Use of continuous glucose monitoring to estimate insulin requirements in patients with type 1 diabetes mellitus during a short course of prednisone

Optimal initial insulin dosage for managing steroid-induced hyperglycemia in hospitalized COVID-19 patients: A retrospective single-center study

Objectives

To determine the optimal initial insulin dosage for controlling hyperglycemia in COVID-19 patients receiving steroids, an area with limited data.

Methods

We retrospectively analyzed 156 COVID-19 patients with steroid-induced hyperglycemia treated with insulin. Patients were categorized by their total daily dose of subcutaneous insulin therapy when starting dexamethasone ⩾6 mg/day or equivalent dose of glucocorticoid: Group A (⩽0.29 units/kg), Group B (0.3-0.49 units/kg), Group C (0.5-0.69 units/kg), and Group B (⩾0.7 units/kg). Treatment failure was defined as mean blood glucose level > 280 mg/dL for two consecutive days after initiating insulin or any blood glucose ⩾ 400 mg/dL.

Results

The mean age was 64 ± 14 years, with 50% male, and a mean body mass index of 26.9 ± 6.9 kg/m2. Most had preexisting type 2 diabetes (62%). Mean admission blood glucose and HbA1c were 233 ± 112 mg/dL and 7.8 ± 2.3%, respectively. Group A had the lowest HbA1c (6.7 ± 1.2%), while group D had the highest (9.8 ± 2.5%). Median daily dexamethasone dosage or equivalent was 36 (IQR 16.72) mg, with no significant differences in among groups. Group A had the lowest treatment failure rate. There were no significant differences in treatment failure rate between Groups B, C, and D. Additionally, there were no statistically significant differences in mean BG across the groups: Group A 232 ± 42 mg/dL, Group B 247 ± 57 mg/dL, Group C 247 ± 61 mg/dL, and Group D 227 ± 67 mg/dL (p = 0.2). Group D had a significantly higher rate of level 1 hypoglycemia (p = 0.008), while no differences in clinically significant hypoglycemia (level 2 or 3) were observed between groups.

Conclusions

Among patients requiring TDD ⩾ 0.3 units/kg/day, there was no significant difference in treatment failure rate between Groups B, C, and D. Group D had the highest rate of level 1 hypoglycemia. This initial insulin dosage for hospitalized COVID-19 patients on high-dose steroid therapy should be personalized.

Prevalence of Steroid-Induced Hyperglycemia in King Abdulaziz Specialist Hospital, Taif City, Saudi Arabia

Background Hyperglycemia is a common side effect of high-dose steroid therapy in hospitalized patients. Objectives To assess the prevalence of hyperglycemia among hospitalized patients receiving steroid therapy. Methods A retrospective study was conducted among 245 patients. The inclusion criteria were patients undergoing steroid therapy and admitted to a single tertiary care hospital due to medical complications or exacerbation of the diseases they were suffering from. Data encompassing patient demographics, admission, discharge dates, comorbidities, medication histories, laboratory results (including blood glucose levels), and documented corticosteroid administrations were meticulously gathered from electronic health records (EHRs). A logistic regression model analysis was done to predict the risk factors of poor glycemic control among hospitalized patients. Results The prevalence of hyperglycemia among the patients who were on steroid therapy was 34.2%. About 70.7% of the patients who required insulin at the time of admission required >17 units, and the insulin requirement was significantly higher among patients who received dexamethasone compared to other steroids (p<0.05). Older age (>65 years) was found to be independently associated with poor glycemic control (p<0.05). Conclusion The study revealed that almost one-third of patients on steroid therapy had hyperglycemia. Monitoring of patients for hyperglycemia after beginning high-dose steroid therapy should be done.

Continuous glucose monitoring (CGM) in a non-Icu hospital setting: The patient's journey

AIMS

Although consistent data support the outpatient use of continuous glucose monitoring (CGM) to improve glycemic control and reduce hypoglycemic burden, and clinical outcomes, there are limited data regarding its use in the hospital setting, particularly in the non-intensive care unit (non-ICU) setting. The emerging use of CGM in the non-critical care setting may be useful in increasing the efficiency of hospital care and reducing the length of stay for patients with diabetes while improving glycemic control.

DATA SYNTHESIS

The purpose of this Expert Opinion paper was to evaluate the state of the art and provide a practical model of how CGM can be implemented in the hospital.

SETTING

A patient's CGM journey from admission to the ward to the application of the sensor, from patient education on the device during hospitalization until discharge of the patient to maintain remote control.

CONCLUSIONS

This practical approach for the implementation and management of CGM in patients with diabetes admitted to non-ICUs could guide hospitals in their diabetes management initiatives using CGM, helping to identify patients most likely to benefit and suggesting how this technology can be implemented to maximize clinical benefits.

Leverage biomaterials to modulate immunity for type 1 diabetes

The pathogeny of type 1 diabetes (T1D) is mainly provoked by the β-cell loss due to the autoimmune attack. Critically, autoreactive T cells firsthand attack β-cell in islet, that results in the deficiency of insulin in bloodstream and ultimately leads to hyperglycemia. Hence, modulating immunity to conserve residual β-cell is a desirable way to treat new-onset T1D. However, systemic immunosuppression makes patients at risk of organ damage, infection, even cancers. Biomaterials can be leveraged to achieve targeted immunomodulation, which can reduce the toxic side effects of immunosuppressants. In this review, we discuss the recent advances in harness of biomaterials to immunomodulate immunity for T1D. We investigate nanotechnology in targeting delivery of immunosuppressant, biological macromolecule for β-cell specific autoreactive T cell regulation. We also explore the biomaterials for developing vaccines and facilitate immunosuppressive cells to restore immune tolerance in pancreas.

Diabetes Technology Meeting 2021

Diabetes Technology Society hosted its annual Diabetes Technology Meeting on November 4 to November 6, 2021. This meeting brought together speakers to discuss various developments within the field of diabetes technology. Meeting topics included blood glucose monitoring, continuous glucose monitoring, novel sensors, direct-to-consumer telehealth, metrics for glycemia, software for diabetes, regulation of diabetes technology, diabetes data science, artificial pancreas, novel insulins, insulin delivery, skin trauma, metabesity, precision diabetes, diversity in diabetes technology, use of diabetes technology in pregnancy, and green diabetes. A live demonstration on a mobile app to monitor diabetic foot wounds was presented.

Outpatient Randomized Crossover Automated Insulin Delivery Versus Conventional Therapy with Induced Stress Challenges

Background: Automated insulin delivery (AID) systems have not been evaluated in the context of psychological and pharmacological stress in type 1 diabetes. Our objective was to determine glycemic control and insulin use with Zone Model Predictive Control (zone-MPC) AID system enhanced for states of persistent hyperglycemia versus sensor-augmented pump (SAP) during outpatient use, including in-clinic induced stress. Materials and Methods: Randomized, crossover, 2-week trial of zone-MPC AID versus SAP in 14 adults with type 1 diabetes. In each arm, each participant was studied in-clinic with psychological stress induction (Trier Social Stress Test [TSST] and Socially Evaluated Cold Pressor Test [SECPT]), followed by pharmacological stress induction with oral hydrocortisone (total four sessions per participant). The main outcomes were 2-week continuous glucose monitor percent time in range (TIR) 70-180 mg/dL, and glucose and insulin outcomes during and overnight following stress induction. Results: During psychological stress, AID decreased glycemic variability percentage by 13.4% (P = 0.009). During pharmacological stress, including the following overnight, there were no differences in glucose outcomes and total insulin between AID and physician-assisted SAP. However, with AID total user-requested insulin was lower by 6.9 U (P = 0.01) for pharmacological stress. Stress induction was validated by changes in heart rate and salivary cortisol levels. During the 2-week AID use, TIR was 74.4% (vs. SAP 63.1%, P = 0.001) and overnight TIR was 78.3% (vs. SAP 63.1%, P = 0.004). There were no adverse events. Conclusions: Zone-MPC AID can reduce glycemic variability and the need for user-requested insulin during pharmacological stress and can improve overall glycemic outcomes. Clinical Trial Identifier NCT04142229.

The Relationship Between Diabetes Mellitus and Cancers and Its Underlying Mechanisms

Epidemiological studies suggest associations between diabetes mellitus and some cancers. The risk of a number of cancers appears to be increased in diabetes mellitus. On the other hand, some cancer and cancer therapies could lead to diabetes mellitus. Genetic factors, obesity, inflammation, oxidative stress, hyperglycemia, hyperinsulinemia, cancer therapies, insulin and some oral hypoglycemic drugs appear to play a role in the crosstalk between diabetes mellitus and cancers. This review summarized the associations between various types of diabetes and cancers and updated available evidence of underlying mechanisms between diabetes and cancers.

Short course corticosteroid treatment and closed-loop insulin delivery system: the experience of the DBLG1 pre-launch

No abstract (brief report).

Insulin management in hospitalized patients with diabetes mellitus on high-dose glucocorticoids: Management of steroid-exacerbated hyperglycemia

Background

Glucocorticoid (GC)-exacerbated hyperglycemia is prevalent in hospitalized patients with diabetes mellitus (DM) but evidence-based insulin guidelines in inpatient settings are lacking.

Methods and findings

Retrospective cohort study with capillary blood glucose (CBG) readings and insulin use, dosed with 50% basal (glargine)-50% bolus (lispro) insulin, analyzed in hospitalized patients with insulin-treated DM given GC and matched controls without GC (n = 131 pairs). GC group (median daily prednisone-equivalent dose: 53.36 mg (IQR 30.00, 80.04)) had greatest CBG differences compared to controls at dinner (254±69 vs. 184±63 mg/dL, P<0.001) and bedtime (260±72 vs. 182±55 mg/dL, P<0.001). In GC group, dinner CBG was 30% higher than lunch (254±69 vs. 199±77 mg/dL, P<0.001) when similar lispro to controls given at lunch. Bedtime CBG not different from dinner when 20% more lispro given at dinner (0.12 units/kg (IQR 0.08, 0.17) vs. 0.10 units/kg (0.06, 0.14), P<0.01). Despite receiving more lispro, bedtime hypoglycemic events were lower in GC group (0.0% vs. 5.9%, P = 0.03).

Conclusions

Since equal bolus doses inadequately treat large dinner and bedtime GC-exacerbated glycemic excursions, initiating higher bolus insulin at lunch and dinner with additional enhanced GC-specific insulin supplemental scale may be needed as initial insulin doses in setting of high-dose GC.

A Practical Guide for the Management of Steroid Induced Hyperglycaemia in the Hospital

Glucocorticoids represent frequently recommended and often indispensable immunosuppressant and anti-inflammatory agents prescribed in various medical conditions. Despite their proven efficacy, glucocorticoids bear a wide variety of side effects among which steroid induced hyperglycaemia (SIHG) is among the most important ones. SIHG, potentially causes new-onset hyperglycaemia or exacerbation of glucose control in patients with previously known diabetes. Retrospective data showed that similar to general hyperglycaemia in diabetes, SIHG in the hospital and in outpatient settings detrimentally impacts patient outcomes, including mortality. However, recommendations for treatment targets and guidelines for in-hospital as well as outpatient therapeutic management are lacking, partially due to missing evidence from clinical studies. Still, SIHG caused by various types of glucocorticoids is a common challenge in daily routine and clinical guidance is needed. In this review, we aimed to summarize clinical evidence of SIHG in inpatient care impacting clinical outcome, establishment of diagnosis, diagnostic procedures and therapeutic recommendations.

Management of Type 1 Diabetes in the Hospital Setting

PURPOSE OF REVIEW

The purpose of this article was to review recent guideline recommendations on glycemic target, glucose monitoring, and therapeutic strategies, while providing practical recommendations for the management of medical and surgical patients with type 1 diabetes (T1D) admitted to critical and non-critical care settings.

RECENT FINDINGS

Studies evaluating safety and efficacy of insulin pump therapy, continuous glucose monitoring, electronic glucose management systems, and closed loop systems for the inpatient management of hyperglycemia are described. Due to the increased prevalence and life expectancy of patients with type 1 diabetes, a growing number of these patients require hospitalization every year. Inpatient diabetes management is complex and is best provided by a multidisciplinary diabetes team. In the absence of such resource, providers and health care staff must become familiar with the features of this condition to avoid complications such as severe hyperglycemia, ketoacidosis, hypoglycemia, or glycemic variability. We reviewed most recent guidelines and relevant literature in the topic to provide practical recommendations for the inpatient management of patients with T1D.

Glucocorticoids and Diabetes

Glucocorticoids (GCs), stress hormones produced by the adrenal gland, are involved in many pathways in physiology and metabolism including glucose homeostasis and inflammation. Excess GC signaling results in Cushing's syndrome and possibly metabolic syndrome. Diabetes, central adiposity, and hyperlipidemia are components of both syndromes. Here, we discuss the mechanisms of GC action, clinical syndromes of GC excess, modulation of glucose homeostasis by GCs, and future treatments for diabetes based on GC signaling.

Co-Managing Patients with Type 1 Diabetes and Cancer

The life expectancy of people with type 1 diabetes is improving and now approaches that of those without diabetes. As this population ages, a growing number will be diagnosed with and treated for cancer. Cancer treatments can drastically affect insulin requirement and glycemic control through multiple mechanisms including high doses of glucocorticoids and targeted therapies that directly interfere with cellular pathways involved in the action of insulin. Patients with cancer frequently also have alterations in gastrointestinal motility or appetite and require supplemental enteral or parenteral nutrition. Few studies have evaluated these patients directly, but data on patients with and without diabetes suggest that glycemic control may play a larger role in cancer outcomes than is often recognized. Collaboration between the treating oncologist and diabetologist allows people with diabetes to receive the most effective therapies for their cancers without undue risk of hypoglycemia or adverse outcomes due to hyperglycemia.

Randomized clinical trial of the efficacy and safety of insulin glargine vs. NPH insulin as basal insulin for the treatment of glucocorticoid induced hyperglycemia using continuous glucose monitoring in hospitalized patients with type 2 diabetes and respiratory disease

AIMS

To investigate the clinical efficacy and safety of insulin glargine compared with NPH insulin as basal insulin for the management of corticosteroid-induced hyperglycemia in hospitalized people with type 2 diabetes (T2DM) and respiratory disease.

MATERIALS AND METHODS

Randomized, two-arm parallel group, clinical trial undertaken from February 2011 to November 2012 on the pneumology ward of the Hospital Regional Universitario de Málaga (Spain), involving 53 participants with T2DM treated with medium/high doses of intermediate-acting corticosteroids. Participants were randomly assigned to receive one single dose of insulin glargine or NPH insulin in three equally divided doses before each meal as basal insulin within a basal-bolus insulin protocol. The intervention lasted six days or until discharge if earlier.

RESULTS

No significant differences were seen between groups during the study in mean blood glucose (11.43±3.44 mmol/l in glargine vs. 11.88±2.94 mmol/l in NPH, p=0.624), and measures of glucose variability (standard deviation 3.27±1.16 mmol/l vs. 3.61±0.99 mmol/l, p=0.273; coefficient of variation 1.55±0.33 mmol/l vs. 1.72±0.39 mmol/l, p=0.200). Results from CGM were concordant with those obtained with capillary blood glucose reading. The length of hospital stay was also similar between groups (8.2±2.8 days vs. 9.8±3.4 days, p=0.166) There was a non significant trend for lower episodes of mild (4 vs. 8, p=0.351) and severe hypoglycemia (0 vs. 3, p=0.13) in the glargine group.

CONCLUSIONS

The results of this study showed that insulin glargine and NPH insulin are equally effective in a basal-bolus insulin protocol to treat glucocorticoid-induced hyperglycemia in people with T2DM on a pneumology ward.

Physician's attitudes towards diagnosing and treating glucocorticoid induced hyperglycaemia: Sliding scale regimen is still widely used despite guidelines

AIMS

Treatment with glucocorticoids for neoplasms and inflammatory disorders is frequently complicated by glucocorticoid induced hyperglycaemia (GCIH). GCIH is associated with adverse outcomes and its treatment has short term and long term benefits. Currently, treatment targets and modalities depend on local protocols and habits of individual clinicians. We explored current practice of screening and treatment of GCIH in patients receiving glucocorticoid pulse therapy.

METHODS

A factorial survey with written case vignettes. All vignette patients received glucocorticoid pulse therapy. Other characteristics (e.g., indication for glucocorticoid therapy, pre-existent diabetes) varied. The survey was held between November 2013 and May 2014 on 2 nationwide conferences and in hospitals across The Netherlands. Pulmonologists and internists expressed their level of agreement with statements on ordering capillary glucose testing and treatment initiation.

RESULTS

Respondents ordered screening for GCIH in 85% of vignette patients and initiated treatment in 56%. When initiating treatment, respondents opt for sliding scale insulin in 62% of patients. Sliding scale insulin was more frequently prescribed in patients with pre-existent insulin dependent diabetes (OR 2.4, CI 1.3-4.2) and by residents (vs. specialists, OR 2.1, CI 1.2-3.5). Sixty-nine percent of clinicians experienced a lack of guidelines for GCIH.

CONCLUSIONS

Clinicians have a strong tendency to screen for GCIH but subsequent initiation of treatment was low. Sliding scale insulin is still widely used in episodic GCIH despite evidence against its effectiveness. This may be due to lacking evidence on feasible treatment options for GCIH.

Artificial pancreas (AP) clinical trial participants' acceptance of future AP technology

BACKGROUND

Artificial pancreas (AP) systems are currently an active field of diabetes research. This pilot study examined the attitudes of AP clinical trial participants toward future acceptance of the technology, having gained firsthand experience.

SUBJECTS AND METHODS

After possible influencers of AP technology adoption were considered, a 34-question questionnaire was developed. The survey assessed current treatment satisfaction, dimensions of clinical trial participant motivation, and variables of the technology acceptance model (TAM). Forty-seven subjects were contacted to complete the survey. The reliability of the survey scales was tested using Cronbach's α. The relationship of the factors to the likelihood of AP technology adoption was explored using regression analysis.

RESULTS

Thirty-six subjects (76.6%) completed the survey. Of the respondents, 86.1% were either highly likely or likely to adopt the technology once available. Reliability analysis of the survey dimensions revealed good internal consistency, with scores of >0.7 for current treatment satisfaction, convenience (motivation), personal health benefit (motivation), perceived ease of use (TAM), and perceived usefulness (TAM). Linear modeling showed that future acceptance of the AP was significantly associated with TAM and the motivation variables of convenience plus the individual item benefit to others (R(2)=0.26, P=0.05). When insulin pump and continuous glucose monitor use were added, the model significance improved (R(2)=0.37, P=0.02).

CONCLUSIONS

This pilot study demonstrated that individuals with direct AP technology experience expressed high likelihood of future acceptance. Results support the factors of personal benefit, convenience, perceived usefulness, and perceived ease of use as reliable scales that suggest system adoption in this highly motivated patient population.

Effect of inhaled corticosteroids on glycemic status

Although the dysglycemic effects of systemic glucocorticoid therapy are well known, the effect of inhaled corticosteroids (ICS) on carbohydrate metabolism is still a subject of debate. The systemic bioavailability of ICS is claimed to be minimal and the side effects negligible. However, some large retrospective cohort studies showed a definite association between ICS use and incident diabetes or worsening glycemic control in pre-existing diabetes. There are no professional-body recommended guidelines on the diagnosis and management of steroid-induced diabetes for the general population. This review aims to evaluate the systemic dysglycemic effect of ICS treatment and to propose a management algorithm.

Management of the hospitalized patient with type 1 diabetes mellitus

Patients with type 1 diabetes mellitus (T1DM) have minimal to absent pancreatic β-cell function and rely on the exogenous delivery of insulin to obtain adequate and life-sustaining glucose homeostasis. Maintaining glycemic control is challenging in hospitalized patients with T1DM, as insulin requirements are influenced by the presence of acute medical or surgical conditions, as well as altered nutritional intake. The risks of hyperglycemia, ketoacidosis, hypoglycemia, and glycemic variability are increased in hospitalized patients with T1DM. Diabetic ketoacidosis and severe hypoglycemia are the 2 most common emergency conditions that account for the majority of hospital admissions in patients with T1DM. The association between hyperglycemia and increased risk of complications and mortality in patients with type 2 diabetes (T2DM) is well established; however, the impact of glycemic control on clinical outcomes has not been determined in patients with T1DM who present without ketoacidosis. To decrease complications associated with insulin therapy, health care professionals must be well versed in the use of insulin because it is a common source of medication error. For non-critically ill, hospitalized patients, subcutaneous insulin given to cover basal and prandial needs instead of sliding scale is the preferred method of insulin dosing. Protocols are available for initiating and titrating insulin doses, as well as for transitioning from an insulin infusion to a subcutaneous regimen. In our review, we identify and discuss special considerations related to inpatient glycemic control of non-ketotic patients with T1DM. Additionally, point differences and similarities associated with the management of patients with T2DM are discussed.

Pathways to quality inpatient management of hyperglycemia and diabetes: a call to action

Currently patients with diabetes comprise up to 25-30% of the census of adult wards and critical care units in our hospitals. Although evidence suggests that avoidance of hyperglycemia (>180 mg/dL) and hypoglycemia (<70 mg/dL) is beneficial for positive outcomes in the hospitalized patient, much of this evidence remains controversial and at times somewhat contradictory. We have recently formed a consortium for Planning Research in Inpatient Diabetes (PRIDE) with the goal of promoting clinical research in the area of management of hyperglycemia and diabetes in the hospital. In this article, we outline eight aspects of inpatient glucose management in which randomized clinical trials are needed. We refer to four as system-based issues and four as patient-based issues. We urge further progress in the science of inpatient diabetes management. We hope this call to action is supported by the American Diabetes Association, The Endocrine Society, the American Association of Clinical Endocrinologists, the American Heart Association, the European Association for the Study of Diabetes, the International Diabetes Federation, and the Society of Hospital Medicine. Appropriate federal research funding in this area will help ensure high-quality investigations, the results of which will advance the field. Future clinical trials will allow practitioners to develop optimal approaches for the management of hyperglycemia in the hospitalized patient and lessen the economic and human burden of poor glycemic control and its associated complications and comorbidities in the inpatient setting.

Challenges and Recent Progress in the Development of a Closed-loop Artificial Pancreas

Pursuit of a closed-loop artificial pancreas that automatically controls the blood glucose of individuals with type 1 diabetes has intensified during the past six years. Here we discuss the recent progress and challenges in the major steps towards a closed-loop system. Continuous insulin infusion pumps have been widely available for over two decades, but "smart pump" technology has made the devices easier to use and more powerful. Continuous glucose monitoring (CGM) technology has improved and the devices are more widely available. A number of approaches are currently under study for fully closed-loop systems; most manipulate only insulin, while others manipulate insulin and glucagon. Algorithms include on-off (for prevention of overnight hypoglycemia), proportional-integral-derivative (PID), model predictive control (MPC) and fuzzy logic based learning control. Meals cause a major "disturbance" to blood glucose, and we discuss techniques that our group has developed to predict when a meal is likely to be consumed and its effect. We further examine both physiology and device-related challenges, including insulin infusion set failure and sensor signal attenuation. Finally, we discuss the next steps required to make a closed-loop artificial pancreas a commercial reality.

Cystic fibrosis-related diabetes in adults: inpatient management of 121 patients during 410 admissions

BACKGROUND

With improved longevity, cystic fibrosis (CF)-related diabetes (CFRD) has emerged as the most common nonpulmonary complication of CF. Patients with CFRD are frequently admitted to the hospital with infections and deterioration of pulmonary function, during which time glycemic control might have an impact on pulmonary function, recovery from infection, and survival.

METHODS AND RESULTS

In an attempt to share our insight into inpatient management of CFRD, this article summarizes the experience of our inpatient glucose management team with hospital management of 121 adult CFRD patients who were hospitalized on 410 occasions at the University of Colorado Hospital between January 2009 and September 2011. This is a retrospective chart review descriptive study of inpatient management of CFRD in our center. Our cohort includes CFRD patients treated with basal and mealtime insulin through multiple daily injections or continuous subcutaneous insulin infusion (CSII), as well as patients receiving steroids or enteral nutrition, which adds complexity to the management of CFRD during hospitalization.

CONCLUSIONS

Multiple hospitalizations and intensive inpatient management of CF are integral elements of treatment. Inpatient therapy for CFRD requires a customized approach that is uniquely different from that of type 1 or type 2 diabetes. Our experience highlights clinical circumstances such as irregular food intake, high dose steroid therapy, and supplemental tube feeding. For many patients, it is possible to continue CSII therapy during hospitalization through a combination of mutual trust between the patient and hospital staff and oversight provided by the glucose management team.

Insulin pump use in adults with type 1 diabetes--practical issues

In recent years continuous subcutaneous insulin infusion pumps have become widely adopted in many parts of the world in the treatment of type 1 diabetes in adults. A comprehensive summary of all aspects of pump therapy is beyond the scope of this article, and in this review we will focus on several practical issues that in our experience are of clinical importance in the care of patients using insulin pumps. These include: benefits and risks of pump therapy, including the use of pumps to limit hypoglycemia; individual patient considerations in choosing between pump therapy and multiple daily injections; common pump-specific etiologies of erratic glucose control, including routine clinical practices that can assist with the detection of these problems; and the use of different pump bolus types for prandial insulin coverage.

Automatic Detection of Stress States in Type 1 Diabetes Subjects in Ambulatory Conditions

Two levels of control are crucial to the robustness of an artificial β-cell, a medical device that would automatically regulate blood glucose levels in patients with type 1 diabetes. A low-level component would attempt to regulate blood glucose continuously, while a supervisory-level, or monitoring, component would detect underlying changes in the subject's glucose-insulin dynamics and take corrective actions accordingly. These underlying changes, or "faults," can include changes in insulin sensitivity, sensor problems, and insulin delivery problems, to name a few. A multivariate statistical monitoring technique, principal component analysis (PCA), has been applied to both simulated and experimental type 1 diabetes data. The objective of this study was to determine if PCA could be used to distinguish between normal patient data, and data for abnormal conditions that included a variety of "faults." The PCA results showed a high degree of accuracy; for data from nine type 1 diabetes subjects in ambulatory conditions, 33 of 37 total test days (89%), including fault days and normal days, were classified correctly. Thus, the proposed monitoring technique shows considerable promise for incorporation into an artificial β-cell.

Experimental evaluation of a recursive model identification technique for type 1 diabetes

BACKGROUND

A model-based controller for an artificial beta cell requires an accurate model of the glucose-insulin dynamics in type 1 diabetes subjects. To ensure the robustness of the controller for changing conditions (e.g., changes in insulin sensitivity due to illnesses, changes in exercise habits, or changes in stress levels), the model should be able to adapt to the new conditions by means of a recursive parameter estimation technique. Such an adaptive strategy will ensure that the most accurate model is used for the current conditions, and thus the most accurate model predictions are used in model-based control calculations.

METHODS

In a retrospective analysis, empirical dynamic autoregressive exogenous input (ARX) models were identified from glucose-insulin data for nine type 1 diabetes subjects in ambulatory conditions. Data sets consisted of continuous (5-minute) glucose concentration measurements obtained from a continuous glucose monitor, basal insulin infusion rates and times and amounts of insulin boluses obtained from the subjects' insulin pumps, and subject-reported estimates of the times and carbohydrate content of meals. Two identification techniques were investigated: nonrecursive, or batch methods, and recursive methods. Batch models were identified from a set of training data, whereas recursively identified models were updated at each sampling instant. Both types of models were used to make predictions of new test data. For the purpose of comparison, model predictions were compared to zero-order hold (ZOH) predictions, which were made by simply holding the current glucose value constant for p steps into the future, where p is the prediction horizon. Thus, the ZOH predictions are model free and provide a base case for the prediction metrics used to quantify the accuracy of the model predictions. In theory, recursive identification techniques are needed only when there are changing conditions in the subject that require model adaptation. Thus, the identification and validation techniques were performed with both "normal" data and data collected during conditions of reduced insulin sensitivity. The latter were achieved by having the subjects self-administer a medication, prednisone, for 3 consecutive days. The recursive models were allowed to adapt to this condition of reduced insulin sensitivity, while the batch models were only identified from normal data.

RESULTS

Data from nine type 1 diabetes subjects in ambulatory conditions were analyzed; six of these subjects also participated in the prednisone portion of the study. For normal test data, the batch ARX models produced 30-, 45-, and 60-minute-ahead predictions that had average root mean square error (RMSE) values of 26, 34, and 40 mg/dl, respectively. For test data characterized by reduced insulin sensitivity, the batch ARX models produced 30-, 60-, and 90-minute-ahead predictions with average RMSE values of 27, 46, and 59 mg/dl, respectively; the recursive ARX models demonstrated similar performance with corresponding values of 27, 45, and 61 mg/dl, respectively. The identified ARX models (batch and recursive) produced more accurate predictions than the model-free ZOH predictions, but only marginally. For test data characterized by reduced insulin sensitivity, RMSE values for the predictions of the batch ARX models were 9, 5, and 5% more accurate than the ZOH predictions for prediction horizons of 30, 60, and 90 minutes, respectively. In terms of RMSE values, the 30-, 60-, and 90-minute predictions of the recursive models were more accurate than the ZOH predictions, by 10, 5, and 2%, respectively.

CONCLUSION

In this experimental study, the recursively identified ARX models resulted in predictions of test data that were similar, but not superior, to the batch models. Even for the test data characteristic of reduced insulin sensitivity, the batch and recursive models demonstrated similar prediction accuracy. The predictions of the identified ARX models were only marginally more accurate than the model-free ZOH predictions. Given the simplicity of the ARX models and the computational ease with which they are identified, however, even modest improvements may justify the use of these models in a model-based controller for an artificial beta cell.