Abstract
Table of Contents
Examination of the Accuracy and Stability of an Arterial Sensor for Glucose Monitoring in a Porcine Model Using Glucose Clamp Technique
Medical University of Graz Endocrinology and Diabetology Graz, Austria
Intravascular glucose sensors have the potential to improve and facilitate glycemic control in critically ill patients and in contrast to subcutaneous devices might overcome measurement delay and accuracy issues.
This study investigated the accuracy and stability of a novel in-vivo biosensor for arterial glucose monitoring tested in a hypo- and hyperglycemic clamp experiment in pigs. In total, 12 sensors were tested over 5 consecutive days in 6 different pigs. Samples of sensor and reference measurement pairs were obtained every 15 minutes. The intended use of the sensor allows 96 hours of sensing.
1337 pairs of glucose values (range 37-458 mg/dl) were available for analysis. The systems met ISO 15197:2013 criteria in 99.2%. Fulfilment of ISO 15197:2013 was 100% for glucose <100 mg/dl (n=414) and 98.8% for glucose >100 mg/dl (n=923). The mean absolute relative difference (MARD) during the entire glycemic range of all sensors was 4.5 ± 4.1%. The MARDs within the hypoglycemic (<70 mg/dl), euglycemic (70-180 mg/dl) and hyperglycemic glucose ranges (>180 mg/dl) were 5.8 ± 5.2%, 3.8 ± 3.1% and 5.0 ± 4.3%, respectively. Sensors indicated comparable performance assessed by MARD on all days investigated (day 1, 3 and 5). None of the systems showed premature failures.
In a porcine model, the performance of the biosensor revealed a promising performance. The transfer of these results into a human setting is the logical next step.
Implementation of an Electronic Order Set to Improve the Perioperative Management of Patients with Diabetes Mellitus
Mills Peninsula Medical Center Burlingame, California
Multiple studies have demonstrated the benefits of optimizing glycemic control in the perioperative period. Several professional organizations have published guidelines that recommend routine perioperative blood glucose testing and treatment with insulin to target blood glucoses (BG) under 180 mg/dL. However, practical tools to implement these guidelines are lacking. This study aims to implement established guidelines by piloting a novel electronic order set to optimize perioperative glycemic control in the adult, non-pregnant patient with diabetes mellitus (DM) in a community hospital setting.
A novel electronic order set and workflow was developed and piloted for adult patients with known DM undergoing inpatient surgery. 6-month baseline and 6-week pilot data were obtained including frequency of perioperative BG monitoring, glycemic control, frequency of hypoglycemia and use of insulin for hyperglycemia.
At baseline, the frequency of BG monitoring was 85% pre-op, 33% intra-op, 21% PACU and 94% in the 24-hours after surgery. During the pilot, the frequency of BG monitoring increased to 95% pre-op, 59% intra-op, 48% PACU and 98% in the 24-hours after surgery. Fast-acting insulin was administered more frequently for hyperglycemia at all phases of care, including 78% in pre-op compared to 44% at baseline. Mean BG decreased in all phases of care including in pre-op from 160mg/dL to 140mg/dL and in the 24-hours after PACU from 173 mg/dL to 157mg/dL without significant increases in hypoglycemia. Severe hyperglycemia rates in the 24-hours after surgery decreased from 10.1% to 4.7%.
Implementation of a novel electronic perioperative diabetes management order set and workflow significantly improved the BG monitoring and glycemic control in patients with known DM undergoing inpatient surgery in a community hospital setting.
Reduction of Hospital Hypoglycemia with eGMS and Quality Programming Across 180 US Hospitals
Wake Forest School of Medicine Winston-Salem, North Carolina
Hospital hyperglycemia effects up to 40% of all inpatients and has been associated with worse outcomes, including increased length of stay, surgical site infections and cost. There are very few publications focused on glycemic benchmarking that look at a large number of hospital in the United States. This lack of a true benchmarking can lead to confusion among clinical and financial leadership in the hospital as to when to begin a glycemic improvement project. Because eGlycemic Management Systems (eGMS) has been in use now for over 10-years, and thought of as best practice in the critical care setting, we measured the clinical impact of an eGMS in a large number of hospitals across the US.
This Retrospective Quality Improvement Study evaluated the nation-wide safety of eGMS IV insulin management in 180 hospital and across 590 hospital units. Data was collected from the EHR and Lab systems from each hospital over a 24-month timeframe (12/10/17 – 12/10/19). Glucose targets ranged from a low end 80 mg/dL highest at 200 mg/dL in the varied hospital units. Primary objective was to measure impact on glycemic care, initially with a primary focus on patient safety as evidence by severe and mild-moderate hypoglycemia.
The utilization of eGMS was observed in 108,535 patients who received 3,132,493 blood glucose test during their hospital stay. Mild to moderate hypoglycemia measured as blood glucose events <70 mg/dL was 0.55% and Severe hypoglycemia measured as blood glucose events <40 mg/dL was 0.015%.
This data suggest hospitals can benchmark verses safe and low rates of hypoglycemia eGMS had among this large number of patients, hospital units and hospitals in this study.
3 Things in 30 Minutes
UC Davis Medical Center Sacramento, California
To coordinate the timing of point of care (POC) testing, insulin administration and meal delivery in the inpatient setting.
Policy: Established policy to reflect best practice including finger stick blood glucose level check before administering insulin and as ordered by the physician. Pre-meal finger stick blood glucose levels should be drawn within thirty (30) minutes before the patient starts eating the meal.
Tray delivery: Coordination of breakfast tray delivery with Dietary, including identifying patients receiving insulin by the tray ticket with “Med Alert” printed on the ticket. Dietary will request the HUSC to alert nursing that meal trays have arrived, and patients requiring insulin will need finger stick and mealtime insulin prior to eating. Trays are not delivered directly to the patient so that testing and mealtime insulin can be accomplished prior to tray delivery.
Education: Robust nursing education on the initiative “3 Things in 30 Minutes”.
Reporting: Create a dashboard to track POC to insulin administration.
The average POC glucose to insulin administration time has decreaseddramatically with the overall hospital average going from 51.69 minutes in January of 2017 to 25.37minutes in July of 2019-a 50 % improvement in insulin timing.
Improvement in insulin timing and overall awareness with importance of insulin administrations is paramount to decreasing adverse events and overall glycemic control.
Managing Diabetes in the Hospital with an Insulin Pump and or Continuous Glucose Monitor (CGM): Nurse's Perspective
UCSD, Rady Children's Hospital
San Diego, California
Describe the experience of nurses caring for hospitalized patients who are managing blood glucose levels with an insulin pump and/or CGM. Propose actions based on the results of this study, to enhance the care of the patient managing diabetes in the hospital with an insulin pump and/or CGM.
Qualitative descriptive study. Setting: Pediatric inpatient care units of a 524 bed Pediatric Academic Hospital. In October 2019, 104 nurses from medical, surgical, OR/PACU units participated in this study by completing a self- administered questionnaire.
Although 70% of nurses have cared for at least one patient on insulin pump during hospitalization, almost 50% were uncomfortable to manage a patient wearing an insulin pump. About 80% of nurses were unable to retrieve insulin pump's settings such as targets, basal rates, insulin to carbohydrate ratio or sensitivity/correction factor. About 50% of the nurses did not establish goals and expectations with the patients and their caregivers regarding the management of their insulin pump and about 70% was unaware of the hospital insulin pump policy. Regarding the use of CGM in the hospital, 70% of nurses did not feel comfortable to manage a patient wearing a CGM in the inpatient setting and did not establish goals and expectations with the patients and their caregivers regarding the uses of CGM while hospitalized.
Nurses reported having knowledge gaps regarding the use of insulin pump and/or CGM in the inpatient setting. Implementing a comprehensive insulin pump and CGM training program that addresses these practices gaps would help mitigate this problem.
Utilization of Computer-Guided Insulin Dosing Decreases Hypoglycemia Adverse Drug Events, Length of Stay and Costs at Large Pacific Northwest Health System
CHI Franciscan Tacoma, WA. USA
Inpatient hypoglycemia is associated with poor clinical outcomes and prolonged length of stay (LOS). In January 2018, in response to above-average incidence of hypoglycemia adverse drug events (HADE), as reported through Washington State Hospital Association’s (WSHA) Quality and Safety Program, a large Pacific Northwest health system launched a glycemic management software platform, the eGlycemic Management System®(eGMS®), for IV and SubQ insulin titration. This study aimed to compare glycemic outcomes, LOS and cost per case when patients were managed with eGMS® versus routine care.
We conducted a retrospective review of the most recent quarter’s HADE (BG < 50 mg/dl after hypoglycemic agent administration) reported to WSHA compared to HADE reported the quarter before eGMS®. After three months’ eGMS® utilization, our Strategy Department compared impact of eGMS® versus routine care on LOS and cost per case for patients with diabetes ranked in the top 5 inpatient stay diagnoses and for patients admitted with DKA and HHS.
WSHA-reported HADE decreased 44%, from 7.1% in Q4-2017 (before eGMS® implementation) to 4.0% in Q3- 2019 (p<0.000). BGs in the target range (70-180 mg/dL) decreased 9.7%, from 71.1% to 64.2% (p<0.000). Length of stay decreased by 0.6 days and cost per case by $500 for patients whose insulin therapy was guided by eGMS® (compared to routine care). Among the subset of DKA and HHS, LOS decreased by 1.4 days and cost per case by $1000.
Utilization of eGMS® across the health system resulted in decreased HADE, LOS and cost per case among insulin- requiring patients. Health systems struggling to achieve glycemic control should consider integrating computer- guided insulin dosing into their glycemic care programming.
Preventing Hypoglycemia in Hospitalized Patients: Closing the Loop by Including Patients
Valleywise Health Phoenix, AZ
To reduce hypoglycemic events through a proactive surveillance using a multidisciplinary data-driven approach, by also empowering patients to make educated decisions at bedside.
Patients with or without diabetes who are on insulin, may be at risk of experiencing hypoglycemia in the hospital setting. Common causes of hypoglycemia are improper insulin prescription, inappropriate management of first episode of blood glucose below target, interruption of nutrition, and nutrition-insulin mismatch. The process implemented redesigned the meal tray delivery process in order to shorten the time between point of care testing (POCT), meal tray delivery and insulin. Dietary staff notified patients’ point of care technician about patient ordering meal. POCT, insulin administration and meal tray delivery goal time were set to <45 minutes. Dietary staff placed a 3x5 “meal tray tent card” at eye level to encourage patients to call when meal tray was delivered as insulin may be needed before consuming meal. Data was captured by dietary and quality departments. Physicians performed evaluated charts, protocols and management order sets.
Severe hypoglycemic events where reduced by 51%. Severe hypoglycemia was defined as POCT <50mg/dL due to electronic report restrictions. Patients were reminded and empowered to learn and make educated decisions at bedside. Meal tray orders by patients before POCT glucose tests 52%. Average time of 16.6 minutes from tray order to delivery at bedside.
In order to avoid hypoglycemic events, the coordination of point of care glucose monitoring, insulin administration and meal tray delivery should be coordinated through multidisciplinary efforts. The inpatient environment can also provide a great learning environment for patients and their family.
Multidisciplinary Approach to Decreasing Hypoglycemia in an Urban Academic Hospital
UChicago Medicine Chicago, IL
The goal was to minimize iatrogenic hypoglycemic events in adults at an 800-bed hospital. Creation of a multidisciplinary program examining factors that contribute to hypoglycemia assured the thoroughness of our ability to identify practice changes to achieve this goal.
Using quality improvement methods, we sought to address the incidence of medication related hypoglycemia in hospitalized patients by implementing evidence-based practice changes. Following an extensive literature review and internal practice audit, we identified important components of a comprehensive plan to trial. Interventions were trialed over time using Plan-Do-Study-Act (PDSA) cycles on a pilot unit to optimize performance and improve outcomes. Final approaches included provider and patient education; initiating a new meal time insulin process; and enhanced multidisciplinary communication, e.g. huddles, order sets; pharmacy scoring tool; and real time alerts.
Targeted interventions such as patient and staff education, door signage for every patient receiving insulin, timely insulin administration, insulin ordering guidelines and alerts have all contributed to a substantial reduction in hypoglycemic events. In addition, input from all involved disciplines including ancillary dietary and nursing staff was crucial to success. Continued attention to real time feedback and analysis of events provides sustained improvement.
Characteristics of Patients Identified at High- Risk by a Hypoglycemia Prediction Model
University of Florida Gainesville, Florida
Hypoglycemia in hospitalized patients due to inappropriate insulin administration is one of the most serious and common adverse drug events and will soon become a Medicare measure of hospital quality. Our group previously developed an automated algorithm for identifying patients at high risk for drug-induced hypoglycemia. We have also recently developed a Nurse Practitioner led Inpatient Diabetes Service (IDS). This quality improvement study aimed to perform an initial evaluation of patients identified as high-risk by the hypoglycemia prediction algorithm in the 1,000-bed Shand’s Hospital, University of Florida.
The hypoglycemia prediction algorithm (heat-map) runs nightly on all adult patients prescribed with diabetes medication therapy in the Shand’s Hospital. A cohort of eligible patients was observed between October 1st and 31st of 2019. We collected data patients at high risk of hypoglycemia (top 10% of the hypoglycemia risk based on the algorithm), patients seen by the IDS team, and hypoglycemic events defined as glucose of
We observed 163
Those patients with drug-induced hypoglycemic events were more frequently reported on the heat-map prior to the event which implies the predictive value of the model. Approximately 25% of patients received care from the IDS team. This observation suggests an opportunity to use the heat-map tool to recommend the IDS consult and provide quality insulin management for patients at a high risk of hypoglycemia.
Technology Reduces the #1 Adverse Medication Safety Events (AMSE) in the Hospital
AdventHealth Waterman, Tavares, FL
AdventHealth Waterman, a 269 bed- community hospital, realized an opportunity to lower the number of dangerous hypoglycemic events. Using insulin to control blood glucoses while avoiding hypoglycemia is an essential but often challenging aspect of quality inpatient medical care. Hypoglycemia defined as a blood glucoses (BG) < 70 mg/dl has been identified as the third most common adverse medication safety event (AMSE) resulting in frequent patient harm. An interdisciplinary glycemic team identified and led implementation of glycemic management improvements to reduce hypoglycemia related AMSE.
A retrospective analysis was completed to compare baseline pre-glucommander ((Pre-GM) April 1, 2016 – March 31, 2017)data to post-glucommander ((Post-GM) April 1, 2018-March 31, 2019) data. During the pre-glycemic management improvement period the hospital used EndoTool, an IV insulin software to support the management of severe hyperglycemia. To achieve effective glucose control, a tool to support IV and subcutaneous insulin management was identified as the top priority. Revising order sets and providing physician/nurse education were the hypoglycemia reduction strategies used in conjunction with the implementation of the Glucommander eGlycemic Management System (eGMS).
After implementation of Glucommander eGMS the percent of patient days (%PDs) <70 mg/dL decreased significantly (7.19%) compared to the Pre-GM period. Glucose delta between admission and discharge showed a statically significant difference between the Pre-GM to Post-GM average admission BG compared to the average discharge BG. Mortality comparison between the two groups showed statistically significant reduction (41%) in the Post-GM patients.
Implementation of eGMS showed significant improvements in hypoglycemia rates, hospital glucose control and mortality. This study supports the efficacy and safety of eGMS.
Emergency Department mHealth Improves Outcomes for Patients with Comorbid Depression
In this study, we compare the efficacy of a diabetes self-care mHealth intervention in ED patients with and without depression.
Patients with HbA1C>8.5 were enrolled during their ED visit. They and a patient-designated supporter were registered in a mHealth program to improve diabetes self-care. Participants received text messages in English or Spanish, three times daily for six months. At enrollment, we collected demographics, Patient Health Questionnaire-9 (PHQ9) and World Health Organization-5 (WHO5) quality of life assessments, sBP, and HbA1C. Measures were repeated at 6-months. Patients were categorized by PHQ9 score. T-test, Fisher exact, and ANOVA were used to compare depressed and non-depressed groups.
We enrolled 166 patients, and 130 patients were depressed. Enrolled patients were 51% female, 93% Latino, and 70% Spanish-speaking, with no differences based on depression. There was no difference in baseline HbA1C, sBP, or weight for depressed patients. Baseline WHO5 scores were higher for non-depressed patients, 79.8 (95%CI 73.4- 86.2) vs 54.6 (95%CI 49.8-59.4) p<0.001. 97 patients followed up at 6 months, with no difference in follow up rates by depression. There was a decrease in combined group sBP (95%CI: -4.26 to -17.7) p=0.0017, with no difference by depression. Depressed patients decreased HbA1C by an average of 1.74 (95%CI 1.19-2.29) compared to an increase of 0.04 (95%CI -0.85 to 0.77) in non-depressed patients, p<0.001. Reductions in HbA1C varied by depression severity (F= 4.29, p=0.003). Those with mild (95%CI 1.31-3.47), moderate (95%CI 0.308-2.12), and severe (95%CI 1.07-3.71) depression improved the most.
Overall patients had mean improvements in sBP, and patients with depression showed HbA1C reductions. ED-based mHealth chronic disease interventions may be particularly effective for depressed patients.
Accuracy Analysis of an Autonomous System to Personalize Blood Glucose Prediction for T1DM Patients with Real World Data
University of Sao Paulo, Rebeirao Preto, Sao Paulo, Brazil
BGL predictive algorithms can improve T1DM treatment preventing glucose excursions. An autonomous system based on neural networks was developed to personalize blood glucose predictions based on blood glucose, insulin infusion, nutrient intake and heart rate in real world scenario's.
20 T1DM patients were monitored with flash glucose monitors, activity trackers and a mobile app (GlucoTrends) to collect meal and insulin data. Personalized prediction models were trained independently, without requiring specific settings for each individual. Patient's characteristics: 11 males and 9 females, age: 32.4 (SD:10.5), BMI: 26.0 (SD:3.8), BGL: 159.1 (SD:34.0), under the following therapies: 45% under fixed doses, 40% carbohydrate counting and 15% insulin pumps. Patients were monitored during on average 29.3 days (SD: 7.9), and the last 20% of measurements were reserved for evaluation. Prediction accuracy for 1-hour prediction horizon was evaluated by Clarke Error Grid (CEG) compareding to BGL measured with flash monitor.
The percentage of predictions within AB zones of CEG for all-day and for night-time only periods are 92.9% and 94.0%, respectively (Figures 1 & 2). Patients using Insulin Pumps (Analog insulin), Fixed Doses (Human insulin) and CArbohydrate counting (Analog insulin) had 97.5% (SD:0.7), 93.6% (SD:4.5) and 90.0% (SD:8.2) predictions within AB zone on average, respectively (Figure 3).
Interprofessional Inpatient Diabetes Care: Good for Medical Practice but Great for Medical Education
Vanderbilt University Medical Center Nashville, TN, USA
Medical students, regardless of their specialty of choice, will have to care for patients with diabetes during their internship. Use of multidisciplinary diabetes care has been shown to be effective in diabetes treatment, especially regarding diabetes technology. Interprofessional medical education is an effective method to teach healthcare delivery and foundational knowledge. The use of interprofessional teaching to prepare medical students for multidisciplinary diabetes care and to approach diabetes technology/devices has not been previously reported.
Medical educators developed a four-week immersion course teaching practical and foundational diabetes knowledge with an interprofessional approach. In addition to learning foundational science, medical students received practical instruction from our multidisciplinary team in the management of blood glucose and diabetes technology in situations such as critical illness, tube feedings, total parental nutrition, and encountering insulin pumps/continuous glucose sensors in inpatient settings. Student feedback was collected at the end of each course offering via anonymous questionnaire and direct feedback.
During the 2017-2018 and 2018-2019 academic years, 100% of students felt the course helped them work more collaboratively in in an interprofessional environment, felt their skills of collaboration had increased, and felt more confident about their ability to care for patients with diabetes in an inpatient setting. All students developed the ability to teach patients basal/bolus insulin, insulin pens, carbohydrate counting, calculating insulin doses, and gained familiarity with insulin pumps.
Vanderbilt’s Diabetes Immersion course prepares students to treat diabetes mellitus as an intern and helps them develop ability to address diabetes technology/devices. This results in high degrees of student satisfaction and fosters an environment of interprofessional collaboration. This approach could be replicated in other institutions and in other disciplines.
Basal-On-Board (BOB) Appears to Contribute to Significant Glucose Readings < 70 mg/dl (HYPO) in Hospital Settings
Monarch Medical Tech., LLC Charlotte, North Carolina
The stress of illness results in increased insulin resistance and exacerbation of hyperglycemia. As insulin sensitivity improves, it is difficult to estimate the required decrease in exogenous insulin to avoid HYPO and excess BOB may place patients at risk. BOB is defined as the difference in the reduced basal dose. This retrospective study was designed to evaluate HYPO that is likely attributable to BOB in hospitalized patients utilizing an electronic glucose management system (eGMS) for dosing.
Data from three hospitals utilizing eGMS for basal-bolus insulin dosing were reviewed (3,896 patients) and further stratified based on the decrease in basal dose. Two periods were reviewed: period 1 is the time following a reduction in the dosing model basal insulin until administration of the new, lower basal dose (time at risk of BOB on HYPO) compared to period 2 which is the subsequent 24 hours on the new dosing model.
After a lowered dosing model, patients who required a greater than 4 unit decrease in basal insulin experienced a significantly higher rate of hypoglycemia in period 1 vs. 2. Without adjustment for BOB, 27% of all HYPO was observed in period 1 related to BOB in the face of a reduction in basal insulin dose responding to down trending glucose values.
To address the subsequent HYPO following a reduction in basal insulin dose in an environment with reduced insulin requirements such as hospitalization, options include a further, temporary reduction in the bolus dose or carbohydrate supplementation. These adjustments may be difficult to predict, but eGMS algorithms present an opportunity to make dosing adjustments in real-time to help mitigate HYPO.
Improved Hospital Glycemic Control with eGMS and High-Reliability Strategies
Sentara Virginia Beach Virginia Beach, VA, USA
Improve utilization of standard order sets for subcutaneous insulin management of hyperglycemia and diabetes, with the elimination of sliding scale and increased use of basal/bolus management. In addition, monitor outcomes such as hyperglycemia and hypoglycemia in subcutaneous insulin patients.
Sentara Virginia Beach convened a define, design implement improvement team, including executive leadership, CDEs, pharmacy, provider and nursing champions. The team improved integration of an eGlycemic Management System(eGMS), Glucommander, by standardizing workflow and education, revising order sets, and introducing high reliable strategies. Several key strategies included simplification of order sets, a clinician driven process to allow nurses to automatically start eGMS for hyperglycemia, daily monitoring of all insulin-requiring patients, and establishing super users within each unit.
The utilization of eGMS increased 4.5 times, with subsequent increase in basal/bolus insulin therapy for medical- surgical patients, from an average of 46 per month on eGMS in the 4 months prior to order set changes, increasing to 206 per month in the 4 month period through October 2019. As utilization increased, hypoglycemia and hyperglycemia both reduced. In a similar 4-month period (June to Sept 2018 compared to July to Oct 2019), patients on the eGMS had total BG % for < 40 mg/dl drop from 0.14% to 0.07%, < 70 mg/dl from 2.31% to 1.77%, and severe hyperglycemia >300 mg/dl from 7.33% to 4.78%.
Standardizing care, with evidence-based practice, technology and high reliable strategies resulted in increased use of basal-bolus insulin management, with improvements in both hyperglycemia and hypoglycemia.
Improvement in Glycemic Control Using Algorithm-Guided Basal-Bolus Insulin Therapy in Hospitalized Patients with Type 2 Diabetes
Division of Endocrinology and Diabetology, Department of Internal Medicine, Medical University of Graz Graz, Austria
Type 2 diabetes (T2D) treatment in hospitalized patients is often insufficient, resulting in sustained hyperglycemic or hypoglycemic events due to insulin dosing errors. Therefore, an algorithm-based decision support system (GlucoTab®, decide Clinical Software GmbH, Graz, Austria) was developed, to guide basal-bolus insulin therapy (BBI) in inpatient treatment of T2D. We aimed to analyze the effect of GlucoTab® BBI compared to standard care on glycemic control throughout inpatient stays on a general ward.
In a retrospective analysis, we obtained data of 153 patients with T2D admitted to a general medicine ward over 12 months for various admission indications. We compared patients treated using the algorithm-based BBI therapy to patients receiving standard diabetes care provided by treating physicians.
Among the 153 cases, there were 43 episodes with continuous algorithm-based therapy lasting a minimum of 3 days. 26 patients were treated using the algorithm continuously for at least seven days (maximum duration: 20 days). Overall there were 379 days of algorithm-based BBI treatment and 779 days of standard diabetes care including any given insulin therapy, resulting in a mean daily glucose of 159.9 ± 40.6 vs. 177.1 ± 50.8 mg/dL. For algorithm- treated patients we compared the first day of algorithm therapy with day 3 (n=43; 190.4 ± 58.3 vs. 161 ± 35.3 mg/dL) and day 7 (n=26; 195.1 ± 66.8 vs. 150.2 ± 31.2 mg/dL).
Algorithm-guided BBI treatment safely titrates insulin dosage leading to a rapid improvement of mean daily glucose within three days. There was less glycemic variability, resulting in a decrease in hyper- and hypoglycemic events. Our analysis supports the use of algorithm-based decision support systems in hospitalized insulin-treated T2D patients.
Sustained Hypoglycemia Reduction with IV Insulin via eGlycemic Management System
Grady Memorial Hospital Atlanta, GA, USA
Critically ill patients with hyperglycemia are often managed with continuous insulin infusions. Critical Care and step down patients were started on an eGlycemic Management System (eGMS), GlucommanderTM, to help safely lower hyperglycemia. The eGMS system was put in place in December 2016. Prior analysis, from 12/16 to 10/18, showed low rates of severe hypoglycemia, < 40 mg/dl, with eGMS compared to usual care, 0.02% vs 0.44%, as well as hypoglycemia <70 mg/dl, 0.32% vs 2.44%. Interventions are often difficult to sustain.
Retrospective comparison of patients with hyperglycemia managed by usual care compared to a similar group placed on eGMS.
From 10/31/18 to 11/1/19 there were a total of 2028 patients with persistent hyperglycemia, >180 mg/dl. 669 patients (32%) were managed with eGMS and 1359 patients (68%) with hyperglycemia were managed with usual care, including either subcutaneous insulin, no insulin, or a paper insulin protocol. Hypoglycemia rates were lower in patients on eGMS. Severe hypoglycemia, < 40 mg/dl, was 0.01% in patients on eGMS and 0.33%, for usual care. Rates of < 70 mg/dl was 0.30% vs 2.18% for eGMS vs usual care, respectively. Both groups had similar rates of 70-180 mg/dl.
eGMS system safely lowers blood glucose, with a reduction in hypoglycemia. A similar group of patients managed with usual care experienced higher rates of hypoglycemia. These results have been sustained for almost 3 years.
Patient Safety Improvements with IV Insulin Compared to Subcutaneous Insulin in the ICU
Grady Memorial Hospital Atlanta, GA, USA
In the ICU setting, compare the use of continuous insulin infusions to similar patients on long-acting basal subcutaneous insulin, on patient safety-related to hypoglycemia and efficiency outcomes, such as time to target.
Retrospective comparison of 100 patients on continuous IV insulin via eGlycemic management system, Glucommander, to 100 patients on long-acting insulin. Patients with DKA were excluded, and all patients were >18 years old. Both groups were on therapy for at least 48 hours and grouped according to the initial insulin therapy chosen.
Patients on continuous IV insulin had significant reductions in hypoglycemia, p-value < 0.05, compared to a similar group on subcutaneous basal insulin. Patients on continuous IV insulin had no reported severe hypoglycemia rates, and rates of < 70 mg/dl based on all blood glucose testing was 0.1%. The patients on subcutaneous basal insulin initially had rates of severe hypoglycemia of 1.5% and < 70 mg/dl of 3.1%. The time to target was 18.6 hours in the subcutaneous group on average, and 7.1 hours in the IV insulin group.
For the management of critically ill patients with hyperglycemia, continuous IV insulin is preferred to subcutaneous insulin.
Tracking Inpatient Dysglycemia Management Performance in Clinical Units over Time Using Capillary Blood Glucose (CBG) Data / Metadata and Process Control Charts: A Description of a Method and Demonstration of Residual High Management Performance 4 Ye
Queen Elizabeth University Hospital, Glasgow Glasgow, Scotland
Time to repeat (TTR) CBG following identification of an index hypoglycemic CBG, and time to demonstrated resolution of hypoglycaemic episodes are useful measures of guideline adherence and management efficacy. We aimed to extend these techniques to track performance over time, allowing identification of changes in performance retrospectively and ultimately in real-time. We investigated whether there was residual positive impact of an intensive multimodal educational intervention (NHS Scotland Think, Check, Act) focused on dysglycemia management in 2012-2014 within non-specialist clinical units.
CBG measurements / metadata were extracted from a large teaching hospital in Scotland Mar18- Dec19. Hypoglycemia was investigated at thresholds of 4mmol/l(72mg/dl) and 3mmol/l(54mg/dl), TTR at 60 minutes, and hypo resolution at 60 minutes. Process control charts were constructed for each clinical unit. At each hypoglycemic event, a score of +1 was awarded for success (TTR or resolution within time threshold), or -1 for an unsuccessful outcome. The gradient of the plot represents performance level, with change of gradient signifying time and magnitude of performance change.
Process control charts were successfully generated for all clinical units. Non-specialist (vascular surgical) clinical units that had previously undergone educational intervention demonstrate significantly better performance in TTR, and hypoglycaemia resolution than all other clinical units in the hospital, and better performance than specialist diabetes units.
The process control chart is a useful technique to measure, visualize and generate testable comparisons of clinical management performance both over time (intra-unit) or between units. In this application of the method we have demonstrated a significant residual high level of performance in measures of inpatient hyperglycemia management 4 years after the end of the intervention.
Alternative Inpatient Hypoglycemia Treatment
Pomona Valley Medical Center Pomona, California
High-quality hospital care for diabetes requires both hospital care delivery standards, often assured by structured order sets, and quality assurance standards for process improvement. A hypoglycemia prevention and management protocol should be adopted and implemented by each hospital or hospital system. This study was to evaluate an alternative hypoglycemia treatment protocol in the hospital setting.
Data for all hypoglycemia events-blood glucose via POCT less than 70mg/dL are entered into an event reporting system and time of recheck and treatment given was evaluated.
The results demonstrated treatment with the PVHMC adopted protocol, which includes 30gm for awake and alert patients and D50% when indicated and a recheck within 30mins was optimal over smaller treatment of 15gm and recheck within 15mins.
A hypoglycemia prevention and management protocol should be adopted and implemented by each hospital or hospital system per the American Diabetes Association Standards. This study was to evaluate an alternative hypoglycemia treatment protocol in the hospital setting. Our results indicate, a protocol with 30gm for awake and alert patients and D50% when indicated and a recheck within 30mins is ideal.
Improving Hospital Glucometrics, Workflow and Outcomes with a Computerized Intravenous Insulin Dose Calculator built into the Electronic Medical Record
Clinical Fellow, Metabolism, Endocrinology and Diabetes Ann Arbor, Michigan, United States of America
The goal of this project is to improve the quality and efficiency of insulin infusion therapy using a computerized insulin calculator incorporated into the electronic medical record (EMR). This is expected to improve safety and clinical outcomes by reducing wide glucose fluctuations and rates of hypoglycemia.
In coordination with another institution, we adopted their validated insulin calculator and coordinated with our EMR provider to incorporate their calculator into our workflow in a stepwise fashion. Employing an aggressive training program and computer simulation prior to significant elbow-support at the time of institution, we successfully integrated the insulin calculator in our cardiovascular surgery intensive care unit with plans to expand hospital wide. We evaluated the glucometrics before and after implementation as well as nursing satisfaction following calculator implementation.
Following calculator implementation, we noticed that the calculator had a tendency to rapidly uptitrate in response
to persistently high glucoses. This uptitration resulted in hypoglycemia and prompted us to transition to a percentage change in multiplier coefficient as opposed to absolute values. Following this adjustment, hypoglycemia and hyperglycemia rates decreased relative to baseline and a greater percentage of blood glucose readings were within goal. The percentage of blood glucose values below 54 decreased from 0.044% to 0.041% since implementation. The percentage of blood sugars between 70-180 increased from 88.4% to 91.1% and the percentage of blood sugars greater than 180 decreased from 11.22% to 8.57%. Nursing feedback was positive overall.
By utilizing an aggressive education campaign championing super-users and making adjustments to the calculator based on early problems that were encountered, we were able to improve glycemic control and limit glucose variability at our institution.
Improving Identification and Documentation of Patients Utilizing Insulin Pumps at MD Anderson Cancer Center
University of Texas, MD Anderson Cancer Center Houston, TX, USA
Patients utilizing insulin pumps at MD Anderson are seen across various clinical settings and there was no clear policy in place to dictate the evaluation management of these patients during inpatient admission. We developed a multidisciplinary group to create an insulin pump policy with goal to improve appropriate identification of patients utilizing an insulin pump during inpatient stay.
We established a multidisciplinary team of stakeholders including Nursing, Diagnostic Imaging, Clinical Effectiveness, EPIC support, Pharmacy, Endocrinology, Perioperative Service, Nursing informatics and Clinical Informatics with focus on use of insulin pump during inpatient admission. The inpatient insulin pump policy was put into effect on June 2017 and included documentation of patient and provider insulin pump agreement form, and insulin pump record to document insulin boluses administered via patient through insulin pump. The policy also clearly outlined the roles of nursing, primary team provider and Endocrinology consulting service in management of patients with insulin pump to give clear delineation of responsibilities, including specific instances when Endocrinology service needed to be notified to ensure patient safety. External Medical Devices section was created in electronic health record for central location to document presence of insulin pump.
Since the initiation of the insulin pump policy the number of patients identified with insulin pump related diagnosis on their problem list has increased from 92 prior to policy implementation to 269 patients in 2019 (192% increase). Of admitted patients, there was insulin pump identified as a diagnosis on hospital problem list in 23 patients prior to policy implementation in comparison to 67 in 2019 (191% increase). There has been an increase in safety event reporting surrounding insulin pumps due to increased awareness.
The Cure for Inpatient Harm
Digital Hospital, Inc. San Jose, CA, USA
Modeling a hospital inside a surrounding society as a thermodynamic system in order to apply existing external intensive force variables to reduce the entropy of inpatient harm.
Research societal and hospital literature to identify any past scientific experiments that used existing societal intensive force variables to affect the behavior of large internal systems.
Specific experiments have been found which prove the existence of an intensive force variable in society which can be applied to hospitals. A further experiment was revealed that proved the effectiveness of a particular societal intensive force variable in reducing inpatient harm in hospitals.
There are external societal forces which when applied to hospitals can cure inpatient harm and those forces have been proven to be effective.