Low-Cost Blood Glucose Monitoring System with Printed On-Demand Test Strips for

Implementation in Resource-Poor Settings

Joseph Wilson

South Carolina Governor’s School for Science and Mathematics

ABSTRACT

More than 285 million people worldwide are diabetic and require daily blood glucose monitoring. As glucometers have evolved, they have become more accurate (~3% variance), but test strips can be expensive for patients, especially those without health insurance. In addition, in developing countries that rely on donated medical supplies, matching meters and strips are not always available to patients. The goal of our project is to design a low-cost meter and strip system that can be used in resource poor settings when standard meters or strips are not available. Our strategy is to create test strips that may be printed on-demand by a standard inkjet printer. To print the enzyme, we used emptied color-ink cartridges. Glucose oxidase, horseradish peroxidase, and o-dianisidine dihydrochloride are inserted in the printing wells of the cartridge. These enzymes catalyze a glucose reaction whose final products elicit a color change. The enzymes are printed using a template in Microsoft Word. By varying the color in the templates, we can select the amount of each enzyme applied to the paper. To read the strips, we designed a low-cost glucometer using LED lights, a photodetector, and an amplifier that outputs the absorbance, which is processed by an Arduino microcontroller to determine the glucose concentration based on a standard curve. For proof of concept, the strips were tested using glucose solutions of varying concentrations (0-450 mg/dl). The absorption measurement was able to distinguish between glucose solutions with 25 mg/dl accuracy.

INTRODUCTION

Diabetes is not only a problem in first-world countries, but it is also becoming a silent

killer in resource-poor third-world countries. The number of cases in developing countries is

expected to more than double in the next 20 years, expected to explode from 115 million cases in

2000 to 284 million in 2030 (World Health Organization). Of the current 285 million cases of

diabetes worldwide, over 70% of cases occur in low to middle income countries. People must

first be diagnosed with diabetes; however, only half of the patients in the third-world are given a

diagnosis. Even if patients are diagnosed, they must monitor their blood glucose many times

daily and take the appropriate action to either raise or lower their sugar, which becomes very

challenging in resource-poor settings, where glucometers are in short supply. If patients cannot

monitor their blood glucose levels, the rate of complications and mortality rises exponentially

(“Diabetes Facts”).

Developing countries heavily rely on donated supplies, which can be problematic for

those who suffer from diabetes. Supplies for glucose monitoring are not always available, and

when they are, the quantity is limited, which does not guarantee treatment and monitoring for all

individuals. Even if hospitals and clinics receive testing supplies for those who are diabetic, these

supplies may not be usable. Most glucometers sold in developing countries require a specific test

strip. Many times, clinics will receive one brand of glucometer and another brand of test strip,

rendering both technologies virtually useless for the patient. In addition, due to issues of enzyme

stability, the strips have a limited shelf life and may expire before they are donated. The lack of

these resources results in many medical complications that could be easily prevented with access

to testing supplies.

To solve this problem, a cost-effective means of monitoring blood glucose must be

designed to be implemented in resource-poor settings to simplify the treatment and diagnosis of

diabetes. A device such as this would provide an accurate indication of blood glucose levels in

places that could not obtain these data before. This device would allow for doctors to prescribe

medications and treatments for those with diabetes and help prevent complications and

hospitalizations that take up needed bed space in clinics and hospitals.

A cost-effective means of monitoring blood glucose would reduce premature deaths from

unmonitored cases of diabetes. This device would provide doctors and patients with an effective

and obtainable means of monitoring blood sugar on a regular basis to manage diabetes. One must

take many steps in one’s treatment of diabetes, but to start taking these steps, one must be able to

monitor one’s blood glucose levels. The development of a system such as this is the biggest step

in helping patients in the developing world manage living with diabetes.

The goals in this project are (1) to develop a method to print test strips on-demand on

regular paper using a regular inkjet printer with modified cartridges filled with enzyme and (2) to

produce a user-friendly colorimetric meter for the strips that may be easily assembled in the

third-world that is accurate in the range of 0 to 450 mg/dl of glucose. These two elements would

combine to create an inexpensive, easily-obtainable method of managing and treating diabetes in

resource-poor settings. The glucometers could be assembled near the clinic and test strips could

be printed as needed in the clinic lab. The device would provide an accurate qualitative measure

of blood glucose and would be able to show the user if he or she were out of the normal range

and needed to take action. Ideally, this device would provide different kinds of feedback,

whether the patient is in a normal or abnormal range.

The design idea is to use a regular thermal inkjet printer to “print” enzymes onto paper on

an as-needed basis. By using a printer and template to print the strips, a uniform amount of

enzyme will be applied. Though the device must be cost-effective, it must also be reliable. The

use of a printer provides a means of maintaining quality and consistency among the strips and is

simple to use. In this study, the Epson Workforce 30 and H.P. Deskjet 500 printers were used,

which would be readily available in developing countries because of the low cost of this inkjet

technology. The Epson is a newer model whose cartridges are harder to modify than the H.P.’s

cartridges. The Epson has separate color cartridges while the H.P. has one color cartridge with

different compartments for magenta, cyan, and yellow respectively. The Epson has a chip to

monitor ink levels and render cartridges unable to be reused after being emptied, which initially

caused problems; however, the older H.P. cartridges do not contain this chip technology.

Most meters commercially sold today detect glucose levels through an electrochemical

mechanism where the glucose concentration is converted into a voltage or current signal using

special sensor strips (Wang). Recently, microfluidic paper-based analytical devices (mPADS)

have been developed specifically for use in developing countries. These systems tend to be

colorimetric. Paper-based systems are important because paper is widely available, affordable,

compatible, and easily shows a color change because of its white color. The mPADS have

complex systems of hydrophilic microchannels surrounded by hydrophobic barriers to control

the amount of blood allowed to react with the enzyme. The length, width, and height of the

channels are determined by the type of paper used. The reagents used for running the assay of

glucose to determine concentration are then printed onto the paper with an inkjet printer.

Quantitative colorimetric detection of different analytes using mPADS has been achieved

through the process of reflectance detection, where the amount of light reflected off the surface

of the test strip is a function of the concentration of the analyte (Whitesides). A camera captures

the reflected light and the intensity of the color is used to calculate concentration based on a

calibration curve. This method uses the same color change that traditional colorimetric

biosensors utilize. Whitesides et al. successfully quantified glucose concentration in urine using

these methods, which is shown below in Figure 1 (Whitesides).

Figure 1. mPADS for analysis of glucose in urine from Whitesides et al. A) Patterning of paper shown using Waterman red ink to illustrate integrity of hydrophilic channel. B) A complete mPAD after depositing the reagents. The left bulb was prepared for glucose detection, the right bulb was being used for protein detection assays. C) Positive assays for glucose seen by the red color on the left of the mPAD. D) The left portion depicts results of paper based glucose assays using a range of concentrations in artificial urine. E) Analytical calibration plot for glucose concentration.

A colorimetric biosensor was utilized. A biosensor is an analytical device that uses

specific biochemical reactions to detect compounds in a biological sample. This is usually

accomplished by converting a biological response into an electrical signal (Chaplin). A glucose

biosensor that operates based upon oxidation-reduction reactions between glucose, glucose

oxidase (GOx), horseradish peroxidase (HPOD), and O-dianisidine dye was used. The reaction is

shown here:

glucose+H 2 O+O2 gluconic acid+H 2O2

H 2 02+o−dianisidine0−dianisidine+H 2O (reduced form) (brown color)

The final reaction between o-dianisidine, hydrogen peroxide, and horseradish peroxidase

produces a color change (detectable at a range of about 400-600nm) based upon how much

glucose is present in solution. This subtle difference in color and absorbance between different

glucose concentrations is the basis of the colorimetric biosensor in this research.

This enzymatic technique has become commonplace in glucose monitoring systems. A

photodetector was used to output a difference in voltage, which may then be converted into a

measure of absorbance. Using a standard curve obtained from running assays of glucose from

concentrations of 0 mg/dl to 450 mg/dl, an absorbance may be converted into a concentration. A

color change in blood occurs because of the reaction between blood glucose, enzymes, and a dye.

This change in color and also absorbance will be detected by a photodiode in the glucose meter.

The meter must meet specific performance specifications. The glucometer must be able

to accurately show blood glucose levels that someone who is diabetic may experience. This

ranges from lows close to 0 mg/dL to as high as 450 mg/dL. The test strips also must be able to

have a decent shelf life. At minimum, they must be viable for 24 hours after the enzyme

solutions are deposited.

METHODS AND MATERIALS

Peroxidase

Glucose Oxidase

30 Minutes

Figure 2: The color changing reactions at time 0 and 30 minutes.

Dilutions of glucose were made from 45% glucose solution (450 g/L) (MediaTech Inc.)

in distilled water. A stock of 450 mg/dL was prepared using the 45% glucose solution and

distilled water. This stock was left to sit and mix for two hours. From the stock of 450 mg/dL,

concentrations from 0-450 mg/dL (by 25 mg/dL) were created for the purpose of creating a

standard curve.

Using a procedure based upon Sigma’s Enzymatic Assay of Glucose, solutions were made

for testing. Sodium acetate buffer (50mM) was prepared by adding 3.402g of sodium acetate

trihydrate (Sigma) to 500mL of deionized water. The pH was then adjusted to 5.1 using

hydrochloric acid (Acros Organics).

O-dianisidine solution (0.21 mM) was prepared by dissolving 20mg of o-dianisidine

dihydrochloride (Sigma) in 8mL of purified water in a vial protected from light.

GOx solution (0.8 unit/mL) was produced by adding 6.94 mg of Type II glucose oxidase

from Aspergillus Niger (17,300 units/g ) (Sigma) to 150mL of cold 50mM sodium acetate buffer.

POD (60 units/mL) was prepared by adding 46.6mg of Type II horseradish peroxidase

(193 purpurogallin units/mg sold) (Sigma) to 150mL of cold water.

A standard 96-well plate was obtained to run assays of differing concentrations of

glucose to obtain a calibration curve. Assays were run using different amounts of glucose, dye,

enzyme, and buffer, which is shown in Figure 3; furthermore, each combination was run in

duplicate or triplicate.

Assay # Amount of Enzyme

Amount of Buffer

Amount of Dye

Amount of Glucose

1-5 200µL * * 10µL

6-9 11µL 282µL 7µL 10µL

10 50µL 203µL 7µL 10µL

11 75µL 150µL 10µL 10µL

12 100µL 100µL 10µL 10µL

Figure 3: The relative amounts of enzyme in each well for all of the assays.

* These assays were run with the Sigma Enzymatic Assay of Glucose kit, which combined dye, buffer, and enzyme.

The buffer was first deposited in well A1 and advancing until A12 was reached. Rows B, C, E,

F, and G were filled in the same way. Row D was skipped to ensure that overflow would not

contaminate the different samples in vertically adjacent wells. This filling procedure was

repeated with GOx, HPOD, dye, and different concentrations of glucose. Each concentration of

glucose would be in three wells, all vertically adjacent to one another in order to run triplicate

samples on each plate. Concentrations of glucose were increasing in increments of 25 mg/dL

starting at 0 mg/dL in wells A1, B1, and C1. When adding glucose, pipette tips were switched

between each concentration of glucose to avoid contamination.

Once everything was added to the 96-well plate, the enzymes were allowed to react for

30 minutes at room temperature before being put into the spectrophotometer. The

spectrophotometer was set to read at 500 nm and also a spectrum of 400-600 nm by 20 nm. At 45

minutes, the plate was again read at 500 nm. All of this data was imputed into Microsoft Office

Excel 2010™ from the Gen5™ spectrophotometry program. This data was graphed using an

average of each concentration and formulating a line of best fit. The R2 value was also examined

for the accuracy of the curve. The curve from the ninth assay is shown in Figure 4.

0 100 200 300 400 5000

0.10.20.30.40.50.60.70.80.9

Glucose Assay at 500nm

(Concentration [in mg/dl])

Abso

rbtio

n

Figure 4: The standard curve obtained from the data of the 9th Assay of glucose. The Arduino microprocessor was programmed using the equation y=.0017x + .0121

to convert light

absorbance into glucose concentration, with an R² value

Before printing test strips, different designs, combinations of enzymes, and types of paper

were tested. The main problem encountered was the viscosity of blood and how blood could be

applied to the strip in a sterile manner. To help influence the flow of blood to the correct area on

test strips, various hydrophobic materials were used and tested. To test patient-to-strip delivery

systems, a hydrophobic sheet was obtained, such as contact paper or parafilm. An absorbant

paper, such as filter paper, was backed with the hydrophobic sheet. Then designs were cut into

the filter paper without cutting away the hydrophobic surface using an X-ACTO knife. A

mixture of corn starch and water with a viscosity similar to that of blood was applied using a

gloved hand. With a sweeping motion, a finger prick was simulated and the distance the corn

starch and water mixture moved over a thirty minute period was observed. After obtaining a

design that worked, strips were printed using the printer.

To make test strips with the printer, modified cartridges with the ink removed were used

to print enzyme onto regular paper. An old H.P. Deskjet 500, which had ink cartridges that were

easy to clean and fill, was first used. A newer Epson Stylus Workforce 30 was then used because

it is currently widely available in the market and is relatively cheap. The Epson has cartridges

that are very hard to clean, fill, and that reset themselves after they become empty. Therefore,

instead of cleaning and emptying the Epson cartridges, Epson-compatible cartridges that were

reusable, refillable, and came empty were purchased for ease of use. However, for both printers’

cartridges the same care techniques were used.

Before inserting enzyme and using the cartridges, they were cleaned to prevent clogging

and get rid of extra salts and proteins. The cartridges were immersed in a 1:1 deionized water

solution of rust inhibitor for 10 minutes (Burnishine Products). The cartridges were immersed in

a 1:4 in deionized water instrument lubricant solution for 30 minutes (Burnishine Products). The

cartridge was put in a beaker full of deionized and sonicated for 15 minutes. After printing, the

cartridges were rinsed out thoroughly. They were then put it in the same lubricant solution as

before for 30 minutes and sonicated for 15 minutes.

To print test strips, the same procedure was followed; however, the templates for printing

changed over time. To print, the printer was first powered on and given time to warm up. The

desired printing surface was placed in the printer’s tray (wax paper, regular paper, or filter

paper). On Microsoft Word, a template was either created or selected from premade templates

for printing test strips. An example of the test strip design is shown in Figure 5:

The cartridges were filled with 10 mL of glucose oxidase, peroxidase, and o-dianisidine

dye and inserted into the printer. Glucose oxidase was in the magenta cartridge, peroxidase was

in cyan, and o-dianisidine was in yellow. In Microsoft Word, the file was selected to print on the

printer. The printer was warmed up again and the cartridges moved to the “ready” position. In

the old printer, the paper feed mechanism had to be bypassed by pulling on it. This tricked the

printer into thinking regular sized paper was inside it, so that the printer would actually print on

paper or wax of any size. For multiple copies, the paper feed mechanism would have to be

manually bypassed each time.

Once test strips were printed, glucose could then be applied to the strips and the

glucometer could be used to find the absorbance of the strip. The absorbance directly relates to

the level of glucose. To test the absorbance of the strips, 5µL of glucose solution is first applied.

After waiting a few minutes and observing a color change on the strip, the strip is put on the

photodiode of the glucometer. On the other photodiode is a strip with glucose solution and all

Figure 5: The design of our test strips. The enzyme is printed in blocks on the ends of the strip. The wax paper causes blood to move and interact because of capillary action.

enzymes except for glucose oxidase. When the glucometer runs, an LED light shines onto the

strip and photodiode. The photodiode compares the absorbance to the other diode, which acts as

a control, and outputs the absorbance to a LOG102 amplification chip. The amplification chip

then outputs to an Arduino microprocessor which uses the equation determined from the

standard curve and converts the absorbance to a concentration of glucose.

RESULTS

A calibration curve with an R2 value of .9571 was obtained through the running of assays

and their analysis in Microsoft Excel 2010. An R2 value of over .9 is said to be statistically

significant. Because this value is over .9, the glucometer should be relatively accurate in a range

of 0-425mg/dl of glucose and is precise to 10mg/dl intervals.

Test strips have been created by layering filter paper and contact paper. Test strips were

designed so that blood would wick down the test strip in a controlled manner. This way, the

amount of blood on the strip initially would not matter unless it was an amount too small to be

detected. Test strips have been successfully printed using the printing process described

previously. The test strip design is currently a 5mm wide and 20mm long filter paper and

contact paper strip with printed o-dianisidine, glucose oxidase, and horseradish peroxidase.

When fetal bovine serum is applied to the test strips, the color changing reactions occur

in about thirty minutes. The glucometer will read the difference in voltage and then convert this

into a concentration of glucose that is, on average, only 6mg/dL off. However, the readings have

a standard deviation of over 40.0mg/dL.

DISCUSSION

The color-changing reactions have been taking 30 minutes to complete, which is too

much time because the patient has to take action quickly. Modern meters take seconds to give the

user a reading.

In addition, modern meters are accurate up to concentrations of 600mg/dl and are very

precise.

This glucometer and test strip method is relatively inexpensive, utilizing cheap materials

such as filter paper, contact paper, and parts widely available.

Even though this meter does not currently comply with ISO standards, it may be

employed in developing countries when other meters are not available. The creation of a system

such as this is integral for resource-poor nations.

This device is to be used as a bridge for those patients who are waiting for more accurate

testing supplies.

Although this meter-strip system is currently not as accurate as standard commercial

systems and further tests are required to improve the design, it is still accurate enough to inform

patients of their relative range of blood glucose (low, normal, high). This can allow them to still

take appropriate action for raising or lowering their blood glucose. Future work will focus on

optimizing the strip design to shorten the time necessary before a measurement is made and to

decrease the variability between measurements. In addition, further testing will be performed to

assess the stability of printed strips. The hope is that this system can be implemented in resource

poor settings where glucometers are in short supply and help decrease the incidence of diabetes

related complications in these settings.

Acknowledgments

I would like to acknowledge Dr. Delphine Dean of the Biomedical Engineering

Department at Clemson University for serving as my mentor and all of her help throughout the

research process. I would also like to acknowledge Kayla Gainey and Kelsey Byrd as my

partners on this project.

Literature Cited

“Diabetes Facts”. World Diabetes Foundation Online. Web. 26 January, 2012. <

http://www.worlddiabetesfoundation.org/composite-35.htm>

“Diabetes cases could double in developing countries in the next 30 years”. World Health

Organization Online. Web. 26 January, 2012. http://www.who.int/mediacentre/news/

releases/2003/pr86/en/

American Diabetes Association. Diabetes Care. Web. 11 July, 2012.

Chaplin, Martin. “What are biosensors?”. London Southbank University Department of

Engineering and Science. Web. 23, April 2012.

<http://www.lsbu.ac.uk/biology/enztech/

biosensors.html>

Wang, Joseph. “Electrochemical biosensors”. American Chemical Society: Chemical Reviews,

2008, 108(2): 814-825.

Whitesides, George. “Diagnostics for the Developing World: Microfluidic Paper-Based

Analytical Devices. Analytical Chemistry, 2010, 82(1): 3-10.