Errors in using devices
The increased complexity of modern diabetes care creates both challenges and potential hazards for both patients and providers. It is not only defective or malfunctioning devices that present a problem. For many of the devices now in use, neither care provider nor patient nor their family may understand the use of the devices they are using in more than a superficial way. The provider often cannot answer many questions posed by the patient, and may not be able to troubleshoot or even identify when a medical device is malfunctioning or being misused. Too often the patient is not properly evaluated to ascertain whether they can safely use the device they have been given, and errors, sometimes with serious consequences, can and do result.
The most complex instrument most of our patients may encounter is the external insulin pump. Insulin pumps have improved greatly since the earliest commercial products, now more than 40 years ago. The number of different varieties of pumps is quite large, from the very simple pumps to the most complex external insulin pumps. Most adult patients who are considered for therapy have Type 1 diabetes, many have neuropathy or retinopathy, many have had diabetes for some time, and some have failed alternative therapy, which for most is intensive insulin therapy. There is often great pressure to intensify the therapy further, a reasonable approach, but the choice of therapy needs to be individualized.
Careful choosing of appropriate candidates for insulin pump therapy is essential, but at some clinics and hospitals, particularly in the U.S., it may not be done thoroughly. The patient must, for example, be both willing and able to perform acceptable self-blood glucose monitoring in order to use an insulin pump safely, either depending upon point-of-care (POC) glucose meters alone, or using mainly continuous glucose monitoring systems (CGMS). A thorough evaluation of the patient’s cognitive function as well as their psychological status is needed. In addition, the patient should receive sufficient education about the requirements of self-management with an insulin pump so they can decide whether they are willing and able to use this device safely. Unfortunately, if any of these three components of evaluation is incomplete, (cognitive evaluation, psychological assessment, and education), errors may occur.
In our experience in the U.S., failure in each of the three areas is not uncommon. Patients rarely volunteer for a neurocognitive exam, yet patients with poor executive function or impaired power of attention or memory will make many errors when using a complex insulin pump. Patient who have a depressive disorder or those with substance abuse or other serious emotional disorders often may have serious problems with insulin pumps. And patients who have not been educated on the proper use of the pump are at risk for severe episodes of hypo- or hyperglycemia, particularly in the first week of use. Some of the best insulin pump programs in Europe include not only intensive education prior to unsupervised use of the insulin pump, but also reevaluation at six months or sooner, and periodically thereafter to make sure the pump user is still able to perform the needed tasks safely and the glucose control is optimal. Retraining of capable pump users is especially important during pump upgrades. Comorbid conditions affecting the patient may have an adverse effect on the suitability of pump use and should be carefully evaluated. An example of this situation would be a patient who develops Stage IV chronic renal failure, develops depression and mild cognitive dysfunction. Therapy will need to be simplified for safety, and the insulin pump may no longer be optimal.
Many patients initially assume that an insulin pump, a complex, highly-engineered device, will not fail. But they do. Judith Cope reported examples of failures of insulin pumps in the U.S. and the hazards they presented to the patient.
There have been pump failures due to pump underdelivery of insulin, and the more dangerous problem of overdelivering insulin. In the past five years, several manufacturers of insulin pumps have issued recalls due to defective pump components. In one instance more than 1.1 million pump infusion sets were voluntarily recalled because a defective reservoir could potentially lead to insulin overdelivery.
Among the causes of pump failure were software deficits, user interface issues, or mechanical or electrical failure. Pump design flaws are not uncommon, and may result in repetitive failure or present latent flaws that become evident only in certain settings, as for example an alarm that should warn the patient about a tube blockage, but because of a defective reservoir, fails to alarm. Unfortunately, at present, insulin pump manufacturers, despite the recommendation of the supervisory agencies, do not use open-source insulin pump software, which would greatly speed external evaluation and correction of errors in the software.
External insulin pumps are potentially a highly efficient and flexible tool for glycemic control, a rapidly developing technology of potentially enormous value for our patients. In order to most effectively use insulin pumps, we need to provide a substantial infrastructure if the pumps are to be used safely. We recommend creating a culture of safety, focused on the use of insulin pumps, that provides highly educated providers at the point of care, and provides an extensive education program for both providers and patients, and frequent testing to be sure that the pump management is done safely.
Recommendations for developing a culture of safety
First, develop a curriculum, a training program for the providers who will be teaching the patients. The curriculum should include all that is to be taught to the patients. Ascertain that the providers both have learned the material well and observe their teaching style and skill before releasing them to work. The curriculum should include fundamental principles of insulin use, beginning with the explanation of why basal and bolus dosing is used and algorithms that are useful to help calculate basal doses, and methods to estimate insulin sensitivity factors, insulin/carbohydrate ratios, basal rates at different times of the day and night, and adjustments of dosage that will be needed for a wide variety of common situations, from sick-day rules to adjustments for exercise. A very common error in clinical insulin pump use is insulin overdosage because the patient failed to calculate how much of earlier dosing of insulin is still “on board” and will still lower glucose levels. When one calculates the dosage needed to compensate for an elevated glucose level or an upcoming meal, the patient needs to make sure that the amount of insulin still remaining from the previous bolus is accounted for. The term most often used to describe this error is “stacking” of insulin doses, where the second or third dose of insulin is close enough to the first dose that one needs to reduce the subsequent dosing so as not to overdose oneself. This is not a simple concept for many patients. The use of diagrams or other visual aids is often helpful to illustrate this point.
However, it is also important to explain to patients when to contact support staff for expert guidance. Explaining the conditions where the patient should call for assistance is key and needs to be individualized. The vulnerabilities of a pregnant patient with diabetes on an insulin pump are quite different from that of a healthy adult male with moderate insulin sensitivity and insulin deficiency, and the needs for support are different. We strongly recommend developing, training, and appropriately staffing the support sites that will be available: on-line, on the phone, and in person, for the patient.
Many patients will grasp some concepts readily and stumble with other concepts. Attempt to customize the education for each patient, as well as identifying common areas of error of most patients and emphasize both the generally troublesome issues and issues specific to the patient in the initial education.
Allow flexible scheduling for patient education and be sure that their technical skills are adequate in handling pump components such as proper catheter insertion, programming of the pump, use of the settings, changing of the reservoir, replacing batteries, significance of pump alarms, choosing the insertion site, proper cleaning of the insertion site, priming, button pushing, etc.
Both patient and family, if possible, should receive training and education. And most importantly, in the event of pump failure for any reason, the patient should have insulin syringes or pens and insulin available, and instruction as to how much insulin to administer while waiting to speak to the trained personnel who will be available to advise them further while pump delivery is interrupted.
It is not uncommon for severe problems with insulin pump use to develop in the first two weeks, when the patient is least familiar with their device. When there is a question as to whether the pump is functioning properly and the patient has marked hyperglycemia, it is most helpful if the patient has a checklist as to how to respond, whom to contact, and what information to collect about pump function. Patients need to be instructed as to what to do if there is any evidence suggesting pump malfunction, and to understand the common and uncommon issues that may occur with the pump model that they possess.
Point-of-Care (POC) glucose meters
Point-of-care glucose (POC) meters are used by nearly all patients who must monitor their glucose levels. Even those who use CGMS for continuous monitoring of their interstitial glucose levels use a POC glucose meter to calibrate their CGMS at least twice daily. Unfortunately, POC glucose meters, although a mainstay of self-management, may be a source of inaccurate glucose readings that may be large enough in their variance from the true blood glucose levels that patients and healthcare providers may make serious therapeutic errors that harm patients.
There are many reasons as to why POC glucose meter data may lead to injurious errors and, in order to reduce the likelihood of errors in care, one must first understand why errors are so frequent.
Part of the problem has been, in recent years, due to the lax regulatory standards to which the glucose meters have been held. Both in the United States, where the Food & Drug Administration (FDA) sets the regulations, and worldwide, where the ISO-15197 regulations hold sway, until 2014, previous regulations only required 95% of the readings to be within ±20% from the true analytic glucose value. Up to 5% of the readings could be anything at all! In other words, these standards allowed the proliferation of meters that would, at times, provide glucose readings that were so far removed from the true blood glucose readings so as to create a clinical outlier – that is, a misleading result that commonly leads to errors in insulin dosing or under-treatment of low glucose levels. Although newer regulations promise somewhat tightened standards, the general public still often is using meters that range widely in analytic accuracy.
POC glucose meters, for the most part, use either glucose oxidase (GO) or glucose dehydrogenase (GDH) methodology to measure capillary glucose levels. Each method has its vulnerabilities. The meters using glucose oxidase enzyme methodology, for instance, are very susceptible to variations due to hypoxia or marked altitude changes, which falsely elevate measured glucose levels. High O2 tensions result in falsely lowered glucose levels. However, glucose oxidase methodology is less susceptible to drug and chemical interference.
On the other hand, glucose dehydrogenase enzyme methodologies are not sensitive to oxygen fluctuations, but can be profoundly affected by interferences. Perhaps the best documented evidence of catastrophe due to this vulnerability occurred in 2009 when the FDA reported 13 deaths that occurred when the unwary clinicians were treating patients on peritoneal dialysis. The POC glucose meter that was used was an ACCU-CHEK, which used a form of GDH methodology using the PQQ coenzyme, which falsely reports maltose as glucose. The peritoneal dialysate contained icodextrin, which is slowly metabolized to maltose in the body. The increasing levels of maltose, falsely reported as high glucose levels by the GDH-PQQ type glucose meter, led the clinicians to overdose the patients with insulin until coma and eventual death occurred. The problem was discovered when clinicians realized that the simultaneous central laboratory glucose levels were in the severely hypoglycemic range, but sadly, this error was repeated in different centers until the problem became more widely recognized.
There are other common problems due to glucose meter dysfunction. Many meters in common use do not perform as well as their manufacturers claim. Skeie and Kristensen showed that many meters do not properly adjust to changes in hematocrit and give falsely elevated glucose levels with anemia. And recently, there were recalls of meters from two manufacturers because very high glucose levels (≥ 1024 mg/dl) would be falsely reported as very low glucose levels, since their software would begin recounting from zero when the glucose levels reached 1024 mg/dl, resulting in a post-analytic error, faulty reporting of a correctly analyzed result.
Freckmann has shown that 40% of the approved meters he studied performed more poorly than the low IOS standard 15197 set prior to 2013. Also, he noted that many approved meters performed much better in certain glucose ranges than in others, and most were less accurate in the hypoglycemic range. However, there are a number of new meters that have performed much better, with Total Analytic Error ~4-5%. It is probably prudent for both providers and patients to encourage the use only of more accurate, well-validated meters, learn their vulnerabilities, and act accordingly.
Since many meters need regular calibration, in safer clinical systems, regular education is provided to all who use the meters, metrics are kept on the calibration and performance of each meter and each user. This maintenance of metrics on glucose control is very useful in increasing the usefulness and safety of these meters, and to identify clinical problems in glycemic control. However, when the patient receives a new meter, the patient should receive retraining. Patients with any cognitive dysfunction are particularly apt to find a new meter an unwanted challenge.
It is often assumed that technique in using meters is so straightforward that less attention needs to be paid to this aspect. But, that would be a serious mistake. There are abundant data showing that user error contributes greatly to the inaccuracy of glucose meters. For example, failure to wash fingers before obtaining a capillary sample, a pre-analytic error, can lead to overestimation of blood glucose levels by up to 35%. Also, failure to dry the finger after washing it can cause hemodilution lowering blood glucose by up to 28%. Moreover, some meters are very sensitive to drop size and may be hard to read or use. Despite these very large errors, it is surprising that so little attention is paid to the technique of the use of POC glucose meters. Moreover, many patients have no idea as to the magnitude of error they are adding to the measurement of glucose they are seeking by their poor technique.
It is worthwhile to recall that from a practical standpoint, although a manufacturer may claim that under ideal conditions their instrument may have a total analytic error of less than 10%, what is clinically important is the total error, the sum of the pre-analytic error, analytic error, biologic variability, and post-analytic errors. If the pre-analytic errors are large, as for example failure to wash hands before obtaining a sample, or failure to dry the finger before pricking the finger, the pre-analytic errors may result in a total error that dwarfs the analytic error and a true glucose of 120mg/dl may be reported as much as 60-100 mg/dl or more higher or lower than the true value.
Table 1: Summary of the effect of interfering substances on glucose measurement (Click to enlarge)Also, the storage of meters and strips is another potentially major cause of errors. The strips may degrade with excessive heat or cold or humidity or exposure to air. These kinds of errors can lead to serious mistakes in care, and those providing a program of diabetes care should be advised to spend time during patient and staff education to inform people so as to avoid such mistakes in the future.
Table 2: Effect of preanalytic errors on glucose management (Click to enlarge)Probably the most controversial issue surrounding the limitations of POC glucose meters is whether they should be used at all in critical care settings. There are studies which highlight the weaknesses of most POC meters in critical care areas, both in the operating suite and in ICU’s. In hypotensive states, during cardiovascular collapse or during surgery, capillary values using POC glucose meters appear often to be far less accurate than either the central laboratory glucose values or point-of-care glucose levels obtained using blood-gas methodologies, and many experts recommend not using POC glucose meters unless they are completely validated in the specific settings for which they are intended. There is even limited data to show that some meters will provide falsely low glucose readings in the presence of diabetic ketoacidosis, with levels showing up to 300mg/dl lower than the true values.
Because POC glucose meters are so widespread in their use and essential to many patients in their daily self-care, a modern diabetes management program would be well-advised to provide an infrastructure and training programs to support continue use of POC glucose meters. We recommend, whenever possible, to test the meters used by patients first. Kristensen, Skeie, Heineman and others strongly recommend that an independent institute be developed to test all meters and strips and to retest them to see if their accuracy decays with time. We agree that such an institute, if independent, would be a great advance.
Robust education programs for both patients and providers are key. Also, special populations will respond to different educational strategies. For example, Mykityshyn showed that older patients forgot much of the education on POC meters within one week, and did much better when the education was repeated one week later and the teachers relied more on visual aids in teaching.
Self-blood glucose monitoring has revolutionized diabetes self-management and together with the advances in continuous blood glucose monitoring, has changed the life of the patient with diabetes for the better. But the vulnerabilities of POC glucose monitoring are complex, and in order to provide better and safer diabetes management, we need to do more to educate patients and providers to deal with the latent errors inherent in the use of each POC glucose meter.
Grunberger G et al. Consensus Statement by the American Association of Clinical Endocrinologists/American College of Endocrinology Insulin Pump Management Task Force. Endocr Pract. 2014;20(5):463-489
Cope JU, Samuels-Reid JH, Morrison AE. Pediatric use of insulin pump technology: a retrospective study of adverse events in children ages 1-12 years. J Diabetes Sci Technol. 2012;6(5):1053-1059
Hellman R. “Interfering Factors in Quality of Glucose Measurement”. In: Grassi G, Bruttomesso D (Eds) Technological Advancements in the Treatment of Type 1 Diabetes (Frontiers in Diabetes). Karger, 2015, Vol 24, 63-80.
Freckmann G et al. System accuracy evaluation of 27 blood glucose monitoring systems according to DIN EN ISO 15197. Diabetes Technol Ther. 2010;12(3):221-231
Scott JG, Bruns DE, Boyd JC, Sacks DB. Tight glucose control in the intensive care unit: are glucose meters up to the task? Clin Chem. 2009;55(1):18-20
Mykityshyn AL, Fisk AD, Rogers, WA. Learning to use a Home Medical Device: Mediating Age-Related Differences with Training. Human Factors. 2002;44(3):354-364
US Food and Drug Administration: Public health notification: potentially fatal errors with GDH-PQQ glucose monitoring technology. 2009. http://www.fda.gov/medicaldevices/safety/alertsandnotices/publichealthnotifications/ucm176992.htm
Kristensen GB et al. Standardized evaluation of nin instruments for self-monitoring of blood glucose. Diabetes Technol Ther. 2008;10:467-477.
Dungan K et al. Glucose measurements: confounding issues in setting targets for inpatient management. Diabetes Care. 2007;30:403-409
Pitkin AD, Rice, MJ. Challenges to glycemic measurement in the perioperative and critically ill patients: a review. J Diabetes Sci Technol. 2009;3:1270-1281
“High-alert” medications and patient safety. Int J Qual Health Care. 2001;13:339-340
Hellman R. “Glycemic Variability in the Use of Point-of-Care Glucose Meters”. Diabetes Spectrum, Vol 25, No 3, Summer 2012, pp 135-140
Kristensen GB et al. Standardized evaluation of nine instruments for self-monitoring of blood glucose. Diabetes Technol Ther. 2008;10:467-477
Dungan K et al. Glucose measurements: confounding issues in setting targets for inpatient management. Diabetes Care. 2007;30:403-409
Tang Z et al. Oxygen effects on glucose meter measurements with glucose dehydrogenase- and oxidase-based test strips for point-of-care testing. Crit Care Med. 2001;29:1062-1070.
Hellman R. Glucose meter accuracy and the impact on the care of diabetes in childhood and adolescence. Pediatr Endocrinol Rev. 2011;8:200-207
Cembrowski GS et al., Tran DV, Slater-Maclean L, Chin D, Gibney RTN, Jacak M. Could susceptibility to low hematocrit interference have compromised the results of the NICE-SUGAR trial? Clin Chem. 2010;56:1193-1195
Heinemann L. Measurement quality of blood glucose meters: is there a need for an institution with an unbiased view? J Diabetes Sci Technol. 2007;1:178-180