Abstract
Early Hyperglycaemia After Initiation of Glucocorticoid Therapy Predicts Adverse Outcome in Patients with Acute Graft- Versus-Host Disease
Medical University of Graz, Institute of Internal Medicine, Department of Endocrinology and Diabetology Graz, Austria
Objective:
Due to first-line treatment with high-dose glucocorticoids (GC), steroid-induced hyperglycaemia develops frequently in patients with acute graft-versus-host disease (aGvHD) which is potentially impacting on their outcome.
Method:
We performed a retrospective analysis on 104 patients who received systemic GC for aGvHD and thoroughly investigated the consequences of aberrant glucose metabolism. In particular, we focused on glucose parameters early after initiation of GC.
Result:
With a median of 50 (range 4-513) blood glucose measurements during GC treatment, increasing mean, median and maximum glucose levels as well as the need for insulin treatment were associated with decreased overall survival (OS) in simple and multiple survival analysis. Early hyperglycaemia, as defined by mean blood glucose levels >125mg/dl during the first three days of GC therapy, was also found to be highly associated with adverse outcome (hazard ratio (HR) of 2.5 for death and HR of 3.26 for death due to non-relapse mortality in a competing risk analysis). A score based on early hyperglycaemia and non-response to GC within 7 days allowed the identification of three risk groups: patients with both risk factors (n= 32) had an inferior OS at 5 years of 4.1% as compared to 75.4% in patients with none (n =25). Patients with one risk factor (n=47) had a 5-year OS rate of 32.0% (p=0.0002 for trend).
Conclusion:
We identified early hyperglycaemia after GC initiation as a prominent risk factor for adverse outcome in aGvHD patients. A score based solely on early hyperglycaemia and lack of response to GC was highly predictive for survival in these patients.
Operational Benefits of Full eConnectivity in Glucose Monitoring in Point of Care Setting – Experience with Nova Statstrip
The Hospital for Sick Children, University of Toronto Toronto, Ontario, Canada
Objective:
Recent published studies in our laboratory indicate that introduction of improved point of care glucose monitoring using Nova Statstrip results in significant improvements in clinical decision making in neonatal intensive care and improved test utilization. In the current study, we studied the benefits of full eConnectivity of Point of care testing (POCT) glucometers across the hospital to improve workflow and capture data seamlessly and without error.
Method:
Along with the implementation of a more precise and accurate meter, Toronto Hospital for Sick Children has made a fundamental shift in the provision of point of care testing by implementation of complete eConnectivity. Previously, meters, results, quality control, and operators were all managed on paper. With the implementation of the Nova Statstrip glucosemeter, vendor supplied middleware was implemented which allowed electronic management of the meters, quality control and operators but still required the manual (paper) reporting of glucose results. The final step to full connectivity occurred with the implementation of vendor neutral middleware (AegisPOC) that allowed for all of the earlier middleware functionality, and the addition of automatic glucose meter resulting to the patient electronic medical record (EMR).
Result:
The combination of improved quality management via the mitigation and/or elimination of pre- and post-analytical sources of error and the ability of clinicians to have access to immediate centralized data as is offered with full connectivity was found to provide additional improvements to clinical outcomes and lead to improved test utilization and less need for duplicate testing in the central laboratory.
Conclusion:
Hospital-wide eConnectivity solutions enhance the benefits of bedside glucose monitoring and significantly reduce the rates of patient error including pre-analytical and post-analytical errors.
Prevalence of Psychiatric Disorders in Patients Admitted for Diabetic Ketoacidosis and Their Risk of Readmission
Memorial Family Medicine Residency Sugar Land, Texas
Introduction:
Physicians should identify risk factors for hospital readmission in order to improve patient care and reduce healthcare costs. Our goal is to determine if psychiatric disorder is a risk factor for recurrent DKA admissions and whether inpatient social work consultation helped lower risk of readmission.
Methods:
A retrospective chart review was performed on patients admitted for DKA between 1/2014-12/2015. We compared the incidence of psychiatric disorder, psychiatric diagnoses, rate of readmission twelve months after the index hospitalization, duration of hospitalization, effect of social work consultation, and patient characteristics. Chi-square, T-test, Odds ratio, and Relative risk were used to calculate statistical significance.
Results:
A total of 91 patients were admitted for DKA, of which, 43% were Black, 43% Latino, 9% White, and 5% Asian, p=0.02. The most common causes of DKA were medication noncompliance (65%), new diagnosis of diabetes (13%), sepsis (12%), and suboptimal treatment (5%). Out of 91 patients, 32 (35%, p=0.0003) had at least one diagnosis of psychiatric disorder including Depression (29%), Alcohol abuse (16%), Cannabis abuse (16%), and Cocaine abuse (10%). Patients with psychiatric disorder had higher odds of readmission (OR 6.53, 95 % CI: 2.5116.98, p=0.0001) and mean annual readmissions (1.22/year vs 0.39/year, p=0.0008) predominantly due to medication noncompliance (84%, p=0.004). There was no significant difference in A1C (12.5% vs 12.6%, p=0.44) and duration of hospitalization (5 vs 4 days, p=0.17) in patients with and without psychiatric disorder. There was no significant difference in incidence of DKA readmissions (38% vs 27%, RR 1.43, 95% CI: 0.71-2.89, p=0.32) among patients who did and did not receive social work consultation.
Conclusion:
Diabetic patients with comorbid psychiatric disorder have higher odds of DKA readmission within 12 months following the index hospitalization largely from medication noncompliance. Unfortunately, social work consultation had no effect on the readmission rate.
Two-Step Validation of the Performance of Blood Glucose Monitoring Systems for Assessment of Dysglycemia in Pediatric Patients
Department of Clinical Laboratory, Children’s Hospital of Fudan University
Shanghai, China
Objective:
In China there has been little validation of blood glucose monitoring systems (BGMS) used in neonatal and pediatric patients. Interest is now growing in ensuring accurate blood glucose measurements are obtained that enable safe and effective decision making for managing dysglycaemia. The aim of this study was to use an IDMS traceable reference method in a two-step validation approach to evaluate the clinical risk associated with using BGMS.
Method:
The calibration of the laboratory glucose oxidase reference method was verified using NIST standards 917c and 965b. As a first validation step, precision and specificity studies were performed on two BGMS. For the specificity studies the influence of interfering factors such as ascorbic acid, lactose, galactose, P-hydroxybutyrate, N-acetylcysteine, glutathione and hematocrit (19.8-65.5%) on the accuracy of glucose measurements. For the second validation step residual venous blood collected from pediatric patients was assessed with the BGMS demonstrating better specificity when compared to the reference method. The performance of the BGMS was assessed according to the performance criteria of Chinese consensus guidelines and CLSI POCT 12-A3.
Result:
Both BGMS showed a good correlation to the ID-LCMS aligned reference method. The accuracy of one of the BGMS was affected by ascorbic acid, lactose, galactose, N-acetylcysteine and glutathione which represented a clinical risk in pediatric patients. The accuracy of the other BGMS was unaffected. The BGMS Blood Glucose Monitoring System achieved 100% concordance to the Chinese performance criteria and 100% (< 5.56 mmol/L) and 98.1% (> 5.56 mmol/L) concordance to CLSI POCT12-A3 guidelines.
Conclusion:
This two-step validation approach combined with using a true definitive IDMS aligned reference method is a useful tool for identifying clinical risk associated with BGMS.
Expectation Meets Reality! Computerization of In-Hospital Diabetes Management Significantly Reduces Medication Errors
Joanneum Research GmbH, HEALTH - Institute for Biomedicine and Health Sciences, Graz, Austria
Objective:
We aimed to compare the attitude of users to medication error prevention of paper-based protocols and computerized systems, providing workflow and decision support in diabetes management, with the real occurrence of medication errors in a before and after study design.
Method:
Seventy-nine hospitalized patients with type 2 diabetes mellitus were treated with an algorithm driven basal-bolus insulin regimen in a before and after study. Patients were treated either based on a paper-based workflow protocol (PaperG: 37 patients) or by using the GlucoTab® system (CompG: 42 patients), a mobile computerized system that provides workflow and decision support. Health care professionals (HCPs) were invited to complete a short questionnaire which was designed to assess how users experienced the performance of either paper-based protocol or computerized system. The attitude of the users was compared with the real occurrence of medication errors in the study.
Result:
At the end of the study 12 (PaperG) and 17 (CompG) HCPs filled in the questionnaire. For PaperG, 8 of 12 HCPs answered that the protocol will help to prevent errors (1 did not answer). For CompG 16 of 17 HCPs answered that the system will help to prevent errors compared to standard care (1 did not answer). 11.1% of insulin dose calculations in PaperG were erroneous and the odds ratio of a hypoglycemic event following an insulin dosing error was 3.1 (95% CI 1.4-6.9). Dose calculation errors were entirely avoided in CompG.
Conclusion:
Fewer users in PaperG expected that the protocol prevents medication errors and a significantly higher number of medication errors was identified compared to CompG. Only computerization can be regarded as a safe measure to significantly reduce medication errors in complex therapies.
Multi-site Assessment of the Conformance of Hospital Blood Glucose Monitoring Systems to the October 11, 2016 FDA Guidance Entitled “Blood Glucose Monitoring Test Systems for Prescription Point-of-Care Use”
Nova Biomedical Waltham, Massachusetts, USA
Objective:
Accuracy is a key requirement for Blood Glucose Monitoring Systems (BGMS) used in critical care settings. Inaccurate results can lead to potentially fatal errors with insulin dosing. The FDA has recently published bedside glucose guidelines with tighter accuracy performance standards compared to previously published guidelines (ISO and CLSI). The aim of this study was to assess the concordance of three BGMS to the FDA performance criteria using central laboratory glucose measurement as the reference.
Method:
Method correlation was performed using leftover venous whole blood hospital patient specimens for BGMS and the corresponding plasma samples for the central laboratory analyzer. Accuracy interference studies were performed with a range of known BGMS interference factors. Parkes Error Grid (PEG) analysis was performed to assess the clinical risk associated with these interferences. The glucose results from the three BGMS were assessed according to the new FDA1 accuracy performance criteria.
Results:
The accuracy of two BGMS was affected by some of the interference factors which showed a high clinical risk in the PEG analysis. The accuracy of the other BGMS was unaffected by the interference factors. The two BGMS affected by interferences also failed to meet the new FDA performance criteria (BGMS 1 = 88% within ±12 mg/dL for glucose values <75mg/dL and 67.2% ±12% for glucose values > 75mg/dL, n= 264, BGMS 2 = 98.6% within ±12 mg/dL for glucose values <75mg/dL and 92% ±12% for glucose values > 75mg/dL, n=262). BGMS 3 was unaffected by interference and met FDA performance criteria (100% within ±12 mg/dL for glucose values <75mg/dL and 99.5% within ±12% for glucose values > 75mg/dL, n=526)
Conclusion:
BGMS differ in their accuracy when tested in hospitalized patients and with known accuracy interference factors. In this multisite assessment only one BGMS achieved the FDA performance criteria and demonstrated no clinical risk in PEG analysis.
Multi-site Evaluation of a Hospital POC Blood Glucose Monitoring System in Critically Ill Patients on Intensive Insulin Therapy
Department of Medical and Scientific Affairs, Nova Biomedical Waltham, Massachusetts, USA
Objective:
Accurate glucose measurement is important in maintaining glycemic control in critically ill patients on intensive insulin therapy (IIT). Inaccurate glucose readings can lead to insulin dosing errors and increased risk of hypoglycemia. This multi-center study was undertaken to validate the use of a BGMS in an extensive critically ill patient population comparing performance to IDMS aligned reference methods.
Method:
715 arterial and venous blood specimens from 603 critically ill patients on IIT were analysed using StatStrip BGMS and an IDMS NIST aligned reference laboratory glucose method. Markers associated with BGMS interference were also measured. BGMS glucose results were assessed according to CLSI1 and FDA2 accuracy performance criteria. Bland Altman bias plot analysis was performed on the interference markers to determine their influence on the accuracy of results.
Result:
The critically ill patient population comprised of an extensive range of severe clinical conditions with abnormal ranges of biochemical interference factors including haematocrit, pH, sO2, pCO2, pO2, sodium, potassium, calcium, lactate and triglycerides. Bias plot analysis did not show any influence of these factors on the accuracy of BGMS glucose measurement. Patients received a wide range of polypharmacy medication none of which influenced the accuracy of BGMS glucose results. The performance of the BGMS met CLSI (95.5% within ±12 mg/dL for glucose values <100mg/dL and 95.5% within ±12.5% for glucose values > 100mg/dL) and FDA performance guidelines (95.6% within ±12 mg/dL for glucose values <75mg/dL and 95.4% ±12% for glucose values > 75mg/dL)
Conclusion:
In this clinical risk assessment of BGMS performance in a challenging critically ill patient population, no known clinically significant interference was observed and the BGMS assessed demonstrated substantial equivalence to an IDMS traceable plasma hexokinase aligned laboratory reference methods validating its performance for use in patients on IIT.
Improving the Quality of Diabetes Care: A Survey Analysis of an Informatics Intervention
Wake Forest Baptist Medical Center, Department of Internal Medicine, Section of Endocrinology Winston Salem, NC, USA
Objective:
Diabetes mellitus is a chronic disease that often requires glucose testing and insulin use for monitoring and treatment. This requires many supplies that come in variable sizes, brands, compatibility, and insurance coverage. Provider and patient frustration with delayed testing and treatment often occurs if the supplies are not ordered properly.
Method:
We developed an informatics intervention in the electronic medical record for selected outpatient clinics to ease the burden of ordering diabetic supplies. Orders for generic diabetic supplies (pen needles, needles/syringes, lancets, lancing devices, alcohol swabs, glucometer, control solution, test strips) were grouped together with a description for the pharmacy to dispense based on patient preference, insurance coverage, and availability. Baseline and 6-month follow-up surveys were obtained from providers.
Result:
Pre-intervention, 27.2% (9/33) of respondents were either satisfied or very satisfied with the process of ordering diabetic supplies; post-intervention, 100% of respondents (28/28) were either satisfied or very satisfied. A further 96.4% (27/28) of respondents stated the informatics intervention made it easier for them to order diabetic supplies, and 100% of respondents (28/28) stated the intervention was either helpful or very helpful in the treatment of patients with diabetes.
Conclusion:
The informatics intervention was well-received among users as evidenced by pre and post intervention survey data. The majority of providers felt the intervention was helpful and made the process of ordering diabetic supplies easier, subsequently increasing provider satisfaction in the process. Given the success of the intervention in the clinics, we subsequently expanded use of the intervention to include inpatient discharge orders.
Pre-Clinical Automated Venous Blood Glucose Monitoring over 72 hours
Cascade Metrix, LLC Fishers, Indiana
Objective:
The study purpose was to inform the design and assess pre-clinical performance of an ex-vivo automated venous blood glucose monitoring system developed by Cascade Metrix.
Method:
At user defined measurement intervals, the system withdraws blood from the subject’s vascular line for measurement across a flow-through glucose sensor (BST Biosensor Technology GmbH, Germany), followed by complete reinfusion of the blood using non-heparinized saline. Before each measurement the system performs an automated calibration by running a 100 mg/dl glucose solution across the sensor. The current study was performed at Purdue University’s Pharmacology Testing Facility (West Lafayette, IN). Two young pigs (2 months old, each weighing ~ 30 kg) were tethered to two alpha devices placed adjacent to movement-responsive caging stations from BioAnalytical Systems, Inc. (West Lafayette, IN). Central venous catheters were surgically placed in the jugular vein of the pigs and the device was connected to catheter hub using specialized zero dead volume luer locks. Both the devices were controlled from a Windows Tablet. The study period was 72 hours.
Result:
The device was able to perform the automated sampling routine for the entire 3-day period with intermittent glucose measurements that were validated against a reference analyzer. On day 3, an intravenous glucose tolerance test was done on the pigs and the glucose spikes were tracked by the device.
Conclusion:
The study demonstrated the reliability of a fully automated sampling and glucose sensing in tethered pigs over a 3-day period without the need for heparin infusion. The study identified changes in the disposable design and the software user interface. Implementation of these changes is supporting a human investigational device application.
Evaluation of the Accuracy of the YSI Glucose Analyzer Relative to an IDMS Aligned Perchloric Acid Hexokinase Reference Glucose Method
Division of Clinical Biochemistry, Royal University Hospital Saskatoon, Saskatchewan, Canada
Objective:
The YSI glucose analyzer has been used as a reference method to assess whether blood glucose monitoring systems (BGMS) will meet accuracy expectations outlined by regulatory organizations. The influence of reference method accuracy on the outcome of BGMS assessments has recently been published. The objective of this study was to evaluate the accuracy of the YSI glucose reference analyzer compared with an IDMS aligned perchloric acid (PCA) hexokinase glucose reference method.
Method:
The effect of hematocrit and plasma water concentration on the performance of the YSI analyzer relative to the PCA hexokinase method was assessed using whole blood specimens. Linear regression analysis was conducted to determine the extent that hematocrit, plasma water concentration and glucose concentration predicted the performance of the YSI instrument.
Result:
YSI accuracy was influenced by both hematocrit and plasma water concentration in a glucose concentration dependent manner relative to the PCA hexokinase method. Prediction models based on the regression analysis indicated low hematocrit levels (down to 15%) can introduce up to a 15% positive bias over a glucose range of 36 to 270 mg/dL when glucose is measured with the YSI. At high hematocrit levels (up to 75%) up to a -18% bias can be introduced with YSI glucose measurements. At a 70% hematocrit and plasma water changes from 91% to 98%, glucose concentration dependent biases ranging from -16% to +3.1%.
Conclusion:
The susceptibility of the YSI analyzer to hematocrit and plasma water concentration emphasizes the need to understand biases that can influence reference glucose methods to ensure reliable evaluation of glucose meters and other whole blood glucose testing devices used within community, hospital and critical care units.
Increase in Time in Target when Using a Basal-Bolus Algorithm for Insulin Dosing during Hospital Stay
Medical University of Graz
Graz, Austria
Objective:
Insulin therapy and the use of clinical decision support systems to achieve glycemic control in hospitalized patients with hyperglycemia are recommended by clinical guidelines. The aim of this evaluation was to assess time in target by continuous glucose monitoring (CGM) achieved by a basal-bolus insulin algorithm over the course of hospital stay in patients with type 2 diabetes (T2D).
Method:
30 patients with T2D (12 female, age 67 ± 11 years, HbA1c 79 ± 2 6 mmol/mol, BMI 32 ± 6 kg/m2, diabetes duration 14 ± 11 years, creatinine 1.3 ± 0.5 mg/dL) were treated with GlucoTab®, a mobile decision support system providing automated workflow and suggestions for insulin dosing to health care professionals, during hospital stay. Insulins glargine U300 and glulisine were used for basal-bolus therapy. Additionally to blood glucose measurements using a standard point-of-care device blinded CGM (iPro2, Medtronic) was performed throughout the study. CGM was calibrated four times daily.
Result:
Mean total daily insulin dose was 63.8 ± 39.8 U. A total of 49,846 CGM values were collected. Mean daily sensor glucose was 152 ± 21mg/dL. Percentage of CGM values in the ranges was as follows: 100-140mg/dL (42.0%), 70-180mg/dL (80.2%), >180mg/dL (19.0%), >300mg/dL (1.5%). Percentage in the hypoglycemic range were low: <70mg/dL (0.77%), <60mg/dL (0.35%) and <50mg/dL (0.15%), respectively. When comparing the first vs. last treatment day, time in target 70-180mg/dL (61.8% vs. 85.2%) increased, whereas time in hyperglycemia >180mg/dL (37.1% vs. 14.2%) and hypoglycemia <70mg/dL (1.2% vs. 0.6%) decreased.
Conclusion:
Basal-bolus insulin therapy using a long acting basal insulin analogue safely establishes glycemic control as assessed by CGM. Over time percentage of values in target increases without increasing hypoglycemia risk.
Implementation of an Electronic Glycemic Management System “A Hospital Case Study”
Kaweah Delta Health Care District Visalia, CA USA
Objective:
Hyperglycemia effects up to 40% of all hospitalized patients and is associated with poor outcomes, including increased mortality, length of hospitalization and surgical site infections. The standard of care in critically ill patients is intravenous (IV) insulin, and for non-critical patients is subcutaneous (SubQ) basal bolus insulin therapy (BBI). Despite best practice guideline recommendations, there was a persistence of sliding scale insulin (SSI) use across our hospital due to many challenges, including lack of diabetes and endocrinology expertise. An electronic glycemic management system (eGMS) was selected to facilitate the conversion of our system to BBI.
Method:
This Retrospective Quality Improvement Case Study evaluated the safety and efficacy of usual care (UC) with IV and SubQ insulin compared to a nurse-directed eGMS using Glucommander (GM), an insulin-dosing algorithm integrated with our hospitals electronic health record. Primary objectives were to analyze the adoption rate of BBI and measure impact on glycemic outcomes, initially with a primary focus on patient safety as evidence by hypoglycemia trends.
Results:
Patients on UC had 7% of hypoglycemic events (% of BG<70mg/dL) compared to GM at 1.74%. UC, ICU only hypoglycemia (% of BG<70mg/dL) was 3% compared to GM at 1.5%. UC, hospital wide hyperglycemia (% of BG>180mg/dL) was 29% compared to GM at 30%. UC, ICU only hyperglycemia (% of BG>180mg/dL) was 30% compared to GM at 18%. Percent of hospital patients who were prescribed a BBI regimen was 5% at baseline compared to 96% using GM.
Conclusion:
The percent of patients prescribed best practice BBI significantly improved using GM. GM also improved patient safety by reducing hypoglycemia 50-75% hospital wide while significantly reducing hyperglycemia in the ICU setting compared to baseline.
Stress Hyperglycemia, but Not Diabetes, Increases In-Hospital Mortality
Hospital Israelita Albert Einstein Sao Paulo, SP, Brazil
Objective:
To evaluate the influence of diabetes (DM) and stress hyperglycemia (SH) on in-hospital mortality after adjusting for glycemic control, acute and chronic comorbidities.
Method:
This retrospective review included 53757 non-obstetrical adult (> 18 years) admissions, between 2 and 30 days in a 600-bed tertiary hospital (2010 to 2013). We compared the estimated risk of death among inpatients with DM (n=11737; 21,8%) and SH (n=4068; 7,6%) and normoglycemic nondiabetic inpatients (NL; n=37952; 70,6%), after adjustment to glycemic control (admission and average glucose, standard deviation and incidence of any hypoglycemic episode), patients’ characteristics (sex and age), acute stress factors (need of surgery, emergency admission, ICU admission), presence of bacterial infection or sepsis, and chronic comorbidities (congestive heart failure, end-stage renal failure and chronic obstructive pulmonary disease), using generalized estimating equations. Results were expressed in hazard ratio (confidence interval 95%), p value <0,05.
Result:
Compared to NL, the hazard ratio of mortality in DM and SH groups were 3,8 (3,2-4,5) and 17,4 (14,8-20,5) respectively, booth p<0,05 in the unadjusted model. After adjustment, the hazard ratio in DM and SH were 1,0 (0,81,2) and 3,3 (2,6-4,0). The difference remained significant only in SH group.
Conclusion:
In these population, DM and SH was associated to higher mortality risk compared to NL. However, after adjustment to patients’ characteristics, glycemic control, acute stress factors and chronic comorbidities, only stress hyperglycemia was associated with increased in-hospital mortality risk.
The Effects of a Clinical Decision Support Software Program Stability Requirement on Glycemic Outcomes
University of Missouri - Kansas City and Shawnee Mission Medical Center Overland Park, Kansas
Objective:
The purpose of this DNP project is to evaluate a quality improvement project implementing a clinical decision support software program stability requirement and the effects on blood glucose control in hospitalized diabetic and hyperglycemic adult patients following discontinuation of an intravenous insulin regimen.
Method:
This retrospective analysis of a before and after cohort will evaluate the percent of mean blood glucose values within the range of 70-180 mg/dL, percent of patient days with any blood glucose <40 mg/dL, <70 mg/dL, >180 mg/dL, and >300 mg/dL in patients on intravenous insulin, during the day of transition, and up to three days following discontinuation of intravenous insulin in both cohorts: those transitioned based on provider discretion and those transition after the software update that requires patients to meet a stability requirement prior to transition.
Result:
A mean comparison via independent t-test will be used to determine if there is a statistically significant difference between the patients transitioned under provider discretion and the patients required to meet the stability requirement prior to transition by comparing each of the primary outcomes listed above for each of the treatment groups, with an anticipated n=200. IRB determination of Not Human Subjects Research was received January 31, 2017 and data collection and analysis is ongoing. Demographic data will be analyzed for group differences utilizing the chi-squared statistic.
Conclusion:
Within clinical practice, clinical decision support software programs are frequently updated by the creators of the software, without the implications of these updates entirely known. Evaluation of the glycemic outcomes associated with this software update are important in establishing whether this quality improvement project resulted in improved outcomes for patients transitioned from intravenous to subcutaneous insulin regimens.
A Pilot Study Evaluating Timing of Pre-meal Glucose Monitoring, Prandial Insulin Administration and Meal Consumption
University of Chicago Chicago, Illinois
Objective:
Many hospitals, including ours, have transitioned from scheduled mealtimes to flexible meal delivery at times most convenient for patients. Point-of-care glucose monitoring, however, has not followed this trend and most often still occurs at fixed times with vital signs. A review of insulin-associated hypoglycemic events at our institution revealed that many episodes of hypoglycemia could be attributed to prandial and correction (sliding) scale insulin being administered for blood glucose levels that were collected one to two hours before the insulin administration and meal consumption. These values were potentially higher than the blood glucose level when the patient actually ate, leading to excess insulin dosing and postprandial hypoglycemia. In response to this, a multi-disciplinary, interdepartmental team designed a pilot study to assess the effectiveness of appropriately timing preprandial glucose monitoring to fewer than thirty minutes before insulin administration and meal consumption.
Methods:
Two units were selected and registered nurses and assistants were given an in-service by diabetes clinical nurse educators. Patient education materials were provided and all patients receiving insulin had a sign placed on their doors to alert the food service attendent sto leave the trays outside prior to informing the unit services coordinator that the food had arrived. If the blood glucose level had not been checked within the previous thirty minutes, it was rechecked prior to insulin administration.
Results:
Data continue to be collected but nurses report administering less prandial insulin.
Conclusions:
Despite a change in routine and potential short-term disruption in work-flow, it is possible to match point-of-care blood glucose monitoring with mealtimes and potentially reduce the incidence of iatrogenic hypoglycemia.
Accurate Glycemic Monitoring in Adult Critically 1ll Patients Decreases Exposure to Hypoglycemia: Clinical Impact of an Autocorrecting Blood Glucose Monitoring System
University of California, Davis, School of Medicine Davis, California
Objective:
Glycemic dysregulation occurs frequently in the critically ill. Although the use of tight glycemic control remains controversial, the dangers of glycemic variability, hypoglycemia, and hyperglycemia are undisputed. Recent studies suggest accurate blood glucose monitoring in critically ill patients reduces glycemic variability and frequency of hypoglycemic events. The objective of this study was to determine the impact of accurate blood glucose monitoring in critically ill adult patients.
Methods:
We conducted a retrospective study of adult (age>18 yrs) ICU patients from 2014-2016. From Jan 2014-Oct 2015, patients were treated with blood glucose monitoring system (BGMS)-1. In Nov 2015-Apr 2016, BGMS-2 was employed incorporating a sensor that autocorrects for interferences. Paired BGMS samples were compared to the lab. Demographics, medications, hypoglycemic events, and lab results were recorded, and length-of-stay (LOS). The average cost per ICU day was calculated and compared between BGMS-1 and 2 patients.
Results:
A total of 1,034 patients’ charts were stratified into BGMS-1 (n=689) and BGMS-2 groups (n=345) respectively. Demographics were similar between cohorts. BGMS-1 reported higher mean±SD bias vs. BGMS-2 when compared to the lab (6.2±8.5 vs. -2.3±6.1 mg/dL, P<0.001). Mean insulin rates (4.3±2.8 vs. 2.7±1.3 U/hr, P<0.001), and frequency of hypoglycemic events (35.7% vs. 15.2%, P<0.001) were significantly higher in BGMS-1 treated patients. BGMS-1 patients also took significantly longer to achieve glycemic control (9.7±3.8 vs. 5.2±4.4 hrs, P<0.001) and spent less time under the glycemic control interval (57.9% vs. 80.1%, P<0.001). Mean LOS for BGMS-1 vs. -2 (21.3±4.7 vs. 17.8±5.5 days, P=0.038) was significantly difference with a cost differential of $5,950 per patient.
Conclusions:
Accurate blood glucose testing using BGMS-2 reduces insulin rates, glycemic variability, and hypoglycemic events. BGMS-2 patients achieved glycemic control faster, maintained normoglycemia longer and experienced shorter LOS—resulting in an average savings of $5,950 per patient.
Accurate Glycemic Monitoring in Children with Severe Burns: Evaluation of an Autocorrecting Blood Glucose Monitoring System in a High-Risk Population
University of California, Davis, School of Medicine Davis, California
Introduction/Hypothesis:
Burn injury results in significant hypermetabolism and places patients at risk for stress-induced hyperglycemia. Although the use of tight glycemic control remains controversial, the dangers of glycemic variability, hypoglycemia, and hyperglycemia are undisputed. Recent studies suggest accurate autocorrecting blood glucose monitoring systems (BGMS) used in severely burned patients reduces glycemic variability and frequency of hypoglycemic events. The objective of this study was to determine the impact of accurate BGMS testing in severely burned children.
Methods:
We conducted a retrospective study of children (aged<18 yrs) with severe burns (>20% total body surface area) receiving intensive insulin therapy guided by a non-correcting (BGMS-1) or autocorrecting (BGMS-2) glucose meter. A total of 122 patient charts from 2001 to 2014 were reviewed. Paired BGMS samples were compared to the lab. Demographics, medications, hypoglycemic events, and lab results were recorded, and length-of-stay (LOS). The average cost per ICU day was calculated and compared between BGMS-1 and 2 patients.
Results:
Sixty-three patients received intensive insulin therapy using BGMS-1 and 59 via BGMS-2. Patient demographics were similar between the two groups. Mean insulin infusion rates (5.1±3.8 U/hr; n=535 paired measurements vs. 2.4 ±1.3 U/hr; n = 511 paired measurements; p<0.001), and frequency of hypoglycemic events (90 vs 12; p<0.001) were significantly higher in BGMS-1-treated patients. BGMS-2 patients achieve glycemic control more quickly (5.7±4.3 vs 13.1±6.9 hr; p< 0.001) and stayed within the target glycemic control range longer compared with BGMS-1 patients (85.2%±13.9% vs 57.9%±29.1%; p<0.001). Mean LOS for BGMS-1 vs. -2 (22.1±3.4 vs. 19.6±4.8 days, p=0.042) was significantly difference with a cost differential of $4,820 per patient.
Conclusions:
Accurate blood glucose testing using BGMS-2 reduces insulin rates, glycemic variability, and hypoglycemic events. BGMS-2 patients achieved glycemic control faster, maintained normoglycemia longer and experienced shorter LOS—resulting in an average savings of $5,950 per patient.
Clinical Impact of Using an Electronic Glycemic Management System at a Large Academic Medical Center
Grady Health System Atlanta, GA USA
Objective:
The benefits of optimal glycemic control in hospitalized patients have been widely studied and strategies should focus on controlling hyperglycemia and avoidance of hypoglycemia. The primary objective was to compare hypoglycemia in critically ill patients at a large academic medical center receiving insulin therapy by usual care vs electronic Glycemic Management System (eGMS). Secondary objectives: additional glycemic measures.
Method:
Retrospective, single center comparison of patients receiving insulin therapy by usual care (UC) prior to and after implementation of eGMS.
Result:
For patients on UC (n=2,079) 32,850 blood glucose (BG) values were evaluated. The average initial BG was 241.48 mg/dL (SD = 205.35), 26.3% of patients experienced hypoglycemia <70mg/dL (n=547) and 8.13% of patients had severe hypoglycemia (BG <40mg/dL; n=169). The % of BGs <70mg/dL was 5.73% (n = 1,881 BGs) and <40mg/dL was 1.11% (n = 363 BGs). Patients on eGMS (n=90) had 7,615 BG values evaluated, with a higher average initial BG of 274.94 mg/dL (SD = 99.29; P-value< 0.0001, significantly less patients with hypoglycemia <70mg/dL at 12.2% (n=11 patients, p<0.0001) and significantly less hypoglycemic events <70mg/dL at 0.28% (n = 21 BGs; P-value < 0.0001). Severe hypoglycemia (BG<40) was also significantly lower with the use of eGMS at 2.22% (n = 2 patients; P-value < 0.0001) and 0.03% (n = 2 BGs; P-value < 0.000) for patients and events, respectively. The average final BG for eGMS was also significantly lower than UC, at 160.76 mg/dL (SD = 48.74; P-value < 0.0001) versus 180.21 mg/dL (SD = 98.16).
Conclusion:
eGMS can significantly reduce adverse effects associated with insulin therapy in hospitalized patients while effectively optimizing BG control compared to usual care.
NIST Calibration Alignment is Essential When Selecting a Laboratory Reference Method for Evaluating POC Blood Glucose Monitoring Systems
Department of Clinical Laboratory, Children’s Hospital of Fudan University Shanghai, China
Objective:
It is important that evaluations of Blood Glucose Monitoring Systems (BGMS) are standardized to a true and traceable definitive reference method. We assessed the calibration of our laboratory hexokinase method with NIST standards before commencing an assessment of the performance of BGMS and found unexpected results.
Method:
We tested twelve glucose levels (1.39 - 55.56 mg/dL) prepared from NIST standard SRM917c and four glucose levels of NIST SRM965b (1.836, 4.194, 6.575, and 16.35 mg/dL) on our laboratory analyser using hexokinase and glucose oxidase methods for glucose measurement. Testing was performed at several time points and the sample aliquots were also tested at another laboratory in Shanghai. During the assessment period, an engineering fault occurred with the laboratory analyser which required maintenance and repair.
Result:
The mean % bias for all twelve NIST SRM917c samples tested in two test runs using the hexokinase method prior to the engineering fault was <1%. After analyser maintenance, the mean % bias for all twelve samples was 6.9% and 11.9 % in two runs. For NIST SRM965b samples a mean % bias of 0.4%, 3.9% and 2.6% for three test runs prior to the engineering fault and 15.1%, 9.3%, 10.1%, 8.8%, 9.0%, 9.7% for six test runs after correction of the engineering fault was obtained. The NIST samples tested using the glucose oxidase reagents and at the other laboratory site continued to give expected target values showing low mean % bias indicating that the pattern of results seen with the hexokinase reagents was not due to instability of the NIST standards.
Conclusion:
It is important that NIST standards are used to assess the calibration of a reference glucose method before commencing an evaluation of the performance of blood glucose monitoring systems.