The differential diagnosis for toxic-metabolic encephalopathy is extensive.1 Hypoglycemia is a common culprit and is frequently seen in hospitalized patients, particularly diabetics. The lower limit of a normal fasting serum glucose value is 70 mg/dL. Symptoms are typically not present until serum glucose is less than 50 to 55 mg/dL and are classified into 2 categories.2 Neurogenic symptoms, which are mediated by adrenergic and cholinergic outflow, include tremor, palpitations, anxiety, sweating, hunger, paresthesias, hypertension, and tachycardia. Neuroglycopenic symptoms, which are caused by the toxic effect of hypoglycemia on neuronal cell bodies, include cognitive impairment, behavioral changes, focal neurologic deficits, seizures, and coma. Most cases are reversible if recognized and treated early. However, as in this case report, sustained hypoglycemia can result in permanent neurologic injury. Care providers frequently rely on point-of-care glucose meters to check glucose levels, but there are certain instances when point-of-care glucose meters can be inaccurate.
The IRB was contacted for permission to publish this report, but they do not review single case reports.
A 65 year-old woman with end-stage renal disease secondary to hypertensive nephropathy was admitted for deceased donor renal operation. Her medical history was significant for coronary artery disease not amenable to percutaneous coronary intervention, chronic anemia secondary to renal disease, and a history of cerebrovascular accident in 2005 with minimal residual weakness of her right upper extremity. She had been receiving continuous ambulatory peritoneal dialysis. She was previously receiving hemodialysis but was converted to continuous ambulatory peritoneal dialysis because of multiple graft failures and graft thromboses. There was no history of diabetes mellitus. Her medications included amlodipine, losartan, prazosin, clopidogrel, simvastatin, aspirin, epoetin, ranitidine, docusate, and calcium supplements. She received peritoneal dialysis the day before surgery.
On the evening of her surgery, at 2100 hours, she was awake, alert, and neurologically intact. She was taken to the operating room and underwent insertion of an indwelling radial arterial catheter, induction of anesthesia and tracheal intubation, and placement of a right internal jugular central venous catheter. The operation proceeded smoothly without major intraoperative events. At the conclusion of surgery, the trachea was extubated, and the patient arrived in the postanesthesia care unit (PACU) at approximately 0230 hours. She was awake and alert in the PACU and received small doses of IV fentanyl and hydromorphone for pain. A finger stick blood glucose checked with a Roche ACCU-CHEK Inform 1 point-of-care glucose meter revealed a value of 340 mg/dL. A continuous IV insulin infusion was started at 0415, at 7 units per hour. She was subsequently discharged from the PACU to a floor bed in stable condition.
A few hours later on postoperative day #1 (POD #1), she was seen by the primary service and was awake, alert, and oriented. In the afternoon of POD #1, the primary service was called by the nurse for increased somnolence. The patient’s IV hydromorphone patient-controlled analgesia pump was discontinued, and IV naloxone was administered without improvement in mental status. She was transferred to the intensive care unit. The patient had not received any tacrolimus, and all other antirejection medications were held. Given her increased blood urea nitrogen, hemodialysis was initiated. The neurology service was consulted. On their evaluation of the patient on POD #1 at 1900, they found her to be unresponsive to painful stimuli with a left upgoing toe and slight hyperreflexia on the left but without other asymmetric findings. Per their recommendations, the patient’s trachea was intubated for airway protection. A computed tomography scan of the head did not show any acute process. A magnetic resonance image of the brain on POD #2 at 0300 revealed “extensive restricted diffusion involving the subcortical and periventricular white matter of cerebral hemispheres including internal capsule, corticospinal tracts, and splenium of the corpus callosum.” An electroencephalogram revealed diffuse slowing without evidence of epileptiform activity. All of these findings were thought to be consistent with toxic-metabolic encephalopathy. At 0600 on POD #2, the patient was noted to have a serum glucose of 43 mg/dL, and the insulin infusion was discontinued, followed by initiation of an IV 40% dextrose infusion for management of hypoglycemia.
From POD #0 to POD #2, a standard insulin protocol had been titrated using point-of-care glucose meter testing. During this period, point-of-care glucose meter values were 150 to 200 mg/dL while laboratory glucose tests revealed values of 20 to 40 mg/dL (Table 1). Care providers primarily used point-of-care glucose values to titrate the insulin infusion, but repeated laboratory glucose readings were obtained because of the inability to reconcile discrepant results. Multiple different glucose meters were used during this period as well as strips from various batches. She received 24 hours of IV regular insulin totaling 115.9 units. Once the discrepancy was recognized, the insulin infusion was discontinued, and a continuous IV 40% dextrose infusion was started. She suffered severe hypoglycemic brain injury. The patient never recovered neurologic function and underwent subsequent tracheostomy and gastrostomy. She had a prolonged hospital stay complicated by multiple medical conditions, including intermittent sepsis from infected decubitus ulcers, renal graft failure, and pneumonia. She was eventually discharged to a care facility with plans for withdrawal of care.
Point-of-care glucose meters have revolutionized diabetic management. Their readings are often presumed very accurate, but the “gold standard” continues to be the laboratory hexokinase test, known as the serum glucose test. If accurate, point-of-care glucose meters should correlate to within 15% of the serum glucose test value. If a discrepancy occurs, an investigation should be made. There are different factors that can lead to point-of-care glucose meter errors; these include the presence of different sugars as in this case (discussed below), hematocrit <34%, hyperbilirubinemia, hyperuricemia, increased partial pressure of oxygen in the sample, as well as the presence of ascorbic acid or acetaminophen.3
Point-of-care glucose meters use different enzymes and cofactors that react with the glucose in a capillary blood sample. The reaction of the sugar with the cofactor produces an electrical current proportional to the concentration of sugar in the sample. Common enzymes include glucose dehydrogenase (GDH) and glucose oxidase. Common cofactors include pyrroloquinolinequinone (PQQ) and nicotine adenine dinucleotide. The ACCU-CHEK point-of-care glucose meter used in this case utilized a GDH-PQQ system. GDH systems have been replacing glucose oxidase systems because of their accuracy in the setting of differing oxygen partial pressures, whereas the glucose oxidase systems are less reliable when the oxygen partial pressure is variable. However, glucose oxidase only reacts with glucose, whereas GDH-PQQ reacts with glucose as well as other sugars including maltose (a metabolite of icodextrin), xylose, and galactose.4
Icodextrin is a starch-derived, water-soluble glucose polymer that acts as a colloidal osmotic agent in peritoneal dialysis solution.5 It is a component of extraneal peritoneal dialysis solution, which is produced by Baxter. Its advantage is that it promotes sustained ultrafiltration across the peritoneum for approximately 14 hours when compared with older dextrose dialysis solutions. However, 20% to 40% of the icodextrin is absorbed through the lymphatics. Once absorbed, icodextrin can persist in the blood for up to 14 days after treatment. Once absorbed into the blood, serum amylase converts icodextrin to maltose, which reacts with GDH-PQQ test strips in point-of-care glucose meters. Multiple studies have shown that the presence of icodextrin in the serum falsely increases point-of-care glucose meter readings by approximately 60 ± 30 mg/dL.
Aside from peritoneal dialysis solution, other products containing icodextrin include immunoglobulin preparations (examples include Octagam 5%, Gamimune N 5%, WinRho SDF liquid, Vaccinia Immune Globulin, and HepaGam B), disease-modifying antirheumatoid drugs such as Orencia (abatacept), chemotherapeutics such as Bexxar (tositumomab), and Adept adhesiolysis solution (used in abdominal surgery to lyse adhesions). Because of the diversity of products containing icodextrin, it is important to recognize which patients are at risk for hypoglycemic injury due to falsely increased point-of-care glucose meter readings. Staff education is an important aspect of patient safety. If available, an alternative glucose meter should be used that does not use a GDH-PQQ enzyme system. If an alternative glucose meter cannot be obtained, the patient should have only laboratory hexokinase readings to determine his/her serum glucose levels.
Although similar cases have been described, there are multiple reasons why it is imperative to reemphasize this issue in our changing medical environment.6 First, a seemingly small oversight can result in potentially devastating consequences for the patient, thus requiring vigilance and increased awareness. Even at a major academic medical center such as ours, many members of the Department of Nephrology and our own Department of Anesthesiology were unaware of glucose meter interferences such as this one. Because our center uses high concentration glucose-containing peritoneal dialysis solutions, and not icodextrin-containing solutions, many care providers, both physicians and nurses, trainees and attendees, were unaware of the interaction. Second, because the nursing staff typically uses glucose meters to measure patients’ glucose levels on the inpatient units, we as physicians are less familiar with the shortcomings of these devices but no less responsible for the care of our patients. Although members of the Department of Anesthesiology at our center receive in-service training on these devices, issues related to uncommon interactions are often not emphasized. In addition to being cognizant of all possible interactions ourselves, we need to share this interaction with other physicians who may not need to undergo in-service training for point-of-care glucose meters. Finally, in the current era of increasing quality control measures such as the Surgical Care Improvement Project, there is greater pressure to standardize care. There is also increased scrutiny, with potential financial implications, for deviating from said standardizations. This may lead to an increased reliance on point-of-care glucose meters in the perioperative setting, as well as protocol-directed therapy, without constant senior clinician input. To avoid harm to our patients while optimizing care, all practitioners need to be aware of the potential errors in our medical devices. Furthermore, as electronic medical records become more universally available, flagging of medications not to be used (in this case containing icodextrin) can occur.
In the wake of this tragedy, multiple changes have been made at our institution to promote awareness and improve safety. Lectures covering this topic were given to the residents of multiple departments, and in-service programs were given to the nursing staff in the intensive care units. Ideally, these will be repeated especially for new employees. Bulletin boards and educational newsletters highlighted the potential for inaccuracy of point-of-care testing devices. With these ongoing educational interventions, both the responsible physicians and nurses should be aware of important drug interactions and the standard-of-care testing. In addition, the hospital has purchased new point-of-care glucose meters that use a glucose oxidase system so that the results are not affected by maltose. To improve safety until these machines were available, changes were made to currently used glucose meters; a pop-up screen appeared with every use of the machine that warns of the interaction with maltose-containing solutions.