March 16-17, 2010; Gaithersburg, MD Day 2 Full Notes Draft

Executive Highlights

If you can believe it, day two in Gaithersburg, MD at the FDA SMBG accuracy public hearing, brought even more nuance to the discussion at hand than the day before. With over 350 attendees and several vocal audience members, including the accomplished Dr. Tony Furnary (author of the Portland Protocol), willing to contribute to the conversation, we feel a lot was gained from this meeting, even if clear answers were not reached for all the questions posed. While the second day’s conversation was dominated by considerations specific to glucose control in the hospital, issues such as better labeling and education still permeated the discussions. Certain concepts (such as the definition of tight glycemic control) were pulled up by the roots and re-examined in detail. We left day #2 with a strong sense that the FDA will be thinking much more broadly about the concept of accuracy with these devices than has been the case historically. We were also impressed to see the FDA looking forward to and planning for the future—real-time glucose monitoring (intravascular) in the hospital garnered substantial attention and the FDA is clearly keen on better understanding how to prepare to evaluate these devices. Below we bring you our full notes from the day, which represent a broad range of perspectives—from academic clinicians such Drs. Irl Hirsch and Rich Bergenstal, to payors, litigators, consumers, nurses, and patient advocates. Dr. David Klonoff closed the meeting with his view that the diabetes community wants better analytical and clinical performance from meters, better incorporation of factors which can improve SMBG above and beyond meter and strip performance, better labeling and reporting of interfering factors, and more consensus on glycemic control for patients in the hospital. His closing remark was quite powerful for some: “You can’t always get what you want, but if you try some times, you find you get what you need,” borrowed of course, from Mick Jagger. We believe coming out of this meeting that the FDA is likely to move toward 15% accuracy for home blood glucose monitors and for hospital meters, with the bounds for hypoglycemia not yet clear; we believe that some improvement will undoubtedly be required for hypoglycemia in the hospital, which underscores the need for CGM in the hospital even further. This directional guess is our best guess only; by no means did the FDA announce this was its plan or expectation.

 

 

Highlights

• Irl B. Hirsch, MD, (University of Washington Medical Center, Seattle, WA) served as moderator for the session of the meeting concerning tight glycemic control. He introduced the session by reminding the audience of the historical legacy of the current discussion—we have come a long way since the days of fractional urinary testing. He reviewed the context of the current tight glycemic control debate, briefly discussing the major studies informing the discussion (Leuven I & II, work by Dr. Krinsley and Dr. Furnary, and NICE SUGAR). While hypoglycemia has been consistently associated with tight glycemic control and there is clear consensus that hypoglycemia is detrimental, the tight glycemic control debate reaches much deeper complexities than is implied in this simple analysis. Dr. Hirsch introduced the question of whether or not we need better technology in order to truly evaluate the impact of tight glycemic control (by uniform and consistent application in avoidance of hypoglycemia) and questioned if there is a future role for CGM in the ICU. He brought his introduction to a close by listing several questions on which the FDA hopes to gain insight throughout the session, outlined below.
• Carol Benson, MS, MT (ASCP), (FDA Division of Clinical Chemistry and Toxicology Devices, Gaithersburg, MD) discussed the current regulatory challenges facing the agency when it comes to SMBG device review. She stressed that there are consequences stemming from the fact that SMBG manufacturers seek FDA clearance for marketing these devices for both lay users and HCPs. Under CLIA provision 493.5, all devices cleared by CLIA for home use will be waived for further review (meaning, if a device is approved for home use, the device can then be used in any setting without any further review required by the FDA). Interestingly, the accuracy required for the CLIA waiver application is tighter than the requirements of ISO 15197 that the FDA uses (95% of results must be within for values greater than or equal to 75 mg/dl or mg/dl for values under 75 mg/dl). The 5% of values not discussed in ISO requirements is considered in the CLIA application process and manufacturers are required to perform flex studies in which the device is stressed to potential user-generated extremes. Regulatory challenges arise when meters designed for use by health care providers only are cleared for lay-users when manufacturers choose the CLIA route. The FDA applies the ISO- 15197 accuracy criteria to all other reviewed meters that do not go through the CLIA waiver and the trial requirements are slightly more robust in this review compared to CLIA. Regardless of the regulatory route, the labeling for the device will be the same for both lay-users and HCPs, with an additional section added for HCPs only. At initial approval, three lots of the system are required to be tested; then after the system is on the market it is at the manufacturer’s discretion to test lots. There is major concern that the end-user may not perform quality control as diligently as suggested, unless quality control lockouts are built into the system. There are also unique issues for meters used in the hospital such as that labels are rarely kept with the mobile devices in the hospital setting and they can spread infection if not cleaned properly.
• Irl B. Hirsch, MD (University of Washington, Seattle, WA) delivered a very important reality check to the meeting—it is understandable that hypoglycemia is an overwhelming immediate concern; however, we cannot allow our concern over hypoglycemia cause us to ignore the very clear detrimental impact of hyperglycemia, which appears to be a significantly more prevalent problem. To drive home this point, Dr. Hirsch cited a study of nearly 3000 patients collected during a three-year review of a tertiary care hospital (Journal of Hospital Medicine, 2007), suggesting that 25% of patients have a mean blood glucose of 200 mg/dl in the first 24 hours after admission, while 20% of patients maintain glucose over 200 mg/dl for their entire stay, and 21% of patients have glucose above 200 mg/dl within the 24 hours before discharge (several of which were over 300 mg/dl). In the same study, less than 1% of bedside measures were less than 60 mg/dl and o.2% were less than 40 mg/dl. Dr. Hirsch also reviewed striking data indicating that not only subcutaneous insulin, but sliding scale, is still being used in major hospitals—it appears that intravenous insulin infusion is an underutilized tool in glucose management in the hospital. In general, Dr. Hirsch believes insulin is not being used intelligently in the hospital and hyperglycemia is nonchalantly treated. The key to good glycemic control is a well-crafted protocol, supported by all clinicians and stakeholders. Beyond a particular protocol, it is key that all stakeholders invest in a similar insulin use philosophy and remain committed to open communication and ongoing education. Dr. Hirsch went on to describe the evolution of his own university hospital’s insulin philosophy and success in achieving good glycemic control in hospitalized patients. He concluded by emphasizing that despite a current concern about hypoglycemia, the real problem remains untreated hyperglycemia in the hospital. He believes this situation stems from a lack of attention to glycemia in general and intimidation of insulin use due to a lack of training and concern over hypoglycemia. He also noted that although the more accurate, the better (and he called specifically for more accurate meters), the issue becomes complicated because if meters could get to 95% accuracy but it took five minutes to perform the test, it would not be worth it due to the value of nurse’s time. In the panel discussion, Dr. Hirsch called for coming up with an agreed upon set of definitions, particularly for terms like tight glcyemic control and hypoglycemia.
• Richard Bergenstal, MD (International Diabetes Center at Park Nicollet, Minneapolis, MN, President of Medicine and Science of the ADA), began his talk bygiving context to the ADA’s current recommendation for SMBG measurements to be accurate within 5% of true measures. This recommendation, published in 1993, followed a 1986 recommendation of 10%, which was not met in the seven-year interim. Dr. Bergenstal clarified that the results of DCCT, released in June of 1993, informed this recommendation with its clear message that good glycemic control can improve eye, kidney, and nerve disease, but also increases the risk of hypoglycemia. He reminded the audience that it is clear the majority of diabetes patients are not achieving good control in the US. While Dr. Bergenstal is not sure whether SMBG error significantly contributes to poor control or not, he argues that we need to be using SMBG more in concert with lifestyle and therapy adjustments. He believes endeavoring to encourage good use of this data is equally as important as improving accuracy of the devices. In his opinion, there is clear room for improvement in accuracy for meters in the hospital setting, but it is not the only way to improve glycemic care in the hospital. After alluding to the high prevalence of dysglycemia in the hospital setting, Dr. Bergenstal briefly reviewed the work of Drs. Furnary, Krinsley, and Van den Berghe, as well as the results of the VISEP and NICE SUGAR trials, to highlight the dangers of both hyper- and hypoglycemia. He emphasized that good insulin infusion protocols are critically important for maintaining good control, as is effective use and training on the protocols and tools used in glycemic control in the hospital. The first step to protecting against hypoglycemic in the hospital, in Dr. Bergenstal’s opinion, is to consider tighter accuracy standards for monitoring glucose in the hospital—he does not have great confidence in the currently available technology’s ability to appropriately detect hypoglycemia. He concluded by discussing the recently CE Mark approved in-hospital intravascular glucose monitoring system by Edwards Lifesciences and DexCom, GlucoClear. If this system continues to prove itself safe and accurate, Dr. Bergenstal believes it could hold great promise for helping the medical community to determine if striving for normoglycemia in the hospital is appropriate. Dr. Bergenstal’s final words were perhaps the most important: clinicians, and educators often do not look to guidelines published by the FDA and CLIA to guide their practices; it will be important to partner with other key diabetes organizations to spread the word. Notably, when asked why the ADA hasn’t written a letter to FDA like AACE has, asking for a 10% accuracy threshold, Dr. Bergenstal said that ADA is thinking about the best way to advocate for both better meter accuracy and better use of the glucose data available to patients and clinicians. Perhaps it is not a letter to the FDA but a group that could take the important information generated from this FDA meeting and the many questions raised and help translate them into meaningful action steps to improve hospital and outpatient use of glucose monitoring data. ..
• Although not a scheduled speaker, we thought Dr. Tony Furnary’s comment during the session about not losing sight of the importance of tight glycemic control was very timely. Effectively, Dr. Furnary said that while it’s important the FDA hear the disadvantages of hypoglycemia, they should also recognize that many lives can be saved by avoiding hyperglycemia. “But even worse are the deaths that have occurred due to hyperglycemia. For patients with diabetes the mortality rate is 3.6%, that is six million patients (cardiac patients). In Portland, our mortality was 0.9%. Either I’m the best surgeon in the world or the fact is we keep our patients at 70-110 mg/dl and that made the difference. If you calculate the number of patient’s lives saved over 10 years there are 150,000 (2.7% of 6 million) that would have suffered from hyperglycemia and died from it. We cannot lose sight of that. That is a huge difference, way bigger than 100 deaths associated with SMBG use. We were able to accomplish this with current hand held meters. With a 20% error, we did not kill anyone with hypoglycemia. I think it’s fantastic that we get better meters, but it’s not just the meter. You cannot put the entire weight of tight glycemic control on the meter. The protocol matters tremendously. Less than 5% of patients go below 60 mg/dl and the average time below 60 is 30 minutes with our protocol. I know you cannot regulate a protocol but you also cannot put the entire weight of this burden on the meter.”
• Dieter Mesotten, MD, PhD (University Hospitals Leuven, Leuven, Belgium) began by remarking that he felt like he imagines Bill Gates would feel if he were asked to promote Apple’s iPad, as he was asked to present a discussion on why tight glycemic control may not be appropriate in the hospital. Really, Dr. Mesotten works with Dr. Greet Van den Berghe and firmly believes there are benefits to tight glycemic control for critically ill patients, albeit, under very specific conditions. He reviewed the well-known Leuven I & II, VISEP, GLUCOCONTROL, and NICE SUGAR studies, focusing on the key differences between the trials. Dr. Mesotten’s message was that the success of tight glycemic control in the Leuven studies was achieved in a very specific setting. With only three sites and a very high study inclusion rate (95% of the SICU, 60% of the MICU, and 68% of the PICU participated), the hospital staff participating in the study became very skilled and experienced in the protocol early in the study. Furthermore, Dr. Van den Berghe’s commitment and passion, along with the extreme dedication of the nursing staff, cannot be overlooked. In the Leuven studies, blood gas analyzers were almost exclusively used. In both studies, blood was only taken from an arterial sample. Finally, insulin infusion was given through a central line syringe pump. The other studies lacked several of these specific conditions; notably, NICE SUGAR did not specify what blood glucose meters were used and there was great variability throughout the study. The key to understanding what makes good glucose control, in Dr. Mesotten’s opinion, is understanding what is happening at the bedside—protocols, algorithms, and nurse experience all make critical differences. He went on to describe the variability inherent in blood glucose measurements by point of care devices, highlighting the “spread” inherent in these meters exceeds the upper and lower parameters of narrow control targets (this concept of overlap has been discussed extensively in criticisms of NICE SUGAR). Dr. Mesotten boldly concluded that capillary blood sampling and all current point of care meters are inadequate for achieving tight glycemic control and that they do not belong in critical care units or ICUs. While he is strongly in favor of the scientific concept of tight glycemic control, he is not comfortable with the current methodology. In the panel discussion, he noted that most of his colleagues in Europe were using 140-150 mg/dl as an upper limit for tight glcyemic control and that the lower limit was 80 mg/dl, concluding with “… They would never tolerate 180 mg/dl.”
• Ellen H. Ullman, MSW (Children with Diabetes Foundation/Close Concerns, Boca Raton, FL), patient advocate, presented the results of an online survey that she conducted to gather data on how meters are used at home and what consumers look for in a meter. There were 512 surveys collected, with 36% of the responders type 1 patients, 49% of responders with a child with type 1 diabetes, 13% with type 2, and 3.2% with LADA. 77% of responders were female. Just 63% of patients replied that they had been trained how to use the meter by a HCP. The majority (90%) of the sample had insurance coverage. Several interesting findings were revealed about the typical practices of these SMBG users in this unscientific survey. We take these results with the full knowledge that those likely to respond to such a survey and the population composition suggests these patients are likely more educated and proactive about their management than the typical patient of diabetes. The majority of patients were checking their blood sugar between 5-12 times per day. Strikingly, 41.1% of the participants said they never use control solution, while another 14.5% said they only use it once per year. 57% of respondents replied that they had forgotten to change the code on their meters in the past and 20% had knowingly used expired strips. Only 20% of respondents wash and dry their hands every time they test. Over one-third of patients keep strips or their meter in the car, 62% of respondents did not know the temperature range of their meters, and 33% did not know the low to high range oftheir meters. Over 45% of patients have had to use more than one strip when testing because they did not trust the accuracy of the first result. As far as how consumers acquired their devices—30% acquired the device from health care providers, 27.2% directly from a pump company, and 22.2% at a local pharmacy. When asked what they considered most important regarding features of their device, 77.5% responded that accuracy was most important – patients could choose “most important” among various factors and were not limited to one; blood sample size and cost of strips were also highly important to consumers. Strikingly, almost half (46%) of patients believe their meters have error below Representing patients, Ullman advocated for accuracy for all (inside the hospital and out) within 10%, better labeling (including tools to help patients compare meters), and for old meters to be phased out once better ones become available. We note we would like to better understand tradeoffs between accuracy improvements at various levels (15%, 12.5%, and 10%). One commenter at the end noted that mail order houses are a major problem because they do not educate well regarding issues like control solution.
• Jim Rollins, MD, PhD, MSHA (Director Centers for Medicare and Medicaid Services, Baltimore, MD) delivered an overview of Medicare’s position on home glucose monitoring. Blood glucose monitoring devices and supplies fall under the jurisdiction of Social Security Act 1862 1a as durable medical equipment (DME). The device is judged on the basis of its ability to improve health outcomes for the Medicare population and for its generalizeability to the Medicare population. According to Dr. Rollins, there are forty four million beneficiaries of Medicare, 85% of which are 65 and older, 14% on disability, and 1% with end- stage renal disease. There are two levels of Medicare coverage: national coverage determination (NCD) and local coverage determination (LCD). The NCD and LCD processes are fairly similar and are evidence based. Peer reviewed literature is relied upon for decision-making. After the review process, which usually takes nine months, a memorandum is produced discussing coverage or non-coverage. Typically, the NCD defines the device that can be used as well as patient criteria, whereas the LCD defines accessories, supplies, and utilization criteria for the device. Currently, insulin dependent patients are allowed 400 test strips per three months while non-insulin dependent patients are covered for 100 strips per three months (we note that Dr. Rollins said “per month” but we assume he meant “per three months”). Dr. Rollins noted that patients could obtain greater coverage if valid clinical documentation is submitted by the treating physician/provider. According to Dr. Rollins, Medicare spending on blood glucose monitoring devices and supplies reached $2 billion in 2009; we note that although this would appear to be a very high percentage of global and US revenue (just under 25% of global blood glucose monitoring revenue and approximately 60% of US blood glucose monitoring revenue), this includes payments to Medco (Liberty), pharmacies, etc. We do believe that such payments will continue to be under pressure; one does wonder why the government wouldn’t try to contract with glucose monitoring companies directly, since Medco margins are probably above glucose monitoring company margins at this stage, which seems wrong given the value is only pass- through. When asked about reimbursement of in-patient blood glucose monitoring, Dr. Rollins said only that he assumed that reimbursement would be part of the DRG (“that’s all I can say.”) When asked about CGM coverage for the Medicare audience, Dr. Rollins said that at this time, decisions about CGM were made at the level of local contractors, and that CGM coverage was for only a subset of patients who were on pumps, though “72-hour use” could be available “to guide therapy or investigate nocturnal hypoglycemia. But that decision is not on a national level. You have to justify it at a local level.” When asked about the use of CGM and insulin dosing, Dr. Harper said that the encouraging development was in artificial pancreas progress, where “…investigators are evaluating the use of CGMs in alerting patients that could be hypoglycemic, especially at night …” Dr. Hirsch added that CGM wasn’t yet ready to drive insulin dosing but he expressed enthusiasm and said “…we are getting closer.” Finally, when asked about potentially higher reimbursement for higher accuracy meters, Dr. Rollins said “that is handled by a different office.”
• Jack R. Bierig, JD (Sidley Austin LLP, Chicago, IL) delivered the keynote address of the meeting, reviewing the liability issues hospitals can face related to SMBG device use in their health care centers. Bierig discussed interfering substances through the lens of the widely publicized non-glucose specificity problem of the GDH-PQQ strip chemistry. He cautioned the FDA to take special care in the phrasing of their safety advisories and warnings, suggesting that the FDA’s safety advisory released August 13, 2009 went too far in declaring that HCPs should, “avoid using GDH-PQQ test strips in health care facilities”. Malpractice liability arises when the hospital or practitioner owes a duty to the patient, there is negligence in the performance of that duty, and that negligence is the proximate cause of the injury or death of the patient. In defining negligence as it would be related to SMBG use in the hospital setting, Bierig highlighted the ADA’s, “safe and rational glycemic management relies on the accuracy of blood glucose monitoring using point of care glucose meters, which have several important limitations.” According to Bierig, the key consideration in liability lies in these limitations of SMBG devices. Importantly, Bierig noted off-label use such as for the achievement of tight glycemic control would not necessarily constitute grounds for malpractice—as long as the use is clinically validated and the limitations of the device are observed, malpractice will not be an issue. Bierig suggested that in addition to the interfering factors discussed extensively on day one, other limitations include system limits, equipment malfunction, and patient misidentification. A critical opinion given in Bierig’s discussion was his point that all meters are not created equal—HCPs must be aware of the limitations of the particular meter (and its chemistry/technology) used in the hospital, and must make adequate provisions considering these limitations. These adequate measures will include initial and maintenance of education as well as diligent observation of quality control measures.
• Dawn Hanson, MS MLS (ASCP), DLM (St. Agnes Hospital, Baltimore, MD) discussed the steps hospitals must take in the spirit of risk mitigation as related to SMBG devices. Overall, hospitals must adhere to standards of the Joint Commission. Hanson offered insight into how SMBG use is handled in her hospital, but discussions throughout the day would suggest that there is likely considerable nuance in hospitals across the country. Hanson, like many others throughout the day, called for better labeling that keeps the device mobility and high staff turnover rate at hospital in mind. She also suggested that manufacturers take care to ensure meters can easily be disinfected without harming the meter’s function or technology. At the conclusion of her talk, she discussed a wish list she has as a nurse at a major hospital: standardized meters, smarter meters, and smarter software. Ultimately, she seeks help to improve quality control (such as color-changing solutions when the bottle of control fluid expires or individually packaged strips). In her opinion and experience, no result is better than a wrong result—meters should automatically shut down if a strip is expired, or it has not been calibrated/coded in a certain period of time, etc..
• As alluded to above, the panel discussion during day #2 was fascinating.
  • There were some differences on how to define tight glycemic control in the hospital, to be sure. Commented Dr. Tony Furnary: “In defining TGC we have to look at what outcome measure we want to improve. Just to eliminate infection, you just have to be below 180 mg/dl. If you want to affect mortality, you need to be at least below 150 mg/dl, but between 80-120 mg/dl is ideal. If you want t0 reduce transfusion below 140 mg/dl, inreducing arrhythmia we haven’t found a lower limit. We are at 70 mg/dl as a lower limit because we can get down to almost no arrhythmia with that. How tight is tight depends on the population and outcomes you are looking at.” Dr. Irl Hirsch countered that for some patient populations, tight glycemic control does not make a difference; Dr. Furnary pointed out that there were six factors affecting glycemic control in the hospital: “the meter, the pump, the insulin delivery method, the protocol, the person interacting with protocol, and the patient. Every single one of these effects how well we do glycemic control and the outcomes. The meter is just one of those.”
  • The discussion then focused on whether different targets for different populations were possible; Dr. Furnary pointed out that this was being done in Portland and that it was possible with the same protocol. Dr. Bergenstal suggested targets on the ICU vs. targets on the floor but noted that this was complicated; he did say that he thought no one should be treated to less than 80 mg/dl. Thankfully from our view, Dr. Mesotten pointed out weaknesses in the VISEP study and argued for evidence-based medicine. Dr. Harper said this highlighted the non-consensus on how to move forward. They discussed what kind of clinical trial might be useful.
  • Dr. Hirsch asked whether it might be useful to have a more expensive strip for one population vs. another, and Elizabeth Mann RN, MS, CCRN, CCNS, said that this was not necessary, that the NovoSure StatStrip enabled 10% at most levels of glycemia.
  • When questioning what to do next, Dr. Furnary said that linking clinical outcomes to approval of a CGM or POC device is not what the FDA should be about. He argued that frequency of measurement was what enabled strong outcomes from his group. “The more frequently you measure, the more you get rid of variability.” He then advocated that tight glcyemic control could be done with current devices if they were used frequently enough, but “What about everyone else? This is why we need glucose monitoring devices. Continuous information gives us much more information to prevent hypoglycemia. I would rather have a measurement every minute that has a 10-12% deviation than a measurement every hour that has a 5% deviation. This will help me steer where we need to go! How to approve a CGM device. You don’t approve it based on the outcomes, you approve it based on how good it is—based on how good it is at telling you what the blood sugar is. Most of the devices I see are at least as good as if not better than current the current POC (point of care) MARD.” Dr. Harper asked if interstitial CGM would be accurate enough, which surprised us; it is well known that interferences make this impossible currently. Dr. Bergenstal noted that there are some intravenous devices in very early release and in development. “With Edwards/Dexcom’s GlucoClear, those 19 patients were introduced to call kinds of drugs and they held up with very good MARD. That’s where I think we are headed. We need a glucose monitor on the floor.” Dr. Hirsch pointed out that CGM is dependent upon POC calibration, but Dr. Furnary said of the six systems he was familiar with, you could calibrate with a lab measurement.
  • Dr. Barry Ginsberg then noted that for patients in whom systems worked, the lag time wasn’t a big problem due to “partial duplicates.”
  • Worryingly, Dr. Hirsch said that the top complaint he heard about related to insulin therapy on the wards was that nurses didn’t have time “for the minimal … testing we do now.” From our view, from a patient perspective, this is very sub-optimal. If it is increasing workload too much, it will not be done, even when the value has been shown. Dr. Mesotten said simply that technology should improve this; we assume CGM will, and he said it still has to prove its value. When discussing a large randomized scale, CGM arose; the question of what endpoints would be used was posed.
  • The discussion floated back to the floor and the fact that one-hour measurements would not “cut it.”
• David C. Klonoff, M.D., F.A.C.P. (Mills-Peninsula Health Services, San Mateo/ University of California at San Francisco, San Francisco, CA) closed the meeting by reviewing discussions from the previous two days. He shared what he believes this meeting has communicated: the community wants better analytical and clinical performance from meters, better incorporation of factors which can improve SMBG above and beyond meter and strip performance, better labeling and reporting of interfering factors, and more consensus on glycemic control for patients in the hospital. Dr. Klonoff concluded by offering a list of ten themes that resonated for him and suggesting that that the infamous lyrics from Mick Jagger apply well here, “you can’t always get what you want, but if you try some times you find you get what you need.”

 

 

Detailed Commentary and Discussion

Keynote Address

Liability Issues in the Use of Blood Glucose Meters

Jack R. Bierig, JD (Sidley Austin LLP, Chicago, IL)

Bierig delivered the keynote address of the meeting, reviewing the liability issues hospitals can face related to SMBG device use in their health care centers. Bierig discussed interfering substances through the lens of the widely publicized non-glucose specificity problem of the GDH-PQQ strip chemistry. He cautioned the FDA to take special care in the phrasing of their safety advisories and warnings, suggesting that the FDA’s safety advisory released August 13, 2009 went too far in declaring that HCPs should, “avoid using GDH-PQQ test strips in health care facilities”. Malpractice liability arises when the hospital or practitioner owes a duty to the patient, there is negligence in the performance of that duty, and that negligence is the proximate cause of the injury or death of the patient. In defining negligence as it would be related to SMBG use in the hospital setting, Bierig highlighted the ADA’s, “safe and rational glycemic management relies on the accuracy of blood glucose monitoring using point of care glucose meters, which have several important limitations.” According to Bierig, the key consideration in liability lies in these limitations of SMBG devices. Importantly, Bierig noted off-label use such as for the achievement of tight glycemic control would not necessarily constitute grounds for malpractice—as long as the use is clinically validated and the limitations of the device are observed, malpractice will not be an issue. Bierig suggested that in addition to the interfering factors discussed extensively on day one, other limitations include system limits, equipment malfunction, and patient misidentification. A critical opinion given in Bierig’s discussion was his point that all meters are not created equal—HCPs must be aware of the limitations of the particular meter (and its chemistry/technology) used in the hospital, and must make adequate provisions considering these limitations. These adequate measures will include initial and maintenance of education as well as diligent observation of quality control measures.

• Bierig discussed interfering substances through the lens of the widely publicized non-glucose specificity problem of the GDH-PQQ strip chemistry. This example highlights how essential review of the packaging and labeling of the meter is in identifying limitations of the devices. He cautioned the FDA to take special care in the phrasing of their safety advisories and warnings, suggesting that the FDA’s safety advisory released August 13, 2009 went too far in declaring that HCPs should, “avoid using GDH-PQQ test strips in health care facilities”. This absolute declaration is problematic for facilities that may have made a major investment in that type of meter, especially if the facility has made proper provisions to recognize the limitations of the meter and use it in appropriate settings only. He balanced this criticism by applauding the same advisory for clearly listing the patients at risk for the contraindication and making specific recommendations for steps to be taken to address the risk.
• System limits, or limitations inherent to the technology, must also be taken into account by health care facilities. While the manufacturer has the responsibility to design stable system, there are technological limits disclosed by the manufacturers that HCPs have a responsibility to recognize such as temperature limitations, storage conditions, and expiration dates. Bierig noted that some systems do build in failsafe mechanisms to prevent over-ride of system limits; however, not every product has the same failsafe features. Hospitals should be aware of the system limits and fail safe features of the meters used in the facility.
• Equipment malfunction can occur and hospitals have to protect against it by conducting quality control in accordance with manufacturer instructions, or if not specified, on a daily basis. Meters and test strips fall under waived testing requirements in CLIA guidelines, which Bierig noted he does not believe is particularly appropriate, but in any case, this means that federal law does not require periodic testing. Good self-initiated quality control is necessary to protect against negligence with SMBG use in hospitals.
• Patient misidentification is one of the most serious issues a hospital must guard against. The Joint Commission requires that the blood glucose measurement is noted in the chart of the right patient to protect against inappropriate therapeutic action based on misidentification of patients.
• According to Bierig, accuracy is well within the limits of standard care if the limitations of the technology are respected. He went on to recommend the 10 commandments to protect avoid malpractice:
  • For meters using GDH-PQQ, educate staff and patients about potential for falsely elevated glucose results in the presence of non-glucose sugars and ensure the meter is not used in patients who are receiving contraindicated therapy or have a contraindicated condition.
  • Be aware of the manufacturer’s instructions for proper use, storage, and handling of strips for all meters and form policy to enforce these instructions.
  • Train all responsible personnel on the proper use of meters, provide proper documentation, and disseminate safety advisories from the FDA or changes in labeling from the manufacturer.
  • Consider incorporating safety alerts into computer entry systems, patient profiles, and charts.
  • Know the hematocrit levels at which the meter will not function properly and do not use meter on patients outside of this range.
  • Perform quality control on each meter as recommended in the package insert or once per day if not specified.
  • Consider periodic verification of glucose meter results with laboratory based glucose assays, particularly in cases where meters are being used to make decisions about glycemic control.
  • Implement a system to ensure there are no patient misidentification errors, build in fail- safe mechanisms when possible.
  • If there is any concern about using a particular meter in a particular patient, do not use the meter for the particular patient until you are satisfied that use on the patient is safe.
  • Generally know the limitations of any meters in use as discussed in the package insert and make sure that use conforms to these limitations.

 

Questions and Answers

Q: What about when there are contraindications and label indications, how does malpractice work with that? Where is the line drawn?

A: Acting contrary to a package insert is not in and of itself malpractice. But a seasoned attorney would introduce that as evidence. It would be up to the defendant to explain why, although required by the FDA, it does not reflect standard of care. The package insert would be a very important piece of evidence in that kind of malpractice suit.

Q: There have been deaths from other substance interferences such as dopamine, renal failure, etc., but the FDA did not actually release a public alert on that—how does that impact the malpractice case?

A: The fact that the agency has not released an alert would certainly be used by the defendant in the case. If the information is in the labeling though, that would be a difficult case to argue.

Q: There are several maltose-containing products that are so low dose they do not interfere with GDH-PQQ.

A: I may have gone too far myself, I accept that. Thank you.

Q: How do you define “widely publicized”?

A: I don’t define it. If there are studies out there that are reliable (published in JAMA, NEJM, Journal of Clinical Endocrinology), these will be used as the basis for defense. You have to understand these are very complicated medical questions that will be decided by a jury of people who do not have a real clue of what is going on. In law when we speak of a jury of peers, those peers are truck drivers, students, unemployed people—these individuals often do not understand the nuances and sometimes they do not even understand the basics.

Q: You are talking about hospital level malpractice, but what about on the level of the practitioner?

A: It is a real problem on the practitioner level. For those who see many people who use SMBG and have very different meters, I think it’s probably an extremely good idea for the physician to put out a one to two page information sheet for the patient, in which the practitioner generally describes various meters and points out limitations.

 

Session 3: Tight Glycemic Control

Moderator's Introduction of Session 3

Irl B. Hirsch, MD, (University of Washington Medical Center, Seattle, WA)

Dr. Hirsch served as moderator for the third session of the meeting concerning tight glycemic control. He introduced the session by reminding the audience of the historical legacy of the current discussion— we have come a long way since the days of fractional urinary testing. He reviewed the context of the current tight glycemic control debate, briefly discussing the major studies informing the discussion (Leuven I & II, work by Dr. Krinsley and Dr. Furnary, and NICE SUGAR). While hypoglycemia has been consistently associated with tight glycemic control and there is clear consensus that hypoglycemia is detrimental, the tight glycemic control debate reaches much deeper complexities than is implied in this simple analysis. Dr. Hirsch introduced the question of whether or not we need better technology in order to truly evaluate the impact of tight glycemic control (by uniform and consistent application in avoidance of hypoglycemia) and questioned if there is a future role for CGM in the ICU. He brought his introduction to a close by listing several questions on which the FDA hopes to gain insight throughout the session.

• Dr. Hirsch highlighted several questions to consider throughout the session:
  • What is the estimate of the percentage of hospitals that use tight glycemic control in the US?
  • What factors must be considered as we try to balance risks and benefits of tight glycemic control?
  • How do we balance the need for faster turn around time with more accurate results?
  • What determines if a patient will be placed on a tight glycemic control regimen or not?
  • How often do users in hospitals or at home read labeling so that they are familiar with the limitations of the meters and are hospital staff properly trained and certified in regards to these limitations?
  • How do limitations in hospitals influence the ability to follow tight glycemic control?
  • How do reimbursement issues affect the way SMBG devices are used?
  • What incentives can be developed by payors to reward more accurate SMBG device innovation?

 

FDA Perspective: Regulatory Challenges for Safe Use of Blood Glucose Meters in Hospital Settings

Carol Benson, MS, MT (ASCP), (FDA Division of Clinical Chemistry and Toxicology Devices, Gaithersburg, MD)

Benson discussed the current regulatory challenges facing the agency when it comes to SMBG device review. She stressed that there are consequences stemming from the fact that SMBG manufacturers seek FDA clearance for marketing these devices for both lay users and HCPs. Under CLIA provision 493.5, all devices cleared by CLIA for home use will be waived for further review (meaning, if a device is approved for home use, the device can then be used in any setting without any further review required by the FDA). Interestingly, the accuracy required for the CLIA waiver application is tighter than the requirements of ISO 15197 that the FDA uses (95% of results must be within for values greater than or equal to 75 mg/dl or mg/dl for values under 75 mg/dl). The 5% of values not discussed inISO requirements is considered in the CLIA application process and manufacturers are required to perform flex studies in which the device is stressed to potential user-generated extremes. Regulatory challenges arise when meters designed for use by health care providers only are cleared for lay-users when manufacturers choose the CLIA route. The FDA applies the ISO-15197 accuracy criteria to all other reviewed meters that do not go through the CLIA waiver and the trial requirements are slightly more robust in this review compared to CLIA. Regardless of the regulatory route, the labeling for the device will be the same for both lay-users and HCPs, with an additional section added for HCPs only. At initial approval, three lots of the system are required to be tested, then after the system is on the market it is at the manufacturer’s discretion to test lots. There is major concern that the end-user may not perform quality control as diligently as suggested, unless quality control lockouts are built into the system. There are also unique issues for meters used in the hospital such as that labels are rarely kept with the mobile devices in the hospital setting and infection can be spread if devices are not cleaned properly.

• The 2008 guidance published by the FDA for in vitro diagnostic devices (IVD) waiver applications requires more robust performance trials than over-the-counter lay-user CLIA applications, requiring a minimum of 360 samples, at a minimum of three study sites, to be tested by typical patients in waived settings, and tested over a minimum of two weeks. Compared to pharmaceutical programs required for new drug approvals, these requirements seem paltry. The FDA uses the minimum ISO 15197 accuracy standards in reviewing new meters. The CLIA standards are more stringent; 95% of results must be within for values greater than or equal to 75 mg/dl or mg/dl for values under 75 mg/dl. Under FDA review, the 5% of outliers not covered by ISO standards are not evaluated. On the other hand, the CLIA review considers outlier data and subjects devices to flex studies. Both sets of standards are lower than the current CAP (College of American Pathologists) requirements for lab based standards that call for 10% or 6 mg/dl (whichever is greater).
• Benson stressed to the audience that interferences are cumulative and add on top of meter inaccuracy factored into current ISO and CLIA standards. Interferences such as hematocrit abnormalities, concomitant drugs (dopamine, acetaminophen), physiological conditions (lipidemia), and environmental effects (extreme temperature, altitude, and humidity) could all each contribute their own margin of error. Ultimately, the effect is cumulative on the patient and the total error a patient is exposed to is largely unknown, but likely exceeds the ISO minimum requirements.
• Benson highlighted several perceived required improvements for safe and effective use of meters in the hospital setting: tighter accuracy criteria, technological improvements, less vulnerability to interference, reduced lot to lot variability, increased quality control lock outs, and formalized cleaning procedures to disinfect mobile meters.

 

Questions and Answers

Q: What should the standard be? Should the standard be changed?

A: We are asking what are appropriate acceptance criteria for use in hospitals? I think we believe that they need to be tighter.

Q: Can you comment on labeling of the system kits in terms of accuracy, could we have some sort of rating scale with, “proven accuracy, high accuracy, etc.”?

A: We don’t like to have manufacturers infer that their device is better than another device as those comparison studies are not usually done. We are comparing the particular device to a standardized reference and cannot make statements about direct comparisons. We include the table from ISO in the current labeling.

Q: I think the reason why you feel you need to tighten standards in the hospital setting is the severity of illness and comorbidities we see there. But I would like to submit that there are many settings outside of the hospital in which patients are equally vulnerable—stage five renal failure, long term care, and multiple co-morbid conditions. How do you justify that accuracy does not need to be improved for all of these people as well?

A: I think we are asking that all accuracy standards be tightened.

Q: Should the performance criteria be equally tight for home use and hospital use?

A:

Audience member comment: There are challenges to making that happen. The issue of how to properly educate people in their use is one we have just begun to grasp. As a person with type 1 diabetes who has to think about getting into their car, I think providing people with something that is more accurate will save lives. We need to grapple with this even though it is more difficult.

Q: Should interferences be part of the accuracy specification or not? I would argue that in an ICU setting even more strongly yes. If you are not taking interferences into account then I don’t know what we are doing in the ICU.

A: I would agree with you.

Q: I’d like a more definitive answer on labeling. If we do a submission with comparative labeling, would we be asked to remove it?

A: If we allowed that we would be getting into superiority claims. We like to use the standard recommended table from the ISO criteria. The data we typically receive do not give companies a right to claim that they are more accurate.

Q: We would urge the FDA to contemplate requiring that new systems should have 95% of readings within 12.5%. I think it is doable but companies would have to stretch to produce this product.

A: Thank you.

Q: How can we better train and communicate? Is there any evidence on what practices work well and what doesn’t work as well so that we can have some guidance about how to best go about these things?

A: We would be really interested in hearing about that at the FDA. I don’t personally have information, but we are very interested in learning more about that topic.

Q: I’m not sure it is reasonable to say that manufacturers should select the most accurate lots for the hospital and therefore lower the lots of the ones going out to home users.

A: I don’t think that has been suggested at all. I think the suggestion is that the best lots that make that meet the new standard should be the only ones to be distributed. It would be interesting what people think are acceptable criteria for manufacturing lot release.

Q: There has been some work published by the Cleveland Clinic on the issue of training. If one chooses to place the trainee in front of a computerized learning tool, they will fail much more frequently compared to personalized, four on one, hour-long training

activities. The old rule still works very well: it is best to see it done, do it yourself, and then teach it to others to really learn how to do something.

A: Thank you.

Q: We may need to have two different sets of standards. I don’t think the error rate of 12.5% would be enough in the hospital, but that doesn’t mean we can’t do better for patients in the field with this stricter level of accuracy.

A: Thank you.

 

Payer Perspective: Reimbursement Issues Associated with Glycemic Control

Jim Rollins, MD, PhD, MSHA (Director Centers for Medicare and Medicaid Services, Baltimore, MD)

Dr. Rollins delivered an overview of Medicare’s position on home glucose monitoring. Blood glucose monitoring devices and supplies fall under the jurisdiction of Social Security Act 1862 1a as durable medical equipment (DME). The device is judged on the basis of its ability to improve health outcomes for the Medicare population and for its generalizeability to the Medicare population. According to Dr. Rollins, there are forty four million beneficiaries of Medicare, 85% of which are 65 and older, 14% on disability, and 1% with end-stage renal disease. There are two levels of Medicare coverage: national coverage determination (NCD) and local coverage determination (LCD). The NCD and LCD processes are fairly similar and are evidence based. Peer reviewed literature is relied upon for decision-making. After the review process, which usually takes nine months, a memorandum is produced discussing coverage or non-coverage. Typically, the NCD defines the device that can be used as well as patient criteria, whereas the LCD defines accessories, supplies, and utilization criteria for the device. Currently, insulin dependent patients are allowed 400 test strips per three months while non-insulin dependent patients are covered for 100 strips per three months. Dr. Rollins noted that patients could obtain greater coverage if valid clinical documentation is submitted by the treating physician/provider. According to Dr. Rollins, Medicare spending on blood glucose monitoring devices and supplies reached $2 billion in 2009; this is higher than our estimates, and represents just under 25% of global blood glucose monitoring revenue and approximately 60% of US blood glucose monitoring revenue.

• SMBG devices are categorized as durable medical equipment (DME) by Medicare. DME is characterized as a device that can withstand repeated use, is used primary for medical service, is not useful to a person in absence of illness, and is appropriate in a person’s home. (As a sidenote – some people with diabetes would term diabetes a condition rather than an illness while others would call it a disease – we note that these definitions can be challenging!)
• There are two separate Medicare benefits used to cover SMBG devices: the DME benefit or the diabetic supply benefit. Prior to the Balanced Budget Act of 1997, Medicare used to only cover SMBG devices for insulin-dependent patients of diabetes, but now covers SMBG devices for non-insulin dependent patients with diabetes as well.
• The NCD for approved SMBG devices specifically states that the patient must be diagnosed with diabetes and the treating physician is required to state that the patient or caregiver has been trained to use the device and it can be safely used in the patient’s home.
• The LCD for approved SMBG devices specifically requires patients to have a diagnosis of diabetes, the supplies must be ordered by the treating physician who must maintain records of these orders, and the patient or caregiver must be trained how to use the device. The patient is also required to maintain records of the results of their home testing and must have been to see their treating physician within one year prior to the dispensing of supplies. In order to get more supplies, the physician must state that the beneficiary has nearly exhausted the supply of previously dispensed materials.

 

Questions and Answers

Q: One of the things exploding right now is the number of people with type 1 diabetes who are reaching the age 65 and entering the Medicare population. I think there are a lot of issues with this that we are not prepared to deal with. We have these patients on pumps with very frequent testing. They will likely need over the Medicare limits for strips per day, but the documentation to request more is a pain. Is there any plan to ease the burden of documentation as more of these kinds of patients enter the Medicare population?

A: CMS would be willing to consider modifying documentation requirements. If you’ve got a patient using 6-7 strips a day, does that patient need to have all that documented? There are things that can be put in place to accommodate that.

Q: Have you considered using electronic medical records of meter downloads? I plead with CMS and industry to push the technology to make this documentation much easier.

A: I think it is excellent technology that CMS along with many other commercial insurers are willing to embrace.

Q: Can you comment on the in-patient tight glycemic control with regard to Medicare payment to the hospital?

A: I cannot address that. I would make the assumption that reimbursement would be part of the DRG, that’s all I can say.

Q: You mentioned that training is needed to receive coverage. What type of training for SMBG are you referring to?

A: Working with a doctor or educator. A lot of physician offices are associated with a CDE. That is sufficient. They need to document the patient or caregiver has been trained. We do require documentation now that the patient has been shown to be proficient in using the device.

Q: Does Medicare have a policy on reimbursement for CGM at this time?

A: Actually someone in the audience might be able to answer both of those questions. One of my colleagues, Dr. Beth Kohler, has been responsible for some of the policy specifically addressing the use of glucose monitoring devices.

Dr. Beth Kohler: Medicare makes its decisions at two levels. At a national level, the decision is consistent across the country, but many decisions for Medicare are made at the level of the local contractor. This is a reflection of the historical development of the Medicare program in which there were concerns about the centralization of medical care delivery and decision-making. At this time, decision about CGM is made at the level of local contractors. The local contractors have significant amounts of experience with monitoring and our population. We have a national decision for insulin pumps. And insulin pumps go to only a limited number of patients who fit certain criteria. In general, the local contractors have provided CGM only to that subset of patients who are on pumps. They may choose to also provide CGM on a shorter interval, such a short 72-hour stay so a physician can obtain information to be able to guide therapy or investigate nocturnal hypoglycemia. But that decision is not on a national level. You have to justify it [at a local level].

Q: The current use of CGM is designed not necessarily to drive insulin dosage but as an alarm system for hypoglycemia, correct?

A:

Dr. Harper: Currently CGM’s are only approved for tracking and trending. They are not approved for any replacement of SMBG. The encouraging development is in developing the artificial pancreas, where investigators are evaluating the use of CGMs in alerting patients that could be hypoglycemic, especially at night.

Dr. Hirsch: My bottom line is we are not quite there yet, but we are getting closer. I’m excited about the future but we’re not quite ready to do that.

Q: Is there any potential for higher reimbursement for higher accuracy meters? What would that require?

A: I have nothing to do with reimbursement; that is handled by a different office, so I cannot address that question.

 

Advantages of Tight Glycemic Control in Hospital Settings

Richard Bergenstal, MD (International Diabetes Center at Park Nicollet, Minneapolis, MN)

Dr. Bergenstal, President of Medicine and Science of the ADA, began his talk by giving context to the ADA’s current recommendation for SMBG measurements to be accurate within 5% of true measures. This recommendation, published in 1993, followed a 1986 recommendation of 10%, which was not met in the seven-year interim. Dr. Bergenstal clarified that the results of DCCT, released in June of 1993, informed this recommendation with its clear message that good glycemic control can improve eye, kidney, and nerve disease, but also increases the risk of hypoglycemia. He reminded the audience that it is clear the majority of diabetes patients are not achieving good control in the US. While Dr. Bergenstal is not sure whether SMBG error significantly contributes to poor control or not, he argues that we need to be using SMBG more in concert with lifestyle and therapy adjustments. He believes endeavoring to encourage good use of this data is equally as important as improving accuracy of the devices. In his opinion, there is clear room for improvement in accuracy for meters in the hospital setting, but it is not the only way to improve glycemic care in the hospital. After alluding to the high prevalence of dysglycemia in the hospital setting, Dr. Bergenstal briefly reviewed the work of Drs. Furnary, Krinsley, and Van den Berghe, as well as the results of the VISEP and NICE SUGAR trials, to highlight the dangers of both hyper- and hypoglycemia. He emphasized that good insulin infusion protocols are critically important for maintaining good control, as is effective use and training on the protocols and tools used in glycemic control in the hospital. The first step to protecting against hypoglycemic in the hospital, in Dr. Bergenstal’s opinion, is to consider tighter accuracy standards for monitoring glucose in the hospital—he does not have great confidence in the currently available technology’s ability to appropriately detect hypoglycemia. He concluded by discussing the recently CE Mark approved in- hospital intravascular glucose monitoring system by Edwards Lifesciences and DexCom, GlucoClear. If this system continues to prove itself safe and accurate, Dr. Bergenstal believes it could hold great promise for helping the medical community to determine if striving for normoglycemia in the hospital is appropriate. Dr. Bergenstal’s final words were perhaps the most important: clinicians and educators often do not look to guidelines published by the FDA and CLIA to guide their practices; it will be important to partner with other key diabetes organizations to spread the word.

• According to Dr. Bergenstal, a recent report by the National Health Service suggests that even non-insulin using patients can potentially benefit from SMBG use, if the data derived from testing is used to inform their lifestyle and therapy choices.
• Accuracy could be a particularly important issue in the hospital setting because the population is exceptionally vulnerable. Patients with diabetes and impaired glucose tolerance compose a large percentage of patients in the hospital (typical 12-25% in the hospitalized population, but up to 50% in cardiothoracic surgery patients). Dr. Bergenstal emphasized that both high and low blood sugar are detrimental to a patient’s help, and stressed that good glucose control is critical in the hospital setting.

 

Questions and Answers

Q: The AACE wrote a letter to the FDA specifically asking for a 10% accuracy threshold. Why hasn’t the ADA considered a similar type of request? Also, in terms of requiring different levels of accuracy for different populations, it seems reasonable to have a consensus recommendation across the board to make sure everyone is properly covered.

A: The ADA is thinking about this and is interested about the output of this meeting. It is probably time for us to have this discussion again. It may be time to seriously consider that.

Q:

Dr. Tony Furnary (Portland): I think it’s important that the FDA hear what these real human advantages are, because we have heard all about the disadvantages of hypoglycemia. I hate that there have been deaths. But even worse are the deaths that have occurred due to hyperglycemia. For patients with diabetes the mortality rate is 3.6%, that is six million patients (cardiac patients). In Portland, our mortality was 0.9%. Either I’m the best surgeon in the world or the fact is we keep our patients at 70-110 mg/dl and that made the difference. If you calculate the number of patient’s lives saved over 10 years there are 150,000 (2.7% of 6 million) that would have suffered from hyperglycemia and died from it. We cannot lose sight of that. That is a huge difference, way bigger than 100 deaths associated with SMBG use. We were able to accomplish this with current hand held meters. With a 20% error, we did not kill anyone with hypoglycemia. I think it’s fantastic that we get better meters, but it’s not just the meter. You cannot put the entire weight of tight glycemic control on the meter. The protocol matters tremendously. Less than 5% of patients go below 60 mg/dl and the average time below 60 is 30 minutes with our protocol. I know you cannot regulate a protocol but you also cannot put the entire weight of this burden on the meter.

A: Thank you for your comment.

Why Tight Glycemic Control May Not Be Appropriate in Hospital Settings

Dieter Mesotten, MD, PhD (University Hospitals Leuven, Leuven, Belgium)

Dr. Mesotten began by remarking that he felt like he imagines Bill Gates would feel if he were asked to promote Apple’s iPad, as he was asked to present a discussion on why tight glycemic control may not be appropriate in the hospital. Really, Dr. Mesotten works with Dr. Greet Van den Berghe and firmly believes there are benefits to tight glycemic control for critically ill patients, albeit, under very specific conditions. He reviewed the well-known Leuven I & II, VISEP, GLUCOCONTROL, and NICE SUGAR studies, focusing on the key differences between the trials. Dr. Mesotten’s message was that the success of tight glycemic control in the Leuven studies was achieved in a very specific setting. With only three sites and a very high study inclusion rate (95% of the SICU, 60% of the MICU, and 68% of the PICU participated), the hospital staff participating in the study became very skilled and experienced in the protocol early in the study. Furthermore, Dr. Van den Berghe’s commitment and passion, along with the extreme dedication of the nursing staff, cannot be overlooked. In the Leuven studies, blood gas analyzers were almost exclusively used. In both studies, blood was only taken from an arterial sample. Finally, insulin infusion was given through a central line syringe pump. The other studies lacked several of these specific conditions; notably, NICE SUGAR did not clearly specify what protocol was used to measure blood glucose and there was great variability throughout the study. The key to understanding what makes good glucose control, in Dr. Mesotten’s opinion, is understanding what is happening at the bedside—protocols, algorithms, and nurse experience all make critical differences. He went on to describe the variability inherent in blood glucose measurements by point of care devices, highlighting the “spread” inherent in these meters exceeds the upper and lower parameters of narrow control targets (this concept of overlap has been discussed extensively in criticisms of NICE SUGAR). Dr. Mesotten boldly concluded that capillary blood sampling and all current point of care meters are inadequate for achieving tight glycemic control. While he is strongly in favor of the scientific concept of tight glycemic control, he is not comfortable with the current methodology.

• According to Dr. Mesotten, major differences exist in the design and interventions implemented in the Leuven, VISEP, GLUCOCONTROL, and NICE SUGAR studies. The blood sample was always taken from an arterial source in the Leuven studies, while in NICE SUGAR, samples would be taken from capillary, arterial, and venous sites. In the Leuven studies, the blood gas analyzer was almost exclusively used (the Hemocue device was used in the MICU in the Leuven studies) for blood glucose measurements, whereas diverse measures were used in NICE SUGAR.
• Choice of monitoring device could have significant impact on the ability to achieve tight glycemic control. Dr. Mesotten cited work by Duncan Young (ISICEM 2009) investigating the 95% confidence intervals of blood gas analyzers. The most accurate measures can be obtained by a blood gas analyzer, which has a confidence interval of 14 mg/dl. The second most accurate are HemoCue devices, with a confidence interval of 19.8 mg/dl. Laboratory measurement has a confidence interval of 21.6 mg/dl. Strikingly, point of care over the counter strips have an average confidence interval of 37.8 mg/dl, while mixed methods with compounded variability and error (such as was the case in NICE SUGAR) results in a confidence interval of50.4 mg/dl. These confidence intervals take on real meaning when we consider that the tight glycemic control target is typically limited to a very narrow 30 mg/dl range (80-120 mg/dl). When capillary blood samples are taken, over one-quarter of the readings will be outside of this target glucose range.
• Dr. Mesotten suggests SMBG devices may perform better in the ambulatory space than in the ICU due to a higher level of interfering factors in the ICU. In particular, Dr. Mesotten notes potassium levels need to be carefully titrated when using insulin therapy in the ICU.

 

Questions and Answers

Q:

Dr. Tony Furnary (Portland): Although I agree with your conclusion I disagree it cannot be done with current devices. Arterial is the best way to go, but we have done it other ways for 18 years now, longer than at Leuven. If anyone reads the NICE SUGAR paper – 90% of all the deaths and 91% in the control group were associated with a DNR withdrawal of care, and occurred after 90 days of admission. I don’t care how much glucose you give before 90 days, it’s not going to effect the family’s decision to withdraw care after 90 days.

A: Thank you.

Q: When I data mined measurements done within 15 minutes of arterial v. point of care, POC biased high by 15-20% would have resulted in artifactual hyperglycemia. This will be published in the Journal of Clinical Chemistry soon.

A: I look forward to the data.

Current Practice and Experiences with Tight Glycemic Control in Hospital Settings

Irl B. Hirsch, MD (University of Washington, Seattle, WA)

Dr. Hirsch delivered a very important reality check to the meeting—it is understandable that hypoglycemia is an overwhelming immediate concern; however, we cannot allow our concern over hypoglycemia cause us to ignore the very clear detrimental impact of hyperglycemia, which appears to be a significantly more prevalent problem. To drive home this point, Dr. Hirsch cited a study of nearly 3000 patients collected during a three-year review of a tertiary care hospital (Journal of Hospital Medicine, 2007), suggesting that 25% of patients have a mean blood glucose of 200 mg/dl in the first 24 hours after admission, while 20% of patients maintain glucose over 200 mg/dl for their entire stay, and 21% of patients have glucose above 200 mg/dl within the 24 hours before discharge (several of which were over 300 mg/dl). In the same study, less than 1% of bedside measures were less than 60 mg/dl and o.2% were less than 40 mg/dl. Dr. Hirsch also reviewed striking data indicating that not only subcutaneous insulin, but sliding scale, is still being used in major hospitals—it appears that intravenous insulin infusion is an underutilized tool in glucose management in the hospital. In general, Dr. Hirsch believes insulin is not being used intelligently in the hospital and hyperglycemia is nonchalantly treated. The key to good glycemic control is a well-crafted protocol, supported by all clinicians and stakeholders. Beyond a particular protocol, it is key that all stakeholders invest in a similar insulin use philosophy and remain committed to open communication and ongoing education. Dr. Hirsch went on to describe the evolution of his own university hospital’s insulin philosophy and success in achieving good glycemic control in hospitalized patients. He concluded by emphasizing that despite a current concern about hypoglycemia, the real problem remains untreated hyperglycemia in the hospital. He believes this situation stems from a lack of attention to glycemia in general and intimidation of insulin use due to a lack of training and concern over hypoglycemia.

• Hyperglycemia remains a persistent problem in the hospital, stemming from unintelligent use (and fear) of insulin inspired by concern over hypoglycemia. Dr. Hirsch discussed a study by Board et al., (Journal of Hospital Medicine, 2009), suggesting that hyperglycemia is quite common in the first three days of a patients stay in the ICU—up to 70% in patients treated with subcutaneous insulin and 50% in patients treated with intravenous insulin. Patients outside of the ICU might be doing even worse. A surprising 23% of patients did not receive a glucose measurement within eight hours of admission in this study. Another paper (Journal of Hospital Medicine, 2006) suggests sliding scale insulin was still being used in 90% of patients in a major teaching hospital in Boston, despite the fact that sliding scale insulin has been associated with poor glycemic control and more hypoglycemia.
• Dr. Hirsch described the evolution of the culture of insulin therapy at his hospital at the University of Washington. After several “near-misses” in 1992 associated with sliding scale insulin, the culture of insulin use began to change. Between 2001-2002, six insulin infusion protocols were developed and eventually, a decision was made to standardize all insulin orders. Remarking that Dr. Van den Berghe’s targets were too ambitious for his hospital, Dr. Hirsch noted that his hospital decided to target 100-180 mg/dl with both intravenous insulin infusion and subcutaneous insulin protocols (insulin infusion used mostly in the ICU, subcutaneous more common in ambulatory population). For a number of years, this protocol consisted of the column method. With this culture of insulin use in place, he saw dramatic reductions in hyperglycemia in both the ICU and the non-ICU patients, with hypoglycemia falling below 5% in both settings. Dr. Hirsch’s message in outlining this history is that the most important tool a hospital can use in treating glycemia is to teach staff how to think about insulin in both inpatient and outpatient settings. In comparing the protocol used at his hospital to the Glucommander computerized protocol, average glucose was found to be 117 mg/dl in the UW protocol and 103 mg/dl in the Glucommander protocol. The Glucommander appears to perform better when glycemia reaches 200 mg/dl and above, but there was very little difference in hypoglycemia. He emphasized that these results are unique and that his institution has come a long way in reaching this point and there is still room for improvement.

 

Questions and Answers

Q: I have a problem with you calling tight glycemic control 100-180 mg/dl. That is not really tight, and you are tolerating a lot of hyperglycemia. We have to improve methodological techniques to compare moderate control to actual tight glycemic control.

A: Your point is well taken. We have a problem with nomenclature here. When you look at the data especially in the ICU you see most people are above 180 mg/dl, or actually above 200 mg/dl. If you look at tight glycemic control of 80-110 mg/dl, and compare to the real world, I think my target is more reasonable. If we try to get where you recommend I don’t think it’s realistic with our current tools and with our current understanding of how to do this—I think it is too difficult. I’m in total agreement with you in many ways.

Q: The group I work with has identified patient data from over 500 hospitals using a particular SMBG on the market right now and our data supports you. Our n= 50 million, 29 million of which are in the ICU. We did a poster on this for Dr. Klonoff’s DTT meeting in San Francisco. We are seeing a higher percentage of glucose values over 180 mg/dl, about 25% of patients are just sitting in the hyperglycemic range. We are seeing hyperglycemia as a major problem and hypoglycemia is not as big of a problem. We are seeing a median range of 150-160 mg/dl for all these hospitals

A: I appreciate you comments, thank you.

Q: What type of SMBG are you using? Do you think that current SMBG with the accuracy they have are sufficient for the type of results you are getting, how much better do they need to get?

A: Our hospital is using the Roche Accu-Chek system. I’d like to see them become more accurate, as Dr. Furnary’s point is a very good one. I’m putting up a flag for the issue of GDH-PQQ, there are little pockets in the hospital that sometimes have problems. In the cath-lab I had a patient on a contraindicated maltose drug going in and it was almost not corrected. Independent of interference I would definitely like to see better accuracy. The 20% for those sensitive to 1-2 units of insulin is a big and real concern. Insulin sensitivity is key. The more accurate, the better. Where the issue becomes complicated is in trade-offs. How long will it take and how much will it cost to become more accurate? The time it takes the nurse to perform the test is a big issue in the hospital—that wasted time is more costly than the strip itself. If we could get accuracy down to 5% but it took five minutes to perform the test, it might not be worth it. All this has to be taken into account.

 

Session 3 Panel Discussion

Richard Bergenstal, MD (International Diabetes Center at Park Nicollet, Minneapolis, MN), Irl B. Hirsch, MD (University of Washington, Seattle, WA), Carol Benson, MS, MT (ASCP), (FDA Division of Clinical Chemistry and Toxicology Devices, Gaithersburg, MD), Dieter Mesotten, MD, PhD (University Hospitals Leuven, Leuven, Belgium), & Patricia Beaston, MD, PhD (FDA, Office of Device Evaluation, Gaithersburg, MD)

Questions and Answers:

Q: What are your top-level concerns about what you heard this morning?

A:

Dr. Beaston: There doesn’t seem to be a consensus on what tight glycemic control consists of. What are the goals will help to inform the needs for performance. There doesn’t seem to be a consistent use of hypoglycemia when these trials are done. Certainly they can use different targets, but there needs to be standardization to compare outcomes. I understand that nurses are really busy and you don’t want to have to wait five minutes, but that is an extreme, what if you are trading 5-10 seconds to one minute for a jump from 20% to 10% accuracy? If you don’t have time to do it right the first time, how do you have time to do it over? If you’re not sure the measure is any good, you have to spend time to repeat it, what is the trade off for that time lost? There are a bunch of clinical issues going into how this will be figured out that people haven’t even discussed yet.

Dr. Hirsch: We are talking about nomenclature, definitions. We don’t call what we do tight glycemic control (TGC), but it has been popularized as such. We need to make sure when we are talking about a topic that we are all talking about the same thing. There is a definition of biochemical hypoglycemia at 70 mg/dl, severe hypoglycemia “requires assistance of another person”. We cannot find where the definition of severe hypoglycemia in the ICU being associated with less than 40 mg/dl originated from. I think as a group we need to come up with an agreed upon set of definitions. I’m curious if anyone else has comments on that point?

Dr. Mesotten: When we talk about TGC, we have to go back to all the clinical trials that have been done. When we talk about tight control, tight means a very narrow range that is 80-110 mg/dl, all the others are intermediate. We determine this by looking at the lowest risk association in the J-shaped curve of glycemia and mortality.

Q:

Dr. Hirsch: We looked at some of the US data on how we deal with control, what is your impression of TGC in the EU or your country?

A:

Dr. Mesotten: I don’t want to go into a politically incorrect comparison. We are definitely in favor of TGC. Most of my colleagues in Europe modify TGC, move upper limit to 140-150 mg/dl, but leave lower limit at 80 mg/dl. They never tolerate 180 mg/dl.

Q:

Dr. Hirsch: You will measure twice a day even for someone receiving four shots a day in the hospital?

Dr. Mesotten: In clinical practice yes. I agree it is not appropriate, but yes, if you are not on the surgical ward, that is what happens.

Dr. Furnary: This country is different than Europe. The US has average value of 161 mg/dl in the hospital, for Europe the average is in the 130-140 mg/dl range. In defining TGC we have to look at what outcome measure we want to improve. Just to eliminate infection, you just have to be below 180 mg/dl. If you want to affect mortality, you need to be at least below 150 mg/dl, but between 80-120 mg/dl is ideal. If you want t0 reduce transfusion below 140 mg/dl, in reducing arrhythmia we haven’t found a lower limit. We are at 70 mg/dl as a lower limit because we can get down to almost no arrhythmia with that. How tight is tight depends on the population and outcomes you are looking at.

Dr. Hirsch: For some patient populations, TGC does not make as much difference.

Dr. Furnary: There are six different things affecting glycemic control in the hospital: the meter, the pump, the insulin delivery method, the protocol, the person interacting with protocol, and the patient. Every single one of these effects how well we do glycemic control and the outcomes. The meter is just one of those.

Dr. Hirsch: The definition of TGC is relevant. From what I’m hearing, is it possible that TGC is dependent upon the population being discussed so those numbers could also potentially vary depending on the population.

Dr. Furnary: Yes, we saw this in the VISEP trial. Patients with sepsis are at higher risk for hypoglycemia. One of our target levels is for sepsis, and that is 130-165 mg/dl, because they are more susceptible to hypoglycemia.

Q:

Dr. Hirsch: Are we sophisticated enough as a medical community to have different targets for different populations? I don’t think we do that very well on the outpatient side.

Dr. Furnary: We do it already. Yes you can have different target ranges, but you must have the same protocol.

Dr. Bergenstal: Can we say ICU v. floor, perhaps? But it’s different based on patient type, not just unit and floor. I hope we could sort that out. We are struggling with that on the outpatient side as well. We should not treat anybody to less than 80 mg/dl I think.

Dr. Beaston: I took care of a lot of transplant patients on my fellowship, they were all very different patients. You had to figure out type of transplant, fed v. not fed, stage of immunosuppression. It gets very complicated. For people who do this all the time you can get into tune with what patients need. With the understanding if something changes, you must be made aware (changes of immuno-suppressant drug, for instance). It’s not a one-size fits all, but it’s a process you need to have standardization with the ability to individualize care.

Dr. Mesotten: We need to think scientifically. There is a difference in RCT from implementation studies. From implementation studies it would be difficult to identify patients who need different targets. VISEP was underpowered in general. We can’t draw conclusions. Non-evidenced based conclusions. From a general point of view I think it is very dangerous to go back to the years when the physician decides what is best on a case-by-case basis. I think we need to use evidence-based medicine. We should really think scientifically, a randomized trial is much stronger.

Q:

Dr. Harper: This discussion highlights the importance of what is needed in this area. Evidence-based medicine is obviously the highest bar and highlights non-consensus in the area on what to do. Hopefully we can come out of this meeting with some consensus. The reality for us is that we are being asked what studies are necessary to show my device would be useful for this situation. We would like input on how we can try and demonstrate the performance on a level that is necessary for this patient population and for this use. What types of studies are really needed here?

Dr. Beaston: We cannot assume any patient is represented by any one clinical trial.

Q: The issue about individual patient targets is exactly right, different patient populations definitely need different targets. We have been working with a computer support system. I think it is possible to program them today for different populations of patients. What is the best target? With our hematocrit correction, we have stopped having occult hypoglycemia and minimized glycemic variability. The accuracy of a glucometer is very valid and absolutely necessary for safe glycemic control.

Q:

Hirsch: Would you be willing for one set of patients to use a more expensive strip that takes 3-5 minutes, but in another population not that tight control, using a meter of different accuracy?

A:

Elizabeth Mann, RN (Army Burn Center): I’ve tested all five available glucometers. The new four-channel glucometer (NovoSure StatStrip) is almost identical to the reliability of the laboratory test and it takes only six seconds. You can correct the other glucometers and get within 10% at most levels of glycemia. We can do this now and there is no extra time and no extra cost.

Dr. Beaston: Are you using capillary blood for a sample?

Mann: No, we only use arterial or venous blood in the burn center.

Q: We’re curious, what do we need to have developed that we don’t have now, or do we have what we already need in front of us?

A:

Dr. Furnary: I want to restate I don’t think that the focus of these two days is proof of concept or disproof or what target level TGC should be at. I think linking clinical outcomes to an approval of a CGM or POC device is not what the FDA is all about.

Dr. Harper: That depends on what manufacturers are trying to claim. If someone comes in and says this device is to do TGC in the ICU, then yes we do look at that. The artificial pancreas is a great effort, but there are stages in between that may come more quickly and we have to prepare for that. We need to figure out what the best studies will be to evaluate that technology. I’ve heard conflicting statements here.

Some people say we have too much data and it doesn’t help the algorithms. Others say if we have the rate of change and trend, that it helps. We need to understand what is best.

Dr. Furnary: What is different about what Dr. Van den Berghe does and what I do from everyone else? The reason is frequency of measurement. With point of care, the most frequent is every hour. In Portland, if a patient gets low their blood sugar is being measured every 30 minutes. The more frequently you measure, the more you get rid of variability. Close to 15 minutes apart, the cluster analysis of what goes on gives you a better reading. The more frequently you measure you will eliminate imprecision. Yes, TGC can be done with current devices if used frequently enough. What about everyone else? This is why we need glucose monitoring devices. Continuous information gives us much more information to prevent hypoglycemia. I would rather have a measurement every minute that has a 10-12% deviation than a measurement every hour that has a 5% deviation. This will help me steer where we need to go! How to approve a CGM device. You don’t approve it based on the outcomes, you approve it based on how good it is—based on how good it is at telling you what the blood sugar is. Most of the devices I see are at least as good as if not better than current the current POC (point of care) MARD.

Q:

Dr. Harper: CGM may need to be equally accurate, but it may need to tolerate more imprecision?

A:

Dr. Hirsch: The CGM is dependent upon the POC calibration.

Dr. Furnary: I think out of the six I am familiar with, it is a correct statement, but only correct if you get your baseline to calibrate the device. You can calibrate by a lab measurement. You can calibrate with anything you want to calibrate with.

Dr. Harper: Do you think the interstitial CGMs are going to be accurate enough for this use or do we need a whole blood CGM?

Dr. Furnary: I think we need whole blood honestly, what we have now for interstitial has been out for 5-6 years. We haven’t used them.

Dr. Bergenstal: There are some intravenous devices in very early release and in development. With Edwards/Dexcom’s GlucoClear, those 19 patients were introduced to all kinds of drugs and they held up with very good MARD. That’s where I think we are headed. We need a glucose monitor on the floor.

Dr. Hirsch: Yes if you are doing intravascular, you can have fewer problems than with interstitial testing.

Dr. Ginsberg: A couple years ago I presented at DTT talking about partial duplicates. Since CGM is a frequent measurement, the values we get every 5 minutes are not independent. Patient’s blood glucose does not change that quickly. If I get a value now of an error 15% and another in 5 minutes with 15% error, the error together is now only 11%. Five minutes from then error goes down 9%. Plus I’m getting a trend and error is going down. For patients in which continuous systems do work, I don’t think a 15 minute lag is a problem. It’s something to think about in how we judge these systems.

Comment from audience: I think we need to figure out what technology the rest of us need to come close to doing such a good job as Dr. Furnary. I agree that the more frequent the testing, assuming it is reliable, the closer we will be able to do this. CGM is probably the way to go and I think vascular monitoring is the way to go in that setting; totally different for the setting of the average diabetic walking around on the street.

Dr. Hirsch: The number one complaint I hear about when we talk about insulin therapy on the wards they tell us they don’t have time for the minimal amount of testing we do now.

Dr. Mesotten: I agree with you. There is ample literature suggesting that measuring at least very hour is very harmful to nurses and it increases their workload too much. I think we have to improve the technology to help that. CGM still needs to prove its value in the clinical setting. In the meantime we have to do something. There is still disagreement there.

Q: I agree science tells all and a large randomized trial can do a lot. What is a large randomized trial? Could we put a large scale randomize CGM on non diabetic patients and figure out what is normal and what we should be striving for and use this technology? What about using standard deviation and variability?

Dr. Hirsch: if we were going to do a CGM study in the ICU, would the endpoint always have to be mortality? I’m very concerned once we start doing prospective randomized, multi-center CGM we have to be very careful where we set the bar. The power to see the number of mortalities would be so high. There are other endpoints than mortality and I think we are talking about hypoglycemia as an outcome. We need to have hard endpoints—variability is a statistic.

Dr. Furnary: This is something we do in the ICU. I disagree with that. Glycemic control is not just for patients in the ICU. A lot of data shows that the length of time you are hyperglycemic, the length of time, impacts outcomes. To limit glycemic control to a patient population in the hospital is like saying it doesn’t matter “out there”. It would be like giving less than a full course of antibiotics. There is a duration of care that matters. I believe that glycemic control should be afforded to all patients including those on the floor. Speaks to why we need a very accurate CGM on the floor so nurses’ time is not taxed. That is what I think we need.

Dr. Bergenstal: I think we definitely need control on the floor, it’s just a question of what the level of control is. We probably are as aggressive as any place with our triple insulin orders on the floor, and we can get 110-150 mg/dl on the floor.

Dr. Furnary: I want the FDA to hear this is not an ICU-only therapy. Let’s just call it glycemic control.

Q: At Medicare, we frequently have sponsors present about the products. The sponsor may also be the vendor for the DME. The info may be derived through those vending activities. The data stream basically appears to be through the sponsor plus or minus the DME vendor. Would the FDA accept data if in fact it were not obtained having gone through an IRB protocol and there was no patient consent?

A:

Dr. Harper: Our studies require IRB and informed consent on human trials.

Q: What about the perspective outside of a University setting? We use the Portland protocol in cardiothoracic patients and our biggest barrier is our glucose meter. It has a lower hematocrit limit of 25% and our patients often do not meet this threshold. The lab turn around we can maybe get it in 5-10 minutes, but once they get out to the floor then it’s a phlebotomy stick and we can’t run insulin drips on the floor for that, they’re getting stuck eight times a day. We need something different. Because our current glucometers at face value meet ISO criteria, there is no incentive to get new technologies, especially in this economy.

Dr. Hirsch: You are saying we need better SMBG that are more accurate for measuring anemic patients? Getting back to Beaston’s question about what is an acceptable increase in time? I don’t know the right answer. Ideally we would like it to be cheap and we would like it to be fast. But there is a real problem, if the nurse can’t use the glucose meter at the bedside due to hypo. [This is] … the most provocative question in the last hour. The real answer is going to be with administrators and not clinicians like myself.

Comment: There is one thought I have, from the hospital side. I have to say there are a lot of good meters out there. The lab directors know about those meters. They are below the 10% MARD. They perform well compared to a YSI. I think the frequency of measurement when you are giving IV insulin is critical. If you are limiting yourself to one hour measurements, that is not the way to go. The question is, what protocol are you going to follow? One hour measurements is not going to do it.

 

How Meters are Used at Home Consumers Choose Meters

Ellen H. Ullman, MSW, (Children with Diabetes Foundation/Close Concerns, Boca Raton, FL)

Patient advocate Ullman presented the results of an online survey that she conducted to gather data on how meters are used at home and what consumers look for in a meter. There were 512 surveys collected, with 36% of the responders type 1 patients, 49% of responders with a child with type 1 diabetes, 13% with type 2, and 3.2% with LADA. 77% of responders were female. Just 63% of patients replied that they had been trained how to use the meter by a HCP. The majority (90%) of the sample had insurance coverage. Several interesting findings were revealed about the typical practices of these SMBG users in this unscientific survey. We take these results with the full knowledge that those likely to respond to such a survey and the population composition suggests these patients are likely more educated and proactive about their management than the typical patient of diabetes. The majority of patients were checking their blood sugar between 5-12 times per day. Strikingly, 41.1% of the participants said they never use control solution, while another 14.5% said they only use it once per year. 57% of respondents replied that they had forgotten to change the code on their meters in the past and 20% had knowingly used expired strips. Only 20% of respondents wash and dry their hands every time they test. Over one-third of patients keep strips or their meter in the car, 62% of respondents did not know the temperature range of their meters, and 33% did not know the low to high range of their meters. Over 45% of patients have had to use more than one strip when testing because they did not trust the accuracy of the first result. As far as how consumers acquired their devices—30% acquired the device from a health care provider, 27.2% directly from a pump company, and 22.2% at a local pharmacy. When asked what was most important in features of their device, 77.5% responded that accuracy was a most important feature. Blood sample size and cost of strips were also highly important to consumers. Strikingly, almost half (46%) of patients believe their meters have error below 10-%. In closing, Ullman advocated for accuracy for all (inside the hospital and out) within 10%, better labeling (including tools to help patients compare meters’ accuracy), and for old meters to be phased out once better ones become available. We note we would like to better understand tradeoffs between accuracy improvements at various levels (15%, 12.5%, and 10%).

• Ullman’s survey suggests that even in educated patients who proactively manage their diabetes, quality control is largely neglected. 41.1% of responders never use control solution, 14.5% use it once per year, 10.8% use it four times per year, and 3.3% were entirely unaware of what control solution is. This is a notable result, as it is likely quality control adherence is even lower in the general population of patients. Ullman noted that according to Diabetes Forecast’s 2010 Consumer Guide, control solution costs $15 per bottle, and each bottle expires 30 days after opening. This is a major area for improvement, according to Ullman. Control solution used to come packaged with meters or strips and Ullman suggests renewing this practice would be a very good idea for manufacturers to consider. Additionally, the cost of control solution is prohibitively high—if diabetes patients were checking the accuracy of strips consistently this could add nearly $200 more per year in cost of managing their diabetes.
• Labeling can be improved in several ways, according to Ullman. A majority of patients (62%) do not know the temperature extremes that are hazardous to strips, and 33% of patients do not know the limits of their meters (so when the measurement is above the limit it reads “high”— if the patient does not know the upper limit of the meter this could be a major problem for insulin dosing estimation). Additionally, better labeling could help patients become more educated consumers. Ullman calls for labeling that will allow patients to compare meters more easily, especially in terms of accuracy. Accuracy is a most important feature for a notable 77.5% of the population Ullman surveyed. Other key features include small blood sample requirements and low costs of strips.
• Ullman brought her discussion to a close by reminding the audience that all consumers are end users, regardless of the setting. A child at a soccer game deserves a higher level of accuracy as much as a patient in the hospital. While accuracy is paramount, better education is critically important as well and will begin to be achieved by better labeling of devices.

 

Questions and Answers

Q: You said two things I am interested in hearing your opinion on. People are interested in the accuracy on the meter. How could we effectively give that information to consumers so that they could make informed decision?

A:

I would love for ADA to include comparison material in the ADA annual consumer report. And on the box. I think we have to be able to compare – we need comparison standards. Limitations - it has to be in any user quick reference guide, on the box, not in the small print. You have to use terms people will be familiar with, they won’t know their hematocrit. Up to 10% of people with type 1 diabetes have celiac, and those people are often not diagnosed and may be anemic. It would be great to register each meter so when there is a recall they get an immediate notice.

Q: Thank you for grounding us back in reality. One pharma chain sold 300,000 glucose meters in a year and less than two dozen vials of control solution in the same period. Most diabetics don’t know what to do with control solutions. The answer to that issue is simple— the manufacturers have to figure out a way to make the meter reject the strip if it is no good. I don’t’ think the average diabetic will understand glucose control solution. About 75 cents a strip, all manufacturers have kept these prices up (we note that the manufacturers don’t set the prices and that the price per strip received by manufacturers isn’t anywhere close to this level). If you don’t line that little drop of blood up right, you’ve lost the strip. A lot of diabetics have more than one meter; one at home, one when you travel. When I travel I take two. Unless they are using all the strips frequently for both meters, some strips will become outdated (we were surprised by this comment – it’s unlikely that strips will expire unless meters are used quite infrequently – we don’t think it’s up to companies producing strips to make sure that they last years). These are the realities that people with diabetes deal with every day. It’s really nice for academicians to come in and ask about what the devices are capable of doing, but that is not the reality. The reality is we have 5000 hospitals in the US and most do not do anything that you’ve heard today. They are using sliding scales and they don’t test once an hour, and they don’t know how to calibrate meters they have on the floors.

Q: I think it’s good we hear what patients and care givers are thinking about all of this. But again, looking at that survey, they are more serious about their diabetes. I shudder to think what my patients in Savannah, GA look like. I share concern over control solution. Medicare only gives you one vial for six months. It is hard to teach it too. We haven’t talked about mail order houses, how a lot of patients get their devices and supplies—there is no education with that method. In my opinion, the mail order houses are a big problem.

Dr. Harper: We would really like you to report to FDA on this. We also take trade complaints.

 

Risk Mitigation in Hospitals

Dawn Hanson, MS MLS (ASCP), DLM (St. Agnes Hospital, Baltimore, MD)

Hanson discussed the steps hospitals must take in the spirit of risk mitigation as related to SMBG devices. Overall, hospitals must adhere to standards of the Joint Commission. Hanson offered insight into how SMBG use is handled in her hospital, but discussions throughout the day would suggest that there is likely considerable nuance in hospitals across the country. Hanson, like many others throughout the day, called for better labeling that keeps the device mobility and high staff turnover rate at hospital in mind. She also suggested that manufacturers take care to ensure meters can easily be disinfected without harming the meter’s function or technology. At the conclusion of her talk, she discussed a wish list she has as a nurse at a major hospital: standardized meters, smarter meters, and smarter software. Ultimately, she seeks help to improve quality control (such as color-changing solutions when the bottle of control fluid expires or individually packaged strips). In her opinion and experience, no results are better than wrong results—meters should automatically shut down if a strip is expired, or it has not been calibrated/coded in a certain period of time, etc.).

Wrap-Up; Where Do We Go from Here?

David C. Klonoff, M.D., F.A.C.P. (Mills-Peninsula Health Services, San Mateo/University of California at San Francisco, San Francisco, CA)

Dr. Klonoff closed the meeting by reviewing discussions from the previous two days. He shared what he believes this meeting has communicated: the community wants better analytical and clinical performance from meters, better incorporation of factors which can improve SMBG above and beyond meter and strip performance, better labeling and reporting of interfering factors, and more consensus on glycemic control for patients in the hospital. Dr. Klonoff concluded by offering a list of ten themes that resonated for him and suggesting that that the infamous lyrics from Mick Jagger apply well here, “you can’t always get what you want, but if you try some times you find you get what you need”:

• We may potentially need separate analytical and clinical accuracy standards for different populations (i.e., ICU vs. home-users), generally falling into categories of “accuracy” and “super- accuracy”.
• Presentation of pertinent information needs to be improved and a good error grid is a key component of this.
• Data that is currently not part of the standard criteria should be accounted for and considered in some way (i.e., the 5% of outliers outside of the 95% of measures the accuracy guidelines apply to).
• New labeling is required for both analytical and clinical performance; an interpretable error grid will be enormously helpful in this.

• How interfering substances are measured and dealt with should be standardized. The CLSI could potentially help in such work.
• There should be a standard mechanism for reporting interfering substances. The CLSI could potentially help in such work.
• We have to be prepared for trade offs if we demand higher accuracy—it could come in the form of increased cost or increased testing time.
• More data is needed to help us understand what outcomes are important and what level of performance is necessary to satisfy these outcomes. Dr. Klonoff specifically called for more work in the field of in silico modeling (such as Dr. Breton’s work).
• Processes need to be developed to improve the accuracy of testing; this will involve improving education, not just improving the meters or strips themselves. According to Dr. Klonoff, we will not get to where we want to be in accuracy solely by revising the technology.
• Optimal targets for glycemic control need to be identified and defined.

--by Jessica Swienckowski and Kelly Close