Abstract
The annual Virtual Hospital Diabetes Meeting was hosted by the Diabetes Technology Society on April 14 and 15, 2023, with the goal of reviewing the progress made in the hospital use of continuous glucose monitors (CGMs). Meeting topics included (1) Nursing Issues, Protocols, Order Sets, and Staff Education for Using CGMs, (2) Implementing CGM Programs for Use in the Wards, (3) Quality Metrics and Financial Implications of CGMs in the Hospital, (4) CGMs in the Critical Care Setting, (5) Special Situations: Labor/Delivery and Hemodialysis, (6) Research Session on CGMs in the Hospital, (7) Starting a CGM on Hospitalized Patients, (8) Automated Insulin Delivery Systems in the Hospital, (9) CGMs in Children, (10) Data Integration of CGMs for Inpatient Use and Telemetry, (11) Accuracy of CGMs/Comparison with Point-of-care Blood Glucose Testing, and (12) Discharge Planning with CGMs. Outcome data as well as shared collective real-life experiences were reviewed, and expert recommendations for CGM implementation were formulated.
Keywords: continuous glucose monitoring, glucose, hospital, inpatient, insulin, metrics
Introduction
Continuous glucose monitors (CGMs) have been cleared by the U.S. Food and Drug Administration (FDA) for outpatient use since 1999 and have revolutionized the care of people with diabetes (PWDs), leading to improved glycemic outcomes and quality of life. CGM use in inpatient settings has been considered investigational until the beginning of the COVID-19 pandemic in 2020, when the FDA provided a policy allowing for hospital CGM use. Rapid and widespread adoption generated a body of evidence, both from clinical research trials as well as real-life experience, evaluating various aspects of CGM use, including safety, efficacy, and impact on glycemic outcomes, as well as barriers to adoption and integration within complex hospital systems. During this CGMs in the Hospital meeting hosted by the Diabetes Technology Society on April 14–15, 2023, international experts in technology and hospital diabetes management discussed a wide variety of topics impacting specific aspects of hospital CGM use as outlined in Table 1. The following report summarizes the key points of each presentation along with recommendations outlining how institutions with experience using inpatient CGMs are introducing this technology into patient management.
Table 1.
Agenda of the Meeting with a List of Session Topics.
|
Friday, April 14, 2023
Session 1: Nursing Issues, Protocols, Order Sets, and Staff Education for Using CGMs Session 2: Implementing CGM Programs for Use in the Wards Session 3: Quality Metrics and Financial Implications of CGMs in the Hospital Session 4: CGMs in the Critical Care Setting Session 5: Special Situations: Labor/Delivery and Hemodialysis Session 6: Research Session on CGMs in the Hospital Saturday, April 15, 2023 Session 7: Starting a CGM on Hospitalized Patients Session 8: AID Systems in the Hospital Session 9: CGMs in Children Session 10: Data Integration of CGMs for Inpatient Use and Telemetry Session 11: Accuracy of CGMs/Comparison with POC BG Testing Session 12: Discharge Planning with CGMs |
Abbreviations: AID, automated insulin delivery; BG, blood glucose; CGM, continuous glucose monitors; POC, protocols combining point-of-care.
Session 1. Nursing Issues, Protocols, Order Sets, and Staff Education for Using CGMs
Moderator:
Jane Jeffrie Seley, DNP, MPH
Weill Cornell Medicine, New York, NY, USA
Panelists:
Eileen Faulds, PhD, MS, RN, FNP-BC, CDCES
Wexner Medical Center and College of Nursing, The Ohio State University, Columbus, OH, USA
Inpatient CGM use varies significantly from outpatient use where the patient interacts with their real-time data and the provider uses retrospective reports. In the inpatient setting, we need to consider who will be consuming/responsible (eg, nurse, primary team, endocrinologist, pharmacy) for the various data elements (eg, discrete value, alarm, trend arrow, daily trends/patterns). This should help dictate how and what data elements are integrated into the electronic health record (EHR).
Hybrid protocols combining point-of-care (POC) blood glucose monitoring (BGM) and CGMs could serve as a bridge to full non-adjunctive CGM use.
When discussing the future implementation of inpatient CGM use, we should keep in mind the many shortcomings of inpatient POC BGM (eg, mistiming of testing, missed testing, missed hypo- and hyperglycemia) and ensure that the addition of CGMs specifically addresses these gaps even within a hybrid protocol.
Juan Pablo del Rincon, MD
MetroHealth Medical Center and Case Western Reserve University, Cleveland, OH, USA
Although the accuracy of current CGM technology is lower than that of POC blood glucose (BG) measurements, CGMs offer the advantages of a high frequency of data points, alerts for hypo- and hyperglycemia, and data trends.
In the event of changes in a patient’s mental status, clinical suspicion of hypo- or hyperglycemia, or significant changes in hemodynamic status or nutrition, CGM readings should be confirmed with POC BG measurements.
The use of CGMs in the hospital may be particularly useful if combined with initiatives such as the implementation of a glucose telemetry system, with the training of staff on protocols to prevent hypoglycemia.
Gwen Klinkner, DNP, RN, BC-ADM, CDCES, FADCES
UW Health, University of Wisconsin Hospitals and Clinics, Madison, WI, USA
Operational and clinical champions are needed to develop the infrastructure for inpatient CGM use within the EHR as well as other clinical supports (policy, education, etc.).
Certified diabetes care and education specialists (CDCES) must be involved with the development, implementation, and evaluation of CGM integration into practice.
The lack of FDA clearance for inpatient CGM use may require organizations to focus efforts on staff education to increase knowledge about CGMs and to develop policies and documentation tools to support the care of patients using their own CGMs.
Rebecca Longo, ACNP-BC, CDCES
Lahey Hospital & Medical Center, Beth Israel Lahey Health, Burlington, MA, USA
Institutions should each develop a clear protocol for the use of CGMs, including the delineation of roles based on their local resources with role-specific staff education. This may include specialty staff such as CDCES, clinical pharmacists, registered nurse (RN) super-users, bedside nursing staff, hospitalists, endocrinologists, or other trained personnel.
Standardized alert and alarm thresholds with corresponding orders directing nursing action based on alerts and alarms should be developed at institutions utilizing CGMs for glucose monitoring.
Hybrid POC BGM and CGM protocols may be cost-effective for patients utilizing insulin infusions; however, larger studies with cost analysis are needed.
Robert J. Rushakoff, MD
University of California, San Francisco, San Francisco, CA, USA
Specific protocols—new orders—need to be developed on how to make use of glucose levels—levels that 99.99% of providers have never been exposed to in the past.
There needs to be full and transparent automated integration of all the data into EHRs.
There needs to be recognition of the false low glucose values that may be reported.
There need to be specific protocols on when the CGM can and cannot be used (ie, magnetic resonance imaging, etc.).
There needs to be recognition of the huge educational effort (time and cost) to add CGM use into a medical system, and this will be in addition to the continued use of glucose meters—all this at the current time of tremendous hiring freezes and cutbacks in many medical systems.
Inpatient-specific CGM systems should be developed rather than just taking the ever-changing outpatient devices.
The discussion began with a comparison of the use of CGMs in the hospital today with the initiation of POC BGM in hospitals in the 1980s. Much like CGMs now, there were no POC BGM standards for accuracy in the hospital setting then, so protocols were created that addressed quality control, staff training, documentation, and more. While Dr. del Rincon described the frequency of results, hypo- and hyperglycemia trends, and detection of night-time hypoglycemia as the greatest benefits of inpatient CGM use, Dr. Rushakoff felt that sending people home with CGMs will give us the “most for the money.” Nurse satisfaction with CGMs was surveyed by Dr. del Rincon, with responses including high satisfaction, ease of use, and trusted results.
Dr. Klinkner discussed the need for protocols, order sets, and smart text for CGM education and documentation in the EHR and provided EHR screenshots.
The conversation moved on to accuracy; calibration; and managing alerts, alarms, and trend arrows in the critical care setting. Dr. Faulds described a response plan for hypoglycemia alerts, including when to perform a POC BG test, with corresponding suspected causes and remediations.
The final discussion focused on the economics of using CGMs with insulin infusions in the hospital compared to using POC BGM. Nurse Practitioner Longo calculated a cost comparison by looking at multiple variables, including the difference in nursing time (RN or technician), cost of supplies (BG test strips or glucose sensors), and length of time until the desired result is available to dose insulin. Although CGMs were more expensive in the first 24 hours of use because of extra POC BG tests to account for warm-up time and accuracy confirmation, there may be a cost saving for people on insulin infusions after the first 24 hours. Figure 1 illustrates the cost comparison between POC BG tests and CGMs. In summary, this session focused on the role of nursing in inpatient CGM use, showing that nurses were interested and generally enthusiastic about using CGMs in the hospital in the future.
Figure 1.
A cost comparison of POC BG versus CGMs for insulin infusions. Abbreviations: BG, blood glucose; CGM, use of continuous glucose monitors; NA, nursing assistant; POC, point-of-care; RN, registered nurse.
Source: Figure courtesy of Rebecca Longo.
Session 2. Implementing CGM Programs for Use in the Wards
Moderator:
Amisha Wallia, MD, MS
Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Panelists:
Marie E. McDonnell, MD
Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
Many epidemiologic studies have shown that recurrent hypoglycemia during hospitalization is associated with increased mortality both inside and outside of the intensive care unit (ICU). 1 In light of this, a key expected benefit of CGM programs on the wards is to reliably detect hypoglycemia so that we can both understand and prevent it.
During the COVID public health crisis and through published studies, we have confirmed that CGM use is acceptable to nurses, providers, and patients as a novel methodology for measuring POC BG levels.
Implementation challenges include refining patient selection for CGMs so that those at highest risk are prioritized, ensuring that conditions that may reduce accuracy (eg, hypoperfusion) are correctly identified to prevent inappropriate use, and developing clear guidance for patients undergoing surgical and radiological procedures.
Athena Philis-Tsimikas, MD
Scripps Health, San Diego, CA, USA
Upfront preparation for implementation on the wards requires communication; education; and buy-in/approval from hospital administrators, nursing leadership, and providers.
Preparation with plans and operations manuals should include steps that will be taken to identify patients that will have CGMs placed, who will make the placement, and connections to apps/iPads/other devices. Preparation should also outline processes for capturing values, validating devices, use for insulin treatment, and documentation into the EHR.
Conducting on-site education should be implemented for bedside nursing staff and supervisors and provide ongoing support for any troubleshooting. There should be education for providers on how to use the CGM readings for treatment intervention, and we should provide algorithms or treatment guidance plans.
Elias Spanakis, MD
Baltimore VA Medical Center and School of Medicine, University of Maryland, Baltimore, MD, USA
Compared to the POC BGM which represents the current standard of care of glucose testing in the hospital, CGM devices can be a better glucose monitoring system for achieving tighter glucose control and reducing inpatient hypoglycemia.
Concerns about the accuracy of CGM devices in the hypoglycemia range can be offset by the continuous monitoring of the glucose values, allowing CGM devices to be a “safety net” for preventing hypoglycemia—a safety net that currently does not exist in the hospital setting.
CGM devices could be utilized differently by (a) nursing staff, who will be responsible for any immediate treatment actions, and (b) medical providers, who will be adjusting diabetes mellitus (DM) medications retrospectively.
Guillermo Umpierrez, MD
School of Medicine, Emory University, Atlanta, GA, USA
The use of inpatient CGMs overcomes the capillary POC BG testing limitations by providing a more complete glycemic profile, including a greater ability to detect hypoglycemia and severe hyperglycemia.
The results of observational and randomized controlled studies indicate acceptable accuracy with an overall mean absolute relative difference (MARD) less than 16% for glucose values between 70 and 250 mg/dL and with over 95% of glucose readings falling into error grid zones A and B.
The implementation of remote glucose monitoring using real-time continuous glucose monitoring (RT-CGM) to wirelessly transmit glucose data from the bedside to a centrally located monitor at the nurses’ station can improve glycemic control by reducing the frequency and duration of hypoglycemic episodes in insulin-treated people with type 2 diabetes (T2D).
Hypoglycemia and hyperglycemia have been shown in both inpatient and outpatient studies to have deleterious effects on patient outcomes in these settings. CGMs offer an opportunity to augment and/or replace current glucose monitoring inpatient workflows, specifically POC BGM. While CGMs have previously been utilized to potentially augment the detection of hypo- and hyperglycemia, the COVID-19 pandemic introduced the opportunity for replacement of BGM by utilizing CGMs on both individual patients and/or throughout a myriad of inpatient settings (eg, floor, units).
Potential use of CGMs to improve patient safety includes early detection of hypoglycemia, leading to potential early treatment. In a future state, the use of trend information could potentially allow for pretreatment or potential amelioration of hypoglycemia, as has been seen in the outpatient setting. In addition, CGMs offer an opportunity to alleviate the need for POC finger stick testing, but this would entail utilizing available CGM glucose levels as values on which treatment decisions with insulin or other diabetes medications would then be based. The accuracy of CGMs in specialized care settings and patient cases must be considered, and most current models specifically include hybrid use with both POC BGM and CGMs being utilized.
Implementation strategies building on pilot and real-life studies, including procedures, workflows, and EHRs, have been differing, pending the type of utilization for CGMs (augmentation/safety vs. replacement/use for treatment dosing). Figure 2 provides an example of CGM utilization as a standard of care of a hospital set of workflows. Different strategies can be utilized, including a telemetry-like setup and/or CGM-specific dashboards. Specific team-based models, such as a specialized mobile CGM team and/or utilization of diabetes “leads” on individual units and/or floors, could be utilized. Smaller-scale proof-of-concept hybrid protocols have been utilized effectively; larger-scale implementation plans with CGMs as a full replacement are currently lacking but could be formulated and utilized in the near future.
Figure 2.
An example of a hospital set up using CGMs as the standard of care. Abbreviations: APN, advanced practice nurse; CGM, continuous glucose monitor; RN, registered nurse.
Source: Figure courtesy of Athena Philis-Tsimikas.
Further into the future, we could see the utilization of CGM-specific glucose metrics, as well as evaluation of outcomes in relation to CGM glucose metric goals.
Session 3. Quality Metrics and Financial Implications of CGMs in the Hospital
Moderator:
Gregory Gilbert, MD
El Camino Hospital, Mountain View, CA, USA
Panelists:
Andjela Drincic, MD
University of Nebraska Medical Center and Diabetes and Endocrinology Center, Nebraska Medicine, Omaha, NE, USA
Determination of the financial implications of CGM adoption for hospitals is complex. It involves balancing the cost of adoption itself vs. the benefit of CGM impact on glucose control and related health outcomes while also taking into account its impact on quality metrics and associated financial implications.
The cost of adoption involves the cost of CGM technology, including sensors, cost of integration into the EHR system, staff education, and the overall cost of CGM program implementation (processes for validation, calibrations, quality assurance, treatment protocol development, cyber security issues, etc.). The financial benefit may be mediated through its positive impact on clinical outcomes, including mortality, length of stay (LOS), 30-day readmission, estimated hospital admission cost, and in-hospital secondary complications.
Current inpatient glucose quality metrics are all calculated based on capturing data such as POC BG values, and integrating continuous data provided by CGMs may require developing new models.
While CGM adoption holds the promise of providing a net financial benefit to the system mediated by its impact on improving glucose control (in particular, decreasing hypoglycemia rates), in the short term, it may have a negative impact on quality metrics as it will detect more hypoglycemia. Therefore, to encourage adoption, institutions should be shielded from potential associated penalties. Furthermore, benchmarking efforts will have to take into account the “like institutions,” and quality metrics generated by CGM vs. POC BG data may need to be evaluated separately.
It is not the CGM itself but the model of CGM adoption that may have the most impact on quality metrics. Adopting CGMs for targeted high-risk populations with the integrated protocol for real-time intervention (such as hypoglycemia treatment) will likely have the most impact on quality metrics.
Spiros Fourlanos, MBBS, FRACP, PhD
The Royal Melbourne Hospital, University of Melbourne, Melbourne, VIC, Australia
Adverse clinical outcomes in hospitals are strongly associated with fundamental metrics of hyperglycemia.
The association of adverse clinical outcomes in hospitals with new CGM metrics requires investigation.
The use of CGM metrics to provide better context to adverse clinical outcomes in hospitals is also vital to determining any potential health economic benefits of CGMs.
Mary E. Justice, MBA
College of Nursing, The Ohio State University, Columbus, OH, USA
If you focus on improving outcomes for patients (ie, improved visualization/monitoring of glucose values), then financial benefits will come. Davlyatov et al showed that the quality of care is related to a Community Health Center’s financial performance. 2
Given the staffing crisis in health care, we have to look to technology (like CGMs) to aid in improving the workload of caregivers.
The long-term return on investment (ROI) model for innovations like CGMs are both qualitative and quantitative. However, when health system leaders are evaluating capital expenditure, it is important that a positive quantitative ROI can be shown.
Nadine Shehab, PharmD, MPH
Lantana Consulting Group, East Thetford, VT, USA
Glycemic control quality metrics are increasingly being incorporated into national quality reporting and surveillance programs that rely on EHR data. Glucose data from CGM devices are not generally integrated into the EHR. With the increasing utilization of CGMs in the hospital, it will be important that CGM data be integrated into EHR laboratory data sources.
In the absence of glucose data from CGMs, national reporting programs and hospitals could miss information about hospital glycemic control patterns and quality of care. This will require EHR and CGM software vendors to identify solutions for the integration of data sources so that the full picture of glucose control in the hospital can be demonstrated to providers and national quality reporting programs, alongside BG data sources (central lab, POC).
It will also be important to identify quality metrics that take into account hospital CGM utilization patterns, patient populations, and data sources, and, wherever possible, align CGM quality metrics with other local- and national-level quality metrics.
We heard a robust discussion regarding continuous glucose monitoring and the difficulties in getting quality data that do not negatively impact the hospital, quality metrics, and billing. Figure 3 outlines the costs of adoption, potential financial benefits, and implications for quality metrics associated with using CGMs in the hospital. Done correctly, this can be a win-win for the hospital, patients, and providers. Glycemic control programs that take into account the full picture of BG values in the hospital, including patients using CGMs, can potentially improve glycemic control quality in the hospital and lead to fewer hospital days.
Figure 3.
Costs of adoption, potential financial benefits, and implications for quality metrics associated with using continuous glucose monitors in the hospital. Abbreviations: CGM, continuous glucose monitor; EHR, electronic health record; LOS, length of stay; POC, point-of-care.
While the quality criteria have yet to be determined for this new continuous glucose monitoring technology, the Centers for Disease Control, researchers, and early adopters are in the process of creating criteria that can not only predict when glucose might be headed in the wrong direction but also how to course correct and keep within the normal range.
For clinicians wanting to add this technology to their hospital, this technology could potentially reduce LOS and readmission rates, which, if these occurred, would then be of interest to the C-suite. Protocols and procedures need to be created to make sure the new science is used on the appropriate patients and to determine how the hospital monitors the results, likely through some sort of central monitoring.
Clearly, this technology is coming, and experts shared real-life experiences from the United States and Australia. This exciting technology, which will improve the care of PWDs, might ultimately become as commonplace as vital signs in hospitals over the next decade.
Session 4. CGMs in the Critical Care Setting
Moderator:
Elias Spanakis, MD
Baltimore VA Medical Center and School of Medicine, University of Maryland, Baltimore, MD, USA
Panelists:
Georgia M. Davis, MD
Emory University, Atlanta, GA, USA
The potential for use of CGMs in the critical care setting is promising, although further data are needed on its function and reliability in diverse critically ill populations.
There will likely be unknown and unanticipated interferences that may impact the reliability of CGMs in this setting, and a hybrid glucose monitoring protocol may be required.
The glucose data obtained from CGMs compared to infrequent POC BG testing can be helpful for the management of critically ill patients.
Kathleen Dungan, MD, MPH
The Ohio State University, Columbus, OH, USA
The use of CGMs in the ICU has the potential to reduce the need for POC BGM through the use of hybrid protocols using both POC BGM and CGM measures.
Concerns about accuracy can be offset by the near continuous availability of glucose data as well as other measures, including initial and periodic POC BG tests to ensure accuracy/reliability, adjustment of alert thresholds, and staff training.
Current devices are designed for home use and need to be configured to facilitate system-based start-up, integration within clinical workflows, and decision support within the EHR.
James S. Krinsley, MD
Stamford Hospital and Columbia University Vagelos College of Physicians and Surgeons, New York, NY, USA
There is a strong interaction between preadmission glycemic control, reflected by hemoglobin A1c (HbA1c) and quantified using the Nathan formula (estimated preadmission glucose = [HbA1c × 28.7] − 46.7 mg/dL), and mortality of critically ill patients. For patients with normal HbA1c, higher mean ICU BG is associated with higher mortality, but the opposite is true for patients with high HbA1C—high mean ICU BG is associated with lower mortality.
The relationship between preadmission and ICU BG has been best described by a new metric, the glycemic ratio (GR), calculated as the quotient of estimated preadmission BG and mean ICU BG. In a large heterogeneous cohort of critically ill patients, the lowest mortality was seen in patients with GR 0.8 to 1.0—in other words, mean ICU BG was 80% to 100% of the estimated preadmission BG. The relationship between GR and mortality is J-shaped, with higher mortality seen with GR < 0.7 and then increasing at GR > 1.1.
CGMs have a relatively long track record of use in stable, “healthy” outpatients. However, many ICU patients are in shock, have acquired substantial amounts of edema, and are being treated with numerous medications, some with the potential for interference with accurate measurement. Acceptable accuracy must be proven in this challenging environment.
Kristen Kulasa, MD
University of California, San Diego, La Jolla, CA, USA
In the ICU setting, achieving glycemic targets can be labor-intensive because insulin sensitivity can vary from hour to hour.
CGMs can be a valuable tool for frequent glucose monitoring while providing targeted and cost-effective care in an efficient manner for nursing staff.
FDA clearance, hospital-specific device modifications, and new workflows are necessary to ensure the successful integration of CGM devices into the ICU.
Although initial studies showed that tighter glycemic control could lead to reduced morbidity and mortality in the ICU setting, subsequent studies showed increased mortality among those who were intensively treated.3 -5 Several subsequent studies found that glycemic targets should be different among PWDs, compared to people without diabetes. However, a subsequent randomized controlled trial explored this hypothesis without showing a difference in mortality. 6 Figure 4 presents the relationship between mean BG and mortality.
Figure 4.
The relationship between mean BG (mg/dL) and mortality, stratified by HbA1c level. For patients with HbA1c less than 6.5%, higher mean BG is strongly associated with increased mortality. For patients with HbA1c greater than or equal to 8.0%, the opposite relationship is observed. Abbreviations: BG, blood glucose; HbA1c, hemoglobin A1c.
Source: Reproduced from Krinsley et al. 7
Previously conducted ICU studies utilized POC BG as a method of glucose testing, its main limitation being that it is checked infrequently. However, it has been shown that the frequency of glucose monitoring is more important than analytic accuracy.8,9 CGM devices can monitor glucose values every 1 to 5 minutes, which allows early detection and treatment of hypo- or hyperglycemia.
CGM use in the ICU can pose technological challenges, including current CGM systems being designed for the outpatient setting; the “warm-up” time period with the absence of data; CGM sensor failures; lack of CGM integration into the EHR system; and CGM accuracy challenges secondary to extreme temperatures, compression effect, hypotension-hypoperfusion, hypoxemia, or medication interferences.10,11 Issues with accuracy can be addressed by either using them as purely adjunctive to POC BG, where the only concern will be alarm fatigue, or by using hybrid POC BGM and CGM protocols. With these hybrid protocols, 12 CGMs are validated and calibrated until they demonstrate acceptable accuracy, at which point CGMs are used nonadjunctively. Trend arrows can be utilized for predicting hypo- and hyperglycemia and have the potential if utilized to overcome accuracy issues.
Following initial training on CGM devices, nursing staff reported positive experiences and were supportive of using them in the ICU, performing calibrations or even utilizing trend CGM arrows for treatment decisions. However, CGM devices specifically designed for hospitals, with minimal interruptions from imaging studies, are needed.
Session 5. Special Situations: Labor/Delivery and Hemodialysis
Moderator:
Umesh Masharani, MB, BS
University of California, San Francisco, San Francisco, CA, USA
Panelists:
Labor/Delivery
Florence Brown, MD
Joslin Diabetes Center, Beth Israel Lahey Hospital, Harvard Medical School, Boston, MA, USA
Accuracy: CGMs placed on the upper arm may be continued in labor to provide additional glucose data but are not a substitute for BGM. There is insufficient data regarding the accuracy of CGM glucose data in the setting of labor and delivery. It is unknown whether maternal edema at the end of pregnancy and intravenous (IV) hydration affect the accuracy of CGMs. It is also unknown if all CGMs are equal in terms of accuracy, particularly at the glucose targets required in labor. Given limited accuracy data, BGM needs to continue as a standard of care (hourly in labor) and also to confirm CGM glucose values that are below or above targets.
Glucose targets: Glucose targets will vary depending on events/stages of labor and delivery. (A) During induction, prior to the onset of labor, individuals will continue to eat. They will continue standard pregnancy time in range (TIR) targets, except using 70 to 140 mg/dL rather than 63 to 140 mg/dL. Per hospital institutional policy, all glucose levels <70 mg/dL are treated. (B) When nil per os (NPO or nothing by mouth) in early or active labor or preoperative awaiting cesarean delivery, switch the target to 80 to 110 or 120 mg/dL. (C) While NPO postpartum, target glucose should be 100 to 150 mg/dL. D) When meals have resumed after delivery, use standard nonpregnant CGM targets of 70 to 180 gm/dL.
Diabetes management: Once the patient is NPO in early labor or upon presentation for cesarean delivery, standard treatment involves the transition of diabetes management to an IV insulin drip for patients using multiple daily injections or insulin pumps. Sensor-augmented insulin pumps with or without predictive low glucose suspension and automated insulin delivery (AID) systems have not been studied in labor and delivery versus standard of care. However, their continued use may be considered based on patient preference, stability of CGM and BG levels, demonstrated ability to self-manage diabetes under changing clinical circumstances, and if the AID pump algorithm is able to achieve the recommended therapeutic targets pre- and postpartum. Switch to insulin if glucose targets are not met within the hour or if there is an urgent obstetrical condition that complicates the individual’s ability to self-manage their insulin pump.
Studies are needed to evaluate CGM glycemic parameters with maternal and neonatal outcomes, especially neonatal hypoglycemia.
Labor/Delivery
Elizabeth O. Buschur, MD, FACE
The Ohio State University, Columbus, OH, USA
Finger stick glucose testing is the standard of care during labor and delivery. CGMs may be used adjunctively and need to be studied to determine if sensor glucose is an accurate measure of capillary glucose in time of labor and delivery when fluid shifts, volume changes, and dramatic physiologic changes occur.
Dexcom G7, which is approved by the FDA for pregnancy, should be studied during labor and delivery to determine its accuracy during this dynamic time.
CGM sensor placement should be done cautiously before the time of delivery. For example, patients should use the upper arm, as this is an approved site. The abdomen should be avoided in the case of the need for operative delivery. The upper arm where the blood pressure cuff is utilized during labor and delivery should also be avoided to avoid compression of the sensor itself and potential false compression of “low” sensor readings.
Automated insulin delivery pumps/devices are also not approved for use in labor and delivery and can be used off-label at this time if tight glucose targets of labor and delivery are achieved. If not, then it is recommended to switch to IV insulin infusion.
Diet, glucose goals, and insulin needs vary during labor and delivery. Women undergoing planned induction continue to eat until they are in labor. Women who arrive at the hospital in spontaneous labor or are admitted for planned cesarean section are fasting. The women resume eating after the delivery. Glycemic targets need to be adjusted for labor and after delivery. There is also a dramatic change in insulin sensitivity after delivery, when insulin requirements are reduced by 50% to 60%.
CGM use in pregnant women with type 1 diabetes (T1D) is associated with a reduction in maternal LOS as well as in the incidence of gestational age births and neonatal hypoglycemia. Currently, it is recommended that CGMs be used in addition to and not as a substitute for fingerstick glucose measurements during pregnancy, labor, and delivery. The use of CGMs during pregnancy in women with diabetes can help to achieve HbA1c targets but should not be used as a substitute for BGM; CGMs for pregnant women can reduce fetal complications. 13 CGM targets during pregnancy for women with diabetes can be found in Figure 5. During pregnancy, more than 70% of the CGM glucose levels should be in the 63 to 140 mg/dL [3.5-7.8] range. During active labor, when the woman is fasting, the CGM glucose level should be kept in the 70 to 110 mg/dL range to reduce the risk for neonatal hypoglycemia. After delivery, CGM glucose targets revert to the standard nonpregnancy range of 70 to 180 mg/dL. Clinical experience indicates that CGM and hourly fingerstick glucose levels correlate fairly well during labor. Because of the time lag between fingerstick glucose levels and interstitial glucose levels, more frequent fingerstick glucose measurements may be necessary in the event of a hypoglycemic event.
Figure 5.
Continuous glucose monitor targets women with diabetes during pregnancy.
Source: American Diabetes Association. “Clinical Targets for Continuous Glucose Monitoring Data Interpretation: Recommendations From the International Consensus on Time in Range,” American Diabetes Association, 2019. Copyright and all rights reserved. Material from this publication has been used with the permission of the American Diabetes Association. Originally published by Battelino et al. 14
aPercentages of time in ranges are based on limited evidence. More research is needed.
bPercentages of time in ranges have not been included because there is very limited evidence in this area. More research is needed. Please see the pregnancy section in reference 14 for considerations of targets for these groups.
**Includes percentage of values <54 mg/dL (3.0 mmol/L).
During labor, when the woman is fasting, the standard procedure is to transition to an IV insulin infusion. Increasingly, however, women on hybrid closed-loop (HCL) systems are staying on these systems during labor and delivery. There is, however, concern regarding the ability of the patient to self-manage the HCL system while in labor and the event of an obstetric emergency. These systems also may not be sufficiently responsive to the dramatic changes in insulin sensitivity immediately after delivery. The recommendation, therefore, is to put these systems in an open loop immediately after delivery and establish a new insulin profile reflecting the nonpregnant state. The clinical experience indicates that the HCL systems work well during active labor, keeping the glucose levels steady in the target range.
The speakers addressed some technical issues regarding CGM use during labor and delivery. The sensors that continuously report their readings appear to be more reliable compared to sensors that require scanning. The patients usually place the sensors on the arms rather than on the abdomen.
Hemodialysis
Roma Gianchandani, MD
Cedars-Sinai Medical Center, Los Angeles, CA, USA
In patients with end-stage kidney disease (ESKD) and/or dialysis, evaluating long-term glucose control is challenging, as HbA1c, fructosamine, and glycated albumin have limitations in the face of anemia, erythropoietin treatment, malnutrition, and glucose shifts in hemo- and peritoneal dialysates. BGM is important but inconvenient, and CGMs can therefore improve burden and compliance.
Several studies and case reports have shown improvement in glucose control and variability with CGM use in ESKD and hemodialysis. Some criteria to offer these devices may include patients with multiple episodes of hypoglycemia, T1D, T2D on basal or multiple doses of insulin, or T2D on noninsulin agents with glucose variability.
CGM agreement with BGM has been shown in smaller studies. Flash glucose monitoring has been shown to underestimate BG concentrations measured by BGM, especially with large fluid shifts, but this error was easily compensated by a correction factor.
These accuracy findings need to be further evaluated with newer generation devices and in larger studies. Despite this, the benefits of CGMs are tremendous in this very vulnerable population.
Hemodialysis
Connie M. Rhee, MD, MSc
University of California, Irvine, Orange, CA, USA
ESKD patients are at risk for hypoglycemia via multiple pathways, including decreased renal insulin degradation and clearance, decreased hepatic clearance of insulin, impaired renal gluconeogenesis, co-existing comorbidities (protein energy wasting, gastroparesis), uremic toxin accumulation, and the dialytic procedure (loss of glucose in the dialysate, intradialytic glucose shifts into erythrocytes).
While clinical practice guidelines (ie, Kidney Disease Improving Global Outcomes) still recommend HbA1c for monitoring of glycemic control in patients with diabetes- and non-dialysis-dependent chronic kidney disease, the accuracy and precision of HbA1c measurements decline in patients with advanced chronic kidney disease, particularly among patients treated by dialysis. Across these guidelines, there is growing support for the use of CGM metrics, such as TIR and time in hypoglycemia, as alternatives to HbA1c for defining glycemic targets in some subpopulations.
While further research including clinical trials of the efficacy and safety of CGMs in patients with diabetes and dialysis is needed, this diabetes technology holds promise as a more patient-centered glycemic assessment method that has the potential to enhance adherence, improve health-related quality of life, reduce (hypoglycemia) anxiety, and reduce lifestyle interruption in this high-risk population.
The HbA1c value may not be accurate in end-stage renal failure because of the shortened red cell lifespan and the use of erythropoietin. Fructosamine may also not be accurate because of proteinuria. Continuous glucose monitoring may provide an alternate method to assess glycemic control. The comparison of glycemic monitoring metrics in patients with diabetes and ESKD is shown in Figure 6. There are, however, limited data on accuracy.
Figure 6.
Comparison of glycemic monitoring metrics in patients with diabetes and patients with end-stage kidney disease. Abbreviations: Alb, albumin; Glu, glucose; Hb, hemoglobin; Pro, D-proline; RBC, red blood cell.
Source: Reproduced from Narasaki et al. 15 with permission from John Wiley & Sons—Books; permission conveyed through Copyright Clearance Center, Inc.
Continuous glucose monitoring may also provide additional insight regarding the glucose fluctuations that occur during hemodialysis and peritoneal dialysis. Those individuals who are hyperglycemic coming into a hemodialysis session may see a rapid drop of glucose during dialysis followed by rebound hyperglycemia. The latter may occur because of membrane adsorption of insulin and counterregulatory response to the rapid glucose drop during dialysis. Peritoneal dialysis can lead to large fluctuations in glucose levels, reflecting the glucose concentration of the peritoneal dialysis solution, dwell time, and insulin therapy. It is common to see hyperglycemia within minutes of starting peritoneal dialysis.
Session 6. Research Session on CGMs in the Hospital
Co-Moderators:
Elias Spanakis, MD
Baltimore VA Medical Center and School of Medicine, University of Maryland, Baltimore, MD, USA
Andjela Drincic, MD
University of Nebraska Medical Center and Diabetes and Endocrinology Center, Nebraska Medicine, Omaha, NE, USA
Seven abstracts were presented.
Research Presenters:
Ming Yeh Lee, MD, PhD
Division of Pediatric Endocrinology & Diabetes, Department of Pediatrics, Stanford University, Stanford, CA, USA
Establishing Safe Use of Continuous Glucose Monitoring for Inpatient Diabetes Management
Ming Yeh Lee, MD, PhD, Susan Seav, MD, Maja Ivanovic, MD, Loice Ongwela, CNS, CDCES, Yunzi Gu, CNS, CDCES, Rayhan A. Lal, MD, and Michael S. Hughes, MD
Objective: Stanford University Hospital had no protocol for inpatient CGM use, contributing to negative experiences related to glucose monitoring for patients and bedside nurses. The objective of this quality improvement study was to standardize workflow and increase positive inpatient CGM experience for patients and nurses.
Method: Multidisciplinary stakeholders from the Inpatient Diabetes Task Force and the Resident Safety Council (including endocrinologists, residents, nurses, pharmacists, hospital administration, and Epic IT staff) created and implemented standard workflow for inpatient primary providers and nurses to order and use CGMs, document in Epic, and reference clinical guidelines. Staff communication was developed to raise awareness of the protocol. Over six weeks of the study period (two weeks baseline and four weeks postinterventions), Likert scale satisfaction surveys were administered to patients and nurses who used CGM inpatient.
Result: Fourteen patients and 16 nurses participated in the surveys. Prior to a standardized protocol, 40% of patients found interactions with nursing about their CGM to be a positive experience and 40% found their blood glucose were well controlled. These rates increased to 100% postinterventions. At baseline, 80% of patients thought more could be done to improve inpatient CGM experience, which reduced to 67% postintervention. At baseline, 67% of nurses said they would like to use the CGM system more often, 33% felt confident using the CGM system, and 0% of nurses thought the system was well integrated. All three measures increased to 100% postintervention.
Conclusion: Using a standardized inpatient CGM protocol developed by a multidisciplinary team was associated with more positive patient and nursing experience regarding inpatient CGM use. Work is ongoing to optimize nursing workflow with the Epic interface and create sustainable improvement.
Ashley Yoo, BS
Department of Endocrinology, Diabetes and Nutrition, University of Maryland School of Medicine, Baltimore, MD, USA
Relation of Hemoglobin A1c and Hypoglycemia Incidence Using Continuous Glucose Monitoring in Hospitalized Patients with Type 2 Diabetes in the General Wards
Ashley Yoo, BS, Lakshmi G. Singh, PharmD, Isabel Marcano, MD, Sergio Lizama, MD, William H. Scott, MA, John D. Sorkin, MD, PhD, and Elias K. Spanakis, MD
Objective: Inpatient hypoglycemia is associated with poor outcomes. When evaluated by Point of Care-Blood Glucose (POC-BG), Hemoglobin A1c (HbA1c) upon admission showed predictive value in hypoglycemia risk. The use of real-time Continuous Glucose Monitoring (CGM) may measure hypoglycemic events unidentified with standard POC-BG. It is unclear if lower HbA1c values are predictive of inpatient hypoglycemia risk, when evaluated by CGM.
Methods: A secondary analysis of 70 inpatients with insulin-treated type 2 diabetes (T2D) at risk for inpatient hypoglycemia from a randomized controlled trial. All wore Dexcom G6 CGM devices and were grouped by admission HbA1c ≥7% (N = 56) or <7% (N = 14). Study outcomes include CGM-defined hypoglycemic events (<70 or <54 mg/dL), prolonged hypoglycemic events (<70 or <54 mg/dL for >120 minutes), and time below range (TBR) of <70 or <54 mg/dL. Respectively, Poisson, Quasi-Poisson, or linear regressions were used to calculate relative risk (RR) according to the studied outcome and satisfaction of Poisson parameters then adjusted with zero-inflated beta regression analysis.
Results: Compared to patients with HbA1c <7%, those with HbA1c ≥7% had a nonstatistically significant lower incidence of hypoglycemia for both hypoglycemic events <70 mg/dL (RR: 0.66, 95% CI: 0.34-0.43, p = .33) and <54 mg/dL (RR: 0.78, 95% CI: 0.49-1.24, p = .57), as well as prolonged hypoglycemic events <70 mg/dL (RR: 0.83, 95% CI: 0.44-1.58, p = .78) and <54 mg/dL (RR: 0.26, 95% CI: 0.07-1.03, p = .16). Similarly, there was nonstatistically significant less TBR <70 mg/dL (difference: −22.3 minutes, 95% CI: −47.6 to 3.0, p = .43) and TBR <54 mg/dL (difference: −42.2 minutes, 95% CI: −83.8 to 0.6, p = .36).
Conclusion: Patients with T2D and HbA1c ≥7% did not have a statistically significant lower risk of inpatient hypoglycemia compared to those with HbA1c <7% when evaluated using CGM.
Cory DeClue, MS, MD
Vanderbilt University, Nashville, TN, USA
Continuation of Continuous Glucose Monitors in Hospitalized Diabetic Patients Decreased the Incidence of Hypoglycemia
Cory DeClue, MS, MD, and Shichun Bao, MD, PhD
Objective: Continuous glucose monitoring (CGM) devices have been widely used in the outpatient setting with successful glycemic control. However, data on the benefit of CGM use in the hospital are limited. We performed a retrospective study to further elucidate this topic.
Method: Diabetes consults (n = 197) were analyzed in a single-center inpatient glucose management service during January 2023. Patients were included if managed with insulin via multiple daily injections or pumps in the outpatient setting. Patients were separated based on CGM continuation, and point-of-care (POC) glucose levels were recorded over a 72-hour period.
Result: Type-1 (T1DM, n = 17) and Type-2 (T2DM, n = 26) patients had average hemoglobin A1c levels of 9.1 and 8.6%, respectively. Six T1DM and eight T2DM patients had CGM devices continued during their hospitalization. Average POC glucose levels were not significantly different between groups. However, none of the T1DM or T2DM patients continued on CGM devices had hypoglycemia, whereas four T1DM and four T2DM patients not using CGM experienced at least one hypoglycemic event.
Conclusion: CGM devices have great potential to improve diabetes care in the hospital setting. In this study, we observed significantly less hypoglycemia in patients who continued on CGM devices in comparison to those not using this technology, regardless of the type of diabetes or baseline glycemic control. This study adds further credence to the growing body of evidence supporting CGM use in the hospital.
Lauren A. Waterman, MD
Barbara Davis Center for Childhood Diabetes, Aurora, CO, USA
Accuracy of Dexcom G6 Continuous Glucose Monitor during IV Insulin Administration in Pediatric Type 1 Diabetes
Lauren A. Waterman, MD, Laura Pyle, PhD, Lindsey Towers, CCRP, Emily Jost, MPH, RD, CDCES, CSSD, Angela J. Karami, BS, Cari Berget, RN, MPH, CDE, Gregory P Forlenza, MD, R. Paul Wadwa, MD, and Erin C. Cobry, MD
Objective: Continuous glucose monitors (CGMs) play an integral role in the outpatient management of type 1 diabetes (T1D), but little is known about CGM accuracy during pediatric admissions. IV insulin administration during the treatment of diabetic ketoacidosis (DKA) requires frequent glucose monitoring, which may be facilitated with the use of CGMs.
Method: This retrospective chart review of pediatric encounters evaluated the accuracy of Dexcom G6 CGM versus point of care (POC) blood glucose measurements (Nova Biomedical StatStrip [mean absolute relative difference [MARD] 6%]) during IV insulin infusions. Blood glucose values were collected from the medical record. CGM values were obtained from Dexcom Clarity CSV files, using the closest value within 5 minutes of the POC value.
Result: Paired glucose values (N = 1120) from 83 pediatric patients with T1D (median age 12 years) resulted in an overall MARD of 11.8%. During IV insulin infusion (38 patients, median age 14 years, 355 paired values), the MARD was 15.3% with 98.3% of values within the Clarke Error Grid A&B zones (clinically acceptable values). The % within 15/15, 20/20, and 30/30 were 46.7%, 61.5%, and 80.6%, respectively. When receiving IV insulin on the floor, MARD was 18.8% (N = 211) compared to 10.3% in the ICU (N = 144). Admissions for severe DKA (N = 163) (pH <7.15 and/or bicarbonate <5 mmol/L) had a lower MARD when receiving IV insulin compared to nonsevere DKA admissions (N = 138) (10.1% vs 15.5%, p < .001).
Conclusion: While CGM accuracy during IV insulin infusion is less than overall hospital admission accuracy, it remains within the MARD of other published adult hospitalization analyses. This suggests the feasibility of CGM use during IV insulin administration in pediatrics, particularly in the ICU, where close glucose monitoring is crucial.
Gautam Ramesh, BS
Division of Endocrinology and Metabolism, Department of Medicine, University of California San Diego, La Jolla, CA, USA
Variable Accuracy of Two Continuous Glucose Monitors in the Critical Care Setting
Schafer Boeder, MD, Emily Kobayashi, BS, Navyaa Sharma, None, Gautam Ramesh, BS, Kevin Box, PharmD, Amit Majithia, MD, and Kristen Kulasa, MD
Objective: There is growing interest in using continuous glucose monitors (CGMs) for critically ill patients. However, concerns persist about the accuracy of CGM systems in the critical care setting. In this study, we assess the accuracy of two subcutaneous CGM systems that were applied to patients in the ICU.
Method: Two blinded CGM devices—Dexcom G6 Pro (Dexcom) and FreeStyle Libre Pro (Abbott Diabetes Care)—were simultaneously applied to adult participants (n = 40) who required intravenous insulin infusion in the critical care unit. Serum (Lab) glucose was collected every 4 to 6 hours. Point-of-care (POC) glucose (Accu-Chek Inform II, Roche Diagnostics) was frequently collected as part of routine care. During analysis, Lab and POC glucose values were matched with the nearest CGM value (within 5 minutes) for each CGM system.
Result: Including all collected glucose values, there were 1016 matched Lab-Dexcom pairs with a mean absolute relative difference (MARD) 22.9% (range 6.3%-62.6% for individual patients), 2364 POC-Dexcom pairs with MARD 23.2% (4.9%-62.7%), 334 matched Lab-FreeStyle pairs with MARD 25.8% (9.0%-41.8%), and 776 POC-FreeStyle pairs with MARD 27.6% (9.1%-47.0%).
Conclusion: Preliminary analysis suggests that overall CGM accuracy in the critical care setting is poor. However, a wide range of MARD values were observed among patients, with some CGM sensors being highly accurate and others being clearly unusable. The reason for this inconsistency will need further investigation, including analysis of additional collected patient metadata.
Mikkel Thor Olsen, MD
Department of Endocrinology and Nephrology, Copenhagen University Hospital, Hilleroed, Denmark
In-Hospital Assessment of Glucocorticoids and Infection Severity on Glucose Levels Monitored by Continuous Glucose Monitoring in Patients with Diabetes
Mikkel Thor Olsen, MD, Carina Kirstine Klarskov, MD, PhD, Andreas Kryger Jensen, Biostatistician, PhD, Birgitte Lindegaard, MD, PhD, Ulrik Pedersen-Bjergaard, MD, PhD, and Peter L. Kristensen, MD, PhD
Objective: Glucocorticoids (GCs) and inflammation due to infections might induce hyperglycemia. However, it is not known to what degree infection severity influences glucocorticoid-induced hyperglycemia or vice versa. We assessed the effect of infection severity and GC treatment on glucose levels in hospitalized patients with diabetes mellitus (DM) using continuous glucose monitoring (CGM).
Method: We included hospitalized patients with DM and infection from May 2020 until February 2021. The primary endpoint was to visualize and quantify glucose levels during GC treatment depending on infection severity (measured by CRP) in heat maps. Glucose data were obtained by the Dexcom G6 CGM.
Result: Generally, GC treatment was, independent of infection severity, associated with increased glucose levels, especially during the daytime when GC was administered in the morning. However, heat maps suggest that glucose levels in patients treated with GC and with high CRP levels >100 mg/dL had a peak in glucose levels of 4 mmol/L (72 mg/dL) between (17:00 pm till 21:00 pm) and patients with lower CRP levels <100 mg/dL had a similar peak in glucose levels but earlier between (14:00 pm till 18:00 pm), compared to patients not treated with GC but with similar CRP levels, respectively.
Conclusion: Glucose levels peaked at different time points in patients with diabetes and infection depending on high CRP levels >100 mg/dL or lower CRP levels <100 mg/dL.
Ayan Banerjee, PhD
Arizona State University, Tempe, AZ, USA
Developing a Personalized Diabetes Care Simulator for Analyzing CGM-Based Automated Insulin Delivery Systems for In-Hospital Glucose Control in Type 1 Diabetes
Ayan Banerjee, PhD, Ravinderjeet Kaur, MBBS, Yogish Kudva, MBBS, and Sandeep Gupta, PhD
Objective: To develop a simulator for in-clinic glucose control in type 1 diabetes (T1D) using a CGM-enabled automated insulin delivery (AID) system.
Method: Twenty T1D patients were recruited to perform a single-day in-clinic CGM monitoring study and subsequent at-home monitoring. The in-clinic and at-home CGM data was used to identify a Bergman minimal model-based digital twin of each patient. A personalized diabetes care simulator (PDCS) was developed based on the resulting digital twin. A model predictive control was integrated with the PDCS to show the feasibility of using CGM-enabled AID systems in hospital care.
Result: The average time in range (TIR) obtained for the AID system in the PDCS simulator for the in-clinic day was 73% (15%), the time above 180 mg/dL was 23% (10%), and time below 70 mg/dL was 3.9% (2%). There was no statistically significant difference in the glycemic parameters between in-clinic and at-home days.
Conclusion: The diabetes simulator shows that it is feasible to use CGM-based AID systems to control glycemic variation in a clinical setting. The PDCS simulator can be configured to take into account various forms of hospital visits by patients. The in-clinic day is the least risky form of hospital visit, a more rigorous study on different circumstances for hospital visits, such as emergency or scheduled surgery, needs to be undertaken.
Session 7. Starting a CGM on Hospitalized Patients
Moderator:
Eileen Faulds, PhD, MS, RN, FNP-BC, CDCES
College of Nursing and Wexner Medical Center, The Ohio State University, Columbus, OH, USA
Panelists:
Grazia Aleppo, MD
Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
CGM glucose targets specific to patient acuity need to be established through outcome-driven clinical trials to guide protocols and guidelines (ie, perioperative CGM use or immediate postoperative targets may be different from nonacutely ill patients’ CGM targets).
Inpatient CGM use will require revision of inpatient hypoglycemia protocols toward addressing hypoglycemia trends and prevent/reduce its severity.
Nursing staff education curricula are needed to transform the approach to hyperglycemia in the hospital from BG-driven insulin dosing to sensor glucose-guided insulin administration protocols.
Gerry Rayman, MD, FRCP
The Diabetes Centre, Ipswich Hospital, East Suffolk and North Essex Foundation Trust, Ipswich, UK
Given that currently, in the United Kingdom (UK) the use of web-linked capillary BG meters and EHRs are generally limited to large hospitals, starting CGMs on a U.K.-wide basis would be a significant challenge.
In the first instance, this would be best limited to those groups of people who would most benefit from the use of a CGM—these would have to be defined.
Introduction on a U.K.-wide basis would need to be in a programmed, stepwise fashion, learning from early adopters and then rolled out with educational support from other institutions.
Archana R. Sadhu, MD, FACE
Weill Cornell Medicine, Texas A&M Health Sciences and Houston Methodist Hospital, Houston, TX, USA
FDA clearance and cost-effectiveness data are necessary for hospital system stakeholders to invest in CGM technology as a new standard of care over BGM in the inpatient setting.
Real-time integration of CGM data into the EHR in a clinically useful format is an essential next step.
New clinical protocols are needed for translating CGM trends into immediate therapeutic interventions that will improve patient outcomes.
Jagdeesh Ullal, MD, FACE, FACP
University of Pittsburgh Medical Center, Pittsburgh, PA, USA
Each institution or hospital system should design a plan and create protocols to manage CGM data on a real-time basis, employing unit dashboards or sending data to applications on secure handheld devices that alert clinicians to rapid changes in glucose trends.
CGMs in the hospital may be useful for PWDs, clinicians, and providers. Clinicians and providers need to be aware of those groups of patients with hyperglycemia or diabetes in whom CGM data may not be reliable as well as possible drug interference that can cause data to be erroneous.
The risks and benefits of CGM use may have to be balanced with cost to health care institutions as well as insurance coverage vis-a-vis reduction in LOS, emergency department (ED) visits, or hospital readmissions.
Dr. Aleppo opened with a brief discussion of CGM accuracy and the need to augment such discussions to include existing knowledge regarding accuracy in the outpatient setting. Figure 7 highlights the logistics of initiating CGM use in the hospital setting, including a discussion of stakeholders, processes, approvals, training, and protocols.
Figure 7.
Logistics of implementing a CGM program in the Hospital.
Abbreviations: EHR, electronic health record; IT, Information Technology; P&T, pharmacy and therapeutics
Source: Figure courtesy of Grazia Aleppo.
The panel then began discussing inpatient populations that might experience particular benefits from CGMs. Dr. Rayman discussed the use of CGMs in pregnant patients within the United Kingdom and mentioned that health systems should be prepared to manage the growing population of outpatient CGM users admitted to the hospital. The panel later expanded on this topic to discuss initiation by unit type (eg, ICU, Medical-Surgical) and personal vs. hospital-initiated CGMs.
Dr. Ullal discussed nursing diabetes training initiatives and programs within his health system, which includes 25 hospitals, and highlighted the complexity of training amid staff turnover and the high volume of travel nurses. Dr. Sadhu expanded on this point by highlighting how CGMs could potentially relieve the nursing burden and how the economic benefits of this use case should be explored.
The panel received multiple questions regarding the initiation of CGMs in smaller, nonacademic hospitals. Many of these smaller hospitals may not have endocrinology consult services or CDCES, and therefore, the stakeholders and champions may vary by the type of health system and resources available. Dr. Sadhu closed by mentioning that inpatient CGM clinical use, while growing under the FDA COVID emergency provisions, really hinges in the long term on FDA clearance.
Session 8. AID Systems in the Hospital
Moderator:
Irl B. Hirsch, MD
University of Washington Medicine Diabetes Institute, Seattle, WA, USA
Panelists:
Lia Bally, MD, PhD
Inselspital (Bern University Hospital) and University of Bern, Bern, Switzerland
Fully closed-loop systems are designed to automate all insulin delivery on the sole basis of CGMs, without requiring input for mealtime bolusing.
Closed-loop delivery systems can accommodate the rapidly changing insulin requirements in the hospital environment (eg, on/off nutrition support, use of glucocorticoids, surgical stress, etc.)
Closed-loop systems that are simple to use and fully automated have the potential to reduce the workload burden of hospital staff.
Stan Klek MD, CDCES
Division of Endocrinology, NYU Langone Hospital-Long Island, Mineola, NY, USA
Patients should be counseled on the institutional policy for continued usage of HCL/AID systems while hospitalized and sign a collaborative agreement to participate with the hospital staff in co-managing diabetes during admission.
Expert-level clinical staff (endocrinologists or RN CDCES) must be available and consulted on all cases to support and educate patients and other clinical staff in hospital usage of HCL/AID systems.
Insulin pumps and CGMs should be removed or shielded prior to radiologic procedures. Care coordination/communication between floor teams and ancillary staff must take place to ensure that all providers are aware that insulin pumps are in place.
Felicia A. Mendelsohn Curanaj, MD
Weill Cornell Medicine, New York, NY, USA
Institution-specific protocols, including detailed prescriber and nursing responsibilities, should be created for the safe use of AID systems in the inpatient setting.
The accuracy of AID systems may be impaired by medical conditions that arise affecting skin perfusion, blood pressure, or body temperature.
The decision regarding the continuation of an AID system during the perioperative period must take into account the duration of the procedure, recovery time, and any planned exposure to an electromagnetic field.
Francisco J. Pasquel, MD, MPH
School of Medicine, Emory University, Atlanta, GA, USA
There are an increasing number of AID systems and an increase in the number of patients using these technologies. Patients may get hospitalized while using these technologies, and there is a need for protocols and workflows to ensure a safe continuation of AID systems in the hospital.
Research on the use of AID systems in the hospital has shown that this technology is superior to the standard of care. However, several implementation barriers prevent the successful initiation of this technology, particularly for patients who have not used it previously.
Further research is needed in real settings to determine the cost-effectiveness of using AID systems in the hospital.
The status of AID systems in the United States and Europe is different in clinical outpatient practice than in the hospital. The use of these systems is exploding in both the United States and the European Union (although different systems are used) yet, in the hospital, uptake is much slower. Figure 8 illustrates AID system use in an admitted patient. In a survey of 50 participants at the conference (the number of MDs, RNs, PharmD’s, etc. was not disclosed, and information was not collected on how many had glycemic management teams in their hospitals), 92% would allow both pumps and CGM use in their hospitals, while, in a hospitalized patient admitted to a non-ICU bed with an infected foot ulcer, 72% would allow continuation of that person’s AID system. However, if that same patient needed to be transferred to an ICU setting, only 8% would allow that person to stay on an AID system, while 76% would come off both pump and CGM and move to multiple injections and fingerstick glucose management. Many concerns were raised by the panelists about AID system use during surgery, although for relatively minor, short procedures, they could be considered. The problem of the CGM not working with electrocautery in the same area was discussed, meaning that for some of these patients, “manual mode” would be required.
Figure 8.
An illustration of components in an automated insulin delivery system.
Source: Figure reproduced from Bally. 16
For admitted patients already using AID systems, the panel felt strongly that they needed to be self-competent to manage their AID system and their diabetes in general. This point may not be as simple as it sounds if the person making the decision is not familiar with AID systems. In addition, the individual’s baseline cognition to self-manage their diabetes needs to be evaluated. Therefore, communication with a patient’s outpatient diabetes provider is critical.
One point where there was strong agreement was the need for clinical trials with AID systems in the hospital. The panel presented a case of a patient admitted with an AID system starting high-dose steroids. 17 The panelists thought the patient could be continued on their AID system at first but would quickly move to a multiple daily injection insulin therapy, or even an insulin drip, if glucose levels were too high, despite changes in pump settings. Given how the AID systems differ in their algorithms, it is quite possible different AID systems would behave differently, and this is another area where further research would be welcomed.
Session 9. CGMs in Children
Moderator:
Juan Espinoza, MD, FAAP
Ann & Robert H. Lurie Children’s Hospital of Chicago and Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Panelists:
Michael S.D. Agus, MD
Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA
CGMs are powerful tools that have enormous potential in the inpatient setting but need substantial modifications by the manufacturers for successful implementation, including clear display, loud and easily programmable alarms, real-time output to the EHR, and bedside monitors.
CGMs have high-value use cases in both diabetes and nondiabetes settings. Diabetes scenarios include known and new-onset diabetes. Nondiabetes scenarios include ICU hyperglycemia as well as fasting protocols with diagnostic workup for metabolic disorders.
Glucose control in the ICU with IV insulin, guided by a CGM and a computer algorithm, generally does not provide benefit to critically ill children. However, in two distinct hyper-inflamed populations in retrospective analysis, survival benefit was recently demonstrated when randomized to 80 to 110 mg/dL as compared to 150 to 180 mg/dL. This needs to be demonstrated in a prospective randomized clinical trial before implementation in general care.
Eda Cengiz, MD, MHS, FAAP
School of Medicine, University of California, San Francisco, San Francisco, CA, USA
CGM technology has revolutionized outpatient diabetes treatment, and we expect that CGMs will be increasingly used in the hospital for PWDs to detect and prevent hypoglycemia and hyperglycemia while managing patients with and without diabetes.
The implementation of a CGM program in the hospital requires the cooperation of physicians, nurses, hospital leadership, and EHR technology support.
As inpatient CGM use increases, physicians and nurses need to become familiar with this technology and its advantages and limitations. Insights gained from data derived from inpatient CGM use would be used to build advanced decision support to inform future treatment and warning systems to improve inpatient treatment and safety.
Mary Pat Gallagher, MD
Hassenfeld Children’s Hospital at NYU Langone and NYU Grossman School of Medicine, New York, NY, USA
Increased user satisfaction reported in children with diabetes and their families makes continued use of CGMs while hospitalized highly desirable in this population. In addition, there are potential benefits to having the ability to trend glucose levels in other populations, including children experiencing hyperglycemia in the inpatient setting for the first time and children treated with steroids, chemotherapy, and other medications commonly used in children with malignancy and organ transplantation.
Instituting CGMs in the hospital setting for pediatric patients may require separate validation in different age groups: neonate, toddler, school-aged child, and adolescent. To improve outcomes, developing programs that can interface with the EHR and notify providers of measurements outside of the individualized target ranges will be critical.
The pediatric population is at low risk for adverse cardiovascular events associated with hypoglycemia. It will likely require different glycemic targets than those used as benchmarks in the inpatient adult population. These targets may need to be adjusted further according to diagnosis (critical illness, leukemia, or suspected hyperinsulinism in the neonate) and clinical settings (pediatric ICU, neonatal ICU, operating room, or subacute setting).
Rayhan A. Lal, MD
Stanford University, Stanford, CA, USA
We have more data and studies regarding CGM use in the hospital in the adult literature. Often these adult hospitalizations expose the individual to more situations where sensor accuracy may be affected (eg, electrocautery, defibrillation). We should have equity in CGM research between the pediatric and adult inpatient settings.
All methods of glucose measurement will carry some error. What is not clear is how much error (especially those resulting in injury) we are willing to tolerate in the inpatient, monitored setting.
It is challenging for any health care provider to respond to 288 data points per day for even one person. Automation is needed to fully utilize CGMs in a hospital setting.
The effective use of CGMs in the outpatient setting, for both adults and pediatric populations, has become a revolutionary tool in the treatment of diabetes. Yet, the question discussed in this session is whether its use in the inpatient pediatric setting can be equally impactful. The use of CGMs in hospitalized pediatric patients, both with and without diabetes, is becoming more widespread. Figure 9 outlines several potential advantages and challenges associated with using CGMs in pediatric patients. Current inpatient CGMs are those with pre-existing or new diagnoses of diabetes. The use of CGMs in patients without diabetes has also been explored, especially in ICU monitoring of patients with prolonged hyperglycemia and in fasting diagnostic workups for metabolic disorders.
Figure 9.
Potential advantages and challenges of inpatient CGM use in children. Abbreviation: CGM, continuous glucose monitor.
Source: Figure courtesy of Eda Cengiz.
CGM data analysis is fundamentally different in the hospital versus the outpatient setting. Outpatient diabetes management measures, such as TIR, likely will not translate to hospitalized patients. More evidence is needed to define new inpatient CGM metrics that are responsive to both acuity and different diagnoses. The timeliness of inpatient CGM data is a robust fix for previously problematic delays. However, individualized consideration of patient context is extremely important in determining the accuracy and interpretability of its data.
Caution must be taken to avoid thinking the use of inpatient CGM data obviates the need for POC testing. This can be an especially challenging concept for patients and families who use CGMs almost exclusively with only limited home POC BG testing. In some higher-risk use cases, such as oncology patients or those with significant anemia where repeated POC BGM could be consequential, the use of CGMs is particularly attractive. While we have yet to unequivocally prove better outcomes with CGM use in one population versus another, new evidence is emerging to help define appropriate use-case scenarios. For example, the recent study by Zinter et al 18 demonstrated a mortality benefit to tighter BG control in inflammatory conditions in the ICU with CGM use.
Unique challenges of CGMs in the pediatric inpatient setting include the potential for CGM-naive families to become burdened by the amount of data the device generates, possibly leading to decreased parent/child satisfaction. Clearly communicating the capabilities and limitations of this technology is vital to manage expectations. During the COVID-19 pandemic, the use of CGMs in the inpatient setting rapidly increased under the FDA’s enforcement discretion, but most of the data generated was in adults. Best practices in CGM planning, implementation, and management for pediatric inpatients remain poorly defined. Another important consideration is measurement error. All measuring devices have inherent errors that are generally accounted for in clinical practice patterns. As a new use case for CGMs, it is unclear what the appropriate error tolerance should be for inpatient pediatric use and still be considered safe, or if those parameters should be different in children versus adults. For example, mild hypoglycemia is better tolerated in pediatric populations, compared to adults for whom immediate action is necessary, especially in those with cardiovascular comorbidities. Workflows, EHR integrations, protocols, and financial implications are potential barriers to widespread inpatient implementation. Despite these barriers, there are clear advantages to continuing to pursue judicious inpatient use of CGMs among a variety of pediatric patients.
Session 10. Data Integration of CGMs for Inpatient Use and Telemetry
Moderator:
Juan Espinoza, MD, FAAP
Ann & Robert H. Lurie Children’s Hospital of Chicago and Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Panelists:
Mark A. Clements, MD, PhD
School of Medicine, University of Missouri-Kansas City, Kansas City, MO, USA
CGM is FDA-cleared for use in ambulatory care but not for use across in-hospital critical and non-critical care settings, yet real-time CGM and glucose telemetry have been shown to promote early detection and prevention of hypoglycemia and reduce hyperglycemia.
RT-CGM allows healthcare providers to observe and anticipate near-term and longer-term trends in glucose levels, promoting more timely “in-the-moment” and pattern-driven decisions.
Remote real-time glucose monitoring decreases the need for frequent POC checks, reducing staff exposure in infectious settings and increasing staff efficiency.
Juan Espinoza, MD, FAAP
Ann & Robert H. Lurie Children’s Hospital of Chicago and Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
The lack of integration and analysis of RT-CGM data in the hospital setting represents a technical gap.
The next generation of CGMs will more readily support data integration, analysis, and reporting in the inpatient setting.
RT-CGM data in the hospital should leverage existing pathways and architecture used by other bedside physiologic monitors. This typically includes connecting to a hub or gateway, with real-time data stored in an intermediary database. An integration engine then brings the data into the EHR, either ad hoc (when requested) or at scheduled intervals.
Samantha R. Spierling Bagsic, PhD, MSE
Scripps Whittier Diabetes Institute, Scripps Health, San Diego, CA, USA
Within our health system, CGM data integration with our EHR is limited to manual entry of glucose values when a care provider treats based on CGM data. Otherwise, integration of CGM and EHR data is retrospectively done for evaluation purposes outside of the EHR system in cloud-based tools. Integration of CGM data into the EHR automatically is crucial for expanding the use of inpatient CGMs.
Currently, to implement a CGM standard of care program in our health system, different software applications are utilized by multiple different providers to tailor CGM information specific to the provider role. For instance, a centralized (off-site) CGM Advanced Practice Nurse follows near real-time data to recommend insulin dosage changes; bedside RNs use trend data to implement alert checking and follow predefined glucose management algorithms; and a remote monitoring team uses a separate platform designed to prioritize patients by clinical risk and provide alerts based on ongoing or impending severe dysglycemia. Integration of CGMs for inpatient use must consider the different data needs of different providers to curate the information they receive.
Validation of CGMs against standard POC testing remains a challenge for a small subset of patients, and standard consensus validation criteria in the inpatient setting are warranted to determine when CGMs can appropriately be used for clinical management. Additional research is needed to identify who the best candidate is for CGMs and if there are subgroups more prone to inaccuracies against POC BG testing.
Scott Weinstein, JD
McDermott Will & Emery LLP, Washington, DC, USA
Under the 21st Century Cures Act, EHR vendors must deploy Fast Health Interoperability Resources (FHIR)-based Application Programming Interfaces (APIs) to certified EHRs, which will permit hospitals and health systems to connect third-party applications, including potentially CGM applications, to the EHR.
The information-blocking provisions of the 21st Century Cures Act place parameters around the fees that EHR vendors may charge to integrate with EHRs.
Approved clinical decision support mechanisms within EHRs and CGM applications may assist hospitals and health systems in obtaining more detailed data from CGMs by providing them the ability to quickly identify and flag potentially critically relevant clinical information.
CGMs in the outpatient context are relatively ubiquitous and backed by FDA clearance; this is not yet true for inpatient use. Inpatient use of CGMs is attractive for its ability to observe and anticipate glucose trends, thus producing more timely, pattern-driven decision-making in the prevention and treatment of hypo- and hyperglycemia. CGM monitoring decreases the need for frequent POC BG checks, which ultimately improves patient experience, reduces staff exposure in infectious settings, and increases efficiency.
This session began with a deeper dive into Dr. Bagsic’s work. Three years ago, her team developed a system that allows for remote real-time CGM data so that a CGM Standard of Care Program could be developed at Scripps Whittier Diabetes Institute as part of Scripps Health. Figure 10 illustrates the aspects of the CGM Standard of Care Program. The key drivers of their program include a number of different software and platforms used for each end user of the data. In terms of the actual integration with the EHR, that workflow remains a manual process for now. Three years into the program, the takeaway is that with the current state of the technology and approvals, a standard of care program is feasible by using specialized teams to promote scaling.
Figure 10.
CGM as standard of care (CGM as SOC) program at Scripps Health. Abbreviations: APN, advanced practice nurse; CGM, continuous glucose monitor; EHR, electronic health record; RMT, remote monitoring team; RN, registered nurse; SOC, standard of care; T1D, type 1 diabetes; T2D, type 2 diabetes.
Source: Figure reproduced from Spierling Bagsic et al. 19
With the potential for managing an overwhelming amount of data, this particular program formed teams of individuals whose sole responsibility was to receive and review the data that would ultimately inform clinical decision-making at the bedside. With set rules in place and alert recognition of specific metrics, many of the remote monitoring teams are not clinically trained. Central to their roles, however, is a digital dashboard that aids in rapidly prioritizing patients based on clinic risk. From a much broader perspective, this particular program has multiple systems that present curated data for the particular individual that needs it.
The ability to move CGM use into the inpatient space will present opportunities that remove siloed manufacturer/aggregator software solutions to begin setting up automated alert processes. Technical gaps remain given that the current technologies were not designed for the type of real-time or near real-time monitoring that is required in hospitals. Other real-time monitoring devices in use today, such as EKG leads, heart rate leads, and pulse ox sensors all have to conform to specific data standards. For real-time inpatient CGMs to become a reality at scale, CGMs will also need to conform to these standards. Data conformance may become part of the FDA clearance process for inpatient CGMs.
The 21st Century Cures Act requires that certified EHR vendors must deploy FHIR-based APIs. This will then allow hospitals and health systems to connect third-party applications, potentially including CGM applications, to the EHR. If CGM data are being used by clinicians to make decisions and documented in the medical record, it will likely be subject to patient access requests. Given the complex data architecture, it is unclear which data elements should be stored and therefore available to patients and other organizations requesting records. Another potential complication is whether or not inpatient CGM data will be made available via the manufacturers’ portals to patients and their families, as it currently is for outpatient CGMs. These issues will need to be addressed by both industry and healthcare organizations as inpatient CGMs become a reality.
Session 11. Accuracy of CGMs/Comparison with POC BG Testing
Moderator:
Carlos E. Mendez, MD, FACP
Medical College of Wisconsin, Milwaukee, WI, USA
Panelists:
Joseph A. Aloi, MD, FACP, FACE
Atrium Health Wake Forest Baptist and Wake Forest School of Medicine, Winston-Salem, NC, USA
As inpatient CGM use expands, metrics specific to inpatient care versus traditional ambulatory measures for glucose control will need to be developed.
Continuous glucose data supplied by CGMs provides the opportunity for predictive alerts for clinically significant hypo- and hyperglycemic excursions.
The adoption of CGMs in the inpatient setting will not immediately replace traditional POC BGM, but it is an effective addition to our ability to monitor glucose and safely administer insulin.
Shichun Bao, MD, PhD
Vanderbilt University Medical Center, Nashville, TN, USA
CGM glucose correlates well with POC BG values. It is accurate and helpful to use CGMs to monitor glucose trends and assist in diabetes management in hospitalized patients.
CGMs tend to underestimate glucose readings in the first few days of sensor wear, which may help to improve patient safety by increasing the detection of hypoglycemia events.
CGMs should be used as an adjunctive tool for hospital diabetes management.
Rodolfo J. Galindo, MD, FACE
School of Medicine, University of Miami Miller, Miami, FL, USA
Studies reporting the accuracy of newer CGMs in the hospital have shown acceptable performance, particularly in noncritical settings.
Current accuracy studies are not sufficient for regulatory clearance at this time, given their retrospective nature, small sample sizes, lack of standardization of accuracy metrics, heterogeneity of the populations, and use of reference comparators that are not considered to be a gold standard.
Future studies are critically needed since the demand and interest in using CGMs in the hospital is high and expected to continue to increase.
James H. Nichols, PhD, DABCC, FAACC
Vanderbilt University Medical Center, Nashville, TN, USA
There is currently a lack of consensus regarding the optimum comparative method (YSI, glucose meter, central lab method calibrated to Mass Spectrometry) or means for determining CGM accuracy (capillary versus venous sample).
Given that CGM samples from the interstitial fluid provide continuous values, clinical trends may be a more important determinant of CGM reliability than point result accuracy.
While recent standards (the Integration of Continuous Glucose Monitoring Data into the Electronic Health Record) 20 have been released for documenting CGM data in hospital electronic records, a single universal interface is yet to be developed that connects all available CGM devices to any hospital electronic record.
This was an exciting and interesting session feeling like a natural succession after the previous discussions addressing more of the logistical and operational aspects of CGM implementation in the hospital. The pressing question: is the accuracy of CGMs good enough to allow for its routine use in hospitalized patients? Fortunately, each of the panelists had first-hand experience and shared valuable information.
Dr. Aloi, whose recent research involved the use of CGMs in non-critically and critically ill hospitalized patients, stated that, given its unquestionable benefits, especially around the detection of hypoglycemia, the adoption of CGMs in the hospital seemed unavoidable. 21 However, the current operational and logistical challenges place us in a “transition phase” where a complete replacement of fingerstick BGM (POC BGM) seems unlikely in the near future. Along these lines, Dr. Bao shared with the audience the results of her study, where the accuracy of two CGM devices (Freestyle Libre and Libre 2) was evaluated. 22 The relative difference of both CGM devices compared to POC BG showed 98.4% of values falling in acceptable zones A and B of the Clarke Error Grid, as well as an acceptable MARD variability of 21.4 and 17.7% for Libre and Libre 2 CGMs, respectively. Figure 11 depicts the accuracies of these CGMs in hospitalized patients. She concluded that, considering the accuracy together with the clinical utility, implementation of CGMs should be considered at least in conjunction with POC BGM.
Figure 11.
Libre (CGM-1) and Libre 2 (CGM-2) accuracy in adult hospitalized patients. The relative difference between CGM and POC capillary BG remains unchanged regardless of (A) POC BGM value and (B) duration of time on the sensor. 98.4% of values fall in acceptable zones A and B of (C) the Clarke Error Grid. (D) Mean absolute relative difference variability among individual participants is high. The figure presents n = 44 (Libre) and n = 36 (Libre 2) participants, with a total of 676 paired glucose values. Abbreviations: BG, blood glucose; BGM, blood glucose monitoring; CGM, continuous glucose monitoring; POC point-of-care; p = ns, probability is not significant.
Source: Figure reproduced from Wright et al. 22
Dr. Galindo took the opportunity to emphasize the point that traditional methods to test the accuracy of glucose monitors may not necessarily be applicable when it comes to the use of CGMs in the hospital. He suggested that, because CGMs increasingly allow clinicians the possibility of targeting lower overall glycemic targets in a safe manner, collaborations between clinicians/investigators and regulatory bodies are essential to come up with a consensus about which methods and levels of accuracy will be needed before the universal adoption of inpatient CGMs as a replacement of POC BGM can occur.
In agreement, Dr. Nichols expanded on the current requirements that manufacturers are enforced to meet for regulatory approval, including the 15/15 or 20/20 metrics. He highlighted, however, that these metrics were not based on clinical cutoffs and that to adequately evaluate the accuracy of CGMs in the inpatient setting, input from clinicians and regulators may also be necessary. In summary, it was concluded from the session that, given the significant value of using CGMs in high-risk hospitalized patients, efforts should be made to encourage its current use in conjunction with POC BGM at least until further guidance is produced by regulatory bodies.
Session 12. Discharge Planning with CGMs
Moderator:
David Kerr, MuBChB, DM, FRCPE, FRCP
Diabetes Technology Society, Burlingame, CA, USA
Panelists:
Nuha Ali El Sayed, MD, MMSc
American Diabetes Association, Joslin Diabetes Center and Harvard Medical School, Boston, MA, USA
Discharging PWDs on technology from the inpatient setting should be standardized through a timely process that is PWD-centric and team-based.
Include the PWD on technology’s caregivers and support system in the discharge process. Involve community support as needed.
All technology equipment must be available for the PWD/their caregivers at the time of discharge.
Ann Levine, FNP, CNS, BC-ADM, CDCES
Montefiore Medical Center, Bronx, NY, USA
It is important to identify health inequities, including literacy, numeracy, computer/technology literacy, and resources.
We must build bridges for our patients experiencing health inequities to maximize custom resources and search for novel approaches.
We should use glucose sensors to teach and inspire.
Daniel J. Rubin, MD, MSc, FACE
Lewis Katz School of Medicine, Temple University, Philadelphia, PA, USA
Successful postdischarge follow-up with a CGM means glucose data is captured and reviewed.
Barriers to follow-up include low technological literacy, inadequate education, lack of outpatient provider support, high financial cost (inadequate insurance coverage), and lack of supplies.
Facilitators to follow-up include involving a family member/caregiver, inpatient and outpatient education, engagement of outpatient diabetes care provider, determining coverage and costs before discharge, and provision of supplies before discharge.
Jane Jeffrie Seley, DNP, MPH
Weill Cornell Medicine, New York, NY, USA
PWDs who wish to wear their personal CGM during hospitalization should be supported whenever possible, after an initial assessment is done and education is provided on when and how CGM data and trend arrows can be utilized in the inpatient setting.
CGMs should be considered on select high-risk individuals and placed just prior to hospital discharge to facilitate insulin dose adjustments, reduce hypo- and hyperglycemia, and prevent ED visits and readmissions.
When choosing a personal CGM prior to discharge, accommodations need to be made to address health equity concerns, including lack of coverage, smartphones, frequency of use, and ability to perform self-care.
Rates of hospital readmission following discharge for PWDs are recognized to be double compared to individuals without diabetes. 23 With the increasing use of CGMs in hospitals, it is hoped that this technology can contribute to lowering the rate of readmissions as well as preventing hypoglycemia and/or uncontrolled diabetes. Although the indications for in-hospital use of CGMs are expanding, there is currently a paucity of research to guide the selection of PWDs who would benefit from this technology in the postdischarge period (defined for this discussion as the first 30 days after discharge). Several important elements of a safe discharge for a PWD are outlined in Figure 12.
Figure 12.
Important elements of a safe discharge for a patient with diabetes.
Source: Figure courtesy of Nuha El Sayed.
Ideally, prior to discharge, the CGM should be applied at least two days beforehand to provide opportunities for education and to ensure an understanding of the goals of using CGMs. Wherever possible, the in-hospital glucose management team should be involved as early as possible. Indications for CGMs to be continued after discharge include individuals discharged on multiple daily injections of insulin, people started on insulin while in hospital, those individuals where hypoglycemia is a concern, temporary use of steroid treatment in a person with diabetes, the use of enteral feeding, resolving infection (eg, an infected foot ulcer), pregnancy, and for people with a history of recurrent diabetic ketoacidosis.
To close the loop of care, it is imperative that the postdischarge team reviewing the patient is informed of the use of a CGM and is aware of the goals of treatment in the immediate postdischarge period. Obtaining supplies (with help from specialty pharmacists if available) and timely and appropriate insurance coverage is also key. Follow-up visits can be in-person or virtual according to local circumstances. User-support materials should take into consideration issues such as preferred language as well as literacy, numeracy, and digital literacy skills.
In the peridischarge period, glucose targets need to be individualized and shared with all stakeholders (including caregivers). The level of diabetes expertise in the team providing postdischarge care should be taken into consideration. One evidence-based approach is the use of a structured discharge care plan based on the admission HbA1c level. 24
The challenge for the diabetes care community is to ensure that, with the expanded use of CGMs in the hospital, after discharge, PWDs have equitable access to the technology and that their care providers are fully skilled in helping to deliver the evidence-based benefits associated with the use of CGMs.
Abbreviations
APN, advanced practice nurse; Alb, albumin; APIs, application programming interfaces; AID, automated insulin delivery; BG, blood glucose; BGM, blood glucose monitoring; CDCES, certified diabetes care and education specialist; CGM, continuous glucose monitor; DM, diabetes mellitus; HER, electronic health record; ED, emergency department; ESKD, end-stage kidney disease; FHIR, fast health interoperability resources; Glu, glucose; GR, glycemic ratio; Hb, hemoglobin; HbA1c, Hemoglobin A1c; HCL, hybrid closed-loop; ICU, intensive care unit; IV, intravenous; LOS, length of stay; MARD, mean absolute relative difference; NPO, Nil Per Os; PWD, a person with diabetes; POC, point-of-care; Pro, D-proline; RT-CRM, real-time continuous glucose monitoring; RBC, red blood cell; RN, registered nurse; RMT, remote monitoring team; ROI, return on investment; SOC, standard of care; TIR, time in range; T1D, type 1 diabetes; T2D, type 2 diabetes; UK, United Kingdom; FDA, US Food and Drug Administration.
Acknowledgments
The authors thank Annamarie Sucher-Jones for her expert editorial assistance.
Footnotes
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: AW receives research salary support from UnitedHealth Group.
EKS was partially supported by the VA MERIT award (#1I01CX001825) and CSP #2002 from the United States (U.S.) Department of Veterans Affairs. EKS has received unrestricted research support from Dexcom, MannKind Cooperation and Tandem (to Baltimore VA Medical Center and University of Maryland) for the conduction of clinical trials. EKS has received fees from the Medscape (speaker fees) and the Endocrine Society (ESAP 2024-2029 editions).
EF has research, consulting, and speaker disclosures with Dexcom. EF has speaker disclosures with Medscape.
IRB has research with Dexcom and consults for Abbott Diabetes Care, Hagar, and Embecta.
JCE receives funding from FDA, NIMHD, and NCATS and consults for Sanofi.
DK has received research support from Abbott Diabetes Care.
DCK is a consultant for Better Therapeutics, Eoflow, Integrity, Lifecare, Nevro, Novo, Sanofi, and Thirdwayv.
TT, REA, AMY, JH, AD, JJS, GG, UM, GED, and CEM had no disclosures.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The meeting was supported by grants from Abbott, Dexcom and Glytec to Diabetes Technology Society.
ORCID iDs: Tiffany Tian 
https://orcid.org/0009-0003-1417-6445
Rachel E. Aaron
https://orcid.org/0009-0005-5120-2264
Andrea M. Yeung
https://orcid.org/0000-0002-5592-453X
Jingtong Huang
https://orcid.org/0000-0002-3119-9361
Andjela Drincic
https://orcid.org/0000-0001-8365-7662
Jane Jeffrie Seley
https://orcid.org/0000-0003-1582-4320
Amisha Wallia
https://orcid.org/0000-0002-3183-4062
Gregory Gilbert
https://orcid.org/0000-0002-4697-2281
Elias K. Spanakis
https://orcid.org/0000-0002-9352-7172
Umesh Masharani
https://orcid.org/0000-0002-3269-804X
Eileen Faulds
https://orcid.org/0000-0003-2315-1146
Irl B. Hirsch
https://orcid.org/0000-0003-1675-8417
Gigi E. Dawood
https://orcid.org/0009-0008-2612-6687
Juan C. Espinoza
https://orcid.org/0000-0003-0513-588X
Carlos E. Mendez
https://orcid.org/0000-0003-1875-6124
David Kerr
https://orcid.org/0000-0003-1335-1857
David C. Klonoff
https://orcid.org/0000-0001-6394-6862
References
- 1. Turchin A, Matheny ME, Shubina M, Scanlon JV, Greenwood B, Pendergrass ML. Hypoglycemia and clinical outcomes in patients with diabetes hospitalized in the general ward. Diabetes Care. 2009;32(7):1153-1157. doi: 10.2337/dc08-2127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Davlyatov G, Borkowski N, Qu H, Burke D, Bronstein J, Brickman A. Is there an association between quality care and financial performance in community health centers? J Health Care Finance. 2019. Accessed June 9, 2023. http://healthfinancejournal.com/index.php/johcf/article/view/176.
- 3. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med. 2001;345(19):1359-1367. doi: 10.1056/NEJMoa011300. [DOI] [PubMed] [Google Scholar]
- 4. Krinsley JS. Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clin Proc. 2004;79(8):992-1000. doi: 10.4065/79.8.992. [DOI] [PubMed] [Google Scholar]
- 5. Intensive versus conventional glucose control in critically ill patients. N Engl J Med. 2009;360(13):1283–1297. doi: 10.1056/NEJMoa0810625. [DOI] [PubMed] [Google Scholar]
- 6. Bohé J, Preiser JC. Individualized glycaemic management for critically ill patients. Intensive Care Med. 2022;48(1):128-129. doi: 10.1007/s00134-021-06572-2. [DOI] [PubMed] [Google Scholar]
- 7. Krinsley JS, Rule P, Pappy L, et al. The interaction of acute and chronic glycemia on the relationship of hyperglycemia, hypoglycemia, and glucose variability to mortality in the critically ill. Crit Care Med. 2020;48(12):1744-1751. doi: 10.1097/CCM.0000000000004599. [DOI] [PubMed] [Google Scholar]
- 8. Krinsley JS, Roberts G, Brownlee M, et al. Case-control investigation of previously undiagnosed diabetes in the critically ill. J Endocr Soc. 2023;7(2):bvac180. doi: 10.1210/jendso/bvac180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Krinsley JS, Bruns DE, Boyd JC. The impact of measurement frequency on the domains of glycemic control in the critically ill–a Monte Carlo simulation. J Diabetes Sci Technol. 2015;9(2):237-245. doi: 10.1177/1932296814566507. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Galindo RJ, Umpierrez GE, Rushakoff RJ, et al. Continuous glucose monitors and automated insulin dosing systems in the hospital consensus guideline. J Diabetes Sci Technol. 2020;14(6):1035-1064. doi: 10.1177/1932296820954163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Davis GM, Faulds E, Walker T, et al. Remote continuous glucose monitoring with a computerized insulin infusion protocol for critically ill patients in a COVID-19 medical ICU: proof of concept. Diabetes Care. 2021;44(4):1055-1058. doi: 10.2337/dc20-2085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Faulds ER, Boutsicaris A, Sumner L, et al. Use of continuous glucose monitor in critically Ill COVID-19 patients requiring insulin infusion: an observational study. J Clin Endocrinol Metab. 2021;106(10):e4007-e4016. doi: 10.1210/clinem/dgab409. [DOI] [PubMed] [Google Scholar]
- 13. ElSayed NA, Aleppo G, Aroda VR, et al. 15. Management of diabetes in pregnancy: standards of care in diabetes—2023. Diabetes Care. 2022;46(suppl 1):S254-S266. doi: 10.2337/dc23-S015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Battelino T, Danne T, Bergenstal RM, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care. 2019;42(8):1593-1603. doi: 10.2337/dci19-0028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Narasaki Y, Park E, You AS, et al. Continuous glucose monitoring in an end-stage renal disease patient with diabetes receiving hemodialysis. Semin Dial. 2021;34(5):388-393. doi: 10.1111/sdi.13009. [DOI] [PubMed] [Google Scholar]
- 16. Bally L. Die closed-loop-insulintherapie. Schweiz Ärzteztg. 2023;104(1415):70-72. doi: 10.4414/saez.2023.21642. [DOI] [Google Scholar]
- 17. Bally L. Fully Closed-Loop Insulin Delivery in Abdominal Surgery: A Randomised Controlled Two-Centre Trial (CLAB-Study). Published 2022. Accessed June 6, 2023. https://clinicaltrials.gov/ct2/show/NCT05392452.
- 18. Zinter MS, Markovic D, Asaro LA, et al. Tight glycemic control, inflammation, and the icu: evidence for heterogeneous treatment effects in two randomized controlled trials. Am J Respir Crit Care Med. 2023;207(7):945-949. Accessed June 9, 2023. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10111993/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bagsic SRS, Fortmann AL, Belasco R, et al. Real-time continuous glucose monitoring in the hospital: a real-world experience. J Diabetes Sci Technol. 2023;17(3):656-666. doi: 10.1177/19322968231165982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Yeung AM, Huang J, Klonoff DC, et al. iCoDE June 22, 2022 Steering Committee Meeting Summary Report. J Diabetes Sci Technol. 2022;16(6):1575-1576. doi: 10.1177/19322968221119146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Price C, Ditton G, Russell GB, Aloi J. Reliability of inpatient CGM: comparison to standard of care. J Diabetes Sci Technol. 2021;17(2):329-335. doi: 10.1177/19322968211062168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Wright JJ, Williams AJ, Friedman SB, et al. Accuracy of continuous glucose monitors for inpatient diabetes management [published online ahead of print February 7, 2022]. J Diabetes Sci Technol. doi: 10.1177/19322968221076562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. ElSayed NA, Aleppo G, Aroda VR, et al. 16. Diabetes care in the hospital: Standards of care in diabetes—2023. Diabetes Care. 2023; 46(suppl 1): S267–S278. Accessed June 8, 2023. https://diabetesjournals.org/care/article/46/Supplement_1/S267/148051/16-Diabetes-Care-in-the-Hospital-Standards-of-Care [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Umpierrez GE, Reyes D, Smiley D, et al. Hospital discharge algorithm based on admission HbA1c for the management of patients with type 2 diabetes. Diabetes Care. 2014;37(11):2934-2939. doi: 10.2337/dc14-0479. [DOI] [PMC free article] [PubMed] [Google Scholar]