We were recently honored to attend an exciting presentation of an expert panel of clinicians and researchers in type 1 diabetes, convened in Tampa, FL by the Park Nicollet International Diabetes Center. The presentation, 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 Helmsley awarded IDC to study “optimizing glucose reporting, analysis and clinical decision making in type 1 diabetes.” As we understand it, the IDC will use the funds to update its ambulatory glucose profile (AGP), a universal software report for glucose data, with this panel focusing primarily on CGM data, based on the panel’s recommendations. 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 Helmsley Trust’s plan to improve outcomes by optimizing diabetes technology. We heard that the panelists had been quite vocal and passionate in their closed sessions on the day-and- a-half before the final presentations, as alluded to in several inside jokes throughout the presentation. Mostly, however, this seems to have ended in “violent agreement”; when they spoke the panelists were unanimous in supporting standardization, and quite aligned in terms of what a one-page standardized readout should look like: e.g., some sort of modal day as the most prominent graphic, time in range (TIR) as a key number, extreme simplicity overall such that anyone can get a basic sense of the report within 30 seconds of looking at it. (The current AGP modal day averages several days together and represents the data as 24-hour lines for the median, 10th, 25th, 75th, and 90th percentiles. Because the lines never cross, we found the overall display quite easy to look at and – we imagine – interpret clinically. At a glance one could tell where an individual was experiencing (or was at high risk for) hypoglycemia or significant hyperglycemia that would help prioritize clinical decision making. At the same time the spread of the 25th to 75th percentile lines (interquartile range - IQR) gave a quick visual of areas of most glucose variability that could benefit from some discussions about lifestyle adjustments. That said, it was stressed that the ultimate standard needn’t necessarily be the AGP, although many agreed this was a good starting point and praised the IDC for pushing this notion of a standard report for the past 5 years.
The guests of honor at the presentation included representatives from all the major glucose monitoring companies as well as several other diabetes technology players; after Dr. Bergenstal and the panelists presented their thoughts, the industry members had a chance to voice their thoughts. Many of these company representatives agreed with Helmsley’s David Panzirer and Dana Ball that a standardized report would surely expand the market for CGM, leading to better clinical outcomes and financial upside across the industry. (Although the emphasis of the meeting was on CGM, a similar principle could presumably apply in SMBG – both sectors were well represented at the table.) However, other representatives articulated concerns that standardization would be practically difficult to adopt, could stifle innovation in software, and might even be considered a form of collusion. Thankfully all of these challenges appeared surmountable based on the discussion that followed; we were glad to see solutions- oriented thinking by most of those assembled. Of course only time will tell whether those or other difficulties will ultimately stymie the Tampa meeting’s collaborative spirit. But if all goes well, we understand that the plan is f/em>or IDC’s team to create a v3.0 of the AGP per the panelists’ suggestions, study whether the new software improves clinical process measures in 1H12 (tweaking accordingly all the while), and make the final product freely available for industry wide adoption as early as 2H13. We think these efforts could lead to real, near-term progress for patients and providers, and we are eager to track yet another example of how Helmsley is shaking up the world of type 1 diabetes for the better.
Detailed Discussion and Commentary
Final Report and Presentation
David Panzirer (Trustee, Helmsley Charitable Trust) and Dana Ball (Program Director, Type 1 Diabetes Program, Helmsley Charitable Trust)
David Panzirer and Dana Ball kicked off the afternoon with brief, inspiring speeches directed primarily to the assembled business leaders. Mr. Panzirer introduced himself as the parent of a child with type 1 diabetes and spoke on the inefficiencies of current clinical diabetes care – particularly in the way data are downloaded. Given these day-to-day burdens, he said that technology is “the low-hanging fruit” in efforts to improve the lives of people with type 1 diabetes. Mr. Ball then gave background on the Helmsley team’s yearlong, multidisciplinary due-diligence efforts, through which they came to appreciate the importance of all the stakeholders in the community. He stated his belief that through collaboration, everyone can win: “There is enough business for everyone in type 1 diabetes.” Indeed, both men emphasized that the Helmsley Trust’s goal is to make life easier for everyone – patients, providers, and industry – by simplifying clinical care and streamlining the development of drugs and devices. They also made clear their conviction that no meaningful change can occur without the participation of industry.
Richard Bergenstal, MD (Park Nicollet International Diabetes Center, Minneapolis, MN)
Dr. Bergenstal summarized the characteristics of the ideal standardized CGM data report, as discussed by the expert panel in the preceding day-and-a-half. The main display should be some sort of customizable modal day, they decided, with the opportunity to see thumbnails of individual days at the bottom of the report. Must-haves included time in range (70-180 mg/dl was generally accepted as the euglycemic zone); time spent in extreme hypo- and hyperglycemia; average sensor glucose; and glycemic variability (SD, CV, IQR each having their pros and cons from which to select for the initial view still to be determined with more professional and patient evaluation). The panelists agreed that patients and providers alike should be able to make sense of the first page within 30 seconds or less (a strong emphasis at IDC), and that the data acquisition process should be minimally disruptive to clinical flow. They also produced a “wish list” of functions to be developed after the basic standardization is complete.
The IDC/Helmsley were gratified to attract such a renown group of adult and pediatric diabetes experts, clinicians, researchers, technology gurus, technology journal editors and a strong patient advocate as panel members that worked a long day and a half and have promised to keep engaged to see the project through.
Members included: Dr. Bill Tamborlane – Yale; Dr. Satish Garg – Barbara Davis; Dr. David Klonoff – UCSF; Dr. Anne Peters – USC; Dr. Bruce Buckingham – Stanford; Dr. Robin Goland – Naomi Berrie, Columbia; Dr. Larry Deeb – Tallahassee; Dr. Andy Ahmann – Oregon Health & Science; Dr. Henry Rodriguez – U. South Florida; Dr. Rem>oy Beck – Jaeb Center for Health Research; Dr. Tim Bailey – San Diego; Davida Kruger, NP – Henry Ford; Phil Southerland – Team Type 1; Dr. Glenn Matfin, Ellie Strock, ANP, Maggie Powers, PhD, RD, David Wesley, Joan Bissen (project administrator) – IDC staff; Carol Rizzo, Imran Nasrullah and Marie Schiller – Helmsley staff. Dr. Irl Hirsch –U. Washington, gave input and will be part of the follow-up but was not able to attend.
In addition there were guests from the FDA, HL-7, Helmsley staff, IDC staff including Beth Olson, RN (project manager) and Thoughtworks- technology consultants. Representatives from ADA, AACE, AADE, Endocrine Society, JDRF, NIDDK, and Close Concerns/diaTribe, were invited and several participated in the pre-panel overview webinar and others have dialogue scheduled to follow the conference and the white paper output.
- The panelists were in agreement about most aspects of what a one-page, clinically focused CGM report should include. Notable disagreements included exactly where to place the cutoffs for hypo- and hyperglycemia, as well as how to measure glycemic variability. However, we got the impressions that the panelists would be able to come to consensus on these topics in the interest of settling on something, if only to start.
- Time in Range (TIR): The panelists agreed that the default euglycemic range should be defined as 70-180 mg/dl, though some advocated for 140 mg/dl as the upper limit of normal. Dr. Bergenstal pointed out that getting the panelists to define hypoglycemia as<70 mg/dl (as opposed to ≤ 70 mg/dl) was itself somewhat notable, since today even key clinical guidelines are not consistent on this point. He acknowledged that the practical relevance is minor, but we agree with his point that such inconsistencies indicate how far diabetes care is from standardized data reporting.
- Time spent in extreme glucose ranges: Not all glucose values below 70 mg/dl and above 180 mg/dl are similarly worrisome. Thus the panelists recommended that the main report include information about time spent with glucose below 55 mg/dl and above some threshold hyperglycemic value (250 mg/dl if euglycemia is defined as 70-140 mg/dl, 300 mg/dl if euglycemia is defined as 70-180 mg/dl).
- Alternate measures of hypoglycemia: The panelists thought that data on the number of hypoglycemic episodes would not be necessary for the main readout page. However they agreed that such data could be useful for research and should be includedin a supplementary section of the report (envision tabs that one can click to drill down in each area of interest).
- “Glucose exposure”: A simple average of glucose values may be an adequate summary of exposure to blood sugar, the panelists decided. Area under the curve (AUC), a more complicated measurement of glucose exposure, may be interesting for researchers; thus the panelists recommended including it in a supplementary section. Of particular interest is the very easy way the AGP report can report and visualize overnight vs. daytime AUC.
- Glycemic variability (GV): This seemed to be the area of least consensus among the panelists. Standard deviation (SD) is probably the most familiar measurement for patients and providers, but many experts advocated for coefficient of variation (CV) or interquartile range (IQR) on the grounds that those measurements might make more sense. Dr. Bergenstal said that the IDC and Helmsley were open to feedback and opinions on the GV issue. Given the lack of high-quality clinical evidence on GV, we think that building consensus on a single “best” measurement at this point is more a matter of diplomacy than science. We tend to think that getting all clinicians and researchers to measure GV in any consistent way would be an improvement in clinical care, though we fully admit that we are not experts on the statistical subtleties that make each approach more or less valid. We understand the IDC has tended to report the IQR as it top line measure of GV since IQR does not assume the glucose variation is evenly distributed above and below the median glucose value as does the more commonly used SD.
- Data acquisition should disrupt the clinical workflow minimally, if at all. Dr. Bergenstal briefly mentioned several possible ways that the data could be downloaded: at the clinic via some sort of kiosk, at the clinic via a cloud-based technology, or at the clinic from a scanned copy of a sheet that the patient brings in. Another attractive possibility is for patients to be able to easily send the data to the cloud from home. That way the clinician could have the numbers available without waiting any precious visit-time on data acquisition.
- The graphical interface should pass what the IDC’s Dr. Roger Mazze (inventor of the AGP) calls the “30-second rule”: endocrinologists, educators, primary care providers, and patients should all be able to look at the display and determine whether a problem exists within 30 seconds. On that note of consensus, the panelists recommended that the main display to some sort of simple modal day (showing up to two weeks of data) that could be edited by the clinician to show only certain types of days (e.g., weekdays, weekends, days with exercise). The panelists also asked for a clear visualization of the target range and how much time patients spend within, above, and below it. Another must-have is straightforward comparison to previous reports. Providers need to be able to see what changed between visits (or even more frequently if they check in on patients remotely). Dr. Bergenstal noted that ideally, patients would even review their own data readouts on a week-to-week basis between visits. (We think that the low rates of data downloading among patients today have something to do with human nature as much a particular interface problem, though the interface problem is certainly a major one and every little bit will help in this regard; we anticipate that the major benefits of standardization would be at the clinical and educational level.)
- The panelists also generated a “wish list” of functions that would be nice to have once the simplified, standardized software has been developed. These nice-to-have features included: integration with pump downloads; easy marking of insulin time and dose (no matter whether given by pump or MDI); easy marking of meal time and quantity (e.g., grams ofcarbohydrate), easy marking of exercise time, type, and amount; easy marking of illness or other stressors; and integration of downloaded data with electronic medical records (EMR). Dr. Bergenstal said that the panelists tried not to let themselves get too distracted by these possibilities. However, he made it clear that EMR integration, especially, is a big long-term priority for the IDC/Helmsley collaboration.
Richard Bergenstal, MD (Park Nicollet International Diabetes Center, Minneapolis, MN)
The day concluded with a lively group discussion led by Dr. Bergenstal, who solicited comments first from members of the expert panel and then from members of industry. The panelists made passionate arguments for standardization, both to improve clinical care in the short-term but also to make CGM more feasible to teach to patients and providers. The assembled executives voiced unanimous support for standardization in the abstract, but several company representatives aired concerns about, for example, disrupting their current business models and hindering company-by-company innovation. Fortunately all of these issues seemed surmountable given a concerted effort; at the very least the standardization project seems poised to move to the next phase.
- Members of the expert panel outlined several potential benefits of standardizing glycemic data reports. A strong sense of frustration with the current multiplicity of standards was implicit in many comments, and explicit in several others. “Could we at least come up with a common language?” asked one panelist, referring to the different words that diabetes technology firms use for the same concept (e.g., insulin on board vs. residual active insulin) as an example of how wide the gaps between companies are today.
- Multiple panelists emphasized how standardized reports could improve education: of patients, of nurses and primary care providers, and of young diabetes specialists and medical students. The complexity of today’s landscape makes it difficult for anyone trying to learn multiple systems (and difficult for teachers and mentors to keep straight which system they are teaching at a particular time), one expert observed. Another clinician said he wants to know that when he teaches a pupil how to read a glucose sensor download, that instruction will still be useful in 10 years. Someone else on the panel noted that physicians who lead case studies are often wary of company-branded materials. Thus, current company-branded data readouts are seen by some as a non-starter as instruction materials, while a standardized readout would overcome this hurdle. With more education, uptake would likely improve among all parties. And with clinicians and patients more educated about how to leverage glucose monitoring, the rise in usage would have big potential payoff in both glycemic control and quality of life.
- The most immediate expected benefit would be greater efficiency in clinical care. The panelists anticipate that a simplified protocol for data readouts would save significant time during each visit, and even outside of visits (since clinicians would not have to study to keep abreast of multiple different softwares).
- A couple of panelists highlighted the benefits of a standardized protocol for research, especially in conjunction with the T1D Exchange’s clinical registry. We were somewhat surprised not to hear anyone mention the value of standardization for artificial pancreas development – indicative of the panel’s focus on near-term clinical goals.
- Patient representative Phil Southerland mentioned some additional quality- of-life benefits that standardized reporting could bring: for example, enabling friends to compare their numbers and generally building rapport among the patient community.
- Although industry representatives were uniformly positive on the concept of standardization, several raised concerns about how a new industry-wide standard might actually be implemented. Fortunately most of these issues seemed fairly surmountable, at least based on the preliminary discussion at the presentation. The main debate seemed to be as follows: skeptics of standardization thought that it could tie up companies in red tape, while proponents argued that market size would grow more than enough to make up for any temporary restructuring burdens.
- One major set of industry concern was that setting a standard would stifle innovation and/or competition. Many company representatives noted that by adopting a uniform approach, manufacturers would sacrifice a chance to differentiate themselves from the competition. Other industry members were more optimistic, however. These proponents said that a uniform basic report could actually foster innovation by giving companies a platform onto which they could add proprietary features. Moreover, panelists remarked that only ~10% of providers and ~5% of patients perform data downloads today, indicating that current software innovations are being lost to the vast majority of their intended markets. (Another of the expert panelists observed that companies would still be able to innovate with regard to real-time data display and every other aspect of device form and function – features that we tend to think are more important than data display in how patients and providers decide which technology to use.) Some executives also warned that setting a standard now could freeze the way glycemic data are reported prematurely, before controversial areas like glycemic variability are fully understood. However, the panelists and IDC representatives seemed confident that any standard could be updated regularly in light of new research or shifting consensus.
- Many industry members spoke about the need for a formal process or certification that would incentivize companies to adopt a new standard (and, importantly, protect them from allegations of collusion). The consensus was that the FDA would not be the best organization to enforce the new standard, but ISO and CLSI were proposed as alternatives. The best way of all may be through Health Level Seven (HL7), an international standards developing organization that focuses on interoperability of clinical and administrative healthcare data. Dr. Robert Dolin, vice- chair of the HL7 board of directors, had discussed the standardization process with the expert panelists. During the final presentation he did not speak much, but he did convey full confidence that standardization would benefit everyone involved in glucose monitoring and would be legal and feasible through HL7.
- A few regulatory hurdles were hypothesized – for example, the possibility that the FDA might take issue with the curve-smoothing functions used in the AGP. Encouragingly, though, the FDA’s Dr. Courtney Lias indicated that standardized glycemic reporting would likely have a clear path forward if the diabetes care community were united in their efforts.
- This was a very impressive collective effort, which encompassed a range of critical groups: 1) the IDC’s project management and consensus building; 2) HelmsleyTrust’s support and vision; 3) diabetes technology industry leaders (Abbott, Bayer, BD, Dexcom/Sweetspot, Diasend, J&J, Medtronic, and Roche) general support for conceptually embracing the concept; 4) the T1D Exchange’s 67 clinic network dedicated to improving the lives of people with T1D; and 5) patient representatives from Team Type 1 and diaTribe for advocating strongly for patients and their needs. One can be encouraged, to say the least, that this project appears to stand a very good chance of showing that it is possible to optimize glucose reporting, analysis, and clinical decision making in an efficient and effective manner. We look forward to changes ahead on the CGM as well as BGM front in this regard.
-- by Joseph Shivers and Kelly Close