Factors affecting blood glucose monitoring: sources of errors in measurement

Abstract

Glucose monitoring has become an integral part of diabetes care but has some limitations in accuracy. Accuracy may be limited due to strip manufacturing variances, strip storage, and aging. They may also be due to limitations on the environment such as temperature or altitude or to patient factors such as improper coding, incorrect hand washing, altered hematocrit, or naturally occurring interfering substances. Finally, exogenous interfering substances may contribute errors to the system evaluation of blood glucose. In this review, I discuss the measurement of error in blood glucose, the sources of error, and their mechanism and potential solutions to improve accuracy in the hands of the patient. I also discuss the clinical measurement of system accuracy and methods of judging the suitability of clinical trials and finally some methods of overcoming the inaccuracies. I have included comments about additional information or education that could be done today by manufacturers in the appropriate sections. Areas that require additional work are discussed in the final section.

Figure 1.

In each panel, the center of the crosshairs represents the reference value. In the left panel, the individual values, shown in yellow, have a mean value that is the same as the reference value, thus the set of values is accurate, although no individual value is similar to the reference. In the center panel, all values are nearly identical, thus the set of values is very precise and reproducible, although no individual value is similar to the reference. In the right panel, the set of values is both accurate and precise.

Figure 2.

The left panel is a set of measured glucose values shown in yellow circles and compared to the line of identity (reference) shown in green. The set of values is accurate since half of the values are high by 100 and half are low by 100, but none of the values represents the reference values. In the center panel, all measured values are high by 100. The set of values is very precise and reproducible but not representative of the reference values. On the right, the set of values is both accurate and precise, and under these conditions, the individual values must all be similar to the reference values.

Figure 3.

Mean absolute relative deviation is created by taking the absolute value of the deviation from the reference value. Thus a value that is 100 points too high has a deviation of +100 and a value that is 100 points too low is also +100. Each value is calculated as a percent of the reference and then averaged. On the right, if the reference values are 100, 200, 300, and 400mg/dl, the individual errors would be 100%, 50%, 33%, and 25%, respectively, making the MARE 53.3% (the sum, 213, divided by 4, the number of values).

Figure 4.

Example of a hypothetical label for blood glucose strips that would give meaningful data about the accuracy of the system and allow consumers to compare the value of systems.

Figure 5.

Effect of enzyme/mediator on strip accuracy. The current produced by a strip is dependent on the amount of reaction (amount of enzyme/mediator and amount of glucose) and the interaction with the electrode. As seen on the left, strips normally have excess enzyme and mediator, and moderate loss of these components does not significantly affect the reading. In contrast, the area of interaction with the electrode is limited, and loss of coverage of the electrode by even a small amount lowers the reading.

Figure 6.

Mechanism of action of a glucose strip. Glucose oxidase interacts with glucose, taking an electron and forming gluconic acid. The enzyme then passes the electron to water and oxygen, regenerating the enzyme and forming hydrogen peroxide. On glucose strips, a mediator replaces oxygen, accepting the electron and passing it to an electrode to generate the current that is reported as the glucose concentration.

Figure 7.

Effect of temperature on strip accuracy. Glucose strips are fragile and must be stored for limited time under specific conditions. Shown here is the effect of storing strips at 40 °C (104° Fahrenheit) for an extended time.

Figure 8.

A group of mountain climbers tested blood glucose systems atop a 3000 m mountain to measure the effect of altitude and temperature (Figure 9). The glucose-oxidase-based meters overestimated the glucose by 6–15%. The glucose-dehydrogenase-based meters were more accurate at high altitude.

Figure 9.

The same group of mountain climbers measured the effect of the ambient temperature of 8 °C on accuracy, reported here.

Figure 10.

Raine and coworkers intentionally miscoded meters and allowedpatients to measure glucose with them. Substantial errors occurred.

Figure 11.

Glucose in erythrocytes is in equilibrium with plasma glucose but at lower levels. The total blood glucose, therefore, is dependent on the hematocrit.

Figure 12.

Some meters, such as brand 2, in black, compensate well for changes in hematocrit or are not affected. Others, such as brand 1, shown in red, have excessive errors at low and high hematocrit levels.

Figure 13.

Over the decade of the 1990s, meter accuracy did not improve when tested by highly trained nurses.

Figure 14.

As seen in Figure 13, when tested by nurses (orange), meter accuracy did not improve substantially from 1993 to 2004. In contrast, accuracy did improve dramatically when tested by patients (yellow).

Figure 15.

Sugary substances such as cookies (Chips Ahoy) raised glucose readings substantially. Lotions had only a minor effect and soap had almost none.