Establishment of the SingHealth Diabetes Registry

A digital twin model incorporating generalized metabolic fluxes to identify and predict chronic kidney disease in type 2 diabetes mellitus

We have developed a digital twin-based CKD identification and prediction model that leverages generalized metabolic fluxes (GMF) for patients with Type 2 Diabetes Mellitus (T2DM). GMF digital twins utilized basic clinical and physiological biomarkers as inputs for identification and prediction of CKD. We employed four diverse multi-ethnic cohorts (n = 7072): a Singaporean cohort (EVAS, n = 289) and a North American cohort (NHANES, n = 1044) for baseline CKD identification, and two multi-center Singaporean cohorts (CDMD, n = 2119 and SDR, n = 3627) for 3-year CKD prediction and risk stratification. We subsequently conducted a comprehensive study utilizing a single dataset to evaluate the clinical utility of GMF for CKD prediction. The GMF-based identification model performed strongly, achieving an AUC between 0.80 and 0.82. In prediction, the GMF generated with complete parameters attained high performance with an AUC of 0.86, while with incomplete parameters, it achieved an AUC of 0.75. The GMF-based prediction model utilizing complete inputs is the standard implementation of our algorithm: HealthVector Diabetes®. We have established the GMF digital twin-based model as a robust clinical tool capable of predicting and stratifying the risk of future CKD within a 3-year time horizon. We report the correlation of GMF with basic input parameters, their ability to differentiate between future health states and medication status at baseline, and their capability to quantify CKD progression rates. This holistic methodology provides insights into patients' health states and CKD progression rates based on GMF metabolic profile differences, enabling personalized care plans.

Early repeat hospitalization for fluid overload in individuals with cardiovascular disease and risks: a retrospective cohort study

AIMS

Fluid overload is a common manifestation of cardiovascular and kidney disease and a leading cause of hospitalizations. To identify patients at risk of recurrent severe fluid overload, we evaluated the incidence and risk factors associated with early repeat hospitalization for fluid overload among individuals with cardiovascular disease and risks.

METHODS

Single-center retrospective cohort study of 3423 consecutive adults with an index hospitalization for fluid overload between January 2015 and December 2017 and had cardiovascular risks (older age, diabetes mellitus, hypertension, dyslipidemia, kidney disease, known cardiovascular disease), but excluded if lost to follow-up or eGFR < 15 ml/min/1.73 m2. The outcome was early repeat hospitalization for fluid overload within 30 days of discharge.

RESULTS

The mean age was 73.9 ± 11.6 years and eGFR was 54.1 ± 24.6 ml/min/1.73 m2 at index hospitalization. Early repeat hospitalization for fluid overload occurred in 291 patients (8.5%). After adjusting for demographics, comorbidities, clinical parameters during index hospitalization and medications at discharge, cardiovascular disease (adjusted odds ratio, OR 1.66, 95% CI 1.27-2.17), prior hospitalization for fluid overload within 3 months (OR 2.52, 95% CI 1.17-5.44), prior hospitalization for any cause in within 6 months (OR 1.33, 95% CI 1.02-1.73) and intravenous furosemide use (OR 1.58, 95% CI 1.10-2.28) were associated with early repeat hospitalization for fluid overload. Higher systolic BP on admission (OR 0.992, 95% 0.986-0.998) and diuretic at discharge (OR 0.50, 95% CI 0.26-0.98) reduced early hospitalization for fluid overload.

CONCLUSION

Patients at-risk of early repeat hospitalization for fluid overload may be identified using these risk factors for targeted interventions.

Heterogeneity by age and gender in the association of kidney function with mortality among patients with diabetes - analysis of diabetes registry in Singapore

BACKGROUND

We aimed to explore the three-way interaction among age, gender, and kidney function on the risk of all-cause mortality and cardiovascular mortality among patients with type 2 diabetes (T2D).

METHODS

In a retrospective cohort study, patients aged > 40 years with T2D with serum creatinine and urine albumin measured from 2013 to 2019 were included from a multi-institutional diabetes registry. The exposure was estimated glomerular filtration rate (eGFR), outcomes were all-cause mortality (primary outcome) and cardiovascular disease (CVD) mortality (secondary outcome). We applied multivariable cox proportional hazards regression analysis to compute the association between eGFR and mortality.

RESULTS

A total of 36,556 patients were followed for up to 6 years during which 2492 (6.82%) died from all causes, and 690 (1.9%) died from CVD. We observed a significant three-way interaction (p = 0.021) among age (younger, < 65; older, ≥65 years), gender and eGFR for the risk of all-cause mortality. Using age- and gender-specific eGFR of 90 ml/min/1.73m2 as the reference point, the adjusted hazard rate (HR) (95% CI) for all-cause mortality at eGFR of 40 ml/min/1.73m2 was 3.70 (2.29 to 5.99) in younger women and 1.86 (1.08 to 3.19) in younger men. The corresponding adjusted HRs in older women and older men were 2.38 (2.02 to 2.82) and 2.18 (1.85 to 2.57), respectively. Similar results were observed for CVD deaths, although the three-way interaction was not statistically significant. Sensitivity analysis yielded similar results.

CONCLUSIONS

In this T2D population, younger women with reduced kidney function might be more susceptible to higher risks of all-cause mortality and CVD mortality than younger men.

Kidney function trajectories, associated factors, and outcomes in multiethnic Asian patients with type 2 diabetes

BACKGROUND

We examined the trajectory of estimated glomerular filtrate rate (eGFR), associated risk factors, and its relationship with end-stage kidney disease (ESKD) among a multiethnic patient population with type 2 diabetes in Singapore.

METHODS

A follow-up study included 62 080 individuals with type 2 diabetes aged ≥18 years in a multi-institutional SingHealth Diabetes Registry between 2013 and 2019. eGFR trajectories were analyzed using latent class linear mixed models. Factors associated with eGFR trajectories were evaluated using multinomial logistic regression. The association of eGFR trajectories with ESKD was assessed via competing risk models.

RESULTS

Trajectory of kidney function, determined by eGFR, was nonlinear. The trajectory pattern was classified as stable initially then gradual decline (75%), progressive decline (21.9%), and rapid decline (3.1%). Younger age, female sex, Malay ethnicity, lower-income housing type, current smoking, higher glycated hemoglobin, lower low-density lipoprotein, higher triglyceride, uncontrolled blood pressure, albuminuria, cardiovascular disease, hypertension, and higher eGFR levels each were associated with progressive or rapid decline. Compared with the trajectory of stable initially then gradual eGFR decline, progressive decline increased the hazard of ESKD by 6.14-fold (95% confidence interval [CI]: 4.96-7.61)) and rapid decline by 82.55 folds (95% CI: 55.90-121.89).

CONCLUSIONS

Three nonlinear trajectory classes of kidney function were identified among multiethnic individuals with type 2 diabetes in Singapore. About one in four individuals had a progressive or rapid decline in eGFR. Our results suggest that eGFR trajectories are correlated with multiple social and modifiable risk factors and inform the risk of ESKD.

Real-World Systolic and Diastolic Blood Pressure Levels and Cardiovascular Mortality in Patients With Type 2 Diabetes-Results From a Large Registry Cohort in Asia

BACKGROUND

Elevated blood pressure (BP) is associated with increased risk of cardiovascular mortality. However, there is ongoing debate whether intensive BP lowering may paradoxically increase the risk of cardiovascular disease (CVD), especially in patients with type 2 diabetes (T2D). We investigated the association of BP with risk of CVD mortality in patients with T2D.

METHODS AND RESULTS

We used data on 83 721 patients with T2D from a multi-institutional diabetes registry in Singapore from 2013 to 2019. BP was analyzed as categories and restricted cubic splines using Cox multivariable regression analysis stratified by preexisting CVD and age (<65 years versus ≥65 years). The primary outcome was CVD mortality, determined via linkage with the national registry. Among 83 721 patients with T2D (mean age 65.3 years, 50.6% women, 78.9% taking antihypertensive medications), 7.6 per 1000 person-years experienced the primary outcome. Systolic BP had a graded relationship with a significant increase in CVD mortality at levels >120 to 129 mm Hg. Diastolic BP levels >90 mm Hg were significantly associated with CVD mortality in those aged ≥65 years. In addition, diastolic BP <70 mm Hg was associated with a significantly higher risk of CVD mortality in all patients.

CONCLUSIONS

In patients with T2D, clinic systolic BP levels ≥130 mm Hg or diastolic BP levels ≥90 mm Hg are associated with higher risk of CVD mortality. Diastolic BP <70 mm Hg is also associated with the risk of adverse CVD outcomes, although reverse causality cannot be ruled out.

Comparison of existing methods of low-density lipoprotein cholesterol estimation in patients with type 2 diabetes mellitus

Introduction

Elevated low-density lipoprotein cholesterol (LDL-C) is an important risk factor for atherosclerotic cardiovascular disease (ASCVD). Direct LDL-C measurement is not widely performed. LDL-C is routinely calculated using the Friedewald equation (FLDL), which is inaccurate at high triglyceride (TG) or low LDL-C levels. We aimed to compare this routine method with other estimation methods in patients with type 2 diabetes mellitus (T2DM), who typically have elevated TG levels and ASCVD risk.

Method

We performed a retrospective cohort study on T2DM patients from a multi-institutional diabetes registry in Singapore from 2013 to 2020. LDL-C values estimated by the equations: FLDL, Martin/Hopkins (MLDL) and Sampson (SLDL) were compared using measures of agreement and correlation. Subgroup analysis comparing estimated LDL-C with directly measured LDL-C (DLDL) was conducted in patients from a single institution. Estimated LDL-C was considered discordant if LDL-C was <1.8mmol/L for the index equation and ≥1.8mmol/L for the comparator.

Results

A total of 154,877 patients were included in the final analysis, and 11,475 patients in the subgroup analysis. All 3 equations demonstrated strong overall correlation and goodness-of-fit. Discordance was 4.21% for FLDL-SLDL and 6.55% for FLDL-MLDL. In the subgroup analysis, discordance was 21.57% for DLDL-FLDL, 17.31% for DLDL-SLDL and 14.44% for DLDL-MLDL. All discordance rates increased at TG levels >4.5mmol/L.

Conclusion

We demonstrated strong correlations between newer methods of LDL-C estimation, FLDL, and DLDL. At higher TG concentrations, no equation performed well. The Martin/Hopkins equation had the least discordance with DLDL, and may minimise misclassification compared with the FLDL and SLDL.

Associations Between Mean HbA1c, HbA1c Variability, and Both Mortality and Macrovascular Complications in Patients with Diabetes Mellitus: A Registry-Based Cohort Study

Background

We investigate the association between mean HbA1c, HbA1c variability, and all-cause mortality and diabetes-related macrovascular complications in patients with diabetes.

Methods

We performed a retrospective cohort study using patients present in the Singapore Health Services diabetes registry (SDR) during 2013 to 2014. We assessed mean HbA1c using three models: a baseline mean HbA1c for 2013-14, the mean across the whole follow-up period, and a time-varying yearly updated mean. We assessed HbA1c variability at baseline using the patient's HbA1c variability score (HVS) for 2013-14. The association between mean HbA1c, HVS, and 6 outcomes were assessed using Cox proportional hazard models.

Results

We included 43,837-53,934 individuals in the analysis; 99.3% had type 2 diabetes mellitus. The data showed a J-shaped distribution in adjusted hazard ratios (HRs) for all-cause mortality, ischemic heart disease, acute myocardial infarction, peripheral arterial disease, and ischemic stroke, with an increased risk of developing these outcomes at HbA1c <6% (42 mmol/mol) and ≥8% (64 mmol/mol). With the addition of HVS, the J-shaped distribution was maintained for the above outcomes, but HRs were greater at HbA1c <6.0% (42 mmol/mol) and reduced at HbA1c ≥8.0% (64 mmol/mol) when compared to models without HVS. The risk for all outcomes increased substantially with increasing glycaemic variability.

Conclusion

Both low (<6.0% [42 mmol/mol]) and high (≥8.0% [64 mmol/mol]) levels of glycaemic control are associated with increased all-cause mortality and diabetes-related macrovascular complications. Glycaemic variability is independently associated with increased risk for these outcomes. Therefore, patients with stable glycaemic level of 6-8% (42-64mmol/mol) are at lowest risk of all-cause mortality and diabetes-related macrovascular complications.

Longitudinal HbA1c trajectory modelling reveals the association of HbA1c and risk of hospitalization for heart failure for patients with type 2 diabetes mellitus

BACKGROUND

Inconsistent conclusions in past studies on the association between poor glycaemic control and the risk of hospitalization for heart failure (HHF) have been reported largely due to the analysis of non-trajectory-based HbA1c values. Trajectory analysis can incorporate the effects of HbA1c variability across time, which may better elucidate its association with macrovascular complications. Furthermore, studies analysing the relationship between HbA1c trajectories from diabetes diagnosis and the occurrence of HHF are scarce.

METHODS

This is a prospective cohort study of the SingHealth Diabetes Registry (SDR). 17,389 patients diagnosed with type 2 diabetes mellitus (T2DM) from 2013 to 2016 with clinical records extending to the end of 2019 were included in the latent class growth analysis to extract longitudinal HbA1c trajectories. Association between HbA1c trajectories and risk of first known HHF is quantified with the Cox Proportional Hazards (PH) model.

RESULTS

5 distinct HbA1c trajectories were identified as 1. low stable (36.1%), 2. elevated stable (40.4%), 3. high decreasing (3.5%), 4. high with a sharp decline (10.8%), and 5. moderate decreasing (9.2%) over the study period of 7 years. Poorly controlled HbA1c trajectories (Classes 3, 4, and 5) are associated with a higher risk of HHF. Using the diabetes diagnosis time instead of a commonly used pre-defined study start time or time from recruitment has an impact on HbA1c clustering results.

CONCLUSIONS

Findings suggest that tracking the evolution of HbA1c with time has its importance in assessing the HHF risk of T2DM patients, and T2DM diagnosis time as a baseline is strongly recommended in HbA1c trajectory modelling. To the authors' knowledge, this is the first study to identify an association between HbA1c trajectories and HHF occurrence from diabetes diagnosis time.

Non-alcoholic fatty liver disease screening in type 2 diabetes mellitus: A cost-effectiveness and price threshold analysis

Introduction: The cost-effectiveness of screening asymptomatic non-alcoholic fatty liver disease (NAFLD) patients remains debatable, with current studies assuming lifelong benefits of NAFLD screening while neglecting cardiovascular outcomes. This study aims to assess the cost-effectiveness of NAFLD screening among type 2 diabetes mellitus (T2DM) patients, and to establish a price threshold for NAFLD treatment, when it becomes available. Method: A Markov model was constructed comparing 4 screening strategies (versus no screening) to identify NAFLD with advanced fibrosis among T2DM patients: fibrosis-4 (FIB-4), vibration-controlled transient elastography (VCTE), FIB-4 and VCTE (simultaneous), and FIB-4 and VCTE (sequential). Sensitivity analyses and price threshold analyses were performed to assess parameter uncertainties in the results. Results: VCTE was the most cost-effective NAFLD screening strategy (USD24,727/quality-adjusted life year [QALY]), followed by FIB-4 (USD36,800/QALY), when compared to no screening. Probabilistic sensitivity analysis revealed a higher degree of certainty for VCTE as a cost-effective strategy compared to FIB-4 (90.7% versus 73.2%). The duration of expected screening benefit is the most influential variable based on incremental cost-effectiveness ratio tornado analysis. The minimum duration of screening benefit for NAFLD screening to be cost-effective was at least 2.6 years. The annual cost of NAFLD treatment should be less than USD751 for NAFLD screening to be cost-effective. Conclusion: Both VCTE and FIB-4 are cost-effective NAFLD screening strategies among T2DM patients in Singapore. However, given the lack of access to VCTE at primacy care and potential budget constraints, FIB-4 can also be considered for NAFLD screening among T2DM patients in Singapore. Keywords: Cost-effectiveness analysis, fatty liver, screening, liver fibrosis, population health

Trends in diabetes-related complications in Singapore, 2013-2020: A registry-based study

Background

Diabetes mellitus (DM) is a growing global health problem. In Singapore, the prevalence of Type 2 DM is rising, but comprehensive information about trends in DM-related complications is lacking.

Objectives

We utilized the Singapore Health Services (SingHealth) diabetes registry (SDR) to assess trends in DM micro and macro-vascular complications at the population level, explore factors influencing these trends.

Methods

We studied trends for ten DM-related complications: ischemic heart disease (IHD), acute myocardial infarction (AMI), peripheral arterial disease (PAD) and strokes, diabetic eye complications, nephropathy, neuropathy, diabetic foot, major and minor lower extremity amputation (LEA). The complications were determined through clinical coding in hospital (inpatient and outpatient) and primary care settings within the SingHealth cluster. We described event rates for the complications in 4 age-bands. Joinpoint regression was used to identify significant changes in trends.

Results

Among 222,705 patients studied between 2013 and 2020. 48.6% were female, 70.7% Chinese, 14.7% Malay and 10.6% Indian with a mean (SD) age varying between 64.6 (12.5) years in 2013 and 65.7 (13.2) years in 2020. We observed an increase in event rates in IHD, PAD, stroke, diabetic eye complications nephropathy, and neuropathy. Joinpoints was observed for IHD and PAD between 2016 to 2018, with subsequent plateauing of event rates. Major and minor LEA event rates decreased through the study period.

Conclusion

We found that DM and its complications represent an important challenge for healthcare in Singapore. Improvements in the trends of DM macrovascular complications were observed. However, trends in DM microvascular complications remain a cause for concern.

Healthcare Professionals' Views of Factors Influencing Diabetes Self-Management and the Utility of a mHealth Application and Its Features to Support Self-Care

INTRODUCTION

The perspectives of healthcare professionals (HCPs) are pivotal to co-development of self-management strategies for patients with diabetes. However, literature has been largely limited to perspectives of patients within the context of a Western healthcare setting. This study aims to explore factors influencing diabetes self-management in adult patients with diabetes from the perspectives of HCPs and their views of the value of mHealth application for diabetes self-management.

MATERIALS AND METHODS

We conducted focus group discussions (FGD) with purposively selected HCPs in Singapore. All FGDs were audio-recorded and transcribed verbatim. Thematic analysis was conducted using NVivo 12.

RESULTS

A total of 56 HCPs participated in the study. Barriers to self-management included limited patient commitment to lifestyle changes, suboptimal adherence to medication and treatment, patient resistance to insulin initiation and insufficient rapport between patients and HCPs. Patients' perceived susceptibility to complications, social support from family and community, multidisciplinary team care and patient's understanding of the benefits of self-care were viewed as facilitating self-management. HCPs saw mHealth apps as a vital opportunity to engage patients in the self-management of conditions and empower them to foster behavior changes. Yet, there were concerns regarding patient's limited digital literacy, lack of integration into routine electronic system and reluctance.

DISCUSSION

We identified a set of factors influencing self-management in adult patients with diabetes and useful app features that can empower patients to manage their conditions. Findings will inform the development of a mHealth application, and its features designed to improve self-care.

Trends in cardiovascular risk factors and treatment goals in patients with diabetes in Singapore-analysis of the SingHealth Diabetes Registry

Background

Asian populations are at high risk of diabetes and related vascular complications. We examined risk factor control, preventive care, and disparities in these trends among adults with diabetes in Singapore.

Methods

The sample included 209,930 adults with diabetes aged≥18 years from a multi-institutional SingHealth Diabetes Registry between 2013 and 2019 in Singapore. We performed logistic generalized estimating equations (GEEs) regression analysis and used linear mixed effect modeling to evaluate the temporal trends.

Results

Between 2013 and 2019, the unadjusted control rates of glycated hemoglobin (4.8%, 95%CI (4.4 to 5.1) and low-density lipoprotein cholesterol (LDL-C) (11.5%, 95%CI (11.1 to 11.8)) improved, but blood pressure (BP) control worsened (systolic BP (SBP)/diastolic BP (DBP) <140/90 mmHg: -6.6%, 95%CI (-7.0 to -6.2)). These trends persisted after accounting for the demographics including age, gender, ethnicity, and housing type. The 10-year adjusted risk for coronary heart disease (CHD) (3.4%, 95% (3.3 to 3.5)) and stroke (10.4%, 95% CI (10.3 to 10.5)) increased. In 2019, the control rates of glycated hemoglobin, BP (SBP/DBP<140/90 mmHg), LDL-C, each, and all three risk factors together, accounted for 51.5%, 67.7%, 72.2%, and 24.4%, respectively.

Conclusions

Trends in risk factor control improved for glycated hemoglobin and LDL-C, but worsened for BP among diabetic adults in Singapore from 2013 to 2019. Control rates for all risk factors remain inadequate.

Effective Treatment Recommendations for Type 2 Diabetes Management Using Reinforcement Learning: Treatment Recommendation Model Development and Validation

Background

Type 2 diabetes mellitus (T2DM) and its related complications represent a growing economic burden for many countries and health systems. Diabetes complications can be prevented through better disease control, but there is a large gap between the recommended treatment and the treatment that patients actually receive. The treatment of T2DM can be challenging because of different comprehensive therapeutic targets and individual variability of the patients, leading to the need for precise, personalized treatment.

Objective

The aim of this study was to develop treatment recommendation models for T2DM based on deep reinforcement learning. A retrospective analysis was then performed to evaluate the reliability and effectiveness of the models.

Methods

The data used in our study were collected from the Singapore Health Services Diabetes Registry, encompassing 189,520 patients with T2DM, including 6,407,958 outpatient visits from 2013 to 2018. The treatment recommendation model was built based on 80% of the dataset and its effectiveness was evaluated with the remaining 20% of data. Three treatment recommendation models were developed for antiglycemic, antihypertensive, and lipid-lowering treatments by combining a knowledge-driven model and a data-driven model. The knowledge-driven model, based on clinical guidelines and expert experiences, was first applied to select the candidate medications. The data-driven model, based on deep reinforcement learning, was used to rank the candidates according to the expected clinical outcomes. To evaluate the models, short-term outcomes were compared between the model-concordant treatments and the model-nonconcordant treatments with confounder adjustment by stratification, propensity score weighting, and multivariate regression. For long-term outcomes, model-concordant rates were included as independent variables to evaluate if the combined antiglycemic, antihypertensive, and lipid-lowering treatments had a positive impact on reduction of long-term complication occurrence or death at the patient level via multivariate logistic regression.

Results

The test data consisted of 36,993 patients for evaluating the effectiveness of the three treatment recommendation models. In 43.3% of patient visits, the antiglycemic medications recommended by the model were concordant with the actual prescriptions of the physicians. The concordant rates for antihypertensive medications and lipid-lowering medications were 51.3% and 58.9%, respectively. The evaluation results also showed that model-concordant treatments were associated with better glycemic control (odds ratio [OR] 1.73, 95% CI 1.69-1.76), blood pressure control (OR 1.26, 95% CI, 1.23-1.29), and blood lipids control (OR 1.28, 95% CI 1.22-1.35). We also found that patients with more model-concordant treatments were associated with a lower risk of diabetes complications (including 3 macrovascular and 2 microvascular complications) and death, suggesting that the models have the potential of achieving better outcomes in the long term.

Conclusions

Comprehensive management by combining knowledge-driven and data-driven models has good potential to help physicians improve the clinical outcomes of patients with T2DM; achieving good control on blood glucose, blood pressure, and blood lipids; and reducing the risk of diabetes complications in the long term.