Continuous Glucose Monitoring for High-Risk Type 2 Diabetes in the Hospital (Cyber GEMS) (Cyber GEMS)

Continuous Glucose Monitoring for High-Risk Type 2 Diabetes in the Hospital: Cloud-Based Real-Time Glucose Evaluation and Management System (Cyber GEMS)

Given the known serious consequences of uncontrolled blood sugars during hospitalization, this research plans to study an alternative seamlessly integrated continuous glucose monitoring (CGM) system in the hospital to test a dynamic and digitized, team-based approach to glucose management in an underserved and understudied, yet high-risk population. A digital dashboard will facilitate real-time, remote monitoring of a large volume of patients simultaneously; automatically identify and prioritize patients for intervention; and will detect any and all potentially dangerous hypoglycemic episodes in a hospital environment. The study will focus on clinical metrics of glucose control and infection that are in-line with patient priorities and US hospital quality initiatives.

There is strong evidence that poor glycemic control in the hospital is common. Given the known consequences of uncontrolled blood sugars during a hospitalization (e.g., infection, serious neurological and cardiac complications, mortality, longer lengths of stay, readmissions, higher healthcare costs), health systems devote significant resources to developing protocols for improving glucometrics. Despite the widespread use and demonstrated effectiveness of continuous glucose monitoring (CGM) for ambulatory glucose management, CGMs is not routinely used in US hospitals. Therefore, the long-term goal to develop Cloud-Based Real-Time Glucose Evaluation and Management System (Cyber GEMS) is to provide an effective, real-time solution to augment existing processes, to provide a valuable test of real-world effectiveness, while capitalizing on standardized algorithms to facilitate sustainability and scalability to other systems and at-risk populations. The intervention will enable hospital care teams to take immediate steps based on the wireless transmission of glucose data from the Dexcom G6 device, sent to a digital dashboard, where integration with existing real-world hospital processes can provide immediate prioritization to prevent or correct impending hypoglycemia and severe hyperglycemic events. This study is a randomized controlled trial, defined as a Phase II/III definitive clinical trial that in turn establishes efficacy and effectiveness of this intervention. Aim 1 will establish the effectiveness of Cyber GEMS versus Usual Care (UC) in increasing the % time patients are in-range and decreasing % time in hypoglycemia and severe hyperglycemia during hospitalization. Aim 2 will evaluate the effectiveness of Cyber GEMS versus UC in decreasing hospital-acquired infection risk. A digital dashboard will facilitate real-time, wireless transmission of glucose data of a large volume of patients simultaneously; automatically identify and prioritize patients for intervention; and detect potentially dangerous hypoglycemic episodes - all at a reduced burden than current methods of stratification and review. The uninterrupted coverage, and efficient and remote diabetes specialist oversight in Cyber GEMS is a scalable, novel, team-based approach to maximize the use of continuously streaming CGM data for optimal glucose management.

Research Assistants (RAs) will verbally administer baseline survey and insert Dexcom G6 CGM, before unveiling the group assignment. CGM data will be transmitted from bedside iPhone to web-based platforms for: (1) Real-Time Management (via iPad-based FOLLOW app used by bedside RN and Digital Dashboard used by remote monitoring team) and (2) Clinical Optimization (via CLARITY, a Diabetes RN Coordinator will conduct remote clinical management of patients from a central, Scripps Diabetes Hub). A post-CGM satisfaction survey will be administered and compensation provided when CGM is removed prior to discharge or within 2 weeks following discharge. The CGM readings will be used to make recommendations for insulin adjustment and glucose management. After discharge, CGM data will be downloaded from a HIPPA-compliant, web-based CGM data management tool, and saved in Excel. The Data Analyst, blinded to condition, will routinely screen CGM data and merge individual spreadsheets for analysis.

RAs will verbally administer a baseline survey and insert the Dexcom G6 CGM. before unveiling the group assignment. CGM data will be blinded and used for evaluation purposes only. Glucose will be monitored via the hospital's standard POC testing protocol (i.e., prior to meals and at bedtime for patients who are eating, and every 4-6 waking hours if not eating). Glucose management in UC is designed to minimize differences between groups, aside from CGM monitoring, A post-CGM satisfaction survey will be administered and compensation provided when the CGM is removed prior to discharge or within 2 weeks following discharge. After discharge, CGM data will be downloaded from a HIPPA-compliant, web-based CGM data management tool, and saved in individual Excel spreadsheets. The study Data Analyst, blinded to study condition, will routinely screen CGM data and merge individual spreadsheets for analysis.

CGM data will be transmitted from the bedside iPhone to web-based platforms for: (1) Real-Time Management (via iPad-based FOLLOW app used by bedside RN and Digital Dashboard used by the remote monitoring team) and (2) Clinical Optimization (via CLARITY, by which a Diabetes RN Coordinator will conduct remote clinical management of patients from a central, Scripps Diabetes Hub.

CGM data will be blinded and used for evaluation purposes only. Glucose will be monitored via the hospital's standard POC testing protocol (i.e., prior to meals and at bedtime for patients who are eating, and every 4-6 waking hours if not eating). Glucose management in UC is designed to minimize differences between groups, aside from CGM monitoring.

Participants will have their percent time in range calculated following a minimum CGM data collection period of 12 hours and expressed as a percentage where: Percent Time in Range= 100 (Number readings in range (70-200mg/dL)/Total number of readings from CGM). Number of readings will be used in calculation, which scale directly with time.

Our second outcome will be assessed by the same methods as the first, but instead looking at Percent Time in Severe Hyperglycemic Range (>300mg/dL) and Percent Time in Hypoglycemic Range (<70mg/dL).

Rates of hospital-acquired infection are defined as skin wound or surgical site, central line-associated bloodstream infection, urinary tract infection, bacteremia, clostridium difficile infection, or pneumonia not present at admission. Unadjusted incidence rates among study participants will be compared between intervention and control groups via Chi-Square test of two proportions.

Using CGM data, glucose variability will be determined by first calculating the coefficient of variation for each participant, dividing the standard deviation of the glucose readings of that participant, by the mean of those readings and multiplying by 100 to get a percentage. Mean coefficients of variation will be compared between intervention and control groups by a students t test.

Additional metrics of glycemic control will be captured for each study participant from the EMR including: HbA1C. Like primary outcome analyses, group mean differences of each variable will be assessed unadjusted with a students t-test utilized to detect between-group differences.

Additional metrics of glycemic control will be captured for each study participant from the EMR including fasting point-of-care (POC) blood glucose measurements (mg/dL). Like primary outcome analyses, group mean differences of each variable will be assessed unadjusted with a students t-test utilized to detect between-group differences.

To examine enrollment rate, demographic characteristics of eligible patients will be compared between those who enroll versus decline; where continuous measures will be compared between groups will be compared between groups by Chi-Square tests.

To examine generalizability of our sample, distribution of demographics in our sample will be compared to expected distributions of our target population through Chi-square tests.

We do not plan to statistically assess reasons for withdrawal due to an anticipated low number of withdrawals, but all reasons will be recorded and descriptively quantified where applicable.

A generalized linear model will be used to assess whether time on CGM relates to changes in the percent time in primary and secondary outcome ranges over time.

Results of semi-structured interviews will be qualitatively, descriptively analyzed to reveal perceptions of CGM implementation efficacy, challenges, satisfaction, and benefits.

Descriptively assess physicians pre- and post study perceptions and knowledge of and identified barriers to successful inpatient glucose control via the Inpatient Glucose Management Questionnaire (IGCQ).

# of alarms for glucose managed by the Clinical Transfer Center (Cyber GEMS only) will be quantified for percent adherence to protocol by: # of times Clinical Transfer Center notified bedside Registered Nurse (RN) / # of qualifying alarms. Rates of follow-up POC testing at bedside will be analyzed/statistically tested by fitting a linear model with # of alarms.

CGM satisfaction will be determined using a modified CGM Satisfaction Scale. Questions regarding comfort/interruption, given in both arms and mean overall scores compared by unpaired students t-tests.

Number of participants will be continuously monitored by the Data Analyst throughout the study period and tracked against projected numbers for enrollment.

Inclusion Criteria: Documented previous or current Type 2 Diabetes (T2D) diagnosis as defined by either diagnosis in the chart or an HbA1c > or = to 6.5% in the last 90 days Either on subcutaneous (SQ) insulin orders, or greater than two serum or Point of Care (POC) glucose > or = 200 mg/dL in most recent 24 hours of admission Exclusion Criteria: Anticipated length of stay < 24 hours; Current or anticipated ICU placement; Does not speak English or Spanish; Known allergy to adhesives; Current participation in any medication or device research study; Pregnant; Any other condition that Multiple Principal Investigator (MPI) Philis-Tsimikas or the attending physician deems contraindicated