Expert panel calls for standardized reporting of glucose monitoring data, supported by IDC and Helmsley Charitable Trust; joint publication in JDST/DT&T– March 12, 2013

Executive Highlights

  • On March 1, Diabetes Technology & Therapeutics (DT&T) and the Journal of Diabetes Science and Technology (JDST) jointly published a meeting report summarizing the 2012 IDC/Helmsley Charitable Trust Expert Panel on standardizing CGM data download reports.
  • The paper and accompanying editorials summarize the meeting’s key takeaways (a focus on time-in- range, glycemic variability, and hypoglycemia) and propose use of IDC’s Ambulatory Glucose Profile.

On Friday, a fantastic 14-page “meeting report” focused on standardizing the analysis and reporting of glucose monitoring data was jointly (!) published online in Diabetes Technology & Therapeutics and the Journal of Diabetes Science and Technology – the list of authors is a veritable who’s-who in diabetes technology (see below). The paper summarizes the key takeaways and recommendations from a 2012 expert panel on the dire need for uniform glucose data reporting and summary metrics to aide clinicians, researchers, and patients. The initial focus is on CGM data using the IDC’s Ambulatory Glucose Profile (AGP). As a reminder, last year’s expert panel was convened by the Helmsley Charitable Trust and the International Diabetes Center (IDC) in Tampa, FL (March 28-29, 2012). Expert clinicians, industry, FDA, Health Level Seven International, and patient advocate Phil Southerland were present; The paper is a very valuable read and is divided into the following sections: (i) Critical Issues in Diabetes Management (suboptimal control, limitations of A1c, underutilization of CGM, lack of report standardization); (ii) Standardization of Glucose Reporting, Analysis, and Clinical Decision Making (target range, glucose exposure, glycemic variability, hypo- and hyperglycemia); (iii) Proposed Ambulatory Glucose Profile; and (iv) Industry Issues and Consideration. The bullets below contain key excerpts from each section. There were also three excellent commentary pieces published in DT&T: Drs. Sanjoy Dutta and Aaron Kowalski of JDRF (a must-read on why A1c is not everything), Dr. Fran Kaufman (a cautious endorsement), and Dr. Satish Garg (why this is a good first step).

The paper and accompanying editorials are as much of an endorsement of standardizing CGM reports as they are an indictment of A1c. Indeed, time-in-range and hypoglycemia were big, big themes in the paper and the IDC’s Ambulatory Glucose Profile – this is outstanding to see from a patient perspective. We couldn’t agree more with JDRF’s Drs. Aaron Kowalski and Sanjoy Dutta’s accompanying commentary, “Although the HbA1c will likely remain a key metric and measurement in diabetes management, it is not, however, a good marker for diabetes control on a day-to-day basis or for providing insight into strategies to improve glycemic control.” The AGP, by contrast, does both quite well with extensive use of time-in-range, glycemic variability metrics, intuitive graphs, and a real drive to make a useful “one-pager” that clinicians and patients can use to improve outcomes (whether that’s hypoglycemia, hyperglycemia, glycemic variability, or A1c).

The publication also contains valuable example snapshots of the AGP. As patients, we would love to have our CGM data run through this. Aside from the valuable focus on time-in-range and glycemic variability, we’re glad to see normal population reference ranges below every statistic. In most other therapeutic areas, HCPs strive for normalcy. We think the AGP helps HCPs do the same in diabetes. Yes!



  • The meeting report is available online in both Diabetes Technology & Therapeutics and the Journal of Diabetes Science and Technology. Sincere kudos to the two journals for making a joint publication possible. This is the first time this has ever been done and a testament to just how important this issue is. As we understand it, both were pushing up to the March 1 deadline to make this happen. The meeting report and editorials are available for free in DT&T at the links below. Reading the meeting report in JDST seems to require a subscription (in addition to it being buried at the bottom of the page) – we’re not sure why this is, but hope the journal reconsiders.

  • The meeting report and accompanying commentary can be accessed at the following links. All are very quick, engaging reads – we salute the authors for an excellent and concise synthesis of the two-day meeting.

  • The paper proposes use of the IDC’s Ambulatory Glucose Profile, a “simplified” single-page document to be used in clinical practice or in communicating with a patient.” We think it does a great job of this in a number of ways: (i) clear organization; (ii) clear rationale for including things; (iii) a focus on glycemic variability and time-in-range in conjunction with average glucose/A1c; (iv) smart user interface design (clicking in certain places allows for a deep dive); and (v) use of reference ranges to provide context. More detail is provided in the section three below.

  • The publication is a summary of a March 2012 expert panel convened by the International Diabetes Center and supported by the Helmsley Charitable Trust (HCT). That meeting, which also included members of industry, the FDA’s Dr. Courtney Lias, and Vice-Chair of the healthcare standardization organization HL7, represented the first phase of a yearlong, $1.5-million grant that the HCT awarded IDC to study “optimizing glucose reporting,analysis and clinical decision making in type 1 diabetes.” The long-term goal is to develop a standardized, industry-wide report for all CGM data. Such a standardized report has the potential to simplify clinical diabetes care and to increase adoption of CGM among both patients and providers – part of the HCT’s plan to improve outcomes by optimizing diabetes technology. To read our coverage of that meeting, see our report at

  • The paper’s list of authors represents a very impressive list of KOLs in diabetes technology: Drs. Richard Bergenstal, Andrew Ahmann, Timothy Bailey, and Roy Beck, Ms. Joan Bissen, Drs. Bruce Buckingham, Larry Deeb, Robert Dolin, Satish Garg, Robin Goland, Irl Hirsch, and David Klonoff, Ms. Davida Kruger, Drs. Glenn Matfin and Roger Mazze, Ms. Beth Olson, Mr. Christopher Parkin, Drs. Anne Peters, Margaret Powers, and Henry Rodriguez, Mr. Phil Southerland, Ms. Ellie Strock, Dr. William Tamborlane, and Mr. David Wesley.

  • The paper mentions that the following companies were represented at the meeting’s report-out: Abbott Diabetes Care, Animas Corporation, Bayer, BD, Dexcom, Diasend, LifeScan, Medtronic Diabetes, Roche Diagnostics, Sanofi, and SweetSpot Diabetes Care. We were glad to see all the big players present, and look forward to receptivity on adopting a universal standard. We agree that one report standard would help expand the market and utility of the technology, though we would guess it may take some time to win everyone over.


  • Suboptimal control: To support the assertion that control is suboptimal, this section mentions the A1c data from the T1D Exchange Clinical Registry (mean of 8.3%-8.7% in patients ≤25 years, and a mean of 7.7% in older patients. It notes that suboptimal glycemic control is often the result of poor adherence and fear of hypoglycemia.

  • Limitations of A1c: This section does a great job of emphasizing why A1c just cannot say everything: 1) In DCCT, A1c explained only ~11% of the variation in risk between intensive and standard glycemic control patients; 2) A1c is unable to characterize daytime glucose patterns; 3) patients with similar A1c values can have markedly dissimilar patterns of glucose excursions and rates of hypoglycemia throughout the day and overnight. The section also discusses time-in-range and glycemic variability, noting that increased glycemic variability is a strong predictor of hypoglycemia and is also correlated with poor glycemic control. The authors note that it’s time to establish a definition of optimal glycemic control that includes A1c at target (personalized for each individual, but somewhere around 7% for many adults) without any severe hypoglycemia and only a minimal number of very low or dangerously low glucose values. Additionally, it notes that “if CGM becomes the standard research and clinical tool to evaluate and manage glycemic control, a logical glycemic goal would be to maximize time in target range.” We couldn’t agree more.

  • Underutilization of CGM. Only 3% of young patients (≤25 years) in T1D Ex use CGM for their diabetes management; CGM use among older T1DM patients (26-49 years) is slightly higher at 14%. Although underutilization of CGM is often attributed to limited reimbursement and patients’ and parents’ perceptions regarding the complexity and inconvenience of CGM use, clinician reluctance is also a key issue. This may be due to the lack of experience/expertise of clinicians in determining the most appropriate candidates for CGM or in interpreting CGM data. There are also time constraints and the potential disruption of workflow associated with CGM initiation, downloading, and interpretation in clinical practices. Lack of a relatively simple or (at least) straightforward and intuitive statistical and graphic visualization of the glucose data via downloadsoftware is a major contributor to the uncertainty and reluctance of clinicians to incorporate CGM into their practices.

  • Lack of standardization of glucose reporting: In a short and to-the-point section, the paper notes that there is no standardization regarding glycemic statistics, graphical presentation of data, or a common terminology analyses. “Moreover, the sheer diversity and number of reporting options creates such a daunting ‘learning curve’ that many clinicians never invest the time necessary to even start using CGM technology, let alone attempt to become proficient in its use.” We thought this was a great point and wonder to what extent CGM penetration has been limited by this fact alone.


  • Target range: Most panel members (56%) selected 70-180 mg/dl as the default target range. The paper notes that while this is not an ideal or normal glucose range, it represents a target range commonly used in clinical practice and one that promotes realistic and safe expectations. Interestingly, it has been shown that, if 50% of the glucose readings are in this range, the A1c will usually be around 7%.

  • Glucose exposure: Overall, 82% of panel members chose mean glucose of all readings as the metric to which clinicians and patients can most easily relate. Most (59%) participants indicated that mean glucose exposure for specific time periods (e.g., overnight, fasting, 2-4 hours postprandial) could be helpful in evaluating the effects of food, exercise, or insulin. From a patient perspective, this is exactly the kind of useful data analysis that can help make therapy changes. It’s great to see this front and center. Because A1c is not always available when the CGM report is being reviewed, panelists thought it would be appropriate to include the estimated A1c based on the average glucose. In our view, it’s important to note that this will appear next to time-in-range– simply reinforcing the need for and value in using composite targets.

  • Glycemic variability: The majority (69%) of panel members indicated that standard deviation (SD) was the metric most commonly used and understood for assessing and reporting glycemic variability. Percentage of coefficient of variance, derived from SD (100 × SD/mean of observations), was also selected by the panel as one component of glycemic variability to follow. Last, inter-quartile range (IQR) was chosen as a reliable aggregate measure of glycemic variability. Plus, it allows one to easily visualize the time of day or relationship to a meal or medication when there is high glycemic variability, which may need clinical attention. We expect the average HCP might need some education on coefficient of variance and IQR, though SD is fairly easy to understand and compare over time. The question of what glycemic variability metric to use has long been a debate over the years, and we agree that SD is perfectly fine – while the data may not be normally distributed, SD gets us 90% of the way there, and its actually a metric that patients and providers can understand.

  • Hypoglycemia: The discussion of hypoglycemia focused on two main issues: cut points for hypoglycemia and the definition of severe hypoglycemia. The majority (88%) of panel members selected <70 mg/dl as the criteria for “reportable hypoglycemia.” The majority (53%) of participants were satisfied with <55 mg/dl as criteria for more significant hypoglycemia. Following the panel, hypoglycemia and hyperglycemia were divided into three categories of severity: <70 mg/dl (“low”), <60 mg/dl (“very low”), and <50 mg/dl (“dangerously low”). Severe hypoglycemia has been traditionally defined as “requiring assistance from another person.” Seventy-nine percent of the panel felt that there needed to be another subcategory of severehypoglycemia, which they suggested be called “major hypoglycemia” (requiring medical personnel intervention in the home or an emergency center, hospitalization, seizure, coma, or death). This seemed necessary since quantifying severe hypoglycemia can be difficult with the subjective term “requiring assistance.” We’re glad to see this new definition and agree with the rationale. We’ve often heard this as a point of contention in interpreting and comparing the benefits of different therapies and technologies.

  • Hyperglycemia: The paper proposes >180 mg/dl (“high”), >250 mg/dl (“very high”), and >400 mg/dl (“dangerously high”).


  • An example Ambulatory Glucose Profile can be found in figure two of the paper. The link is The picture below is on page 571 of the journal article and page 10 of the PDF.
  • The primary focus of the AGP was to develop a “simplified” single-page document to be used in clinical practice or in communicating with a patient.” We think it does a great job of this – it doesn’t take long to spot where trouble areas are using the modal day graph with interquartile ranges. In the example above, we quickly noticed the following trouble areas: 1) there is room to improve overall glucose (mean: 169 mg/dl); 2) this patient has a fair amount of glycemic variability (an SD of 90 mg/dl is more than half the average glucose); 3) nearly 10% of glucose values are “dangerously low” (based on the modal day report, it looks like overnight and from 4-6 pm are the key times this is happening; 3) there is room to improve postprandial glucose after breakfast and after lunch (these represent the widest IQR timeslots on the modal day report).
  • We are glad to see reference ranges in the statistical summary section for a number of reasons. First, they give context to numbers that can be abstract. That’s useful for clinicians and patients trying to spot problem areas. Second, they are derived from a normal reference population (mean 2 SD), which encourages clinicians to strive for normalcy. The paper notes that “Although patients with diabetes (particularly T1DM) are not expected to achieve completely normal glucose values, this gives a frame of reference.” We agree – for instance, it’s hard to know what glucose exposure (especially AUC) or glucose variability should be without any context. Certainly, reaching glycemic values in people without type 1 diabetes is incredibly challenging, though we think they could be motivating for patients and providers. We wonder if it would be useful to add a reference point for a well-controlled patient with type 1 diabetes (e.g., 6.5% A1c and very little/no hypoglycemia) – this would provide a more realistic goal since normal reference glycemic control might be impossible for many to achieve.

  • We like Visual Display’s focus on quick, immediate takeaways in graphical form. The paper notes, “At a glance, one can observe the time(s) of day when glucose is most consistently low or high and when the most variability is occurring [the width of 25-75 (50% of reading) or 10- 90 (80% of readings) lines] that needs to be addressed. This is an exercise clinicians can do together with patients in a matter of minutes.”
  • The daily view presents a calendar of thumbnail AGPs for each day that is included in the overall profile. Clicking a thumbnail will enlarge it to a full-size one-day AGP with corresponding glucose metrics for that day – awesome user interface design. Having attended many CGM data downloading 101 talks in recent years, we know that many experts like the single- day snapshots, since they provide teachable moments and can hone in on what’s causing problems. IDC is developing a modified daily view as well as a concise modal day AGP view that captures pump download data that is important to health care professionals (e.g., basal rates, insulin-to-carbohydrate ratio, correction doses, carbohydrate intake) as they refine pump therapy.

  • For a comprehensive prospective glucose analysis, it is recommended that patients collect approximately 14 days of CGM data. In a series of analyses of CGM data in type 1 and type 2 diabetes, 14 days of CGM data gave a very accurate, relatively stable reflection of the key glucose metrics in the AGP, and the modal day display was highly reflective of what the display would look like after 30 days of CGM use in most patients. For those not interested in wearing CGM 24/7, this is encouraging to hear. We hope providers will begin prescribing intermittent CGM use for appropriate patients as well.

  • Moving forward, IDC has a lot in the works further refine and optimize the AGP. Piloting is underway to quickly incorporate the AGP dashboard into an electronic medical record this is critical in our view, since EMR integration would make it easy to compare two or more AGPs after a therapy adjustment or between visits/electronic communications. Additional functionality, such as inclusion of data relevant to insulin or other diabetes medication administration, nutrition (timing and carbohydrate content), and physical activity is being explored. Workflow usability studies and patient and provider preference or satisfaction evaluations are being designed utilizing the AGP dashboard.

  • The paper also notes that one key to successful implementation is an enhanced workflow that seamlessly and rapidly acquires glucose data from any device at a clinic visit or over the Internet cloud network – having seen Dr. Bruce Buckingham’s picture of a tangled web of cords, we wholeheartedly agree. We would guess the roadblock on this is an absence of industry standards and the need to redesign devices to accommodate a universal transmission protocol. Certainly not easy, though we hope industry sees the value in standardization and works together to find a solution.


  • Industry representatives were split on how a standardized report would alter incentives to innovate. Some representatives expressed concern that standardization could potentially stifle innovation, while others believed that the AGP dashboard approach, in fact, encourages more innovation because it creates an entry for clinicians to begin interacting with CGM data immediately and thus allows manufacturers to focus on more sophisticated data analysis features and capabilities (e.g., “secondary visualizations”). Many participants (panel members and industry representatives) felt that the AGP dashboard approach was analogous to the electrocardiogram, noting that while several manufacturers produce electrocardiogram systems, visualization of the data is standardized. We agree given how few patients are downloading data as it is – the most innovative software in the world does not matter if no one uses it, and that often seems to be the case with diabetes software. Ultimately, a standard report with a brilliant design like the AGP should encourage more patients to repeatedly download their data and make positive therapy changes.


  • Drs. Aaron Kowalski and Sanjoy Dutta wrote an excellent accompanying commentary entitled “It’s Time to Move from the A1c to Better Metrics for Diabetes Control.” They note that recent advances like CGM have not led to widespread improvements in A1c levels (“we sit on an ‘A1c plateau.’” They cite the “breakdown between therapeutic efficacy and effectiveness,” and “a ripe place to start to address this problem is through standardization of metrics for the most important marker of diabetes management – glucose levels.” The article comes down strongly on using A1c alone, since it gives little information about hypoglycemia or day-to-day variability (e.g., “A mean speed of 55 mph over the past three months of commuting will never reflect times when a vehicle is racing at 100 mph and far in excess of the speed limit or is slowed to 10 mph in traffic congestion. Similarly, the clinician only receives the most basic of information from the HbA1c measurement and masks the occurrence and frequency of dangerous highs and lows”). We couldn’t have said it better. Drs. Kowalski and Dutta also remind readers of the JDRF CGM trial, where patients with an A1c level of 6.4% spent nearly 100 minutes a day – every day – at levels <70 mg/dl – if that’s not a clear and convincing reason to go beyond A1c, we don’t know what is.
  • Dr. Satish Garg’s commentary notes that “Both providers and patients are frustrated on the data reporting not being standardized and that every download report is bulky, slow in report generation, proprietary, and difficult to navigate.” He called the AGP “an important step toward the best use of data by providers and patients to possibly improve healthcare outcomes.” When one combines physicians’ limited time to visit with patients and the low rate of self-downloading among patients, “it is a lost opportunity of using important information to improve diabetes management.”

  • Dr. Francine Kaufman provides a cautious industry perspective in her editorial. She reminds readers of the authors’ assertion that “non-standardization inhibits clinicians and patients from using these reports to maximally improve diabetes outcomes.” Yet, she expressed uncertainty in what standardization can accomplish: “Whether this approach will enable more patients to use and benefit from CGMs has yet to be determined. Whether this will be more facile for clinicians as they attempt to determine patterns and trends and adjust management is also unknown.” In her view, the most clear is that a single set of definitions, metrics, and standard colors and displays should enable those interpreting CGM uploads to save time and improve decision making. Her short editorial ends on a somewhat suggestive note, stating, “Although standardization is good in many respects, keeping the field open and allowing for iterative improvements in uploaded information are imperative.”


-- by Adam Brown and Kelly Close