Since my first detailed report comparing the various Continuous Glucose Monitoring Systems (CGMs) back in 2014, a lot has happened. And not much has changed.
The systems have improved in terms of accuracy, features and ease of use, but the main players remain the same (Medtronic and Dexcom). Access via insurance coverage and professional loaner systems has grown exponentially, yet less than 20% of those eligible for CGM are currently using them. In many cases, insurers make the process of receiving coverage onerous and needlessly complex. This doesn’t even touch on Medicare, which continues to sit idly by with its head up its proverbial butt while older Americans suffer needlessly from dangerous glucose swings. New and improved software programs (plus a brilliant new book called “Practical CGM”) provide guidance on how to interpret/analyze CGM reports, yet few patients bother to look at their own data, and very few healthcare providers have the expertise to convert the reports into useful therapeutic insight to help guide their patients.
So let’s get down to business. How do the latest Medtronic and Dexcom CGM systems compare?
Dexcom’s latest and greatest, the G5, features a transmitter that sends data directly to either a handheld receiver or a mobile phone. Dexcom’s G5 Mobile App displays data on the phone and generates the various alerts; Dexcom’s Clarity App generates reports for retrospective analysis. G5, as well as G4 Platinum, utilizes Dexcom’s up-to-date 505 algorithm for translating subcutaneous electrical impulses into glucose values. Why call it 505? My best guess is that Medtronic copyrighted every other number below 1000. (for some reason, they skipped 505 when naming their various pumps)
Medtronic’s latest CGM features their new-generation Enlite sensor, which has undergone a series of manufacturing improvements in order to provide better adhesive, less trauma upon insertion, and better/more consistent enzyme chemistry. The pump that receives a signal from the Enlite sensor has the ability to react in an automatic fashion. The 530G pump (in the United States) ceases basal insulin delivery when glucose levels drop below a user-defined threshold. Medtronic has also introduced a cell phone app which can display sensor data in real-time with the aid of a Bluetooth transmitter called “Connect”. The 640G pump (in Europe) takes things a step further by temporarily turning off basal insulin delivery when hypoglycemia is anticipated — thereby preventing the low rather than waiting until it has already occurred.
So, which CGM system is best?
Being the fair and impartial (ahem) guy that I am, I spent the past eight weeks putting each company’s latest system to a head-to-head test. Keeping with common practice, I wore each sensor for as long as it was functioning and performing reasonably well. Other than that, I followed all of the rules and regs that each company provides in its Instructions for Use. Simultaneous calibrations were performed two to four times daily prior to meals, when glucose levels were in a relatively steady state. The sensors were stored properly and placed in appropriate/adjacent sites. The sites were prepared appropriately and rotated consistently. Extra adhesive was applied over the sensors at the first sign of loosening. I also avoided using acetaminophen (the pain reliever found in Tylenol products), as both sensors can produce falsely elevated readings when exposed to acetaminophen.
Below are my findings.
Medronic 530G with Enlite
Sensor Insertion & Comfort
Both systems use very thin introducer needles and automated insertion devices. However, insertion of Dexcom is a simpler process requiring fewer steps and less equipment. Dexcom’s built-in adhesive eliminates the need for extra tape for most people
Both Sensors function well past their “approved” life-cycle (7days for Dexcom and 6 days for Medtronic) with a bit of ingenuity (and some extra tape). Based on personal experience and that of our clients, I’ve found Medtronic Enlite sensors last an average of 8-10 days; Dexcom sensors last an average of 12-14. However, that doesn’t mean the sensors perform well at the later stages of use. Towards the end of this report, you’ll find specific data showing how Medtronic’s sensor accuracy falls significantly after 6 days of use, while Dexcom’s performance is fairly stable through day 14 (sometimes beyond).
Both systems offer High & Low alerts as well as rate of change alerts, albeit Medtronic’s rate of change alerts are much more customizable. Only Medtronic offers predictive alerts which provide an earlier warning of potential hypoglycemia (and hyperglycemia). Medtronic also allows th euser to vary the alert settings by time of day – a potential value for those wanting more aggressive or fewer alerts while working or sleeping (or sleeping at work).
Both systems can beep and/or vibrate to alert the user of a potential problem and both have adjustable settings, but the vibrate mechanism on Dexcom’s receiver is considerably stronger, and it can beep much louder (when desired). The Dexcom mobile app has the added benefit of allowing the user to choose from nearly two dozen audible patterns for each type of alert – making it nice to hear somthing other than a “BEEP” for an alert.
As an independent company, Dexcom has done a decent job of “integrating” with a number of devices, including insulin pumps (Animas Vibe, Tandem t:slim G4), cell phones, and smart watches (Apple Watch and Pebble). However, in all cases, integration only replaces the traditional Dexcom receiver with the display of data on these other devices. Medtronic takes things another important step: automated therapy adjustments. Yes, Medtronic’s sensor data displays on their insulin pumps. But, more importantly, the 530G’s “threshold suspend” feature automatically stops the pump’s basal insulin delivery when the glucose falls below a specified threshold (set by the user). The 640G’s “Smartguard” feature temporarily suspends basal delivery when hypoglycemia is approaching.
There is a significant difference in transmitters. Medtronic’s radio transmitter contains a memory chip that stores up to 40 minutes of data and submits i to the receiver in case the receiver was out of transmission range. Dexcom has no such feature. But, Dexcom has much less need for such a feature. The Medtronic signal only travels a few feet and is frequently lost even when in close proximity to the sensor. Dexcom’s signal (radio or bluetooth) tends to pick up pretty well as long as the receiver (or phone) is within 20 feet of the transmitter. Medtronic’s transmitter must be separated from the sensor and charged at least every six days. Dexcom’s transmitter never requires charging. The G4 transmitter tends to last 9-12 months and the G5 transmitter lasts 3 months.
All Dexcom CGM displays, including those in the Dexcom receiver, phone app and linked pumps, are bright/high contrast and full-color. However, the G5 Mobile app has a tendency to “lock up ” from time to time, requiring the user to shut down other programs in order to proceed with data entry. The display on the Medtronic 530G is relatively small, monochromatic (black and white), and low-contrast. The 640G screen has similar quality to Dexcom. However, many people who use an insulin pump prefer to have a receiver/display that can be carried separately for ease of access and detection of alarms. Medtronic’s cell phone app display does offer excellent size and contrast but requires one to carry a “connect” fob in order to send the data from the transmitter to the phone app.
One of the new benefits of CGM use is the ability to have caregivers and loved ones track the data in real-time on their smartphones and be alerted of potential problems. Dexcom pioneered this feature with its SHARE component (available on the G5 and later-model G4 systems). Data may be shared with several “followers” and the alerts are fully customization. Medtronic’s mobile app allows followers to be alerted in the event of a crisis, but data cannot be viewed in real-time.
Dexcom’s download software has some nice features, such as the ability to segment data by day of the week and customize analysis dates. However, Dexcom Studio software does not work on Mac systems or integrate with data from an y pumps or meters. Dexcom also took a major step backwards with its Clarity app/software (necessary for generating reports from the G5 system). Clarity eliminates many of the reporting options and customization features that were popular with Studio.
Medtronic’s Carelink software is web-based, works on virtually all operating systems, is easy to share with healthcare providers, integrates sensor data with pump and meter data, and generates reports that do a super job of revealing post-meal patterns.
As the saying goes, “If something can go wrong, it probably will”. there’s something to be said for a system that doesn’t leave a lot to chance.
Dexcom continues to be very simple to use. Other than learning the sensor insertion (which requires a bit of dexterity) and setting up the mobile app, everything about it is simple. There are very few steps, and the steps are all very logical. You put the sensor on, attach the transmitter, calibrate after 2 hours and it works.
Medtronic’s CGM has a lot of moving parts: finicky transmitters, a multitude of error messages, a lengthy/detailed sensor insertion process and a variety of issues that can impact sensor function and accuracy.
Case-in-point: I do a lot of traveling, so I decided to count up how many items I had to take with me to maintain my Dexcom G5 compared to the Medtronic Enlite (both using the respective mobile apps).
I had to tote 10 items to manage my Medtronic sensor changes, but only 3 for Dexcom (see list and photo below).
Connect device (for transmitting data from pump to phone app)
Charger for Connect device
Dexcom sensor maintenance supplies:
Sensor (with built-in inserter)
Alcohol (necessary to exfoliate skin)
** All the stuff needed to use/change/charge the Medtronic sensor w/cell-phone app (left) and Dexcom (right). The cell phone was needed for both systems.
Here is the most important piece. Most people are willing to put up with some minor inconveniences if the system generates reliable data on a consistent basis.
When reviewing this information, keep in mind that sensor accuracy is best measured in relation to LAB values, not blood glucose meter values. Fingerstick values taken with a home-use meter are intrinsically less accurate than lab values, and this contributes to a greater degree of inaccuracy by the CGM. But, given that not many of us have a fully equipped lab to tote around all day, the best we can do is evaluate the systems based on how they compared to our handy-dandy meters. In my case, I stuck with the meters that are considered to be top-of-the-line for accuracy throughout the comparisons.
Accuracy is best measured by looking at MARD – Mean Absolute RElative Difference. This represents the difference between fingerstick meter values and sensor glucose values, so a lower MARD is better. A calibration reading of 150mg/dl and a sensor value of 120mg/dl represents a MARD of 20% (a 30-point difference on a value of 150).
The bottom line is this: Medtronic still isn’t up to par with Dexcom when it comes to accuracy. But the gap is narrowing.
Note that these MARDs are higher than what is seen when using LAB values for calibration purposes. Looking at rates of accuracy, Medtronic’s sensor was closer to the fingerstick calibration 43% of the time; Dexcom was closer 57% of the time.
There were some interesting accuracy patterns related to day of Sensor use. Unlike earlier versions of both systems, accuracy on “day 1” was not all that bad (19.0% MARD for Medtronic and 14.8% for Dexcom). There was a marked difference in accuracy when sensors were used beyond their approved days of use. While Dexcom maintained its accuracy from week one to week two, Medtronic’s accuracy took a major dive when the sensor was used beyond its “approved” six days.
Dexcom MARDMedtronic MARD
Days 1-7: 13.4% Days 1-6: 16.6%
Days 8-14: 13.2% Days 7-12: 25.6%
Both systems continue to underestimate glucose levels more often than overestimating…which is good for those trying their best to avoid hypoglycemia, but not ideal for those who are striving for the tightest control possible.
Below Calibration: 62% 64%
Above Calibration: 37% 35%
Exact Match: 1% 1%
Clark Error Analysis:
Dexcom: (79% zone A – safest)
(100% zone A + B)
(0% outside zones A & B)
Medtronic: (63% zone A – safest)
(98% zone A+B)
(2% outside Zone A&B)
Take Home Messages…..
Always remember, the comparisons above are based on an “N of 1” (me only). They do not represent the findings from a multi-subject, randomized, controlled study. However, I can say that most of what I experienced reflects what I have seen among countless clients/patients who have worn the various CGM systems.
A few observations worth noting:
Dexcom continues to provide a more accurate, user-friendly system, but Medtronic is a step ahead when it comes to data analysis/reporting and integration with automated insulin delivery. If this was a boxing match, it would probably be a split decision in favor of Dexcom.
CGM accuracy continues to improve, but both systems still underestimate glucose levels overall.
Dexcom sensors appears to maintain their accuracy when used beyond 7-days. Medtronic sensors appear to fall off a cliff accuracy-wise when used beyond six days. Personally, I’m very excited about the “next steps” from each company. Envision a CGM that does not require fingerstick calibration, is simple to use, and integrates easily into daily diabetes management decisions. I guess that means people won’t need my services as much, but that’s OK. I can always pursue my dream of becoming a traffic engineer.