Abstract
As diabetes care and education specialists, we know better than anyone that nothing comes easy. Motivating people with diabetes to take the stairs when a shiny escalator beckons. Training them to know the difference between a “good carb” and a “bad carb.” Fighting with the IT department for the right to download a simple blood glucose meter. The list goes on and on.
Staying on top of technological advances presents a whole new set of challenges for the busy diabetes care and education specialist. ADCES’s Danatech.org website was born specifically to help with this matter.
Staying on top of technological advances presents a whole new set of challenges for the busy diabetes care and education specialist. ADCES’s Danatech.org website was born specifically to help with this matter. Continuous glucose monitors (CGMs) in particular are taking the diabetes world by storm, but successful use hinges on our ability to actually do something useful with the information they generate.
Analyzing CGM data is not a simple task. The reports can be cumbersome and intimidating, propelling many to run for the nearest exit. Does that mean we should give up? Of course not! We’re diabetes care and education specialists. If we can navigate ICD-10 and HCPCS codes, we can do this too.
Despite the complex nature of CGM data, there are ways to access just the right reports, perform quick analyses, and glean sound, meaningful clinical insights in a timely manner. Here’s how.
Breaking Down CGM Reports the E-Z Way
Each CGM manufacturer has its own report-generating software, and there are a number of “midware” programs that can download and display data from multiple devices, including CGMs (Table 1). Is one program better than another? That’s a matter of personal preference. In general, if you’re only looking at CGM data, you can use the manufacturer’s program. But if you want more robust reports and prefer to have contextual information (food, insulin/meds, activity) displayed along with the CGM data, then midware may be a better choice.
CGM Data Reporting Programs.
Each program has the ability to create a variety of different reports. All can produce statistical summaries that can be useful for evaluating general glycemic control and tracking progress from one clinical visit to the next. For example, the mean (average) and median (half of the readings are above and half are below the median) correlate closely with A1C. Standard deviation and coefficient of variation provide some insight regarding glucose variability. A high standard deviation indicates that the patient is experiencing frequent and extreme peaks/valleys. A standard deviation that is less than 36% of the average is generally desirable. And the frequency of high/low alerts or excursions allows us to see how often the patient is experiencing potentially dangerous high and low glucose levels.
But the stats that are emphasized most heavily in the updated American Diabetes Association’s Standards of Medical Care are the percentage of time above, below, and within one’s target range. This is probably the best measure of the quality of glucose management, as opposed to the others that represent quantity metrics. It is important to customize the target range for each person before analyzing CGM data. A range of 70 to 180 mg/ dl is commonly used since it focuses on avoiding hypoglycemia and urine diuresis, but clearly it needs to be individualized based on each patient’s unique goals, risks, and capabilities.
Once you’ve explored some of the key stats, it’s time to dive into the trend graph reports. These offer insight into the specific nature of glucose levels throughout the day and night. And here’s the beauty of it: With a little bit of experience, you should be able to perform an overall evaluation in just a few seconds. Most software can generate 3 types of trend graph reports: (1) individual daily reports, (2) multiple-day overlays, and (3) summary graphs, which take the form of either hourly summaries or an ambulatory glucose profile. Let’s take a look at how different reports can be used.
Daily Trend Graph Report
Individual day trend graphs provide a granular look at glucose values 1 day at a time (Figure 1). Basal insulin doses can be titrated by observing glucose stability overnight and any time there are long intervals between meals and bolus insulin doses. Post-meal management can be evaluated by noting the magnitude of blood glucose peaks that occur after meals. Patterns related to hypoglycemia can also be seen, along with the effects of specific food choices, stressors, and exercise (although this requires some logging on the part of the patient).
Examples of daily trend graphs displayed from a Freestyle Libre download.
The percent of time above, below, and within one’s target range is probably the best measure of the quality of glucose management, as opposed to the others which represent quantity metrics.
The main problem with this type of report is the amount of time it takes to perform the evaluation. It is often necessary to look at 1 to 2 weeks of data to identify patterns, and that means looking at 7 to 14 separate graphs. It can be difficult to detect patterns in this manner unless you’re willing (and able) to go through a lengthy review.
Ambulatory Glucose Profile Report
The ambulatory glucose profile (AGP) report is a standardized report that shows daily ebb and flow in glucose levels (Figure 2). Slight variations of this report are available in just about every CGM reporting program. The report includes a median line, a dark shaded area that includes the middle 50% of readings (25% above the median and 25% below), and a light shaded area that includes the middle 80% of readings (10% of readings are above the high line and 10% are below the low line).
Example of an AGP report (source: www.agpreport.org/agp/agpreports#CGM_AGP).
Looking at the 4 to 5 AM time zone in Figure 2, the median glucose is around 85 mg/dl. The middle 50% of readings fall between 50 and 125, while 10% of readings are above 140, and another 10% are below 40.
The essence of an AGP report is its uniformity and simplicity. When the middle 50% shaded area reaches beyond high and low target levels (180 and 70 in Figure 2), there is reason for concern since a significant amount of time (more than 25%) is spent either above or below one’s target range. In Figure 2, this occurs in the early morning and most of the afternoon/evening. The middle 80% shaded area is more of a “gray area” as glucose is only occasionally outside of the target zone. It is necessary to evaluate at least 2 weeks of data for the AGP report to have validity since single high or low events can produce significant percentages of time out of range when evaluating only a few days at a time. In addition, when the middle 50% zone is uncharacteristically wide (like a pig in a snake), there is a great deal of glucose variability. This may be due to variation in food intake, stress levels, or physical activity or a lack of compensatory insulin/ medication for these situations.
Unfortunately, AGP reports are sort of like a politician at a press conference: They look pretty and sound important, but in the end, they may not offer much in the way of actionable information. Very few CGM users follow a consistent, regimented schedule every day in terms of meal, snack, exercise, and sleep times. For example, an early morning dawn phenomenon may not be captured in the AGP report if a patient’s sleep schedule changes from day to day. Post-meal glucose peaks may not appear for an individual whose meals start and end at different times throughout the week. And it is not possible to identify hyperglycemic episodes that follow hypoglycemic episodes and vice versa.
Sensor Overlay Reports, aka “Spaghetti Graphs”
Sensor overlays reports, also called “spaghetti graphs,” are multicolored and busy, so they can be an eyesore (Figure 3). But nothing is better for gaining quick, meaningful insights.
Example of an overlay report from Dexcom Clarity.
As was the case with AGP reports, sensor overlay reports reveal times of day when glucose levels are often above or below the target range. They give you the opportunity to visually “lop off” rare outliers that can influence statistical summaries. They also show the glucose trend overnight and between meals (for evaluating basal insulin). However, by being able to follow the tracing for individual days, you can also do the following:
■ Evaluate post-meal peaks even when mealtimes vary. Personally, I like to print the overlay report and mark the post-meal peaks with a big black dot and then visualize the dot groupings to assess post-meal management.
■ Determine whether correction doses of insulin are bringing the glucose level down too much, too little, or just the right amount. These can be seen easily by following the individual day tracings following hyperglycemic episodes.
■ Uncover patterns that precede or follow periods of hypoglycemia. Are lows often preceded by highs? Perhaps the correction doses are too aggressive or insulin-on-board is underestimated. Are lows followed by more lows? Perhaps the treatment is too conservative or slow to act or basal insulin doses are too high. Are lows followed by highs? Overtreatment may be taking place.
■ Quantify the precise duration of bolus action. This can be seen by following the sensor tracing after meal or correction boluses are given. See how many hours it takes for the glucose level to cease falling and flatten out.
■ Estimate the effectiveness of meal planning and meal doses of insulin. Are glucose levels consistently elevated, on target, or too low a few hours after eating?
■ Identify the immediate and delayed effects of physical activity.
There are also a few subsets of the sensor overlay report that can provide even greater insight. The “Sensor Overlay by Meal,” called “Meal Bolus Wizard” in the latest version of Medtronic’s Carelink software, lines up the times when meal boluses are delivered so that it’s easy to see the effectiveness of the dose (Does the glucose return to normal 3 hours after eating?) along with the extent of the post-meal glucose excursion, or the high point reached after the meal (Figure 4).
Meal bolus wizard from Medtronic Carelink.
A daily overlay sorted by day(s) of the week is another subset report available when reviewing data in programs such as Glooko, Eversense DMS, and Dexcom Clarity. This report allows for evaluation of glucose patterns on weekends versus weekdays, exercise versus nonexercise days, work versus off days, or any other semiregular event that takes place in a person’s life. For example, you may notice that glucose levels are uncharacteristically elevated on most Wednesday afternoons, which can prompt a discussion to determine the likely cause.
The Proof Is in the Pasta
Comparing an AGP report with a sensor overlay (spaghetti) graph can be a useful exercise. The AGP report in Figure 5 shows that glucose levels are rising in the middle of the night and stable through most of the day. The only conclusion one might draw is that the basal insulin dose is too low during the early part of the night or perhaps that the patient is snacking excessively in the evening. A spaghetti report for the same patient for the same time period (Figure 6) also shows a pattern of rising glucose in the middle of the night, but notice the lows in the evening? The rising glucose between midnight and 4 AM can be traced back to these hypoglycemic events. One might conclude that the patient is rebounding from the lows (overtreating the lows).
Example of a 7-day AGP report from Dexcom Clarity.
Example of a CGM data overlay report from Dexcom Clarity.
It’s interesting to note that on the 1 night when an evening low did not occur (Wednesday), there was no overnight rise afterward. Elimination of the evening lows will likely prevent the nighttime highs. It’s possible that the dinner insulin dose is too aggressive. Or, based on the very rapid decline in the 8 to 9 PM range, the lows may be related to evening exercise—a topic to explore with the individual. Considerable post-meal spikes occur on some but not all occasions, which might prompt a discussion of proper insulin timing or food choices. The 1 day with a significant morning spike (Friday) was preceded by a low at midmorning, indicating possible overtreatment of the low. It’s also noteworthy that the glucose levels on Sunday were much higher and more erratic than the rest of the week, offering another topic for discussion and problem solving.
From the standpoint of yielding useful clinical insight, the spaghetti (overlay) graph rules. It still allows for a quick view of the overall state of glucose control, but it also helps to uncover potential cause-and-effect relationships.
From the standpoint of yielding useful clinical insight, the spaghetti (overlay) graph rules. It still allows for a quick view of the overall state of glucose management, but it also helps to uncover potential cause-and-effect relationships. At a minimum, it lays the framework for productive conversations with the patient.
Avoiding Graphus Overwhelmus
Clearly, there is a great deal that can be learned from evaluating CGM reports. But that doesn’t mean that you must learn and fix everything at once. Focus on a couple of specific, manageable topics at each session. Perhaps you could evaluate glucose levels in a fasting state during today’s session and take a look at post-meal peaks next time. Or key in on patterns surrounding hypoglycemia today and examine the duration of bolus insulin action at the next appointment. There is virtually no limit to the fun you can have! ■
Footnotes
Gary Scheiner, MS, CDE, is owner and clinical director of Integrated Diabetes Services in Wynnewood, PA.