We analyzed 447 glucose test strip batches manufactured from February 2006 to July 2007, using a consensus error grid according to Parkes et al
with the following risk categories:
- No effect on clinical action
- Altered clinical action with little or no effect on clinical outcome
- Altered clinical action—likely to affect clinical outcome
- Altered clinical action—could have significant medical risk
- Altered clinical action—could have dangerous consequences
The greatest concern focuses on errors that fall within risk categories C, D, and E, which all require altered clinical action and are considered extreme events. Errors that fall within risk categories A and B have little or no impact on clinical outcome. The worst-case prediction bias in a mismatch of the glucose test strip code and test strip was estimated for 6 representative glucose levels: 30 mg/dl, 50 mg/dl, 70 mg/dl, 180 mg/dl, 240 mg/dl, and 500 mg/dl. These estimates were calculated by mathematically modeling the influences of the wrong glucose test strip code parameters, the shelf life of test strips, and within-batch variability. These values were then plotted along the consensus error grid ().
Estimated average bias of glucose test strip batches due to a mismatch of the test strip code and the test strip, using four extreme mismatch scenarios at six representative glucose levels.
Factors governing the estimated error rates included the probability of:
- A forced mismatch between the test strip code parameters and the test strip, which represents the largest divergence of performance based upon information from 447 glucose test strip batches
- A customer continuing to use an expired batch of glucose test strips
- The prevalence of blood glucose values <70 mg/dl
The overall probability of a risk category C event is the aggregate of the probabilities of each of these factors coming true. This was computed as the product of the individual probabilities. The only risk category C event that was likely to affect clinical outcome was a low glucose level (i.e., ≤70 mg/dl). The worst-case estimated likelihood of observing a category C error throughout the shelf life was calculated as 16.0 ppm. No values (actual or estimated) fell into risk categories D or E.
In addition to the theoretical modeling described earlier, a study was performed using spiked venous blood with four glucose solutions and two control solutions to assess the performance of extreme mismatches between the test strip code parameter and the test strip. Bias within this study ranged from ±15 mg/dl at glucose levels below 70 mg/dl to ±15% at higher glucose levels; these values fell within area A of the consensus error grid and support the results of the worst-case scenario simulated by the mathematical modeling. Maximum system accuracy requires correct glucose meter coding, reinforcing the need to code in accordance with the labeling instructions. However, this assessment with currently released lots demonstrates that even if the ACCU-CHEK Aviva glucose meter were miscoded, the probability of incorrectly altering a clinical action is low. (Roche Diagnostics. Effect of miscoding on ACCU-CHEK® Aviva system accuracy. Internal data 2007.) A study by Haak et al., sponsored by Bayer, showed that the mean absolute deviation of the ACCU-CHEK Aviva glucose meter is 9.5%, if coded according to instructions, and 10.8% if miscoded. (Haak T, Gerlach H, Krichbaum M, Hermanns N. The effect of incorrect coding of blood glucose meters on the accuracy of blood-glucose self-testing Poster presentation. 2007.)
Advantages of Coding
What are the benefits for the diabetes patient of coding glucose test strips? The electronic code chip containing the batch-specific code can transmit additional information between the glucose meter and test strips ().
Interaction of the electronic code chip, the processor of the glucose meter, and the glucose test strip in the Roche ACCU-CHEK® Aviva system.
This communication allows for easy upgrades to the glucose meter by providing firmware updates on the code chip. The following are examples of major upgrades to Roche's glucose meters conveyed via the code chip:
- The minimal blood volume required for a valid measurement of blood glucose is reduced.
- The reaction chemistry of the glucose test strips is altered.
- The blood glucose test time is reduced.
- The referencing method of the glucose meter is changed from whole blood to blood plasma.
For patients with diabetes, these upgrades offer a clear benefit: they can avail of substantial technological improvements without having to acquire new glucose meters and invest time and effort in learning how to use them. It also significantly reduces the resources needed for retraining diabetes patients, their health care providers, and the diabetes educator teams. The batch-specific code also contains information about the control solution ranges and the expiration date of the glucose test strips. This is a clear advantage for the patient because it significantly enhances the safety of the blood glucose measurement: the glucose meter notifies the user if the expiration date of the test strips is close, so that he or she can replenish the test strips, or automatically rejects test strips that have already expired. This is an important safety feature: if the patient uses expired test strips, the result of the glucose measurement can deviate from the true value and mislead the patient.
The labeling instructions for the Bayer Ascensia® CONTOUR® blood glucose meter are as follows:
It is important not to use the test strips or control solution if the expiration date printed on the bottle label and carton has passed or it has been six months (180 days) since you first opened the bottle. It will help to write the six month discard date on the label in the area provided when you first open the test strips or control solution. (
Glucose test strip bottle of the Bayer Ascensia® CONTOUR® system with handwritten “six month discard date “, printed lot number and expiration date.
In other words, Bayer suggests that patients with diabetes write by hand the “six month discard date” on the bottle containing the glucose test strips and stop using them if the discard date has passed. We think that this manual “coding” process is a potential source of error. In real-life situations, patients with diabetes might forget to note the discard date on the bottle, the handwriting might become illegible over time, or the patients might not notice that the discard date has passed. Ultimately, these situations can result in the use of expired glucose test strips.