A Low-Cost Inkjet-Printed Glucose Test Strip System for Resource-Poor Settings

Abstract

Background:

The prevalence of diabetes is increasing in low-resource settings; however, accessing glucose monitoring is extremely difficult and expensive in these regions. Work is being done to address the multitude of issues surrounding diabetes care in low-resource settings, but an affordable glucose monitoring solution has yet to be presented. An inkjet-printed test strip solution is being proposed as a solution to this problem.

Methods:

The use of a standard inkjet printer is being proposed as a manufacturing method for low-cost glucose monitoring test strips. The printer cartridges are filled with enzyme and dye solutions that are printed onto filter paper. The result is a colorimetric strip that turns a blue/green color in the presence of blood glucose.

Results:

Using a light-based spectroscopic reading, the strips show a linear color change with an R² = .99 using glucose standards and an R² = .93 with bovine blood. Initial testing with bovine blood indicates that the strip accuracy is comparable to the International Organization for Standardization (ISO) standard 15197 for glucose testing in the 0-350 mg/dL range. However, further testing with human blood will be required to confirm this. A visible color gradient was observed with both the glucose standard and bovine blood experiment, which could be used as a visual indicator in cases where an electronic glucose meter was unavailable.

Conclusions:

These results indicate that an inkjet-printed filter paper test strip is a feasible method for monitoring blood glucose levels. The use of inkjet printers would allow for local manufacturing to increase supply in remote regions. This system has the potential to address the dire need for glucose monitoring in low-resource settings.

More disability-adjusted life years (DALY) are lost due to diabetes in low-resource countries than resource-rich countries, despite the fact the diabetes is more prevalent in resource-rich countries.¹ The higher impact of diabetes in low-resource countries is due to a combination of challenges, including limited diabetes education of health professionals and families, insulin availability, and availability of diabetes management supplies.^1,2 Improvements in all of these areas are required to decrease the DALY lost due to diabetes in low-resource countries. Atkinson et al recommend a variety of approaches from increasing health systems’ resources to affordable at-home refrigeration to better procurement of insulin and blood glucose test strips.² These are just a few of the options, but it is clear that the problem is complex and the solution will be as well. While attempts have been made to help alleviate the lack of diabetes management technology and supplies in these low-resource areas through donations and government/nongovernmental organization (NGO) subsidies, these approaches have not proven to be effective and cannot be sustained in the long term.^3,4 Therefore, new technologies and methods to improve the accessibility of devices and consumables are needed to help reduce the burden of diabetes worldwide.

To address this issue, our goal is to design a glucose monitoring system for patients in low-resource areas. The design features absorbance-based glucose test strips manufactured using a standard desktop inkjet printer. The simplicity and low-cost nature of this method allows this strip system to be implemented by local manufacturers in low-resource countries. If implemented, this method could lower price points and simplify supply chains, leading to higher availability of glucose monitoring equipment.

Methods

The test strips work by producing a color-change reaction that scales linearly with the concentration of glucose in the blood. The color change is the result of the reaction between glucose, glucose oxidase (GOX), horseradish peroxidase (HRP), and 2,2’-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS). When blood is applied to a test strip, the D-Glucose is oxidized by GOX to produce D-Gluconic acid and hydrogen peroxide. The hydrogen peroxide and ABTS react with HRP to produce a blue/green product with an absorbance maximum at 414 nm.⁵ This blue/green product can be quantified using standard optical absorbance-based measurement techniques.

To print test strips using a desktop inkjet printer, the color ink cartridges (magenta, cyan, and yellow) are replaced with cartridges containing GOX, HRP, and ABTS solutions. Standard word-processing software is used to create 3 documents. Each document contains the test strip template drawn out in shapes, and the shapes in each document are a single color (cyan, magenta, or yellow) so that only 1 of the solutions (GOX, HRP, or ABTS) is printed each time (Table 1).

Table 1.

Red, Blue, and Green Color Settings to Select Individual Cartridges.

	GOX	HRP	Dye
Red	255	0	255
Green	0	255	255
Blue	255	255	0

The print color adjustment option is turned off in the printer settings to ensure only the desired enzyme or dye is printed, and the photo quality print setting is chosen to provide maximum print resolution. Once these modifications are made, the printer can layer the enzymes and dye onto the filter paper. The printer currently being used is an Epson Workforce 30 (Seiko Epson Company, Suwa, Japan). It has a resolution of 5760 × 1440 dots and deposits a minimum 3 pL per dot.⁶ This particular printer uses a piezoelectric crystal to pressure a controlled droplet of ink out of the print head rather than heat. This may be favorable with the use of enzyme and dye solutions as it prevents the potential enzyme denaturation that heat could cause.⁷ However, it should be noted that several thermal inkjet printers have been found to have minimal effect on protein denaturation.^8,9

The strip has been designed to provide calibration with each test. The sample loading zone is in the center of the strip rather than at the distal end to provide maximum separation between the control and experimental testing zones (Figure 1a). Absorbance is read at each end separately, and the meter process the glucose concentration based on the ratio of these 2 zones. Therefore, any color change on the control side is excluded from the measurement to allow the reading to be the color change caused by the glucose reaction rather than any other potential oxidizers of the dye. Another design feature is the filter paper, which can promote wicking and can help to limit particular mater and cells from contaminating the test area. For these experiments, General Electric filter sheets (G&E product 1093-935 filter sheets, grade 93)¹⁰ were used in our strip design. While wicking is desirable to help the blood sample absorb to the paper and reach the enzyme test site, this same property negatively influences the strength of the strips. Therefore, to help reinforce the strips, the filter was wrapped with clear contact paper, which does not interfere with the absorbance measurement, but which helps to provide rigidity and to prevent the blood sample from seeping through onto the meter during testing. A small slit in the center of the top of the strip is left uncovered to allow for sample placement (Figure 1b).

Figure 1.

(a) Illustration of the zones of the test strips—top view. (b) Illustration of how the contact paper wraps the filter paper (contact paper in black, filter paper in blue) and the layering of solutions (magenta = GOX, yellow = HRP, and cyan = ABTS)—side view.

To calibrate the enzymes and dye solutions for the test strips, 2 dyes were investigated, both with the enzymes GOX and HRP. The enzymes and dye products were purchased from Sigma-Aldrich (Sigma-Aldrich Corporation, St. Louis, MO): GOX (G6125), HRP (P8250), o-Dianisidine (D9143), and ABTS (A1888). The first dye used was o-Dianisidine, which is commonly used for glucose assays. It exhibits a brown color change (wavelength 500 nm). Experiments using this dye were done using solutions with the following concentrations: GOX = 20 units/mL, HRP = 60 units/mL, and o-Dianisidine = 0.005 g/mL. These were chosen based on standard solution protocols.^6,11 All absorbance measurements were performed with the BioTek Synergy4 spectrophotometer (BioTek Instruments Incorporated, Winooski, VT). Several conditions were tested by varying the volumes, buffers, and initially the measurement technique (Table 2). For 2 of the conditions, the Sigma-Aldrich GO assay kit was used for its simplicity and is represented by “kit” (Table 2).¹¹ The o-Dianisidine dye was promising as it produced a linear reaction; however, when it was used to print strips, it was discovered that it was not stable enough. Therefore, ABTS was investigated next. This dye exhibits a blue/green color (wavelength 410 nm). Using the same enzyme concentrations, in-solution assays were repeated to determine the appropriate concentration of ABTS. The final concentration was 0.002 g/mL.

Table 2.

Condition Parameters for Calibration of o-Dianisidine.

	Material measurement	Volume of enzymes/dye solutions (µL)	Diluent	Volume of glucose standard (µL)	Wavelength (nm)
1	Self	200	DI H2O	100	540
2	Kit	200	PBS	100	500
3	Kit	200	PBS	10	500
4	Self	26.25	Sodium acetate (283.75µL)	10	500
5	Self	300	Sodium acetate	10	500

^aDI H₂O, deionized water; Kit, Sigma-Aldrich GO assay kit; PBS, phosphate buffered saline; self, solutions mixed by hand.

Next, following calibration of the dyes and enzymes in solution, the design of the strip with the filter paper was optimized. To allow for easy reading in a plate reader for calibration, 7 µL of each solution was deposited onto 100 mm² circular cutouts of filter paper, and these test strip punch-outs were allowed to dry completely before use. The punch-outs were placed in a 24-well plate, and a 7 µL droplet of each glucose standard was placed on the center of each strip. The droplets varied in concentration from 0-450 mg/dL and absorption was read at 410 nm.

To test with blood, bovine blood was purchased from QuadFive (Materials Bio Inc, Ryegate, MT). Fibrinogen was the anticoagulant used. Bovine blood is reported to have 50-60 mg/dL glucose concentration.¹² However, it is likely that the blood we received was lower than 50 mg/dL as we were unable to determine the concentration using a standard Bayer Contour Next Link glucose meter; the meter gave a “low read” error for several samples of the blood from different batches. An assay was performed using 10 µL of each enzyme solution, 10 µL of dye, 200 µL of deionized water, and 5 µL of bovine blood at varying concentrations in triplicates with a 96-well plate. Therefore, a minimal concentration was assumed and glucose solution was mixed with the blood to provide +50 mg/dL to +400 mg/dL blood in 50 mg/dL increment solutions. In another assay, a 24-well plate was used with pipetted strips in the bottom of each well. Duplicates of each blood sample were pipetted in 5 µL droplets. These same solutions were placed into the cartridges of the Epson Workforce printer and used to print test strips. Small cutout of the printed paper were placed in a 96-well plate and allowed to dry. In triplicates, 5 µL droplets of glucose standards 0-400 mg/dL concentrations were placed in the wells. These were read at 420 nm by the spectrophotometer.

Results

In condition 1 (DI water solvent), the plate reached near saturation at low concentrations of glucose. Using PBS as a buffer (conditions 2 and 3) lead to a linear reaction but too low of absorption at glucose concentrations less than 100 mg/dL. In condition 4 (using sodium acetate), the volumes of the enzymes/dye solution and the glucose standards were lowered to allow for more buffer. This again leads to a linear reaction but low absorption readings. Condition 5 (using sodium acetate and larger dye/enzyme solution volume) was linear and provided the target absorbance of 0.83 at 450 mg/dL with an R² value of .99 (Figure 2).

Figure 2.

Calibration curves with o-Dianisidine dye. Condition 1 used deioinized water as the enzyme/dye solvent. Conditions 2 and 3 used PBS as the solvent. Conditions 4 and 5 used sodium acetate solution.

Strips printed using an inkjet printer with the o-Dianisidine solution showed a visible response to glucose (Figure 3). However, when they were left overnight in an opaque container to prevent exposure to light, the dye oxidized by the next morning without any exposure to glucose. Therefore, ABTS was used as an alternative more stable dye. While further studies will be required for long-term shelf stability of printed ABTS strips, initial results have shown promise. The enzyme/dye solutions were stable when refrigerated for at least 6 weeks. This demonstrates much greater stability than the o-Dianisidine solutions, which had to be used within 1 week. In addition, preliminary shelf-life tests of strips made with ABTS showed that they were stable after 1 month of storage in a closed opaque container at room temperature.

Figure 3.

Printed test strip with o-Dianisidine dye and droplet of 450 mg/dL glucose standard applied at the center. There is a change in color indicating the presence of glucose.

From the calibration of the enzymes and dyes in solution and the concentrations used in them, printing the same distribution of enzymes and dye would require the printing solutions to have the following concentrations: GOX = 0.29 mg/dL, HRP = 1.45 mg/dL, and ABTS = 0.004 mg/dL. The ABTS test strips showed visible changes in color in the blue/green range (Figure 4a). The strips showed increased absorbance at 410 nm with increasing glucose concentrations with a linear fit having an R² = .82 in the range of 0-450 mg/dL. There is deviation from a linear response at the higher glucose concentrations (Figure 4b). However, from 0-250 mg/dL, the response of the strips was very linear (R² = .99; Figure 4c).

Figure 4.

(a) Absorbance of ABTS dye reaction read at 410 nm using glucose standards 0-450 mg/dL. (b) Zoom in of the absorbance of the ABTS dye reaction in the low glucose concentration range (glucose standards 0-250 mg/dL). and (c) Circular test strip punch-outs showing a visible color change after contact with 0-275 mg/dL of glucose solution.

The ABTS dye was effective at detecting glucose concentration differences in bovine blood (Figures 5 and 6). The standard curve with the bovine blood in solution was fairly linear at an R² = .93 (Figure 5). In addition, there was still a visible change in color with blood glucose concentration on the test strips (Figure 6). The ABTS dye remained effective even when printed. The standard curve maintained linearity with an R² = .94 (Figure 7). The slopes of the best-fit linear fit from the data of the strips made using pipetting (Figure 6) and printing (Figure 7) were nearly identical (.0015 and .0014, respectively).

Figure 5.

Graph of average absorbance of ABTS read at 500 nm using bovine blood +50 to +400 mg/dL in solution (mean ± standard deviation). The data fit a linear trend with an R² of .9344.

Figure 6.

Image of ABTS-based test strip punch-outs illustrating color change using bovine blood +50 to +450 mg/dL. (Note: the 250 mg/dL blood sample was not pictured here as it became contaminated during preparation and storage.)

Figure 7.

Graph of absorbance of ABTS read at 420 nm using glucose standards 0-400 mg/dL. The data fit a linear trend line with an R² of .94.

Discussion

These results demonstrate that a filter paper test strip is a potential alternative for standard glucose monitoring test strips. The reactions have demonstrated a linear reaction with glucose standards and bovine blood (Figures 4 and 5). Test strips with bovine blood showed a visible gradient of color change (Figure 6). However, there was some variation in color over the strip test area and in some strips (for instance the 200 mg/dL strip in Figure 6), there was a noticeable red center area. This may cause issues of absorbance measurement saturation. This can be taken into account for changes to the strip design and/or patient guidelines to restrict the volume of blood that reaches the testing area on the strip. The data showed linear responses with similar slopes for both the pipetted and printed strips indicating that the printing process does not affect dye or enzyme efficacy. The color change on the printed strips was not as visually pronounced most likely due to the lower total dye and enzyme amounts on the current printed strip design. To facilitate visual strip reading, it may be necessary to lower the concentrations for the printed enzyme solutions (ie, print a lower amount of enzyme on each strip) to have a greater distinction between low and high glucose concentrations.

The printer cartridges used for the o-Dianisidine dye worked without the need for viscosity modifiers. However, this will vary with ink cartridge manufacture; if the printer cartridges are varied, it may be necessary to include other compounds in the print solution to increase the viscosity of the enzyme and dye solutions. For instance, Risio and Yan have shown that viscosity and surface tension modifications are required for reliable printing with Dimatix DMP 2800 inkjet printer cartridges,^13,14 which is a piezoelectric cartridge system similar to the Epson Workforce 30 printer cartridges used here. Currently, glycerol and methyl cellulose are being investigated as modifiers, but the exact concentrations necessary for printing will depend greatly on the specific printer used. To avoid complications like this in the future, it may be necessary to form a collaboration with the printer producer to ensure the solutions match the cartridge design.

Using desktop printers for manufacturing simplifies the process and makes it suitable for low-resource settings. While GOX-based strips are often thought of being relatively unstable, the enzymes and dye themselves are stable for up to 4 years in their powder forms.^15-18 With this inkjet strip design, the enzymes and dye can be shipped in stable powdered form to help accommodate the inconsistent distribution routes frequently experienced in low-resource countries. Once at the hospital or clinic, they can be mixed to form solutions, placed into empty ink cartridges, and printed onto filter paper. The extra calibration step that is built-in with every test strip helps to overcome the environmental conditions common to low-resource settings. Lack of power in most patient’s homes can lead to a lack of temperature control during storage of the strips. Since the glucose concentration from each strip is calculated from the relative difference of the test and control side, the resultant measurement is robust to changes in dye oxidation with temperature or humidity.

Since the test strips provide a visual indicator of blood glucose level, a meter is not strictly necessary for patient. However, to more accurately assess the glucose level, portable meter to read these test strips has been designed so a patient could use them as standard test strips are used (at home, work, etc). Because the strips use a simple colorimetric assay, meter design is straightforward and can be implemented using off-the-shelf commodity electronics parts. We have designed a portable, low-cost meter using photodiodes, LEDs, a microcontroller, and an LCD screen (Figure 8). It will obtain an absorbance measurement from the strip and calculate the glucose concentration in the sample using the programmed calibration curve. The cost of building an individual meter in very small quantities may be up to $40 depending on the cost of a microcontroller and other parts. However, that cost is expected to decrease to ~$15 or lower if the design is mass produced.

Figure 8.

Image of meter prototype. LCD screen displays reading; once strip is in place, the tray slides into the meter and suspends it between the LEDs and photodiodes.

The strips themselves are very low-cost; they cost only about $0.05/strip to make. Interestingly, the color change on the ABTS strip with bovine was visible (Figure 6). Therefore, the strips may be able to be used by themselves even in situations where an electronic meter is not a feasible device (if the cost is too high, if batteries are not readily available, etc). While it is unlikely that such a color scale eye test will be able to meet the ISO standard for a glucose monitor, this method would allow patients to know if their blood glucose level is very high or very low.

Conclusions

The work so far shows promise for a low-cost glucose meter utilizing inkjet-printed test strips. The strips can be retailed for $0.05 each, a significantly lower price than standard strips cost for self-paying patients. This is due to a combination of the materials and simplified manufacturing of the test strips. By using a simple inkjet printer, the manufacture of the strips themselves can be moved to clinics in the areas where patients typically do not have access to standard glucose testing supplies. In addition, when electronic meters are not available, the color change observed on the strip could be used by itself to help patients their diabetes. Work to obtain regulatory clearance for the meter and strip design is currently ongoing. An institutional review board (IRB) study has been submitted to test diabetic and nondiabetic patients and assess the efficacy of the low-cost meter when using human blood. This test can help confirm if the strips maintain ISO standard accuracy when using human blood.¹⁹ With the prevalence of diabetes in low-resource settings increasing, it is crucial that an appropriate solution be developed for these populations.^2,3 Our inkjet-printed strip design may provide a piece of the puzzle to help with diabetes management and care in these settings by enabling patients to monitor their blood glucose levels at home.