Microsampling tools for collecting, processing, and storing blood at the point‐of‐care

Abstract In the wake of the COVID‐19 global pandemic, self‐administered microsampling tools have reemerged as an effective means to maintain routine healthcare assessments without inundating hospitals or clinics. Finger‐stick collection of blood is easily performed at home, in the workplace, or at the point‐of‐care, obviating the need for a trained phlebotomist. While the initial collection of blood is facile, the diagnostic or clinical utility of the sample is dependent on how the sample is processed and stored prior to transport to an analytical laboratory. The past decade has seen incredible innovation for the development of new materials and technologies to collect low‐volume samples of blood with excellent precision that operate independently of the hematocrit effect. The final application of that blood (i.e., the test to be performed) ultimately dictates the collection and storage approach as certain materials or chemical reagents can render a sample diagnostically useless. Consequently, there is not a single microsampling tool that is capable of addressing every clinical need at this time. In this review, we highlight technologies designed for patient‐centric microsampling blood at the point‐of‐care and discuss their utility for quantitative sampling as a function of collection material and technique. In addition to surveying methods for collecting and storing whole blood, we emphasize the need for direct separation of the cellular and liquid components of blood to produce cell‐free plasma to expand clinical utility. Integrating advanced functionality while maintaining simple user operation presents a viable means of revolutionizing self‐administered microsampling, establishing new avenues for innovation in materials science, and expanding access to healthcare.

trained phlebotomist. While the initial collection of blood is facile, the diagnostic or clinical utility of the sample is dependent on how the sample is processed and stored prior to transport to an analytical laboratory. The past decade has seen incredible innovation for the development of new materials and technologies to collect low-volume samples of blood with excellent precision that operate independently of the hematocrit effect. The final application of that blood (i.e., the test to be performed) ultimately dictates the collection and storage approach as certain materials or chemical reagents can render a sample diagnostically useless. Consequently, there is not a single microsampling tool that is capable of addressing every clinical need at this time. In this review, we highlight technologies designed for patient-centric microsampling blood at the point-of-care and discuss their utility for quantitative sampling as a function of collection material and technique. In addition to surveying methods for collecting and storing whole blood, we emphasize the need for direct separation of the cellular and liquid components of blood to produce cell-free plasma to expand clinical utility. Integrating advanced functionality while maintaining simple user operation presents a viable means of revolutionizing self-administered microsampling, establishing new avenues for innovation in materials science, and expanding access to healthcare. individual experiences an abnormal symptom, they react by seeking medical attention. To confirm a diagnosis, healthcare professionals collect biological samples that are sent out to a laboratory for testing, and days to weeks can lapse before the results are obtained to administer a treatment regimen for the patient. If an individual lacks apparent symptoms, they may suffer from a medical condition without the ability to identify the underlying cause for a substantial amount of time. This lapse in healthcare can be further exacerbated by challenges associated with testing at the point-of-care or in resource-limited settings. As a result, a reactive healthcare system does not benefit patients who require immediate medical intervention to attend to their health issues. Instead, an active system that allows patients to play a role in monitoring their health status or treatment efficacy is desirable. Patient-centric microsampling low volumes of biofluids present an opportunity for a more active healthcare system.
Among the available biofluids that can be collected for health evaluations-including tears, sweat, urine, mucous-blood offers the most complex and diagnostically valuable sample matrix. Blood comprises distinct populations of cells (e.g., erythrocytes, lymphocytes, granulocytes, monocytes, macrophages, and platelets) and a rich carrier fluid (e.g., plasma or serum) containing myriad proteins, ions, dissolved gases, nutrients, and waste products. 1 Analyzing whole blood, or its component parts, can provide a full panel of health and wellness indices for evaluating the cardiovascular system, various organ functionalities, vitamin/mineral levels, and countless other parameters indicative of health status. 2 Clinically established reference values based on patient age or sex aid in making a preliminary diagnosis or determining the stage of a disease. Blood sampling also offers a means to monitor the effectiveness of medications to guide proper treatment. For these reasons, blood is often thought of as the ideal specimen for evaluating the health status of a patient.
Blood is routinely collected by venipuncture in centralized hospitals, local clinics, and even mobile clinics. In these settings, dedicated and highly trained staff are equipped with hypodermic needles, evacuated collection tubes (vacutainers) containing an anticoagulant, and a sharps container for safely disposing biohazardous materials following collection ( Figure 1a). 3 Inclusion of anticoagulant is critical to prevent clotting and maintain hematological indices prior to the analytical process. Collection volumes vary according to the tests ordered by the physician, but generally yield 2-10 ml per tube. 4 While venipuncture offers a highly representative sample of circulating cells and analytical targets, this collection method is often invasive and painful for the patient. 5 The associated large collection volumes can also accumulate a considerable source of waste depending on the final application.
Additionally, venipuncture has the potential to expose phlebotomists to blood-borne pathogens such as viruses (e.g., human immunodeficiency virus [HIV], HBV, HCV) and bacteria (e.g., Treponema pallidum, the pathogen that causes Syphilis) through needle-stick. 5 Due to the elevated risks associated with venipuncture for both patient and clinician, this method is not ideal for microsampling at the point-of-care. 6 An alternative method for collecting smaller volumes of blood with minimized risk of exposure is capillary sampling via a finger stick ( Figure 1b). In the simplest form, retractable single-use lancets can be used to pierce a fingertip, heel, or earlobe to provide a range of expected volumes (e.g., low flow 10-20 μl; high flow 250-500 μl) dependent on needle gauge and penetration depth. 7 Similar to collection by venipuncture, anticoagulants can be introduced to blood collected by finger stick using a pre-coated capillary tube or storage of dried anticoagulant in a porous collection material. Advanced forms of capillary sampling also include microneedle array collection devices such as the TAP device by YourBio Health (Figure 1c), which yields an average of 104 ± 19 μl with just the push of a button. 8 Finger-stick collection represents an attractive method for self-sampling that can be performed anywhere without the need for a trained clinician. Additionally, this method can result in less patient pain and discomfort in comparison to venipuncture, which can allow for more frequent sampling.
While capillary sampling by finger stick offers advantages over venipuncture, different finger stick protocols can result in variations in critical blood parameters. Drop-to-drop variation was evaluated using a hematology analyzer for quantitation of hemoglobin and three-point white blood cell (WBC) differential. 9 A higher degree of variability was reported for hemoglobin, WBC count, and platelet count measurements from a finger stick compared to venous blood. However, the clinical utility of low volume finger stick sampling was reported within instrument variability for volumes equal to or greater than 60-100 μl. In this example, pooling multiple drops of blood from a finger stick provided a more representative sample compared to a single droplet of blood.
The Clinical Laboratory Improvement Amendments of 1988 regulates all clinical laboratory testing of biological samples for diagnostic, prevention, or treatment purposes. 10 Criteria for acceptable performance is analyte dependent, but broadly states that coefficients of variation (%CV) should not exceed 20% of the target value, with some exceptions. 11 For example, quantitation of hemoglobin (±4%), albumin (±8%), platelet count (±25%), and Troponin I (±30%) demonstrate the range of acceptable performance metrics as published by the Center for Medicare and Medicaid Services in 2019. While these criteria are often evaluated and reported during the post-analytical phase, the majority of lab errors occur during the pre-analytical phase and are most frequently caused by hemolysis (40-70%), insufficient sample volume (10-20%), and undue clotting (5-10%). 12 These factors represent an outstanding need for improved tools and methods for collection of biological samples. Specifically, the storage and transportation of blood samples represent two remaining challenges at the point-of-care.
Sample stability with respect to hematological parameters is largely dependent on storage temperature and length of time. 13 The World Health Organization (WHO) recommends samples of whole blood should be stored between 4 and 8 C for a maximum of 24 h and cannot be frozen. 3 To extend storage life, plasma or serum should be separated from the sample and stored at 4-8 C for up to 7 days or frozen (À20 C) for longer periods of time. Additional storage recommendations are available depending on the desired laboratory analyses. 14 These cold-chain storage restraints put a substantial burden on the collection of blood in remote areas and often result in samples being discarded. For applications where immediate testing is not possible, and cold-chain storage is unavailable, preserving the stability of the sample at ambient conditions could navigate the remaining challenges for self-sampling low volumes of blood at the point-of-care or in resource-limited settings.
As complicated and bulky clinical instruments are replaced by user-friendly and compact point-of-care technologies, the collection method and subsequent analytical quality of biofluid samples ultimately determines the clinical relevance of the test result. Recently, several comprehensive reviews have summarized advances in microsampling techniques with respect to therapeutic drug monitoring, 15 biobanking, 16 paper spray mass spectrometry, 17 and lab-on-paper devices. 18 Most technologies have been reviewed with respect to analyte-or technique-specific metrics of performance such as limits of detection and dynamic range as informed by established guidelines for method validation related to quantitative bioanalysis. 19 This endpoint method of analysis is typically determined by the analysis method or instrumentation and does not include critical aspects of device performance such as sample input or output volumes, time requirements, user steps, or sample purity. In this review, we analyze the key aspects of device performance as a function of device design and materials to aid in identifying the correct sampling tool for selfsampling, separating, and storing blood outside of clinical settings for use at the point-of-care. Our search criteria were limited to articles published between 2010 and 2020. Distinction between final sample state (e.g., liquid vs. dry) was indicated for each evaluation. Technologies that require electricity or external equipment-such as syringe pumps-were excluded from this review. Additionally, we survey different methods for passive generation of cell-free plasma and discuss the advantages and limitations of each approach with respect to plasma quality and separation efficiency. Identifying the correct collection tool for targeted downstream testing could address the major sources of analytical errors associated with health diagnostics.

| COLLECTION AND STORAGE OF WHOLE BLOOD AT THE POINT-OF-CARE
Capillary microsampling offers precise and accurate blood volumes stored within an open-ended glass tube often coated with anticoagulant. The tube is filled end-to-end by capillary action upon contact with pool of blood. The resultant blood specimen can be transferred to a plastic microcentrifuge tube or similar low volume container (e.g., microtainer, 20 minivette point-of-care test 21 ) for storage or immediately applied to a point-of-care test (e.g., lateral flow test [LFT] 22 or hand-held analyzer 23 ). Addition of an aqueous solution or buffer to the container washes out the sample and effectively dilutes it to a working volume. Beyond the volumetric advantages of this technique, little is offered with respect to potential for automation or analyte stability. Ultimately, capillary microsampling produces a low volume sample of liquid blood, which requires specific cold-chain storage conditions to maintain analyte stability similar to venipuncture.
In contrast to liquid samples, the power of storing dried blood in a porous matrix such as cellulosic or glass fiber materials-to maintain analyte stability in the absence of cold-chain storage-has been extensively reported. 24,25 In fact, the stability of several analyte classes (e.g., drug metabolites, cytokines, and RNA) is improved upon drying as they are less susceptible to degradation by hydrolysis, photolytic processes, esterase, and RNAase activity. [26][27][28] This technique has been demonstrated with two main form factors: (i) dipstick-style paper strips and (ii) dried blood spot (DBS) cards (Table 1). NOBUTO blood collection strips comprise Advantex type 1 blood sampling paper and have been used for detection of avian influenza virus antibody in waterfowl ( Figure 2a). 29,30 These dipstick-style strips offer little in terms of handling protection to the user or drying considerations as these are essentially strips of raw filter paper. In contrast, the triangle paper dipstick comprises a laser cut Whatman GB003 filter paper inside a matchbook-style case for improved handling and drying (Figure 2b). 31 These strips are placed in contact with a volume of blood and allowed to wick until blood saturates the entire triangle-shaped portion (approximately 20-40 μl) and allowed to dry while attached to the case. Similarly, DBS cards are traditionally affixed to a book-style cardstock material for handling, drying, and recording patient information. DBS microsampling has been widely adopted for clinical use and offers several key advantages. 32,33 Bang pioneered the use of filter paper for quantifying the concentration of glucose from DBS in 1913. 34 Later in 1963, Guthrie and Susi described a simple screening method using DBS for detecting phenylketonuria in large populations of newborn infants. 35 Since then, applications for DBS sampling have extended to over 2000 analytes 36 including the analysis of small molecules, 37 viral infections (e.g., HIV, HBV, HCV), 38 therapeutic drug monitoring (e.g., antiepileptic drugs), 39 anticancer drugs, 40 and toxicology. 41 Traditional DBS cards comprise a sample collection layer imprinted with circular zones to indicate sample addition affixed to a basic cardstock material for patient information ( Figure 2c). Each zone accommodates approximately two to three drops of blood and each card typically contains five zones. Commercially available DBS cards are offered by various manufacturers for general protein analysis (e.g., Whatman 903 and Ahlstrom 226 protein saver cards) 42 and specific nucleic acid stabilization (e.g., Whatman FTA DMPK-A/B). 43 Advantages of DBS cards for blood sampling include (i) ease of self-sampling by direct finger stick, (ii) simplified transportation at ambient conditions, and (iii) ability to archive samples for retrospective analysis (i.e., biobanking). [44][45][46] Additionally, DBS cards are generally inexpensive and require minimal equipment, which permits access to health-related diagnostic data for hard-to-reach populations (i.e., remote settings) and surveillance efforts to monitor population level transmission of infection or track emerging disease. While the benefits of DBS cards for sampling blood at the point-of-care are myriad, they are limited by natural variations in patient hematocrit, which severely complicates quantitative analysis.
The hematocrit is described as the ratio of packed red blood cell (RBC) volume to the total volume of a blood sample following with respect to the quantity of cellular matter (i.e., RBCs, WBCs, and platelets). 48,49 This sampling bias can be further subdivided into (i) area bias and (ii) recovery bias (i.e., matrix effects). 50 The area bias is largely driven by differences in viscosity related to the relative quantities of cells found in samples with different hematocrits. For example, a sample with a higher hematocrit will spread less than a sample with lower hematocrit, resulting in an inconsistent volume of sample contained in the same area of filter paper (i.e., subpunch) ( Figure 3b). This area bias has been reported in multiple studies aimed at evaluating DBS cards. 51,52 The recovery bias is associated with uneven distribution, absorption, and higher evaporation rates of biofluid during the drying process, often resulting in nonhomogeneous blood spots. 53 Both sources of sampling bias can be minimized or eliminated by employing a few basic strategies. 54 Specifically, combining volumetric sample application with whole-spot analysis will produce a precise volume of dried sample independent of the hematocrit value. 55 Additionally, pairing whole-spot analysis with an internal standard can help to address the recovery bias and account for loss of analyte to the extraction process. 56 While these strategies seem simple, they are dependent on accurate volume application with the aid of a metered capillary tube or volumetric pipette, which limits use to trained staff. 57 Poor sampling by DBS can typically be determined visually with respect to the outlined sample application zones ( Figure 4). 58 Here, DBS samples can be

| Volumetric dried blood sampling technologies
The first strategy for volumetric sampling focuses on restricting the overall hydrophilic area, which absorbs and contains the final blood F I G U R E 4 User-errors associated with dried blood spot (DBS) sampling. Examples include (a) application of insufficient volume, (b) uneven sample application with multiple drops, (c) incomplete penetration through the thickness of the filter paper, and (d) coalescence of blood spots due to overfilling. Improper sampling can result in cards being discarded.
sample. The volumetric absorptive paper disk (VAPD) device comprises a precut paper disk surrounded by filter paper to eliminate the need for accurate volume application using pipettes or microcapillaries ( Figure 5a). 59 The precut disk (5.5 mm diameter) is positioned within a circular void in the filter paper using adhesive. The gap between the paper disk and the filter paper is small enough to allow excess blood to flow into the filter paper to ensure an accurate volume is contained within each disk. A similar device, termed VAPDmini, is functionally the same but on a smaller scale (3 mm diameter paper disk The HemaSpot HF (80 μl) and HD (140 μl) from Spot-On Sciences were designed for finger-stick sampling of blood emphasizing safe storage and transport at ambient temperature. 60 The devices comprise eight precut wedges of absorbent paper connected at a center point, desiccant, and a tamper-resistant cartridge with key ( Figure 5b). Sample is applied to the center of the device for distribution to the paper wedges and dried by the surrounding desiccant.
These devices were evaluated with plasma samples from dogs to detect parasitic disease (Visceral leishmaniasis) 61 and human blood samples for HIV-1 drug resistance testing. 62 High sensitivity and specificity were achieved for dog plasma samples; however, genotyping was successful in only 67% of HemaSpots tested within a viral load range of 1000-100,000 copies/ml.
Quantitative sampling by volume restriction has also been achieved using traditional DBS filter papers impregnated with hydrophobic wax barriers to control blood flow and distribution. 63  In contrast, volumetric absorptive microsampler (VAMS) technology deviates from standard dried blood sampling and aims to provide simplified collection of blood by obviating additional tools for accurate volume application. 64 This device comprises an absorbent polymeric tip sized to contain a discrete volume of blood (ca. 10 μl) attached to a plastic handle ( Figure 6a). 65,66 The absorbent tip of the device is dipped into a drop of blood until it is fully saturated. Then, the pen is stored in a special case to ensure the tip does not contact any surfaces, which could alter the contained volume or result in contamination. 67 Initial evaluation by gravimetric analysis yielded an average absorbed volume of 10.6 ± 0.4 μl blood across a hematocrit range of 20-65%. This result was supported by a previous study using radioactive 14 C caffeine, which yielded a similar volume of 10.5 ± 0.1 μl. 68 This device has been extensively evaluated for sampling accuracy in comparison to DBS technologies using a range of target analytes including: phosphatidylethanol, 69 anthelmintic drug moxidectin, 70 tacrolimus trough, 71 and potassium. 72 The devices described above each demonstrate improved volumetric sampling compared to traditional DBS cards through physically limiting the hydrophilic void available for blood collection. In contrast, the following devices employ fixed-volume capillary tubes or integrated microfluidic components to ensure volumetric sample application with simple user interface.
The HemaPEN comprises four end-to-end EDTA-coated microcapillaries (2.74 μl each) and four identical paper disks Two book-style devices also integrate fixed-volume capillary channels to apply discrete volumes of blood to traditional DBS filter papers. The Hemaxis DB10 device comprises four fixed-volume capillary channels (e.g., 5 or 10 μl) and removable filter paper to generate blood spots without needing a pipette ( Figure 6c). 75,76 Blood is collected directly from a finger stick into each individual capillary. Closing the lid of the device brings the filter paper into contact with the capillaries to produce four replicate blood spots. Completely filling the capillaries and removing the entire blood spot ensures volumetric sampling independent of the hematocrit effect. Comparable recovery of phentermine was reported using the microfluidic-based sampling device and blood spots generated with a volumetric pipette. 77 In contrast, the Capitainer-B device offers a single inlet port, which diverts a finger-stick sample to a capillary microchannel designed to hold approximately 13.5 μl of blood independent of the input volume. 78,79 Volume metering is controlled in this device by thin dissolvable PVA films covering the waste pad (to absorb excess sample) and the precut sample pad (Figure 6d). This format was designed to simplify sampling by direct addition of unmetered blood drops. These results support that the Capitainer-B device operates independently of both the hematocrit and applied blood volume.
The ADX Test Card by Accel Diagnostics comprises a microfluidic network to collect, distribute, and analyze blood. 81 This technology leverages the aggregation of magnetic beads in response to the presence of biomolecules to produce quantitative results without an external reader. 82 The size and quantity of aggregates captured by an internal magnet changes the flow of blood inside the microfluidic chip and is proportional to the concentration of target. The sample input volume is approximately 20-80 μl and distance-based results are available after 10-15 min. Alternatively, three-dimensional (3D) blood spheroids have been described using hydrophobic filter papers to overcome some of the limitations associated with traditional DBS cards. 83 Whatman 1 filter paper was functionalized with trichloro(3,-3,3-trifluoropropyl)silane to produce a hydrophobic surface for applying blood samples ( Figure 6e). In this format, the blood droplets dry on top of the paper rather than being absorbed and distributed throughout the porous matrix. Decreasing the interaction of blood in paper simplifies complete recovery of the sample (i.e., no lengthy extraction protocols), eliminates chromatographic effects due to uneven sample spreading, and stabilizes hydrolytically labile chemicals (i.e., only outer layer is exposed to air during drying) such as cocaine and diazepam.
Combining the accuracy of volumetric sample collection with enhanced analyte stability under ambient conditions increases the feasibility of self-sampling low volumes of blood for expanding access to health diagnostics. However, the complex matrix of whole bloodspecifically, the presence of blood cells-imposes a significant limitation on testing capabilities. In fact, many clinically established reference ranges for critical health evaluations are reported for samples of cell-free plasma or serum rather than whole blood. For example, HIV viral load testing requires cell-free plasma due to integrated proviral DNA found within infected T-cells. 84,85 Similarly, RBCs contain high amounts of analytes also found in plasma (e.g., folate, iron, potassium), which can be released through disruption of RBC membranes and artificially elevate measured concentrations and lead to inaccurate test results for nutritional deficiencies, anemia, or even kidney disease, respectively. 86 Ultimately, the presence of blood cells can introduce undesirable analytical interferents and obscure colorimetric read-out techniques, which are common for many rapid point-of-care tests. For these reasons, extending the tools developed for sampling whole blood to include the separation of plasma could greatly increase testing capabilities and improve sample collection at the point-of-care.

| SEPARATION OF CELLULAR AND LIQUID COMPONENTS OF WHOLE BLOOD AT THE POINT-OF-CARE
Methods for separating the cellular and liquid components of liquid blood can be categorized as either active (i.e., requires external instrumentation or sustained mechanical input by the user) or passive (i.e., only requires sample collection without additional user input or external instrumentation) with each approach presenting unique advantages and opportunities for microsampling at the point-of-care. The gold standard method for actively separating blood in a laboratory setting is centrifugation. Large volumes (e.g., ≤10 ml per tube) of whole blood are routinely processed in batches yielding nearly quantitative recovery of cell-free plasma or serum with minimal risk of hemolysis under moderate centrifugal forces (ca. 800 g for 10 min). While centrifugation is highly effective and reproducible, the requirements present several challenges for use at the point-of-care including reliable electricity, bulky size, and cost (e.g., ≥$500). 87 In contrast to active centrifugation, blood cells will passively sediment under gravitational force by simply leaving blood on the benchtop-eliminating the need for electricity. However, natural sedimentation such as this occurs on the order of multiple hours (>4 h) with poor yield (approximately 27% separation). 88 Additionally, liquid plasma requires cold-chain storage to maintain analyte stability, which further complicates sampling in rural or resource-limited settings.
Emerging technologies utilize myriad techniques to address the current challenges associated with processing blood away from centralized laboratories and clinics including: (i) hand-powered devices that output liquid plasma, (ii) passive separation devices that store dried plasma, Alternatively, devices that can separate plasma from cells, and subsequently store plasma in a dried format, can be identified by the term "dried plasma output." Standardizing performance metrics allows for direct comparison between different separation methods. Herein, we define separation efficiency as the ratio of recovered plasma volume to theoretical plasma volume (Equation (1)). For example: an input sample volume of 100 μl at a hematocrit of 45% would yield a theoretical plasma volume of 55 μl. If a hematocrit value is not reported, then a hematocrit value of 50% will be assumed.
3.1 | Active, hand-powered plasma separation devices for liquid plasma output    (Table 3). Each device passively separates plasma and yields cell-free liquid plasma, which can be recovered using volumetric tools such as a capillary tube or micropipette.
Untreated, unmodified PSM oriented for vertical flow provides the baseline for performance. 92 In this example, the PSM is a hydrophilic, asymmetrical polysulfone (Pall Vivid GR) material with a reported loading capacity of 40-50 μl blood cm À2 . 93

| Passive plasma separation devices for dried plasma output in porous materials
Enhanced sample stability over time has successfully been demonstrated for samples of whole blood dried in a porous matrix such as cellulosic paper and silk fibroin. 98 This same approach of dry storage can be extended to samples of cell-free plasma as an alternative to liquid storage to minimize the thermal degradation of proteins in the absence of cold-chain storage. 99 This simple approach of passive separation paired with dried storage has the potential to greatly expand access to routine health analyses-such as plasma viral load testingas demonstrated by a recent population study. 100 The following section includes both commercially available technologies as well as academic demonstrations for plasma separation and storage within a single device ( Table 4). The majority of approaches presented here utilize a form of commercially available PSM and collection pad comprising filter paper to produce a wide range of output plasma volumes with excellent purity (i.e., minimal hemolysis).
The first example employs an assembly of machined polypropylene parts supporting a Vivid GF PSM and collection pad positioned on a movable platform controlled by a screw mechanism (Figure 9a)    ( Figure 10c). 121 The sample input volume of blood scaled with the area of the device. Songjaroen  Recently, a single layer device fabricated from the same glass fiber separation membrane (Whatman LF1) was reported comprising the same geometry and design features as the device described above. 122 In this example, the LF1 membrane was treated with a twostep process: (i) first the membrane was coated with a fluorocarbon layer using pentafluoroethane plasma deposition to render the backside of the material hydrophobic followed by (ii) application of a mask and O 2 plasma etching to imprint the working area of the device for separation and detection within the single layer of LF1. A sample input volume of 9-12 μl blood was applied to the device for the colorimetric detection of glucose and albumin. Fabrication methods presented here are amenable to roll-to-roll manufacturing and represent potential for large-scale production.
Another challenge when fabricating devices with multiple layers of materials is achieving and maintaining conformal contact between layers, and between plasma separation membranes and absorbent materials in particular. Specifically, separation efficiency is greatly affected by the degree of contact between such materials. To address this, a 3D μPAD comprising Vivid GR PSM and Whatman 1 filter paper was fabricated to eliminate the potential gap between the two materials. 123 In this approach, Park et al. printed a sample reservoir and detection zones directly onto the PSM and filter paper assembly by liquid photopolymerization using a digital light processing printer.
This printing method eliminated multi-step assembly processes for streamlined manufacturing. First, the PSM was protected from dissolution in organic solvent (photocurable polymer) using parylene C. Next, the PSM surface was treated with oxygen plasma and superimposed onto the filter paper. Finally, the sample reservoir and detection zones were defined using a 3D printer. The device accommodates a sample input volume of 100 μl of blood for the direct quantitation of glucose using colorimetric analysis.

| Alternative methods for passive separation of plasma without PSM
Agglutination-based separation techniques represent an attractive alternative to PSM alone. Proteins such as wheat germ agglutinin 124 and blood-typing antibodies 125,126 can be used to agglutinate RBCs.
In 2012, Yang et al. reported a single layer μPAD comprising Whatman 1 chromatography paper functionalized with A/B antibodies to yield a dried plasma output. 125 Hydrophobic wax barriers define a central sample addition zone and four auxiliary plasma collection zones containing dried reagents for quantification of glucose in plasma (Figure 11a). An input sample volume of only 7 μl of blood successfully filled the four plasma zones. Eliminating the void volume associated with a standard PSM layer for separation allowed lower volumes of blood to perform the same colorimetric assay. Similarly, a distance-based μPAD featured the same Whatman 1 filter paper with a wax patterned sample addition zone and lateral channel (40 mm Â 2 mm) to yield a dried plasma output. 127 In this device, Recently, a combination approach using agglutination antibodies and interlocked micropillar scaffolds (i.e., "synthetic paper") was reported for separating plasma from whole blood with low protein adsorption. 126 The device comprises a square-shaped sample addition zone (16.8 mm) connected to a long rectangular channel (30 mm Â 2 mm) for plasma separation and dried plasma output.
Agglutination antibodies were spotted onto the sample addition zone to immobilize blood clots within the array of micropillars (50 μm diameter, spaced 100 μm apart) and allowed plasma to enter the lateral channel. An input volume of 90 μl whole blood was added and an average yield of 5-6 μl was obtained for a separation efficiency of 11.1%. Improved protein recovery was reported (>82%) in comparison to traditional PSM due to the low internal surface area of the material.
Traditional microfluidic devices have shown utility for plasma separation using a hand-driven syringe or pipette for sample loading and collection to obviate external pumps. These devices yielded a liquid plasma output for immediate use for biochemical detection and immunoassay applications. 128 However, previous examples suffered from slow separation, poor yield, and contamination from RBCs due to extended separation times. 129,130 A recent report aimed to address these limitations by integrating two sizes of microbeads with a capillary microchannel array for rapid separation of plasma and a liquid plasma output. A plasma separation filter was prepared by compactly stacking microbeads (10 and 100 μm diameters) into a polydimethylsiloxane microfluidic channel by negative pressure. 131 The larger diameter beads were introduced first to block the openings of the capillary microchannels. Next, the smaller beads were stacked to retain the RBCs and allow liquid plasma to enter the array of microchannels. In this configuration, a separation rate of 0.16 μl min À1 was achieved with a sample input volume of 20 μl. Approximately 3.29 μl plasma was collected in 20 min. In comparison, the same channel stacked with 15 μm beads yielded a separation rate of only 0.03 μl min À1 .
While an improved separation rate was demonstrated using two sizes of stacked microbeads, this approach still suffers from lengthy separation times and poor yield.
Another separation mechanism using a microfluidic device employed constriction-expansion channels for inertial cell sorting ( Figure 11b). 132 A narrow sample inlet constricted the blood cells into a tightly packed stream with a liquid plasma output at the periphery.
Gradually increasing the dimensions of the channel created vortices plasma in porous materials offers dual benefits of simple transportation by shipping through the mail and improved stability of myriad analytes at ambient conditions.
A remaining obstacle for expanding access to necessary healthcare diagnostics is the lack tools that offer multiple sample preparations within a single platform. In this review, we detailed materials, approaches, and novel technologies that yield a singular sample composition of either whole blood or cell-free plasma.
Some plasma separation technologies, in part, could feasibly generate unique samples containing plasma or blood cells by utilization of the separation membranes, but acceptable analytical quality of the cellular component has yet to be demonstrated.
Applying the volumetric sampling techniques described here to passive methods of plasma separation could increase the diagnostic utility of the entire biological sample by streamlining collection efforts, reducing patient discomfort, and decreasing preanalytical errors. Overall, these challenges highlight both the significant roles that materials play in the collection, processing, and storage of blood to collect, and also the outstanding need for additional innovation in this space in order to continue to increase access to healthcare.

ACKNOWLEDGMENTS
The project described was supported by the Tufts University Office of the Vice Provost for Research (OVPR) Research and Scholarship Strategic Plan (RSSP). This work was supported in part by a sponsored research agreement from Drummond Scientific.

CONFLICT OF INTEREST
Keith R. Baillargeon and Charles R. Mace are co-inventors on patent applications for technologies related to blood and plasma microsampling devices.

DATA AVAILABILITY STATEMENT
No datasets were generated or analyzed during the preparation of this review.