Ever feel like you’re drowning in numbers? Diabetes care is chock full of numerical data, all designed (presumably) to help a person manage his blood glucose levels, prevent complications, and live life to the fullest. But judging by the problems diabetes still causes, perhaps there needs to be a better way.
And perhaps there is one: continuous glucose monitoring (CGM), which lets people know their blood glucose level practically all of the time, not just when they take a fingerstick reading. The trouble is, CGM systems produce even more numbers, charts, and graphs that can seem like an endless ocean of data unless you know what to do with all that information. And that’s a shame, because even with their shortcomings, CGM systems can be an extremely valuable resource for guiding daily decision making. They can generate information that, if you know how to use it, will allow you to fine-tune your (or your child’s) diabetes management program and stay out of harm’s way.
There are now three CGM systems available that provide real-time blood glucose estimations and warning systems to guard against highs and lows. The two that have been around the longest are the DexCom Seven and the Medtronic Guardian REAL-Time System (which can be used alone or as an integrated part of the MiniMed Paradigm 522 or 722 insulin pump). A third system, the FreeStyle Navigator, from Abbott Diabetes Care, was approved by the Food and Drug Administration in March of 2008. Only the Medtronic Guardian System has received FDA approval for use by children, but any system can be used by a child if prescribed by a physician.
All three systems use a thin metallic sensor inserted just below the skin to detect glucose in the fluid between the cells (called the interstitial fluid) in the fatty tissue just under the skin. Spring-loaded insertion devices make sensor insertion quick and relatively painless, and the sensors do not tend to cause skin irritation the way pump infusion sets can.
The information from the sensor is transmitted by radio signal to a receiver/monitor, which displays an estimate of the current glucose concentration. While the transmitter and receiver/monitor are “durable,” meaning they should last a long time, the life expectancy for the glucose sensors ranges from 3 days to about 14 days.
The DexCom and Medtronic receiver/monitors provide an updated glucose reading every five minutes and can be set to sound an alarm if the glucose level goes above or below a target range set by the user. The Guardian system can also warn the user of glucose levels that are nearing the upper or lower end of the target range. To get the most accurate results, both systems require the user to take fingerstick blood glucose readings at least twice a day and enter those readings into the system. Both systems are generally accurate to within 15% to 20% of fingerstick readings. Both provide line graphs showing glucose trends over the past several hours, and both have compatible software that permits detailed analysis of glucose trends over longer time intervals. (For a fuller look at CGM basics, see the article “Continuous Glucose Monitoring” online or in the March/April 2007 issue of Diabetes Self-Management.)
Go back to the paragraph above. See that “15% to 20%” figure? That means the data generated by CGM are usually only a rough estimate — too rough, in fact, to use for making most insulin-dosing decisions. That’s not to say that CGM users never use their data to determine insulin doses, but you need a sound knowledge of the nuances of the system before doing so.
For example, I find that my CGM data tend to be unreliable during the first 12–24 hours after I insert a new sensor. It often takes time for a sensor to become acclimated to being below the skin, and for fingerstick readings to allow the sensor to “hone in” properly. The CGM data seem to become more precise and reliable the longer a sensor has been in place. Until I start to see a couple of sensor values that are very close to simultaneous fingerstick readings, I just don’t trust the CGM system for immediate decision making.
The accuracy of CGM values must also be questioned if there have been frequent or prolonged data gaps on the receiver/monitor. Signal interference or inconsistent signals may cause erroneous and unreliable information.
There is also an issue known as “lag effect” that must be considered. CGM systems take multiple glucose readings over a five-minute period and average them to generate a single glucose level. This means that, on average, the level you read on your display is 2½ minutes old. And because CGM systems measure the glucose level of interstitial fluid (not blood), there is an inherent 5- to 10-minute delay between fingerstick values and what appears on the monitor’s display. Therefore, if blood glucose levels are rising quickly, the CGM level can be significantly lower than a blood glucose reading done with a fingerstick at the same time. Conversely, if the blood glucose level is dropping quickly, the CGM level can be significantly higher.
Because of this, I only recommend using CGM numbers for immediate insulin dosing if the following are true:
- The last couple of fingerstick values have matched the sensor values closely (within 10%). For example, if the fingerstick reading is 120 mg/dl, the CGM value should be between 108 and 132 mg/dl.
- Your current glucose level is not rising or dropping rapidly. (A look at the graph on your display screen can tell you if this is the case.)
- The sensor has not generated any error messages or significant data gaps for at least the past hour.
Whether or not you use your CGM data for insulin-dosing purposes, the simple act of looking at your monitor has its rewards. Research has shown that checking the monitor 10–20 times per day and wearing the system most of the time (rather than intermittently) tends to produce improvements in HbA1c level (a measure of average blood glucose control) and a reduction in episodes of hypoglycemia (low blood glucose).
Perhaps more valuable than the immediate glucose value is the direction it is headed. If you checked your blood glucose at bedtime and it was 95 mg/dl, you would react differently if you knew it was on the rise rather than if you knew it was dropping. Similarly, seeing a level of 188 mg/dl before exercise would mean more to you if you knew whether your level was rising, falling, or remaining steady.
The Medtronic CGM systems display “up” or “down” arrows to denote both the direction and the magnitude of glucose changes over the past 20 minutes. The DexCom Seven system displays a one-hour trend graph with similar information. Use this information to your advantage! The ability to forecast where your blood glucose will be over the next hour or two can help keep you from straying too far out of your desirable blood glucose range.
For instance, if you see that your glucose level is close to or slightly above normal but dropping quickly, you can either have a snack immediately or be prepared to have a snack soon to prevent hypoglycemia. Similarly, if you see that your glucose level is near or slightly below normal and dropping gradually, you know to have a snack now and to check your monitor again soon.
With a rapidly rising blood glucose level, you may choose to offset the rise with rapid-acting insulin (taking into account any active, unused insulin from previous doses) and check the monitor often to make sure your level doesn’t end up dropping too much. If your blood glucose level continues rising, you may decide to check your urine or blood for ketones. It is this type of frequent fine-tuning that can keep you within your target blood glucose range more of the time.
Both the DexCom and Medtronic systems display three-hour trend graphs on-screen (the Medtronic Guardian also displays a six-hour trend graph). Although looking back at the past three hours of data will not help you fix a problem that happened earlier in the day, it can provide you with information to prevent a recurrence of a problem.
For example, checking the three-hour trend graph a couple of hours after meals can reveal the effect of various types of food. If you have a Medtronic CGM system, you can upload your data to a Web site called Carelink, which has software that stores, organizes, and analyzes all of your numbers; the Sensor Overlay By Meal chart indicates after-meal trends very clearly. Did your blood glucose level spike very high soon after the meal? If so, perhaps you need to take your insulin earlier, get some physical activity after eating, or choose foods that digest more slowly. Did your blood glucose drop soon after the meal, and then rise a few hours later? Perhaps you need to delay your premeal insulin injection or extend the delivery of your bolus (if you use an insulin pump).
For those who inject pramlintide (brand name Symlin) or exenatide (Byetta), the three-hour trend graphs can show whether or not the medicine is effective at minimizing blood glucose spikes after meals without causing hypoglycemia. Although after-meal hypoglycemia with exenatide is rare, it is possible, especially for people who take a sulfonylurea drug such as glipizide, glyburide, or glimepiride. If hypoglycemia occurs, your doctor may reduce the dose of the sulfonylurea. Hypoglycemia after meals is more common with pramlintide. In this case, the solution may be to delay the mealtime insulin injection or to extend the insulin delivery (if using an insulin pump), or to reduce the dose of premeal insulin.
The three-hour trend graph can also reveal blood glucose patterns related to exercise. Checking the graph during and after exercise will show when and how much your blood glucose rises or falls. Insulin and snack adjustments can be made based on the patterns seen on-screen. For competitive as well as casual athletes, these adjustments can enhance performance.
Stress is another concern for many people. In some instances, stress can cause an abrupt or prolonged blood glucose rise. At other times, it can make blood glucose drop. The very inconsistency of the way stress affects blood glucose can cause, well, more stress. I have found that the three-hour trend graph provides a nice “inside look” at the effects of various forms of stress on blood glucose levels. Not only can you see if the stress is making your blood glucose rise or fall (requiring extra insulin or an immediate snack, respectively), but it can also indicate how long the effects last so that you’ll be better prepared the next time around.
The 9-hour graph on the DexCom system and the 12- and 24-hour graphs on the Guardian system (the 24-hour graph is only on the integrated CGM–pump system) play a role in the regulation of basal insulin, particularly overnight. Whether you take insulin by injection or from a pump, the basal component of your insulin program is designed to match your liver’s normal, sustained secretion of glucose into the bloodstream. In other words, the basal insulin should hold your blood glucose level steady in the absence of food, exercise, stress, or mealtime doses of rapid-acting insulin.
By looking at the monitor’s long-term trend graphs starting approximately four hours after eating a meal and taking rapid-acting insulin, you can evaluate the effectiveness of your basal insulin. If your blood glucose level is gradually rising or falling after the food and mealtime insulin have worn off, your basal insulin probably needs adjustment.
In the example here, taken from an insulin pump user, a meal is eaten (and bolus of insulin given) at 11:30 AM. At around 4 PM, the glucose level begins to take a downturn. This indicates that the basal insulin level is too high in the late afternoon.
In this example, taken from a person using injections of glargine (Lantus) as his basal insulin, the blood glucose level is rising through the night from 2 AM until 8 AM. This indicates the need for a higher dose of glargine.
In addition to helping fine-tune basal insulin doses, long-term trend graphs can be used to determine the action curve of your rapid-acting insulin. That is, they can help you find out how long it takes for your insulin to finish working. (They are also valuable for determining how much “insulin on board,” or active, unused insulin, you have in your system.) Action curves can vary from person to person, with times ranging from 2½ to 5 hours. To determine your personal action curve, simply check to see how long it takes for your glucose level to stop dropping after you take insulin for a meal or to correct a high blood glucose level. Once the line graph flattens out, the insulin has pretty much run its course. For instance, in the example here, the insulin action curve is approximately three hours.
Long-term trend graphs are also useful to see the effects of high-fat foods and of intense or prolonged exercise on glucose levels, as well as to see blood glucose patterns during illness. The computer programs that can be used with CGM systems (Medtronic’s Carelink Web site, DexCom’s DM2 software) can be used to examine long-term trends over multiple days. The Sensor Daily Overlay at www.carelink. minimed.com lets users see up to seven days of 24-hour trend graphs superimposed and color-coded. The Hourly Statistics report on DexCom’s software provides statistical averages and graphic displays for virtually unlimited blocks of time. Both systems can reveal overall peaks and valleys in glucose levels over the course of a day, as well as patterns linked to weekends, schedule changes, menstrual cycles, and other variables.
Those lovely alarms
Perhaps the most valuable aspect of CGM systems is their ability to alert the user when blood glucose may be approaching dangerously high or low levels. You might say they act sort of like the highway rumble strips that keep you from veering off the road and into a ditch.
Although they may not detect every low, CGM systems will provide an early warning for the vast majority of them — and much earlier than most people would detect them on their own. In 2004, Dr. Bruce Bode, a renowned diabetes clinician and researcher who practices in Atlanta, Georgia, published one of the first studies on the effectiveness of CGM. He showed that using a CGM system with the low blood glucose alarm turned on reduced the amount of time spent in a hypoglycemic state by nearly 50% compared to using a system without the alarm option turned on. The low alarm makes it considerably safer for someone at risk for hypoglycemia to work, drive, exercise, and generally aim for tighter blood glucose control. Likewise, the high glucose alarm allows a person to be more aggressive at managing after-meal glucose spikes and preventing ketoacidosis, a life-threatening condition associated with very high blood glucose.
High and low alarm thresholds are not the same as the boundaries of your target blood glucose levels. You don’t want to hear an alarm every time your blood glucose goes slightly out of target range (we call these “nuisance alarms”), but you do want to be notified of high or low blood glucose in time to correct the problem and prevent a crisis. When setting the threshold levels for your alarms, keep the lag effect in mind. When blood glucose levels are dropping, the CGM value is likely to be a bit higher than your actual blood glucose level. And when they are rising, the CGM number will probably be a bit lower. For this reason, it is a good idea to set the low alert threshold at 80 mg/dl or higher, and the high alert in the 200–250 mg/dl range. To avoid being woken up by nuisance alarms, many people choose to set their high threshold a bit higher and their low threshold a bit lower during the night.
Now get a good night’s sleep. If you use a CGM system, tomorrow you’ll have a lot of analyzing to do!