Report from Diabetes Technology Meeting
The third annual Diabetes Technology Meeting was held in San Francisco from November 6-8, 2003. In attendance were hundreds of scientists, engineers, and doctors from industry, academia, and clinical practice. During the two-and-a-half days of sessions, speakers shared a variety of topics, including continuous glucose monitoring, other metabolic monitoring, glucose meter and new sensor accuracy, insulin delivery, and various types of computer technologies. While most of the presentations were highly detailed and most suited for engineers and scientists involved in developing new products, the following sessions are worth noting:
- Mary Kimberly, Ph.D., from the Centers for Disease Control and Prevention, gave a session entitled Glucose Monitor Variability Project. The US government funded study, conducted with assistance from five blood glucose manufacturers, looked at the variability of results obtained in commonly used meters. The data presented included five meters, one of which had substantially more variability than the other four meters. Dr. Kimberly was unable to provide information on which meter this was, because the CDC agreed to remain blinded to the meters used during the study.
- In Limitations of Statistics Measures of Error in Assessing the Accuracy of Glucose Sensors, Craig Kollman, Ph.D., from the Jaeb Center for Health Research in Tampa, Florida, examined whether currently used statistics offer correct guidance on whether or not errors in sensors would lead to erroneous treatment decisions. From his presentation it was clear that the Clarke Error Grid, commonly used for blood glucose monitors, can make an inaccurate sensor appear acceptable and that other metrics must be used to evaluate sensor performance, particularly in the hypoglycemic range.
- Dr. Peter Chase, from the Barbara Davis Center for Childhood Diabetes in Denver, Colorado, gave a session entitled Sensor Accuracy: A Limiting Factor? Presenting results from a study of the Medtronic MiniMed CGMS and the Cygnus Glucowatch Biographer, Dr. Chase emphasized the value of these devices in predicting imminent hypoglycemia rather than presenting blood glucose values that can be used to make dosing decisions. (In a pre-conference workshop, Dr. Chase noted that a recent study of his had shown a three-fold increase in detection of hypoglycemia at night in children using the Glucowatch compared with controls.) Finally, Dr. Chase noted that the FDA has only approved home blood glucose monitors for use in insulin dosing decisions, and he said that was a wise position.
- Ben Feldman, Ph.D., from TheraSense presented The FreeStyle Navigator Continuous Glucose Monitor: Complete Results from a Home-Use Trial. The Navigator required slightly more than a one hour warm up period, with finger stick calibrations at one, three, and 24 hours. No additional calibration was needed for the rest of the 72-hour period during with Navigator can be worn. Navigator consists of a sensor component, worn on the arm (for example), and a receiver/display unit that is worn on the belt (or carried in a purse). The display unit shows current blood glucose reading, as well as a trend arrow. Low and high alarms are user configurable.
- Dr. Steve Edelman’s presentation entitled Continuous Glucose Sensor Technology was about a novel, implanted, continuous glucose sensor being developed by Dexcom, a company in San Diego. The Dexcom sensor is about the size of a AA battery, is implanted, and provides glucose readings every five minutes. In clinical trials, patients using the correct implantation technique (minor surgery, but requires a gentle touch) experienced significant clinical benefits, including detecting 99% of glucose excursions below 100 mg/dl, reducing severe lows (40-55 mg/dl) by 47%, reducing highs (250-400 mg/dl) by 25%, and increasing time in the 80-140 mg/dl range by 88%. An updated version of the Dexcom sensor will be reduced in size and be implantable in simpler outpatient procedure.
- In the final session of the conference, participants were asked to vote, using a hand-held electronic voting device, on several questions related to diabetes and technology. When asked when an artificial pancreas would be available for widespread clinical use, the results of the voting was:
♦2 years – 1%
♦4 years – 12%
♦6 years – 25%
♦10 years – 45%
♦Never – 17%
Panel members overseeing the voting session felt more optimistic, generally feeling that some sort of continuous sensor control of a pump was about six years away from clinical availability.