Posted by Sten Westgard, MS
Here's a scenario for you:
Dec. 21, 2010: Patient admitted
Jan. 1, 2011: Physician orders patient to be treated for high blood sugar.
Jan. 2, 11 pm: Physician orders insulin drip to be increased.
Jan. 3: Beside tests show elevated blood sugar; lab tests show very low level
Jan. 3, 4 am: [You become aware of the testing discrepancy.]
What would you do?
Jan 3, 6:06 am: Physician orders second dose of insulin in addition to drip.
Jan 3, 7:46 am: Beside tests still remain high.
Jan 3, 9:20 am: Insulin drip put on hold.
Jan 3, 10:32 am: Staff discovers that test strips malfunctioned; patient now treated for low blood sugar.
Jan 3, [later that] morning: Patient discovered unresponsive due to "prolonged hypoglycemia that met brain death criteria."
This tragedy took place at Advanced Intensive Care Unit in Allentown, Pennsylvania. The patient was being monitored by doctors from Lehigh Valley Hospital-Cedar Crest.
The full story can be found here: Report: Lehigh Valley Hospital gave fatal dose
"Doctors at Lehigh Valley Hospital-Cedar Crest, using faulty blood sugar testing strips, this year administered an apparent fatal double dose of insulin to a patient with extremely low blood sugar levels, a critical state Department of Health report says.
"A nurse monitoring the patient's status from a remote site knew that bedside blood sugar testing strips differed significantly from more reliable laboratory tests of the patient's blood, but failed to warn doctors about the discrepancy, it says.
"Doctors at the hospital ordered insulin for the patient to lower the level of blood sugar, which is the body's fuel, since too much blood sugar can lead to life-threatening conditions. But the lab test indicated that the patient's blood sugar was extremely depressed after hours of receiving insulin, so continuing to administer it was a critical error.
"Additionally, the state report says there was "no documented evidence" that a doctor, a second staffer who was electronically monitoring the patient that morning from the remote site, took the routine step of providing notes on the disparate test results to the attending physician that day, Jan. 3."
Report: Lehigh Valley Hospital gave fatal dose
The point here isn't to condemn a particular glucose meter, or even to blame a nurse or a doctor. But certainly we should point out that glucose meters are waived devices, cleared by the FDA only for screening and monitoring purposes. They are not intended for use in a critical care setting for the adjustment of dosage of Insulin. To make clinical decisions based on those test results is an "off label" use of the device, which carries with it serious additional demands for method validation.
If you use a waived device as it is labeled, basically all you have to do is follow the manufacturer's instructions for use. If you go "off label" you must treat the device as a method of high complexity and perform a full suite of method validation studies, including precision, accuracy, interference, recovery, etc. It's not a trivial thing to do, and while I can't confirm this, I am fairly confident that those glucose meters were not validated in that way.
This is a tragedy, an accident that shouldn't have happened, and a sober reminder that POC devices are not completely benign. They have limited uses and when you push them beyond those limitations, they can be dangerous.
Since labs have responsibility to oversee use of glucose meters, perhaps we should have access to POC results (entered onto patient records and viewable on lab software). Some of our blood gas machines are so linked but not the hand-held glc meters. I'm not sure if we'd have the time available to check POC-lab results anyway. Is there a software "flag"system that can highlight discrepancies? For us, it means integrating several software programs. Not straightforward.
Posted by: Dr Christopher M. Pitt | August 21, 2011 at 08:18 AM