Assessing the Utility, Impact, and Adoption Challenges of an Artificial Intelligence-Enabled Prescription Advisory Tool for Type 2 Diabetes Management: Qualitative Study.

BACKGROUND: The clinical management of type 2 diabetes mellitus (T2DM) presents a significant challenge due to the constantly evolving clinical practice guidelines and growing array of drug classes available. Evidence suggests that artificial intelligence (AI)-enabled clinical decision support systems (CDSSs) have proven to be effective in assisting clinicians with informed decision-making. Despite the merits of AI-driven CDSSs, a significant research gap exists concerning the early-stage implementation and adoption of AI-enabled CDSSs in T2DM management. OBJECTIVE: This study aimed to explore the perspectives of clinicians on the use and impact of the AI-enabled Prescription Advisory (APA) tool, developed using a multi-institution diabetes registry and implemented in specialist endocrinology clinics, and the challenges to its adoption and application. METHODS: We conducted focus group discussions using a semistructured interview guide with purposively selected endocrinologists from a tertiary hospital. The focus group discussions were audio-recorded and transcribed verbatim. Data were thematically analyzed. RESULTS: A total of 13 clinicians participated in 4 focus group discussions. Our findings suggest that the APA tool offered several useful features to assist clinicians in effectively managing T2DM. Specifically, clinicians viewed the AI-generated medication alterations as a good knowledge resource in supporting the clinician's decision-making on drug modifications at the point of care, particularly for patients with comorbidities. The complication risk prediction was seen as positively impacting patient care by facilitating early doctor-patient communication and initiating prompt clinical responses. However, the interpretability of the risk scores, concerns about overreliance and automation bias, and issues surrounding accountability and liability hindered the adoption of the APA tool in clinical practice. CONCLUSIONS: Although the APA tool holds great potential as a valuable resource for improving patient care, further efforts are required to address clinicians' concerns and improve the tool's acceptance and applicability in relevant contexts.

A real-life comparison of hypoglycaemia symptomatology between insulin-treated type 2 diabetes participants with and without impaired awareness of hypoglycaemia.

AIMS: We aim to compare and correlate Gold and Clarke questionnaire scores with hypoglycaemic symptomatic responses between insulin-treated type 2 diabetes participants with and without IAH in a real-life study. METHODS: Insulin-treated type 2 diabetes participants attending an outpatient diabetes clinic in Singapore were asked to complete the Gold and Clarke questionnaires, record capillary blood glucose (CBG) and hypoglycaemic symptoms for 4 weeks. RESULTS: Data were collected from 153 participants (M:F = 98:55) with mean age 61.0 ± 9.4 years, duration of diabetes 19.5 ± 8.8 years and HbA1c 68 ± 17 mmol/mol (8.4 ± 1.5%). Gold and Clarke methods classified 19.6% and 26.8% of participants with IAH, respectively. Using CBG threshold of <3 mmol/L, significantly greater proportion of participants with intact awareness were experiencing autonomic symptoms than those with IAH with either method (Gold: 69% vs. 18%, p = 0.006; Clarke: 85% vs. 46%, p = 0.010). Significantly greater proportion of participants with IAH experienced no hypoglycaemia symptoms than those with intact awareness (Gold: 3.4% vs. 36%, p = 0.015; Clarke: 3.7% vs. 31%, p = 0.031). Participants with IAH had significantly higher rates of severe hypoglycaemia in the preceding year compared to those without (Gold: 17% vs. 3.3%; Clarke: 15% vs. 2.7%, p = 0.012). CONCLUSIONS: Gold and Clarke questionnaires are appropriate tools in ascertaining IAH status in insulin-treated type 2 diabetes participants. This is the first time whereby the hypoglycaemia symptomology has robustly validated the Gold and Clarke questionnaire in insulin-treated type 2 diabetes participants.

Telemonitoring With a Connected Glucose Meter Improves Glycemia Among People With Insulin-Treated Type 2 Diabetes.

BACKGROUND: Delayed initiation and inadequate titration remain critical challenges to optimizing insulin therapy in type 2 diabetes (T2D). We aimed to study whether hemoglobin A1c (HbA1c) can be lowered in people with insulin-treated T2D using telemonitoring. METHODS: This single-center study recruited adults with greater than or equal to six months of diabetes, greater than or equal to three months of insulin therapy, HbA1c ≥8.5% and ≤12.5%, and body mass index (BMI) ≤40 kg/m2. All participants received a connected glucose meter and the accompanying smartphone application. Participants sent weekly blood glucose (BG) diary to their primary endocrinologist via email. Adjustments in insulin doses were communicated to the participants. HbA1c, proportion of BG readings in range (70-180 mg/dL, PIR), below range (<70 mg/dL, PBR) and above range (>180 mg/dL, PAR), and glycemic variability as the coefficient of variation (% CV) were measured at baseline, week 12, and week 24 and compared using repeated-measures analysis of variance (ANOVA) or Friedman's ANOVA. RESULTS: We recruited 40 people (55% women). Mean age was 57.9 years, BMI 27.8 kg/m2, and baseline HbA1c 9.8% (83.7 mmol/mol). Mean HbA1c improved by 1.7%, % CV reduced from 32.9% to 30.7%, PIR increased from 58.8% to 67.1% (all P <.01) by week 24, without any change in PBR. This was achieved with a 0.04 U/kg/d median increase in total daily dose of insulin and 0.9 kg weight gain over 24 weeks. CONCLUSION: Telemonitoring and titration of insulin using a connected glucose meter resulted in significant improvements in glycemia, characterized by a reduction in HbA1c, increase in PIR, and reduction in glycemic variability without any increase in hypoglycemia.

Evaluation of Care Outcomes of Patients Receiving Hyperkalemia Treatment With Insulin in Acute Care Tertiary Hospital Emergency Department.

INTRODUCTION: Treatment of hyperkalemia using intravenous insulin can result in severe hypoglycemia, but regular blood glucose monitoring is not standardized. This study aimed to (i) explore the demographics of adult patients receiving hyperkalemia treatment and (ii) identify the incidence rate of hypoglycemia and associated demographic or clinical characteristics. METHODS: A descriptive design with prospective data collection was used. This study recruited 135 patients who received hyperkalemia treatment in the emergency department. Structured blood glucose monitoring was conducted at 1, 2, 4, and 6 hours after receiving intravenous insulin. Univariate analyses of association between demographic and clinical variables and hypoglycemia outcome were performed. RESULTS: There were 31 hypoglycemic events, with 11.9%, 7.4%, 2.2%, and 1.5% occurring at the 1, 2, 4, and 6 hours after treatment. The logit regression showed no significantly increased risk of hypoglycemia in terms of the demographic and clinical variables. DISCUSSION: The variation in blood glucose response observed in this study combined with the high incidences of hypolycaemia indicated the need for frequent and longer duration of monitoring for patients who were being treated for hyperkalaemia with IDT.

High incidence of undetected low sensor glucose events among elderly patients with type 2 diabetes more than a decade on after the ACCORD study.

OBJECTIVE: Hypoglycaemia leads to significant morbidity and impacts negatively on quality-of-life, especially in elderly people with increased frailty. The aims of this study were to determine the prevalence of low interstitial fluid glucose (IFG) in patients with tightly controlled type 2 diabetes (T2D), and to evaluate whether there were differences in burden of low IFG between sulphonylurea and insulin treated groups. METHODS: A Freestyle Libre-Pro sensor was used for sampling of the IFG continuously. Patients were blinded to the IFG levels. The sensor was returned to the investigators after a 2-week period and the data were downloaded for analysis. RESULTS: There was a total of 69 patients (median age 72 years (IQR = 69-74)) - 40 were sulfonylurea-treated and 29 insulin-treated. In total, 781 low sensor glucose events (<4.0 mmol/L) were detected, of which 254 were very low sensor glucose events (<2.8 mmol/L). Twenty-six out of 29 insulin-treated (89.6%) and 36 out of 40 sulphonylurea-treated patients (90%) contributed to the 781 events of low sensor glucose. Twenty out of 29 insulin-treated (69%) and 26 out of 40 sulphonylurea-treated patients (65.0%) contributed to the 254 very low sensor glucose events. Only 9% of all events were identified by patients. Nocturnal events represented 55.8% of low sensor glucose events and 61.0% of very low sensor glucose events. At a cut-off of <2.8 mmol/L, it was found that the insulin group had a significantly greater number of such events as compared to the sulfonylurea group. CONCLUSIONS: This study demonstrates that elderly patients with tightly-controlled T2D have a significant number of low sensor glucose events which go by undetected.

Factors associated with prolonged length of stay in patients admitted with severe hypoglycaemia to a tertiary care hospital.

Severe hypoglycaemia carries considerable morbidity and potential mortality. We aim to elucidate the factors which were associated with a prolonged length of stay (LOS) among patients with diabetes who were admitted to the hospital with severe hypoglycaemia. Three hundred and four patients were included in the analysis, with a mean age of 70.6 ± 11.3 years, mean glycated haemoglobin of 6.9 ± 1.3% and median LOS of 3 days. Patients with a LOS >3 days had significantly higher Charlson Comorbidity Index (CCI) (4.9 ± 2.1 vs 4.1 ± 2.1, P < 0.01), a lower glomerular filtration rate (GFR) (34.6 ± 31.4 mL/min vs 44.8 ± 28.9 mL/min, P = 0.01) and a higher proportion of these patients suffered from recurrent hypoglycaemia during the admission (38.9% vs 27.7%, P = 0.04). In addition, they had higher white cell counts (11.1 ± 4.8 × 109/L vs 9.3 ± 3.2 × 109/L) and lower albumin concentrations (32.9 ± 6.6 g/L vs 36.8 ± 4.9 g/L). Bivariate analysis showed that the same factors were associated with prolonged LOS. Identification of risk factors associated with prolonged LOS provides the opportunity for intervention to reduce the LOS and improve the outcomes for these patients.

High one-year mortality following hospitalization for severe hypoglycemia among patients with diabetes mellitus: findings of a retrospective cohort study at an acute tertiary care hospital in Singapore.

OBJECTIVES: Little is known about the 1-year short-term mortality rate following hospital admissions with severe hypoglycemia. This study aimed to determine the factors associated with increased 1-year mortality rate following hospitalization in diabetes patients admitted with severe hypoglycemia to the Singapore General Hospital. METHODS: Clinical, biochemical, and 1-year mortality data from diabetes patients who were admitted with severe hypoglycemia in the year 2014 were extracted from institutional medical records. Patients who passed away during the episode of admissions with severe hypoglycemia were excluded from the analysis. The clinical and biochemical factors between patients who survived and those who did not survive within 1 year following admission were compared using logistic regression analysis. RESULTS: Three hundred and four patients (181 female and 123 male) were admitted with severe hypoglycemia in 2014, and the mean capillary blood glucose on admission was 2.3 ± 0.7 mmol/L. Sixty-three (20.7%) patients died within 1-year post-discharge from the hospital. Compared with patients who survived 1-year post-discharge from the hospital, non-survivors were older (69.3 ± 11.0 vs 75.5 ± 11.2 years, p < .001), had longer lengths of stay (LOS) (5.0 ± 7.4 vs 9.0 ± 12.8 days, p = .02), and had a higher Charlson Comorbidity Index (CCI) (4.1 ± 1.9 vs 5.9 ± 2.4, p < .001). Factors associated with increased 1-year mortality risk were age (odds ratio [OR] = 1.06; 95% confidence interval [CI] = 1.03-1.09, p < .01), LOS in hospital (OR = 1.01; 95% CI = 1.01-1.08, p < .01), and CCI (OR = 1.51; 95% CI = 1.31-1.75, p < .01), respectively. CONCLUSIONS: Older diabetes patients with more comorbidities and longer LOS were at increased risk of dying within a year of discharge after hospitalization with severe hypoglycemia. Admission with severe hypoglycemia has important prognostic implications. Healthcare professionals should address hypoglycemia and other health issues during the hospital admissions.

The impact of hypoglycaemia awareness status on regional brain responses to acute hypoglycaemia in men with type 1 diabetes.

AIMS/HYPOTHESIS: Impaired awareness of hypoglycaemia (IAH) in type 1 diabetes increases the risk of severe hypoglycaemia sixfold and can be resistant to intervention. We explored the impact of IAH on central responses to hypoglycaemia to investigate the mechanisms underlying barriers to therapeutic intervention. METHODS: We conducted [15O]water positron emission tomography studies of regional brain perfusion during euglycaemia (target 5 mmol/l), hypoglycaemia (achieved level, 2.4 mmol/l) and recovery (target 5 mmol/l) in 17 men with type 1 diabetes: eight with IAH, and nine with intact hypoglycaemia awareness (HA). RESULTS: Hypoglycaemia with HA was associated with increased activation in brain regions including the thalamus, insula, globus pallidus (GP), anterior cingulate cortex (ACC), orbital cortex, dorsolateral frontal (DLF) cortex, angular gyrus and amygdala; deactivation occurred in the temporal and parahippocampal regions. IAH was associated with reduced catecholamine and symptom responses to hypoglycaemia vs HA (incremental AUC: autonomic scores, 26.2 ± 35.5 vs 422.7 ± 237.1; neuroglycopenic scores, 34.8 ± 88.8 vs 478.9 ± 311.1; both p < 0.002). There were subtle differences (p < 0.005, k ≥ 50 voxels) in brain activation at hypoglycaemia, including early differences in the right central operculum, bilateral medial orbital (MO) cortex, and left posterior DLF cortex, with additional differences in the ACC, right GP and post- and pre-central gyri in established hypoglycaemia, and lack of deactivation in temporal regions in established hypoglycaemia. CONCLUSIONS/INTERPRETATION: Differences in activation in the post- and pre-central gyri may be expected in people with reduced subjective responses to hypoglycaemia. Alterations in the activity of regions involved in the drive to eat (operculum), emotional salience (MO cortex), aversion (GP) and recall (temporal) suggest differences in the perceived importance and urgency of responses to hypoglycaemia in IAH compared with HA, which may be key to the persistence of the condition.

Inpatient glucose management programme in the Asian healthcare setting.

AIM: To implement an inpatient glucose management (IGM) programme in the general medical wards and evaluate its clinical efficacy. METHOD: Consecutive patients admitted to selected medical wards over a 12-month period were included in the IGM programme. All patients with ≥3 capillary blood glucose (CBG) readings (>10.0 mmol/L and/or <4.0 mmol/L) over a 24-hour period were identified electronically. A multidisciplinary team of diabetes care providers would provide specialist input to these patients. The control group comprised of consecutive patients admitted over the preceding 12 months. Outcome glucose measures include mean in-hospital glucose (MHG), mean patient-day glucose (MDG), proportion of CBG readings at predefined cut-offs and length of stay (LOS). RESULTS: Both the MHG and MDG were significantly lower following intervention (10.0±2.4 mmol/L vs 11.2±2.6 mmol/L, P<.001; 10.0±2.3 mmol/L vs 11.2±2.6 mmol/L, P<.001, respectively). Prevalence of hyperglycaemic events, defined by CBG >10.0 mmol/L, was significantly lower at 36.5% versus 51.6% (P<.001). Hypoglycaemic events of CBG <4.0 mmol/L remained infrequent at <1.0% before and after IGM programme. A greater proportion of glucose readings was controlled within the target range of 4.0-10.0 mmol/L (62.6% vs 47.6%, P<.001). With the IGM programme in place, more patients received scheduled CBG monitoring, and a significant shortening of mean LOS by 3.2 days was observed (P=.02). CONCLUSIONS: The IGM programme was effective in improving inpatient glycaemic monitoring and control in the general medical wards, with a significant reduction in LOS observed. These demonstrated the programme's potential to enhance quality and efficiency of patient care.

A study to evaluate the prevalence of impaired awareness of hypoglycaemia in adults with type 2 diabetes in outpatient clinic in a tertiary care centre in Singapore.

BACKGROUND: Impaired awareness of hypoglycaemia (IAH) predisposes affected patients to severe hypoglycaemia. There are few data on prevalence of IAH in adults with insulin-treated type 2 diabetes in Asia. We aim to ascertain the prevalence of IAH among insulin-treated patients with type 2 diabetes in an outpatient clinic in a tertiary care centre in Singapore. METHODS: A total of 374 patients with insulin-treated type 2 diabetes attending the outpatient diabetes clinic in a tertiary referral centre in Singapore were recruited over a 4-month period. Participants completed a questionnaire to document baseline characteristics and assess their hypoglycaemia awareness status, using a combination of the Clarke, Gold and Pedersen-Bjergaard methods. RESULTS: Using the Clarke, Gold and Pedersen-Bjergaard methods, prevalence of IAH in our cohort was 9.6%, 13.4% and 33.2% respectively. Overall, 7.2% of participants suffered from severe hypoglycaemia in the preceding year. The IAH group had more episodes of severe hypoglycaemia across all three methods, compared with the normal awareness group (p < 0.01). There were no significant differences in mean HbA1c, duration of diabetes and insulin treatment between the IAH and normal awareness groups. CONCLUSIONS: IAH is prevalent in adults with insulin-treated type 2 diabetes in Asia, and is associated with significantly increased risk of severe hypoglycaemia.

Description and preliminary results from a structured specialist behavioural weight management group intervention: Specialist Lifestyle Management (SLiM) programme.

BACKGROUND: Specialist Lifestyle Management (SLiM) is a structured patient education and self-management group weight management programme. Each session is run monthly over a 6-month period providing a less intensive long-term approach. The groups are patient-centred incorporating educational, motivational, behavioural and cognitive elements. The theoretical background, programme structure and preliminary results of SLiM are presented. SUBJECTS/METHODS: The study was a pragmatic service evaluation of obese patients with a body mass index (BMI) ≥35 kg/m(2) with comorbidity or ≥40 kg/m(2) without comorbidity referred to a specialist weight management service in the West Midlands, UK. 828 patients were enrolled within SLiM over a 48-month period. Trained facilitators delivered the programme. Preliminary anonymised data were analysed using the intention-to-treat principle. The primary outcome measure was weight loss at 3 and 6 months with comparisons between completers and non-completers performed. The last observation carried forward was used for missing data. RESULTS: Of the 828 enrolled within SLiM, 464 completed the programme (56%). The mean baseline weight was 135 kg (BMI=49.1 kg/m(2)) with 87.2% of patients having a BMI≥40 kg/m(2) and 12.4% with BMI≥60 kg/m(2). The mean weight change of all patients enrolled was -4.1 kg (95% CI -3.6 to -4.6 kg, p=0.0001) at the end of SLiM, with completers (n=464) achieving -5.5 kg (95% CI -4.2 to -6.2 kg, p=0.0001) and non-completers achieving -2.3 kg (p=0.0001). The majority (78.6%) who attended the 6-month programme achieved weight loss with 32.3% achieving a ≥5% weight loss. CONCLUSIONS: The SLiM programme is an effective group intervention for the management of severe and complex obesity.

Evolution and resolution of human brain perfusion responses to the stress of induced hypoglycemia.

The relationship between the human brain response to acute stress and subjective, behavioural and physiological responses is poorly understood. We have examined the human cerebral response to the intense interoceptive stressor of hypoglycemia, controlling plasma glucose at either normal fasting concentrations (5 mmol/l, n=7) or at hypoglycemia (2.7 mmol/l, n=10) for 1 h in healthy volunteers. Hypoglycemia was associated with symptomatic responses, counterregulatory neuroendocrine responses and a sequential pattern of brain regional engagement, mapped as changes in relative cerebral perfusion using [(15)O]-H(2)O water positron emission tomography. The early cerebral response comprised activation bilaterally in anterior cingulate cortex (ACC) and thalamic pulvinar, with deactivation in posterior parahippocampal gyrus. Later responses (>20 min) engaged bilateral anterior insula, ventral striatum and pituitary. Following resolution of hypoglycemia, the majority of responses returned to baseline, save persistent engagement of the ACC and sustained elevation of growth hormone and cortisol. Catecholamine responses correlated with increased perfusion in pulvinar and medial thalamus, ACC and pituitary, while growth hormone and cortisol responses showed no correlation with thalamic activation but did show additional correlation with the hypothalamus and ventral striatum bilaterally. These data demonstrate complex dynamic responses to the stressor of hypoglycemia that would be expected to drive physiological and behavioural changes to remedy the state. Further, these data show that sustained stress and its aftermath engage distinct sets of brain regions, providing a neural substrate for adaptive or 'allostasic' responses to stressors.