Bedside glucose testing using strip technologies constitutes the highest test volume in most hospital-based point-of-care testing (POCT) programs. A 2010 study estimated that in health care settings, 51 percent of POCT involves bedside glucose meters using strip technologies.1 However, since the 2014 published warnings from the US FDA and CMS (both since retracted) on the potential erroneous results produced by glucose meters in critically ill patient populations,2,3 managing the off-label use of bedside glucose testing presents a significant challenge for hospital POCT programs.
Issues Exacerbated in Critically Ill
Originally designed as an over-the-counter product for managing glucose levels in diabetic patients, glucose meters are now used on virtually every hospital patient regardless of medical condition or the limitations specified in the manufacturer’s package insert.4 Glucose meters were initially validated through healthy diabetic lay users but were later marketed for use in different patient populations without additional testing and without the approval of the FDA.4 Between 2004 and 2008, nearly 13,000 serious adverse events were reported to the FDA regarding glucose meters using strip technologies.5 And between 1997 and 2009, the FDA received 13 reports of fatalities associated with glucose meters based on one of the glucose strip technologies, the glucose dehydrogenase pyrroloquinolinequinone (GDH-PQQ).5 These fatalities were reported in critical care settings within the hospital environment.
Glucose measurements using strip technologies may produce erroneous results, including out-of-range hematocrit levels, low oxygen levels, and interferences from drugs and drug metabolites used for the management of patients’ medical conditions.4 These issues can be especially apparent with critically ill patients due to their changing hemodynamic status. Problems can be exacerbated if heterogeneous specimens such as capillary specimens from critically ill patients are used.4 Furthermore, hospital critical care sections have widely adopted glucose meters for Tight Glycemic Control (TGC), a clinical protocol for controlling patients’ glucose levels that was initially developed with glucose results obtained from conventional blood gas analyzers or laboratory analyzers.6
Identifying Our Use of Glucose Meters
Controversy surrounding the use of glucose meters in critical care sections continues to dominate the conversation among laboratorians who oversee POCT programs.7 The debate stems from the fact that there is no clear criteria for defining the “critically ill” and there is little agreement on what level of accuracy glucose meters must provide for safe medical interventions within this patient group. Those who favor the use of glucose meters argue that fast and actionable glucose values at the bedside outweigh the potential risks posed by inaccurate results, especially as glucose results are interpreted in the context of other clinical symptoms.7
At Tripler Army Medical Center (TAMC), a 194-bed federal tertiary care hospital located in Honolulu, Hawaii, we sought to understand how glucose meters are being used in our POCT program in order to align testing practices with the intended use of the meter (as indicated in the manufacturer’s product insert) and, where applicable, to offer alternative glucose testing methodologies that better meet the clinical needs of the patients in their sections.
Within TAMC, the POCT program manages 29 testing sites, 20 of which use one type of glucose meter.8 Of the approximately 600 POCT personnel (including physicians, physician assistants, and a wide range of nursing staff, etc), half use glucose meters in their respective clinical unit. In 2016, glucose testing (utilizing strip methodology) at TAMC totaled over 56,000 tests and accounted for approximately 60% of all POCT testing volume.
Surveying Staff on Meter Use
In order to understand how glucose meters were being used at TAMC, we surveyed physicians and nursing staff in all 20 POCT sections via in-person interviews or through email communications (see TABLE 1 for the list of survey questions). Our survey results indicated that glucose testing fell into three primary use categories:
At the time of our survey in early 2016, all 20 TAMC POCT sites primarily used a capillary specimen to measure glucose via the meters. However, when venous and arterial access lines were available, venous and/or arterial specimens also were used. A summary of glucose testing practices and types of specimens used across the various units within TAMC before and after our education and outreach effort is presented in TABLE 2. Highlighted in red are units that changed their testing practices, whereas units in blue made no change to their use of the meters. With this information in mind, we planned to expand our educational outreach approach with the ICU sections and the OB sections, as we believe the changes made by these sections will have the greatest positive impact on their patient populations.
Primary Indicated Uses and Risks
The clinicians in the adult intensive care unit (AICU), pediatric ICU (PICU), and progressive care unit (PCU) indicated in the survey that glucose meters were primarily used to monitor glucose in diabetic patients in order to begin parenteral nutrition and to adjust insulin dosing. While the ICUs at TAMC do not use TGC, they do follow a standard insulin drip protocol and patients are kept at a target glucose range of 100-150 mg/dL. Patients with diabetic ketoacidosis or hyperosmolar hyperglycemic syndrome (and on insulin drip) have a target glucose range of 150-200 mg/dL.
The second most common use of the glucose meter within TAMC at the time of our survey was to detect neonatal hypoglycemia in the neonatal ICU (NICU), labor and delivery (L&D), mother baby unit (MBU), and antepartum gynecology (APGYN); collectively referred to as the OB (obstetric) sections. TAMC follows protocols published by the American Academy of Pediatrics for the detection of hypoglycemia in newborns.9 Diagnostic criterion for neonatal hypoglycemia is a glucose level ≤ 40 mg/dL.
In both of the above areas, glucose meters were used to guide clinical decisions based on definitive glucose results. As mentioned earlier, this is a concern for laboratorians who recognize the wide total allowable error (TAE) standards used by the FDA for glucose meter approval. The majority of the glucose meters currently on the market have TAE of +15 mg/dL for results that are <75 mg/dL or ±20 percent for results that are >75 mg/dL.8 These error standards are significantly wider than traditional chemistry analyzers used in the core lab, which have TAE of +6 mg/dL or +5 percent.
Weighing the Controversy
Concerns over the accuracy of glucose meters have inspired several simulation studies that explore the potential error in insulin dosing in TGC protocols used in critical care sections.10-11 For example, using a simulation model, one study showed that frequency of errors for insulin dosing of 1-category (or one insulin dose higher or lower than intended) could be as high as 90% and errors for insulin dosing of 2-category (or two insulin doses higher or lower than intended) could be as high as 20% for meters that have 20% TAE. The authors concluded that at 20% TAE, which encompasses the majority of glucose meters in the market, insulin dosage errors could be large enough to lead to hypoglycemia.10
While the results from these simulation studies were disconcerting, other groups took a different approach to assess whether the wider error margins in glucose meters (as compared to core lab chemistry analyzers) present greater adverse patient outcomes. Using retrospective data from 854 point-of-care/core lab glucose test pairs collected within 10 minutes of each other, another study showed that 98 percent of the discrepancies were clinically insignificant. Moreover, no relationship was found between severity of illness and degree of discrepancy. The authors thus concluded that the “overall agreement of point-of-care glucose concentrations to the laboratory results is reasonable.”12 It is worth mentioning that the POCT glucose results used in this study were generated from the same glucose meter as is used at TAMC.
Despite the seemingly conflicting conclusions found in these studies, we proceeded by presenting the accuracy data of our meters along with their limitations (as published in the product insert) to the TAMC hospital leadership. They were receptive to some of our concerns and strongly encouraged that we individually work with the chief of each section to identify alternative testing platforms that could better meet the clinical needs of their section.
Finding an Alternative for Some
After presenting the limitations of the glucose meter to each ICU section chief, most sections emphasized that capillary specimens from their patients were rarely used since these patients often have established arterial and/or venous lines. However, ICU leadership acknowledged the potential for adverse events due to erroneous glucose results and agreed to train their physicians and nursing staff to exclusively use arterial or venous specimens on the glucose meters or send glucose tests to the core lab when patients are deemed critically ill.
When we presented the limitations of our glucose meters to the OB leadership, the chiefs of each section were greatly concerned that our glucose meter can only detect as low as 50 mg/dL. They also were in agreement that +15 mg/dL or ±20% errors posed a greater risk of hypoglycemia to their neonatal and pediatric patients. All four OB sections agreed to replace the glucose meter with a handheld blood analyzer for screening of neonatal hypoglycemia.13 Presenting a tighter error margin of +6 mg/dL or +10%, this device was deemed acceptable by the OB leadership. The original glucose meter is now limited to monitoring diabetic and gestational diabetic mothers in these sections.
After our successes in improving the safety of glucose POCT testing in the critical care and OB areas, we used the same approach with all TAMC POCT sites. Some sections were concerned about the glucose meter accuracy and agreed to relinquish POCT glucose testing altogether (eg, nuclear medicine and outpatient pediatric clinics). The endocrinology metabolic clinic switched from the previous glucose meter to the new handheld blood analyzer to support their insulin tolerance and mixed-meal tolerance tests.
Although some sites modified their practices, others felt satisfied with the limitations of the glucose meters. These sites included the emergency department, inpatient pediatrics, general surgery, ortho neuro vascular surgery, the post-anesthesia care unit, surgical admission center, medical oncology, medical telemetry, psychiatry, and veteran affairs psychiatry (see TABLE 2). The reasons these sections cited in support of the continuation of their glucose testing practices using capillary specimens were numerous, but came down to the fact that their patients are primarily non-critically ill diabetics who have poor venous access. The greater frequency of glucose testing in these patients also rendered venipuncture as too great a risk for anemia and infection. Physicians in these sections nevertheless were advised to use caution when interpreting glucose results from the glucose meter and were urged to confirm with laboratory testing whenever inconsistencies occurred between the glucose meters and clinical symptoms.
Our concern over the safe use of POCT glucose testing with critically ill patient populations inspired us to conduct interviews with our clinicians on how glucose meters were used to make medical decisions. After the survey results were collected and we presented our concerns to the TAMC hospital leadership, and subsequently, the section chiefs, we came to an alternative glucose testing platform that each section could ultimately adopt if deemed appropriate for their clinical needs. Our efforts resulted in half of TAMC POCT sites implementing changes in their glucose monitoring practices that include a combination of using venous specimens instead of capillary specimens or completely switching to other glucose testing methodologies (eg, the new handheld blood analyzer or core lab glucose testing). We expect these measures to help ensure safe testing and support sound medical decisions.
The views expressed in this manuscript are those of the authors and do not reflect the official policy or position of the Department of the Army, Department of Defense, or the US Government.
Uyen B. Chu, PhD, is an active duty medical allied sciences officer in the United States Army. She completed her graduate degree in pharmacology from the University of Wisconsin-Madison before being commissioned in 2015. CPT Chu is currently serving as the deputy chief of the core laboratory and the officer in charge of the point-of-care testing program at Tripler Army Medical Center in Honolulu, Hawaii.
Tiffany N. Heady, PhD, is an active duty medical allied sciences officer in the United States Army and is currently the deputy commander of the Fort Meade Forensic Toxicology Drug Testing Laboratory in Fort Meade, Maryland. She completed her clinical chemistry fellowship at the University of Virginia and previously served as the chief of the core laboratory at Tripler Army Medical Center in Honolulu, Hawaii.
Enter our Sweepstakes now for your chance to win the following prizes:
Just answer the following quick question for your chance to win:
Entries are limited to one entry per person in any active sweepstakes.
Thank you for your entry.