemoglobin of 100 mg dl–1, range 40–230 mg dl–1, n = 69). We also observe that total haemoglobin in blood collected with the TAP device is not lower than in venous blood, which suggests that ISF contamination is minimal.
Clinical application of the device for HbA1c testing
We evaluated the performance of the TAP device for HbA1c determinations in a clinical setting. Results from capillary blood samples collected using the TAP device were compared to those from venous blood samples collected by venepuncture using the DCA Vantage (Siemens) and the Afinion AS100 (Alere). The study was conducted at 3 hospital sites with 143 total participants. The people who performed the blood collections were healthcare workers trained in phlebotomy (for example, nurses, phlebotomists).
HbA1c results for blood samples collected using the TAP device and by venepuncture correlated well (r2 value of 0.988, n = 243) (Fig. 5a). In addition, a Bland–Altman analysis of the data (Fig. 5b) showed the mean percentage bias between HbA1c test results for samples collected using the TAP device relative to venepuncture was −0.1% with a 95% confidence interval of −5.1 to 4.9%. The National Glycohemoglobin Standardization Program (NGSP) sets performance criteria for the measurement of HbA1c38. For a method certification study, the NGSP specifies a minimum of 40 samples for comparison, and requires that at least 37/40 (>93.5%) of HbA1c test results be within ±6% of the reference value. Data from our study showed that greater than 95% of differences between TAP device and venepuncture HbA1c results were within this performance criterion. We therefore conclude that capillary blood samples from the TAP device are equally suitable for performing HbA1c determinations as venous blood samples collected by venepuncture.
We also compared the pain sensation reported for blood collection using the TAP device and venepuncture. For each sample collection, pain sensation scores were recorded by each patient using the Wong–Baker FACES scale, which ranges from 0–10, where 0 corresponds to “no hurt” and 10 corresponds to “hurts worst“39. The range of pain scores reported for the TAP device and for venepuncture were 0–4 and 0–10, respectively. The mean pain scores for blood sampling with the TAP device and venepuncture were 0.4 (n = 481) and 1.5 (n = 162), respectively, with the TAP device being rated significantly less painful than venepuncture (p < 0.001 by two-sided t-test). In a prior study using an earlier iteration of the TAP device, we observed the mean pain score for the device (0.8, n = 80) was also significantly lower than fingerstick (1.9, n = 80) (p < 0.0001 by two-sided t-test).
Dermal response at the TAP device sampling sites was examined and scored immediately following the blood sample collection as well as 20 minutes later. The dermal response scale used for this assessment ranged from 0–7, with 0 indicating no evidence of dermal irritation and 7 indicating a strong reaction spreading beyond the sampling site40. The mean TAP device dermal response score across all subjects was 0.1 immediately after sampling (n = 481) and 0.6 just prior to discharge (n = 479), which falls between the descriptors ‘no evidence of irritation’ and ‘minimal erythema, barely perceptible.’ Similar dermal response scores were observed for the venepuncture sampling sites, with a mean of 0.1 immediately after sampling (n = 162) and 0.5 just prior to discharge (n = 152). In addition, no unresolved issues with the sampling sites were reported by the participants during a scheduled follow-up call five days after testing. Based on these results, we concluded that the TAP device does not produce any clinically significant dermal response.
We have described a device for the automated collection of capillary blood samples. Operation of the device requires only the single push of a button to collect 100 µl of whole blood for diagnostic testing. The sample-collection process is reproducible between different operators across multiple settings. On the strength of the clinical testing presented here, our microneedle-based blood collection device was the first of its kind to receive premarket clearance by the US Food and Drug Administration.
For the patient, our device can improve access to diagnostic information and provide an alternative to more painful and inconvenient forms of blood collection. The design of the device prevents the patient from seeing the microneedles and collected blood, which can reduce anxiety. The use of microneedles and a shallow insertion depth also minimizes any pain sensation. These features may be particularly useful for patient populations that are averse to blood sampling. Home blood-glucose monitors requiring less than 5 µl blood are widely used by laypersons, but it is difficult to reliably collect 100 µl blood from a self-administered fingerstick. Because our device is easy to use and does not require a trained healthcare professional to operate, it has the potential to enable the collection and testing of blood samples beyond the professional healthcare setting, including in the home. The initial commercial version of the device (TAP) is intended for use in professional healthcare settings. However, broader and more convenient access to diagnostic testing has the potential to increase information about disease conditions and improve healthcare outcomes.
The volume of blood collected by this initial version of the device is well suited to many point-of-care instruments. However, many of the larger automated clinical laboratory analysers currently require more blood, primarily due to dead volumes in liquid handling systems. We believe it will be possible to further optimize the device to collect larger volumes of blood even more rapidly. For example, our research has suggested that modifying the geometry and material of the device and the design of the microneedle array can achieve blood flow rates of more than 350 µl min–1. These advancements are being integrated into a future generation of the device that collects 250 µl of blood, which could further expand the compatibility of capillary blood samples with available test panels and systems.
We anticipate that the TAP device will cost about US$5, which is more expensive than the supplies used for venepuncture or fingerstick. However, when comparing the cost of TAP with these sampling methods, one must consider the total clinical and economic costs involved in the pre-analytical workflow. TAP takes minimal set-up time with fewer supplies, and collects a sample in a few minutes, whereas routine venepuncture takes approximately seven minutes41, requires more supplies, and is subject to greater costs when difficult-to-draw patients are considered. Eliminating the need for extensive training means that blood samples could be obtained with TAP more easily and in more locations, which could also improve efficiency and access to care. In addition, TAP standardizes the sample-collection process to minimize pre-analytical error. Because it only takes a single push of a button to collect a blood sample with TAP, human error and variability in technique is minimized. We anticipate that TAP will also provide more subjective benefits, such as reduced patient anxiety and stress, which could lead to improved compliance to prescribed therapies that require blood collection. Because the TAP device simplifies the sample-collection workflow, collects sufficient blood for most point-of-care tests and provides a painless sampling process, we believe that it will be a useful tool to support the current trend of healthcare decentralization.
Theoretical determination of capillary strikes
Monte Carlo simulations were performed using MATLAB (Mathworks). Each trial assessed the number of capillaries (modelled as circles placed randomly in a two-dimensional region) intersected by an array of needles (modelled as 350 µm x 50 µm rectangles). Capillary diameter was assumed to be 15 µm (ref. 33). A 'worst case' capillary density of 14 capillaries per mm2 was used, based on the lowest value found in previous literature33. Inter-capillary spacing was constrained to be >80 µm (ref. 42). Results from 100,000 trials were analysed to assess the distribution of the number of capillaries struck by each needle array design. Probabilities were estimated by normalizing the frequency of a particular outcome against the total number of trials.
Titanium microneedle arrays used for initial device development were produced by photochemical etching of a 50-µm-thick sheet of titanium (Tech-Etch, Plymouth, MA). Individual microneedle arrays were removed from the sheet and placed in a custom fixture to bend the microneedles 90° from the plane of the array support structure. The microneedle arrays used for testing had the same microneedle number (16) and width (175 µm), but varied in microneedle length (750, 850, 950, 1050, 1150 and 1250 µm). A snap-dome-based insertion mechanism was used to deploy the different microneedle arrays into the skin. Each microneedle array was tested 3 times on 5 presumably healthy participants for a total of 15 data points for each microneedle array. Vacuum was applied to the insertion sites using a vacuum cup connected to an evacuated reservoir for 4 minutes. The mass of blood collected was converted to a volume by dividing the mass by the density of blood (1.06 mg µl–1)43. For final development of the TAP device, microneedle arrays are produced from stainless steel.
High-speed video of actuation mechanisms
The motion of each commercially available lancet and snap dome was captured using a Fastcam SA5 high-speed video camera (Photron) and TISLO6 lens (Nikon). The snap dome (Snaptron) was recorded at 65,100 f.p.s., the other lancing devices were recorded at 7,000 f.p.s. The videos were analysed using ProAnalyst software (Xcitex). The software was used to track a specified point on each actuation mechanism in each frame of the video. The velocity and acceleration were then determined for each actuation mechanism using the data output from the software.
Design of anticoagulant-mixing features
The mixer design was optimized by computational fluid dynamics (CFD) simulations using COMSOL Multiphysics (COMSOL, Burlington, MA). For each candidate design, the velocity field was determined by solving the steady-state Navier–Stokes equation, with boundary conditions of zero outlet pressure, constant flow rate of 60 μl min–1, and a velocity of zero at the walls of the channel. The fluid was taken to have a density of 1050 kg m–3 and viscosity of 4.0 cP, values representative of blood. Backpressure generated by the mixer was evaluated by calculating inlet pressure. The steady-state advection–diffusion equation was then solved to calculate the transport of a passive marker, representing the anticoagulant. Heparin has a diffusion coefficient of approximately 1.2 × 10−10 m2 s–1 in blood36. A conservative value of 10–11 m2 s–1 was used in the CFD simulations. The efficiency of mixing, E, was evaluated as the variance of the anticoagulant concentration in the outlet divided by the variance of the anticoagulant concentration in the inlet. With no mixing, E = 1. With perfect mixing, E = 0. We used E < 0.1 as the criterion for adequate mixing. For visualization of mixing in the device, two aqueous solutions containing blue and yellow food colouring were introduced to a device using a syringe-pump-controlled flow system.
Determination of the release profile of lithium heparin
In vitro determination of heparin release was carried out using partial devices built to allow visualization of the flow path (n = 6 without membrane control and n = 8 with membrane control of anticoagulant release). Blood release from the skin was simulated using a syringe-pump-controlled flow system. Water was introduced to test devices at either 25 or 50 µl min–1. Aliquots of the rinse liquid were recovered from the reservoir side of the devices throughout the collection. The concentration of heparin in the aliquots was then measured using a colourimetric assay based on the metachromasia of Azure A44.
Production of the TAP device
Manufacturing of the final TAP device is compliant with good manufacturing practice (GMP) and ISO 13485 requirements. The TAP device is assembled primarily from custom injection-moulded and die-cut components. The injection moulded parts comprise thermoplastic elastomer (TPE), acrylonitrile butadiene styrene (ABS), and polycarbonate materials. Die-cut parts include metals/foils and pressure-sensitive adhesive tapes. Stainless steel microneedle arrays are chemically etched and then over-moulded onto a support post. The device is assembled through a series of manual and semi-automated operations including ultrasonic welding, heat staking, liquid adhesive bonding, and pressure-sensitive adhesive tape bonding. Various jigs and fixtures are used to align components during assembly, and each assembly step includes specific quality control checks to confirm proper fit, orientation, and/or function. The final TAP device assembly step is completed under vacuum, with the primary top and bottom components snapped together to trap vacuum within the vacuum chamber of the device. The assembled device is then packaged in a foil pouch to maintain vacuum throughout its shelf-life. Finally, packaged TAP Devices are sterilized by gamma irradiation.
Clinical study to compare TAP and venepuncture test results
Sample size was chosen based on Clinical and Laboratory Standards Institute (CLSI) guidelines GP34-A and EP09-A3. Due to the fundamental difference between the two collection methods tested (TAP and venepuncture), the participants were not able to be blinded. The order of the venepuncture and TAP blood sample collections for each participant was randomized. Blood collection by venepuncture used 2 ml Vacutainer tubes containing lithium heparin (Becton Dickinson) and 23 G butterfly needles, according to standard techniques. Blood collection with the TAP device was conducted according to the instructions for use. The location of the blood collections (that is, region of the upper arm for TAP sampling or the arm used for venepuncture) was also randomized for each participant. HbA1c was measured using the DCA Vantage Analyzer (Siemens) and the Afinion AS 100 Analyzer (Alere). For linear regression and Bland–Altman analysis, results from both analysers were included in a single dataset.
Pain sensation scores were obtained from each participant immediately after each microneedle array actuation or venepuncture, using the Wong–Baker FACES pain scale39, with 0 corresponding to “no hurt” and 10 corresponding to “hurts worst”. Due to the fundamental difference between venepuncture and TAP device, participants were not blinded to the sample collection method.
During the clinical evaluation, dermal irritation was scored by a member of the study staff immediately following each collection and twenty minutes after the last sample collection for each participant. The following scoring scale was used40: 0 = no evidence of irritation; 1 = minimal erythema, barely perceptible; 2 = definite erythema, readily visible, minimal edema or minimal papular response; 3 = erythema and papules; 4 = definite edema; 5 = erythema, edema and papules; 6 = vesicular eruption; 7 = strong reaction spreading beyond sampling site.
The device optimization studies were conducted on a population of presumably healthy volunteers at Seventh Sense Biosystems. The volunteers provided written consent under an Institutional Review Board (IRB)-supervised protocol (New England Independent Review Board, Needham, MA), or provided written consent after reading a detailed explanation of the test procedure.
The clinical study to compare TAP and venepuncture HbA1c test results was conducted on diabetic and presumably healthy volunteers at three hospital sites, who provided written consent. The protocol was reviewed by the IRB selected for each site (Shulman IRB, The Mary Imogene Bassett Hospital IRB, and Geisinger IRB).
Life Sciences Reporting Summary
Further information on experimental design is available in the Life Sciences Reporting Summary.
Custom code used to perform the Monte Carlo analysis is available at https://github.com/7sbio/microneedle-montecarlo.
The data that support the findings of this study are available within the paper, and further data acquired in the context of this study are available from the corresponding author upon reasonable request. Source data for the figures in this study are available in figshare with the identifier https://doi.org/10.6084/m9.figshare.c.3957568 (ref. 45).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Windich-Biermeier, A., Sjoberg, I., Dale, J. C., Eshelman, D. & Guzzetta, C. E. Effects of distraction on pain, fear, and distress during venous port access and venipuncture in children and adolescents with cancer. J. Pediatr. Oncol. Nurs. 24, 8–19 (2007).
Migdal, M., Chudzynska-Pomianowska, E., Vause, E., Henry, E. & Lazar, J. Rapid, needle-free delivery of lidocaine for reducing the pain of venipuncture among pediatric subjects. Pediatrics 115, e393–e398 (2005).
Hamilton, J. G. Needle phobia: a neglected diagnosis. J. Fam. Pract. 41, 169–176 (1995).
Deacon, B. & Abramowitz, J. Fear of needles and vasovagal reactions among phlebotomy patients. J. Anxiety Disord. 20, 946–960 (2006).
Chan, C. P. Y. et al. Evidence-based point-of-care diagnostics: current status and emerging technologies. Annu. Rev. Anal. Chem. 6, 191–211 (2013).
Yang, X. et al. Simple paper-based test for measuring blood hemoglobin concentration in resource-limited settings. Clin. Chem. 59, 1506–1513 (2013).
Sølvik, U. Ø., Røraas, T., Christensen, N. G. & Sandberg, S. Diagnosing diabetes mellitus: performance of hemoglobin A1c point-of-care instruments in general practice offices. Clin. Chem. 59, 1790–1801 (2013).
Pollock, N. R. et al. A paper-based multiplexed transaminase test for low-cost, point-of-care liver function testing. Sci. Transl. Med. 4, 152ra129 (2012).
Kuriger, R. J. & Romanowicz, B. F. Analysis of blood extraction from a lancet puncture. in Tech. Proc. 2004 NSTI Nanotechnol. Conf. Trade Show 1, 275–279 (Nano Science and Technology Institute, 2004).
Wei, C.-H. et al. Clinical evaluation and alternative site blood glucose testing of the EasyPlus mini R2N blood glucose monitoring system. Clin. Chim. Acta 403, 167–172 (2009).
Cunningham, D. D. et al. Blood extraction from lancet wounds using vacuum combined with skin stretching. J. Appl. Physiol. 92, 1089–1096 (2002).
Heinemann, L. & Boecker, D. Lancing: quo vadis? J. Diabetes Sci. Technol. 5, 966–981 (2011).
Prausnitz, M. R. & Langer, R. Transdermal drug delivery. Nat. Biotechnol. 26, 1261–1268 (2008).
Donnelly, R. F., Singh, T. R. R. & Woolfson, A. D. Microneedle-based drug delivery systems: microfabrication, drug delivery, and safety. Drug Deliv. 17, 187–207 (2010).
Burton, S. A. et al. Rapid intradermal delivery of liquid formulations using a hollow microstructured array. Pharm. Res. 28, 31–40 (2011).
Patel, S. R., Lin, A. S. P., Edelhauser, H. F. & Prausnitz, M. R. Suprachoroidal drug delivery to the back of the eye using hollow microneedles. Pharm. Res. 28, 166–176 (2011).
Coulman, S. A. et al. In Vivo, In Situ imaging of Microneedle Insertion into the Skin of Human Volunteers Using Optical Coherence Tomography. Pharm. Res. 28, 66–81 (2011).
Wang, P. M., Cornwell, M. G. & Prausnitz, M. R. Minimally invasive extraction of dermal interstitial fluid for glucose monitoring using microneedles. Diabetes Technol. Ther. 7, 131–141 (2005).
Ventrelli, L., Marsilio Strambini, L. & Barillaro, G. Microneedles for transdermal biosensing: current picture and future direction. Adv. Healthc. Mater. 4, 2606–2640 (2015).
Li, C. G., Lee, C. Y., Lee, K. & Jung, H. An optimized hollow microneedle for minimally invasive blood extraction. Biomed. Micro. 15, 17–25 (2013).
Ganesan, A. V. et al. Analysis of MEMS-based microneedles for blood monitoring. BioNanoScience 4, 128–135 (2014).
Li, C. G. et al. One-touch-activated blood multidiagnostic system using a minimally invasive hollow microneedle integrated with a paper-based sensor. Lab Chip 15, 3286–3292 (2015).
Shergold, O. A. & Fleck, N. A. Mechanisms of deep penetration of soft solids, with application to the injection and wounding of skin. Proc. R. Soc. Lond. Math. Phys. Eng. Sci. 460, 3037–3058 (2004).
Davis, S. P., Landis, B. J., Adams, Z. H., Allen, M. G. & Prausnitz, M. R. Insertion of microneedles into skin: measurement and prediction of insertion force and needle fracture force. J. Biomech. 37, 1155–1163 (2004).
Crichton, M. L. et al. The effect of strain rate on the precision of penetration of short densely-packed microprojection array patches coated with vaccine. Biomaterials 31, 4562–4572 (2010).
Verbaan, F. J. et al. Improved piercing of microneedle arrays in dermatomed human skin by an impact insertion method. J. Control. Release 128, 80–88 (2008).
Rousche, P. J. & Normann, R. A. A method of pneumatically inserting an array of penetrating electrodes into cortical tissue. Ann. Biomed. Eng. 20, 413–422 (1992).
The Physics of Diaphragm Springs (Häussermann, accessed 7 November 2017); http://www.haussermann.com/pub/pdf/en/tellerfeder.pdf
Braverman, I. M. The cutaneous microcirculation. J. Investig. Dermatol. Symp. Proc. 5, 3–9 (2000).
Gill, H. S., Denson, D. D., Burris, B. A. & Prausnitz, M. R. Effect of microneedle design on pain in human volunteers. Clin. J. Pain. 24, 585–594 (2008).
Davis, M. J., Demis, D. J. & Lawler, J. C. The microcirculation of the skin. J. Invest. Dermatol. 34, 31–35 (1960).
Landis, E. M. The capillaries of the skin. J. Invest. Dermatol. 1, 295–311 (1938).
Humbert, P., Fanian, F., Maibach, H. I. & Agache, P. Agache’s measuring the skin: non-invasive, investigations, physiology, normal constants (Springer, Berlin, Heidelberg, 2016).
Cunningham, D. D. et al. Vacuum-assisted lancing of the forearm: an effective and less painful approach to blood glucose monitoring. Diabetes Technol. Ther. 2, 541–548 (2000).
Simonen, P., O’Brien, M., Hamilton, C., Ashcroft, J. & Denham, J. Normal variation in cutaneous blood content and red blood cell velocity in humans. Physiol. Meas. 18, 155–170 (1997).
Bernstein, H. & Langer, R. Ex vivo model of an immobilized-enzyme reactor. Proc. Natl Acad. Sci. USA 85, 8751–8755 (1988).
Stroock, A. D. et al. Chaotic mixer for microchannels. Science 295, 647–651 (2002).
Rohlfing, C. L., Parvin, C. A., Sacks, D. B. & Little, R. R. Comparing analytic performance criteria: evaluation of HbA1c certification criteria as an example. Clin. Chim. Acta 433, 259–263 (2014).
Wong, D. L., Hockenberry-Eaton, M., Wilson, D., Winkelstein, M. L. & Schwartz, P. Wong’s Essentials of Pediatric Nursing (Mosby, 2001).
Berger, R. S. & Bowman, J. P. A reappraisal of the 21-day cumulative irritation test in man. J. Toxicol. Cutan. Ocul. Toxicol. 1, 109–115 (1982).
Mijailovic, A. S. et al. Optimizing outpatient phlebotomy staffing: tools to assess staffing needs and monitor effectiveness. Arch. Pathol. Lab. Med. 138, 929–935 (2014).
Zhong, J., Asker, C. L. & Salerud, E. G. Imaging, image processing and pattern analysis of skin capillary ensembles. Skin. Res. Technol. 6, 45–57 (2000).
Van Slyke, D. D. et al. Calculation of hemoglobin from bloodspecific gravities. J. Biol. Chem. 183, 349–360 (1949).
Klein, M. D., Drongowski, R. A., Linhardt, R. J. & Langer, R. S. A colorimetric assay for chemical heparin in plasma. Anal. Biochem. 124, 59–64 (1982).
Blicharz, T. M. et al. Dataset for Microneedle-based device for the one-step painless collection of capillary blood samples. figshare https://doi.org/10.6084/m9.figshare.c.3957568 (2017).
We thank M. Gilpatrick, M. Oster, and J. P. Lock for their clinical operations support during internal pilot testing. We also thank the principal investigators and clinical staff at each of the sites that supported the TAP external study: V. Bush (Mary Imogene Bassett Hospital), G. Allen (Roger Williams Medical Center), and Y. Henry (Geisinger Medical Center). We thank W. Fowle (Northeastern University) for providing the SEM image of the microneedle array. We also thank R. Langer for reviewing the manuscript. This work was funded in part by a grant from the Bill and Melinda Gates Foundation (OPP1028750—Point-of-Care Diagnostics).