This article is more than 5 years old. Information may no longer be current.
Diabetes data downloads offer important keys to glycemic management
Click Here to Manage Email Alerts
AUSTIN, Texas — Downloaded data from diabetes devices can offer clinicians the missing clues needed to better understand why a patient has poor glycemic control; however, many clinicians are not reviewing these data due to time and cost concerns, Irl Hirsch, MD, said during a presentation at the American Association of Clinical Endocrinologists Annual Scientific and Clinical Congress.
“Downloading of the technology — from meters, pumps, continuous glucose monitors — needs to be part of the vital signs for every patient with diabetes,” Hirsch, professor of medicine at University of Washington School of Medicine, told Endocrine Today. “A download should be done on everyone.”
There are typically two reasons why many endocrinologists do not download data from these devices, Hirsch said during the presentation. Often, there is no infrastructure to do it properly in the office, leading to poor efficiency and the perception that there is not enough time to download data. Downloading data is also a relatively new trend, Hirsch said, and many endocrinologists were never taught how to do it.
“Resources on how to do this are scarce, it’s still relatively new and changing rapidly,” Hirsch said. “We do this … with the fellows every year, but I can tell you that 50% of the endocrine programs for fellows today don’t have a diabetes clinic that does this on a routine basis.”
In patients with diabetes, the HbA1c alone can be misleading, Hirsch said; hematologic conditions, physiologic state, medications, comorbidities and medical therapies can all result in an altered HbA1c in patients.
“We have become too [HbA1c]-centric, and it’s time to become more glucose-centric, because we now have the glucose data,” Hirsch said. “We’re not fighting the Hba1c. We’re fighting the glucose, both on the hyperglycemia side and the hypoglycemia side.”
A fuller picture
Downloaded data, Hirsch said, will show clinicians the glucose variability — time spent in range, time above range and time below range — that is often hiding behind an HbA1c value. With an HbA1c of 8%, for example, the mean glucose range is between 120 mg/dL and 210 mg/dL, Hirsch said, and downloaded data will show where a patient is from day to day.
“I want to know about the variability,” Hirsch said. “The greater the variability, the more I know that they’re not making their own insulin … and the greater the risk of bad hypoglycemia.”
The data also offer a peek into patient behaviors that, with changes, may easily correct the course for a serious problem, such as the timing of insulin doses, too much basal insulin or missing prandial injections, Hirsch said.
To download data in the office, clinicians need an infrastructure that allows for downloading from various brands of meters, insulin pumps and sensors. Examples include Clinipro (Numedics.com), Glooko/Diasend, Carelink and Tidepool, as well as native software. An even better solution, Hirsch said, is the immediate upload of diabetes data to the cloud, via platforms like Livongo, Accucheck Aviva Connect and Dexcom Clarity.
What to look for
Hirsch listed several data points clinicians should look for when downloading from diabetes devices. For self-monitored blood glucose, a simple average of blood glucose numbers is insufficient; the ideal blood glucose should be two times the standard deviation less than the mean blood glucose, provided the patients’ mean blood glucose is between 120 mg/dL and 180 mg/dL. For CGM, ideal blood glucose should be three times the standard deviation less than the mean, he said.
When downloading insulin pump therapy data, clinicians should look at basic insulin statistics, including the percentage of basal insulin used and the percentages of overrides for the bolus calculator.
“Here’s the rule of thumb,” Hirsch said. “You want to see, ideally, 20% to 25% override or under-ride, because the bolus calculator always sees the glucose as stable. It doesn’t know when there is going to be exercise, it doesn’t know when there is a menstrual period, it doesn’t know when there is stress, and the big thing it doesn’t know is if the glucose level is already going up or down. I want to see that.”
For CGM, clinicians should look at basic blood glucose statistics as well as looking at overall patterns and daily decision making, such as the insulin-to-carbohydrate ratio, to best understand how the patient thinks through each challenge, Hirsch said.
“This is the thing: When I look at a download, the download isn’t any good if I can’t figure out how the patient is thinking,” Hirsch said. “I want to get in the mind of that patient.”
Common patient mistakes
Patients make four common mistakes when managing their diabetes that downloaded data can reveal to a clinician, Hirsch said. These include over-calibration or under-calibration, not looking at the sensor often enough, overreacting to alarms, resulting in “insulin stacking,” and not using SMBG to make decisions.
“Sometimes the Dexcom says you’re low … and you’re not,” Hirsch said. “When there is a question, you should follow up with a finger stick.”
“One needs to see both cumulative and detailed data to understand how to assist their patients,” Hirsch said. “What I do in my clinic may be different from what you want to do. There are a lot of different ways to do that.” – by Regina Schaffer
Reference:
Hirsch I. Applications of Technology to Your Patients. Presented at: AACE Annual Scientific and Clinical Congress; May 3-7, 2017; Austin, Texas.
Disclosures: Hirsch reports receiving consulting fees from Abbott Diabetes Care, Intarcia, Roche Diagnostics and Valeritas.