Development of predictive risk models for major adverse cardiovascular events among patients with type 2 diabetes mellitus using health insurance claims data

Prediction of cardiovascular and renal risk among patients with apparent treatment-resistant hypertension in the United States using machine learning methods

Apparent treatment-resistant hypertension (aTRH), defined as blood pressure (BP) that remains uncontrolled despite unconfirmed concurrent treatment with three antihypertensives, is associated with an increased risk of developing cardiovascular and renal complications compared with controlled hypertension. We aimed to identify the characteristics of aTRH patients with an elevated risk of major adverse cardiovascular events plus (MACE+; defined as stroke, myocardial infarction, or heart failure hospitalization) and end stage renal disease (ESRD). This retrospective cohort study included aTRH patients (BP ≥140/90 mmHg and taking ≥3 antihypertensives) from the United States-based Optum® de-identified Electronic Health Record dataset and used machine learning models to identify risk factors of MACE+ or ESRD. Patients had claims for ≥3 antihypertensive classes within 30 days between January 1, 2015 and June 30, 2021, and two office BP measures recorded 1-90 days apart within 30 days to 11 months after the index regimen date. Of a total 18 797 070 patients identified with any hypertension, 71 100 patients had aTRH. During the study period (mean 25.5 months), 4944 (7.0%) patients had a MACE+ and 2403 (3.4%) developed ESRD. In total, 22 risk factors were included in the MACE+ model and 16 in the ESRD model, and most were significantly associated with study outcomes. The risk factors with the largest impact on MACE+ risk were congestive heart failure, stages 4 and 5 chronic kidney disease (CKD), age ≥80 years, and living in the Southern region of the United States. The risk factors with the largest impact on ESRD risk, other than pre-existing CKD, were anemia, congestive heart failure, and type 2 diabetes. The overall study cohort had a 5-year predicted MACE+ risk of 13.4%; this risk was increased in those in the top 50% and 25% high-risk groups (21.2% and 29.5%, respectively). The overall study cohort had a predicted 5-year risk of ESRD of 6.8%, which was increased in the top 50% and 25% high-risk groups (10.9% and 17.1%, respectively). We conclude that risk models developed in our study can reliably identify patients with aTRH at risk of MACE+ and ESRD based on information available in electronic health records; such models may be used to identify aTRH patients at high risk of adverse outcomes who may benefit from novel treatment interventions.

Precision prognostics for cardiovascular disease in Type 2 diabetes: a systematic review and meta-analysis

BACKGROUND

Precision medicine has the potential to improve cardiovascular disease (CVD) risk prediction in individuals with Type 2 diabetes (T2D).

METHODS

We conducted a systematic review and meta-analysis of longitudinal studies to identify potentially novel prognostic factors that may improve CVD risk prediction in T2D. Out of 9380 studies identified, 416 studies met inclusion criteria. Outcomes were reported for 321 biomarker studies, 48 genetic marker studies, and 47 risk score/model studies.

RESULTS

Out of all evaluated biomarkers, only 13 showed improvement in prediction performance. Results of pooled meta-analyses, non-pooled analyses, and assessments of improvement in prediction performance and risk of bias, yielded the highest predictive utility for N-terminal pro b-type natriuretic peptide (NT-proBNP) (high-evidence), troponin-T (TnT) (moderate-evidence), triglyceride-glucose (TyG) index (moderate-evidence), Genetic Risk Score for Coronary Heart Disease (GRS-CHD) (moderate-evidence); moderate predictive utility for coronary computed tomography angiography (low-evidence), single-photon emission computed tomography (low-evidence), pulse wave velocity (moderate-evidence); and low predictive utility for C-reactive protein (moderate-evidence), coronary artery calcium score (low-evidence), galectin-3 (low-evidence), troponin-I (low-evidence), carotid plaque (low-evidence), and growth differentiation factor-15 (low-evidence). Risk scores showed modest discrimination, with lower performance in populations different from the original development cohort.

CONCLUSIONS

Despite high interest in this topic, very few studies conducted rigorous analyses to demonstrate incremental predictive utility beyond established CVD risk factors for T2D. The most promising markers identified were NT-proBNP, TnT, TyG and GRS-CHD, with the highest strength of evidence for NT-proBNP. Further research is needed to determine their clinical utility in risk stratification and management of CVD in T2D.

Explainable machine learning for predicting conversion to neurological disease: Results from 52,939 medical records

Objective

This study assesses the application of interpretable machine learning modeling using electronic medical record data for the prediction of conversion to neurological disease.

Methods

A retrospective dataset of Cleveland Clinic patients diagnosed with Alzheimer's disease, amyotrophic lateral sclerosis, multiple sclerosis, or Parkinson's disease, and matched controls based on age, sex, race, and ethnicity was compiled. Individualized risk prediction models were created using eXtreme Gradient Boosting for each neurological disease at four timepoints in patient history. The prediction models were assessed for transparency and fairness.

Results

At timepoints 0-months, 12-months, 24-months, and 60-months prior to diagnosis, Alzheimer's disease models achieved the area under the receiver operating characteristic curve on a holdout test dataset of 0.794, 0.742, 0.709, and 0.645; amyotrophic lateral sclerosis of 0.883, 0.710, 0.658, and 0.620; multiple sclerosis of 0.922, 0.877, 0.849, and 0.781; and Parkinson's disease of 0.809, 0.738, 0.700, and 0.651, respectively.

Conclusions

The results demonstrate that electronic medical records contain latent information that can be used for risk stratification for neurological disorders. In particular, patient-reported outcomes, sleep assessments, falls data, additional disease diagnoses, and longitudinal changes in patient health, such as weight change, are important predictors.

Precision Prognostics for Cardiovascular Disease in Type 2 Diabetes: A Systematic Review and Meta-analysis

Background Precision medicine has the potential to improve cardiovascular disease (CVD) risk prediction in individuals with type 2 diabetes (T2D). Methods We conducted a systematic review and meta-analysis of longitudinal studies to identify potentially novel prognostic factors that may improve CVD risk prediction in T2D. Out of 9380 studies identified, 416 studies met inclusion criteria. Outcomes were reported for 321 biomarker studies, 48 genetic marker studies, and 47 risk score/model studies. Results Out of all evaluated biomarkers, only 13 showed improvement in prediction performance. Results of pooled meta-analyses, non-pooled analyses, and assessments of improvement in prediction performance and risk of bias, yielded the highest predictive utility for N-terminal pro b-type natriuretic peptide (NT-proBNP) (high-evidence), troponin-T (TnT) (moderate-evidence), triglyceride-glucose (TyG) index (moderate-evidence), Genetic Risk Score for Coronary Heart Disease (GRS-CHD) (moderate-evidence); moderate predictive utility for coronary computed tomography angiography (low-evidence), single-photon emission computed tomography (low-evidence), pulse wave velocity (moderate-evidence); and low predictive utility for C-reactive protein (moderate-evidence), coronary artery calcium score (low-evidence), galectin-3 (low-evidence), troponin-I (low-evidence), carotid plaque (low-evidence), and growth differentiation factor-15 (low-evidence). Risk scores showed modest discrimination, with lower performance in populations different from the original development cohort. Conclusions Despite high interest in this topic, very few studies conducted rigorous analyses to demonstrate incremental predictive utility beyond established CVD risk factors for T2D. The most promising markers identified were NT-proBNP, TnT, TyG and GRS-CHD, with the highest strength of evidence for NT-proBNP. Further research is needed to determine their clinical utility in risk stratification and management of CVD in T2D.

Incidence, mortality and disability-adjusted life years of acute myocardial infarction in Kazakhstan: data from unified national electronic healthcare system 2014-2019

Background

Cardiovascular diseases contribute to premature mortality globally, resulting in substantial social and economic burdens. The Global Burden of Disease (GBD) Study reported that in 2019 alone, heart attack and strokes accounted for the deaths of 18.6 million individuals. Ischemic heart diseases, including acute myocardial infarction (AMI), accounted for 182 million disability-adjusted life years (DALYs) and it is leading cause of death worldwide.

Aim

The aim of this study is to present the burden of AMI in Kazakhstan and describe the outcome of hospitalized patients.

Methods

The data of 79,172 people admitted to hospital with ICD-10 diagnosis I21 between 2014 and 2019 was derived from the Unified National Electronic Health System and retrospectively analyzed.

Results

The majority of the cohort (53,285, 67%) were men, with an average age of 63 (±12) years, predominantly of Kazakh (38,057, 48%) and Russian (24,583, 31%) ethnicities. Hypertension was the most common comorbidity (61,972, 78%). In males, a sharp increase in incidence is present after 40 years, while for females, the morbidity increases gradually after 55. Throughout the observation period, all-cause mortality rose from 101 to 210 people per million population (PMP). In 2019, AMI account for 169,862 DALYs in Kazakhstan, with a significant proportion (79%) attributed to years of life lost due to premature death (YLDs). Approximately half of disease burden due to AMI (80,794 DALYs) was in age group 55-69 years. Although incidence is higher for men, they have better survival rates than women. In terms of revascularization procedures, coronary artery bypass grafting yielded higher survival rates compared to percutaneous coronary intervention (86.3% and 80.9% respectively) during the 5-year follow-up.

Conclusion

This research evaluated the burden and disability-adjusted life years of AMI in Kazakhstan, the largest Central Asian country. The results show that more effective disease management systems and preventive measures at earlier ages are needed.

Prognostic models for short-term annual risk of severe complications and mortality in patients living with type 2 diabetes using a national medical claim database

OBJECTIVE

Prognostic models in patients living with diabetes allow physicians to estimate individual risk based on medical records and biological results. Clinical risk factors are not always all available to evaluate these models so that they may be complemented with models from claims databases. The objective of this study was to develop, validate and compare models predicting the annual risk of severe complications and mortality in patients living with type 2 diabetes (T2D) from a national claims data.

RESEARCH DESIGN AND METHODS

Adult patients with T2D were identified in a national medical claims database through their history of treatments or hospitalizations. Prognostic models were developed using logistic regression (LR), random forest (RF) and neural network (NN) to predict annual risk of outcome: severe cardiovascular (CV) complications, other severe T2D-related complications, and all-cause mortality. Risk factors included demographics, comorbidities, the adjusted Diabetes Severity and Comorbidity Index (aDSCI) and diabetes medications. Model performance was assessed using discrimination (C-statistics), balanced accuracy, sensibility and specificity.

RESULTS

A total of 22,708 patients with T2D were identified, with mean age of 68 years and average duration of T2D of 9.7 years. Age, aDSCI, disease duration, diabetes medications and chronic cardiovascular disease were the most important predictors for all outcomes. Discrimination with C-statistic ranged from 0.715 to 0.786 for severe CV complications, from 0.670 to 0.847 for other severe complications and from 0.814 to 0.860 for all-cause mortality, with RF having consistently the highest discrimination.

CONCLUSION

The proposed models reliably predict severe complications and mortality in patients with T2D, without requiring medical records or biological measures. These predictions could be used by payers to alert primary care providers and high-risk patients living with T2D.

The health and budget impact of sodium-glucose co-transporter-2 inhibitors (SGLT2is) in the Netherlands

OBJECTIVES

Type-2 Diabetes Mellitus (T2DM) increases both the patient risk of cardiovascular disease (CVD) and renal outcomes, such as chronic kidney disease (CKD). Recent clinical trials of the glucose-lowering drug-class of sodium-glucose co-transporter-2 inhibitors (SGLT2is) have shown benefits in preventing CVD events and progression of CKD, leading to an update of the Dutch T2DM treatment guideline for patients at risk. The aim of this study is to assess the health and economic impact of the guideline-recommended utilisation of SGLT2is in the Netherlands.

METHODS

The patient population at risk was determined by multiplying Dutch T2DM prevalence rates with the total numbers of inhabitants of the Netherlands in 2020. Subsequently, two analyses, comparing a treatment setting before and after implementation of the new guideline for SGLT2is, were conducted. Clinical and adverse event rates in both settings as well as direct healthcare costs were sourced from the literature. Total costs were calculated by multiplying disease prevalence, event rates and costs associated to outcomes. One-time disutilities per event were included to estimate the health impact. The potential health and economic impact of implementing the updated guideline was calculated.

RESULTS

Using a 5-year time horizon, the guideline-suggested utilisation of SGLT2is resulted in a health impact equal to 4,835 quality adjusted life years gained (0.0031 per patient per year) and €461 million cost-savings. The costs of treatment with SGLT2is were €813 million. Hence the net budget impact was €352 million for the total Dutch T2DM population, which translated to €0,57 per patient per day.

CONCLUSION

SGLT2is offer an option to reduce the number of CVD and CKD related events and associated healthcare costs and health losses in the Netherlands. Further research is needed to include the benefits of improved T2DM management options from a broader societal perspective.

Adapted Diabetes Complications Severity Index and Charlson Comorbidity Index in predicting all-cause and cause-specific mortality among patients with type 2 diabetes

INTRODUCTION

Adapted Diabetes Complications Severity Index (aDCSI) is a commonly used severity measure based on the number and severity of diabetes complications using diagnosis codes. The validity of aDCSI in predicting cause-specific mortality has yet to be verified. Additionally, the performance of aDCSI in predicting patient outcomes compared with Charlson Comorbidity Index (CCI) remains unknown.

RESEARCH DESIGN AND METHODS

Patients aged 20 years or older with type 2 diabetes prior to January 1, 2008 were identified from the Taiwan National Health Insurance claims data and were followed up until December 15, 2018. Complications for aDCSI including cardiovascular, cerebrovascular and peripheral vascular disease, metabolic disease, nephropathy, retinopathy and neuropathy, along with comorbidities for CCI, were collected. HRs of death were estimated using Cox regression. Model performance was evaluated by concordance index and Akaike information criterion.

RESULTS

1,002,589 patients with type 2 diabetes were enrolled, with a median follow-up of 11.0 years. After adjusting for age and sex, aDCSI (HR 1.21, 95% CI 1.20 to 1.21) and CCI (HR 1.18, 1.17 to 1.18) were associated with all-cause mortality. The HRs of aDCSI for cancer, cardiovascular disease (CVD) and diabetes mortality were 1.04 (1.04 to 1.05), 1.27 (1.27 to 1.28) and 1.28 (1.28 to 1.29), respectively, and the HRs of CCI were 1.10 (1.09 to 1.10), 1.16 (1.16 to 1.17) and 1.17 (1.16 to 1.17), respectively. The model with aDCSI had a better fit for all-cause, CVD and diabetes mortality with C-index of 0.760, 0.794 and 0.781, respectively. Models incorporating both scores had even better performance, but the HR of aDCSI for cancer (0.98, 0.97 to 0.98) and the HRs of CCI for CVD (1.03, 1.02 to 1.03) and diabetes mortality (1.02, 1.02 to 1.03) became neutral. When aDCSI and CCI were considered time-varying scores, the association with mortality was stronger. aDCSI had a strong correlation with mortality even after 8 years (HR 1.18, 1.17 to 1.18).

CONCLUSIONS

The aDCSI predicts all-cause, CVD and diabetes deaths but not cancer deaths better than the CCI. aDCSI is also a good predictor for long-term mortality.

Effectiveness of statins for the primary prevention of cardiovascular disease in the Asian elderly population

BACKGROUND

Population aging is a global trend, and the elderly have a higher risk of atherosclerotic cardiovascular disease (ASCVD) and related mortality. Statins have been observed to reduce cardiovascular events in patients with ASCVD. However, compared with secondary prevention, the benefits of statins for primary prevention are undetermined among the elderly.

AIMS

This study aimed to evaluate the effectiveness of statins in an elderly population without a history of cardiovascular disease (CVD).

METHODS

The study was carried out using the National Taiwan University Hospital Integrated Medical Database and the National Health Insurance Research Database in Taiwan. Patients aged 65 years and older without a history of CVD were identified between 1 February 2008 and 31 December 2015. New statin users were 1:4 matched to nonusers based on certain variables. The risks of major adverse cardiovascular events (MACEs) and all-cause mortality were estimated using Cox proportional hazards models. Further, we applied marginal structural models to account for time-varying low-density lipoprotein cholesterol (LDL-C) levels.

RESULTS

A total of 2761 new statin users and 9503 nonusers were selected after matching; the mean age was 71.8 years, and 63% were women. At a median follow-up of 4.8 years, statin use was associated with reduced risk of MACEs (hazard ratio [HR]: 0.75; 95% confidence interval [CI], 0.52-0.98) and mortality (HR: 0.72, 95% CI: 0.55-0.93) when accounting for time-varying LDL-C. No significant differences in effect were detected between subgroups.

CONCLUSION

Statin use could be beneficial for the primary prevention of CVD in elderly Asians.

Development and validation of risk prediction models for stroke and mortality among patients with type 2 diabetes in northern China

BACKGROUND

Stroke is one of the leading causes of disability and mortality in patients with type 2 diabetes mellitus (T2DM). Risk models have been developed for predicting stroke and stroke-associated mortality among patients with T2DM. Here, we evaluated risk factors of stroke for individualized prevention measures in patients with T2DM in northern China.

METHODS

In the community-based Tianjin Chronic Disease Cohort study, 58,042 patients were enrolled between January 2014 and December 2019. We used multiple imputation (MI) to impute missing variables and univariate and multivariate Cox's proportional hazard regression to screen risk factors of stroke. Furthermore, we established and validated first-ever prediction models for stroke (Model 1 and Model 2) and death from stroke (Model 3) and evaluated their performance.

RESULTS

In the derivation and validation groups, the area under the curves (AUCs) of Models 1-3 was better at 5 years than at 8 years. The Harrell's C-index for all models was above 0.7. All models had good calibration, discrimination, and clinical net benefit. Sensitivity analysis using the MI dataset indicated that all models had good and stable prediction performance.

CONCLUSION

In this study, we developed and validated first-ever risk prediction models for stroke and death from stroke in patients with T2DM, with good discrimination and calibration observed in all models. Based on lifestyle, demographic characteristics, and laboratory examination, these models could provide multidimensional management and individualized risk assessment. However, the models developed here may only be applicable to Han Chinese.

Evaluation of motor vehicle crashes between scooter riders and car drivers after diagnosis of type 2 diabetes in Taiwan

BACKGROUND

Despite the plausibility that diabetes may increase the risk of motor vehicle crashes (MVCs) due to various diabetes related complications and co-morbidity, findings from epidemiological studies on the relationship between diabetes and MVCs remained inclusive mainly due to heterogeneity in the study design and failure to complete consideration of potential confounders. This study re-visited this putative association with an improved study design.

METHOD

This study employed a controlled before-after study design and included 1,264,280 people aged 18-75 years with T2D newly diagnosed from 2009-2014 and an equal number of age-, sex-, and time-matched controls. The rate ratios (RRs) of vehicle type-specific incidence rates of MVCs in the 1 and 2 years before and after diabetes diagnosis (or the matched dates) were compared between the individuals with type 2 diabetes (T2D) and their matched controls.

RESULTS

The rate of MVCs increased slightly among people with T2D over 1 and 2 years following diabetes diagnosis, with RRs of 1.04 (95% confidence interval [CI]=1.02-1.07) and 1.11 (95% CI=1.09-1.13), respectively. These RRs were comparable to those obtained for controls (1.06 and 1.12, respectively). By contrast, the RRs of scooter crashes were significantly higher in the T2D group than in the control group during the 1 year (1.28 vs. 1.08, p < 0.001) and 2 years (1.32 vs. 1.08, p < 0.001) following diabetes diagnosis.

CONCLUSION

T2D diagnosis was associated with a moderate but significant increase in the risk of MVCs among scooter drivers, but not among car drivers.

Performance assessment across different care settings of a heart failure hospitalisation risk-score for type 2 diabetes using administrative claims

Predicting the risk of cardiovascular complications, in particular heart failure hospitalisation (HHF), can improve the management of type 2 diabetes (T2D). Most predictive models proposed so far rely on clinical data not available at the higher Institutional level. Therefore, it is of interest to assess the risk of HHF in people with T2D using administrative claims data only, which are more easily obtainable and could allow public health systems to identify high-risk individuals. In this paper, the administrative claims of > 175,000 patients with T2D were used to develop a new risk score for HHF based on Cox regression. Internal validation on the administrative data cohort yielded satisfactory results in terms of discrimination (max AUROC = 0.792, C-index = 0.786) and calibration (Hosmer-Lemeshow test p value < 0.05). The risk score was then tested on data gathered from two independent centers (one diabetes outpatient clinic and one primary care network) to demonstrate its applicability to different care settings in the medium-long term. Thanks to the large size and broad demographics of the administrative dataset used for training, the proposed model was able to predict HHF without significant performance loss concerning bespoke models developed within each setting using more informative, but harder-to-acquire clinical variables.

MicroRNA-134-5p and the Extent of Arterial Occlusive Disease Are Associated with Risk of Future Adverse Cardiac and Cerebral Events in Diabetic Patients Undergoing Carotid Artery Stenting for Symptomatic Carotid Artery Disease

There is little known about the prognostic value of serum microRNAs (miRs) in diabetic patients with symptomatic internal carotid artery disease (ICAS) who underwent stent supported angioplasty (PTA) for ICAS. The present study aimed to investigate expression levels of selected miRs for future major adverse cardiac and cerebral events (MACCE) as a marker in diabetic patients following ICAS-PTA. The expression levels of 11 chosen circulating serum miRs were compared in 37 diabetic patients with symptomatic ICAS and 64 control group patients with symptomatic ICAS, but free of diabetes. The prospective median follow-up of 84 months was performed for cardiovascular outcomes. Diabetic patients, as compared to control subjects, did not differ with respect to age (p = 0.159), distribution of gender (p = 0.375), hypertension (p = 0.872), hyperlipidemia (p = 0.203), smoking (p = 0.115), coronary heart disease (p = 0.182), lower extremities arterial disease (LEAD, p = 0.731), and miRs expressions except from lower miR-16-5p (p < 0.001). During the follow-up period, MACCE occurred in 16 (43.2%) diabetic and 26 (40.6%) non-diabetic patients (p = 0.624). On multivariate Cox analysis, hazard ratio (HR) and 95% Confidence Intervals (95%CI) for diabetic patients associated with MACCE were miR-134-5p (1.12; 1.05−1.21, p < 0.001), miR-499-5p (0.16; 0.02−1.32, p = 0.089), hs-CRP (1.14; 1.02−1.28; p = 0.022), prior myocardial infarction (8.56, 1.91−38.3, p = 0.004), LEAD (11.9; 2.99−47.9, p = 0.005), and RAS (20.2; 2.4−167.5, p = 0.005), while in non-diabetic subjects, only miR-16-5p (1.0006; 1.0001−1.0012, p = 0.016), miR-208b-3p (2.82; 0.91−8.71, p = 0.071), and hypertension (0.27, 0.08−0.95, p = 0.042) were associated with MACCE. Our study demonstrated that different circulating miRs may be prognostic for MACCE in diabetic versus non-diabetic patients with symptomatic ICAS. Higher expression levels of miR-134 were prognostic for MACCE in diabetic patients, while higher expression levels of miR-16 were prognostic in non-diabetic patients.

2022 update to the position statement by Primary Care Diabetes Europe: a disease state approach to the pharmacological management of type 2 diabetes in primary care

Type 2 diabetes and its associated comorbidities are growing more prevalent, and the complexity of optimising glycaemic control is increasing, especially on the frontlines of patient care. In many countries, most patients with type 2 diabetes are managed in a primary care setting. However, primary healthcare professionals face the challenge of the growing plethora of available treatment options for managing hyperglycaemia, leading to difficultly in making treatment decisions and contributing to treatment and therapeutic inertia. This position statement offers a simple and patient-centred clinical decision-making model with practical treatment recommendations that can be widely implemented by primary care clinicians worldwide through shared-decision conversations with their patients. It highlights the importance of managing cardiovascular disease and elevated cardiovascular risk in people with type 2 diabetes and aims to provide innovative risk stratification and treatment strategies that connect patients with the most effective care.

Predictive Risk Models to Identify Patients at High-Risk for Severe Clinical Outcomes With Chronic Kidney Disease and Type 2 Diabetes

Introduction/Objective:

Predictive risk models identifying patients at high risk for specific outcomes may provide valuable insights to providers and payers regarding points of intervention and modifiable factors. The goal of our study was to build predictive risk models to identify patients with chronic kidney disease (CKD) and type 2 diabetes (T2D) at high risk for progression to end stage kidney disease (ESKD), mortality, and hospitalization for cardiovascular disease (CVD), cerebrovascular disease (CeVD), and heart failure (HF).

Methods:

This was a retrospective observational cohort study utilizing administrative claims data in patients with CKD (stage 3-4) and T2D aged 65 to 89 years enrolled in a Medicare Advantage Drug Prescription plan offered by Humana Inc. between 1/1/2012 and 12/31/2017. Patients were enrolled ≥1 year pre-index and followed for outcomes, including hospitalization for CVD, CeVD and HF, ESKD, and mortality, 2 years post-index. Pre-index characteristics comprising demographic, comorbidities, laboratory values, and treatment (T2D and cardiovascular) were evaluated and included in the models. LASSO technique was used to identify predictors to be retained in the final models followed by logistic regression to generate parameter estimates and model performance statistics. Inverse probability censoring weighting was used to account for varying follow-up time.

Results:

We identified 169 876 patients for inclusion. Declining estimated glomerular filtration rate (eGFR) increased the risk of hospitalization for CVD (38.6%-61.8%) and HF (2-3 times) for patients with eGFR 15 to 29 mL/min/1.73 m² compared to patients with eGFR 50 to 59 mL/min/1.73 m². Patients with urine albumin-to-creatinine ratio (UACR) ≥300 mg/g had greater chance for hospitalization for CVD (2.0 times) and HF (4.9 times), progression to ESKD (2.9 times) and all-cause mortality (2.4 times) than patients with UACR <30 mg/g. Elevated hemoglobin A1c (≥8%) increased the chances for hospitalization for CVD (21.3%), CeVD (45.4%), and death (20.6%). Among comorbidities, history of HF increased the risk for ESKD, mortality, and hospitalization for CVD, CeVD, and HF.

Conclusions:

The predictive models developed in this study could potentially be used as decision support tools for physicians and payers, and the risk scores from these models can be applied to future outcomes studies focused on patients with T2D and CKD.

Major adverse cardiovascular event definitions used in observational analysis of administrative databases: a systematic review

Background

Major adverse cardiovascular events (MACE) are increasingly used as composite outcomes in randomized controlled trials (RCTs) and observational studies. However, it is unclear how observational studies most commonly define MACE in the literature when using administrative data.

Methods

We identified peer-reviewed articles published in MEDLINE and EMBASE between January 1, 2010 to October 9, 2020. Studies utilizing administrative data to assess the MACE composite outcome using International Classification of Diseases 9th or 10th Revision diagnosis codes were included. Reviews, abstracts, and studies not providing outcome code definitions were excluded. Data extracted included data source, timeframe, MACE components, code definitions, code positions, and outcome validation.

Results

A total of 920 articles were screened, 412 were retained for full-text review, and 58 were included. Only 8.6% (n = 5/58) matched the traditional three-point MACE RCT definition of acute myocardial infarction (AMI), stroke, or cardiovascular death. None matched four-point (+unstable angina) or five-point MACE (+unstable angina and heart failure). The most common MACE components were: AMI and stroke, 15.5% (n = 9/58); AMI, stroke, and all-cause death, 13.8% (n = 8/58); and AMI, stroke and cardiovascular death 8.6% (n = 5/58). Further, 67% (n = 39/58) did not validate outcomes or cite validation studies. Additionally, 70.7% (n = 41/58) did not report code positions of endpoints, 20.7% (n = 12/58) used the primary position, and 8.6% (n = 5/58) used any position.

Conclusions

Components of MACE endpoints and diagnostic codes used varied widely across observational studies. Variability in the MACE definitions used and information reported across observational studies prohibit the comparison, replication, and aggregation of findings. Studies should transparently report the administrative codes used and code positions, as well as utilize validated outcome definitions when possible.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12874-021-01440-5.

Artificial intelligence and diabetes technology: A review

Artificial intelligence (AI) is widely discussed in the popular literature and is portrayed as impacting many aspects of human life, both in and out of the workplace. The potential for revolutionizing healthcare is significant because of the availability of increasingly powerful computational platforms and methods, along with increasingly informative sources of patient data, both in and out of clinical settings. This review aims to provide a realistic assessment of the potential for AI in understanding and managing diabetes, accounting for the state of the art in the methodology and medical devices that collect data, process data, and act accordingly. Acknowledging that many conflicting definitions of AI have been put forth, this article attempts to characterize the main elements of the field as they relate to diabetes, identifying the main perspectives and methods that can (i) affect basic understanding of the disease, (ii) affect understanding of risk factors (genetic, clinical, and behavioral) of diabetes development, (iii) improve diagnosis, (iv) improve understanding of the arc of disease (progression and personal/societal impact), and finally (v) improve treatment.

A Deep Learning Approach to Predict Diabetes' Cardiovascular Complications From Administrative Claims

People with diabetes require lifelong access to healthcare services to delay the onset of complications. Their disease management processes generate great volumes of data across several domains, from clinical to administrative. Difficulties in accessing and processing these data hinder their secondary use in an institutional setting, even for highly desirable applications, such as the prediction of cardiovascular disease, the main driver of excess mortality in diabetes. Hence, in the present work, we propose a deep learning model for the prediction of major adverse cardiovascular events (MACE), developed and validated using the administrative claims of 214,676 diabetic patients of the Veneto region, in North East Italy. Specifically, we use a year of pharmacy and hospitalisation claims, together with basic patient's information, to predict the 4P-MACE composite endpoint, i.e., the first occurrence of death, heart failure, myocardial infarction, or stroke, with a variable prediction horizon of 1 to 5 years. Adapting to the time-to-event nature of this task, we cast our problem as a multi-outcome (4P-MACE and components), multi-label (1 to 5 years) classification task with a custom loss to account for the effect of censoring. Our model, purposefully specified to minimise data preparation costs, exhibits satisfactory performance in predicting 4P-MACE at all prediction horizons: AUROC from 0.812 (C.I.: 0.797 - 0.827) to 0.792 (C.I.: 0.781 - 0.802); C-index from 0.802 (C.I.: 0.788 - 0.816) to 0.770 (C.I.: 0.761 - 0.779). Components' prediction performance is also adequate, ranging from death's 0.877 1-year AUROC to stroke's 0.689 5-year AUROC.

Development and validation of a claims-based model to identify patients at risk of chronic thromboembolic pulmonary hypertension following acute pulmonary embolism

OBJECTIVE

Chronic thromboembolic pulmonary hypertension (CTEPH) is a rare disease that often follows pulmonary embolism (PE). Screening for CTEPH is challenging, often delaying diagnosis and worsening prognosis. Predictive risk models for CTEPH could help identify at-risk patients, but existing models require multiple clinical inputs. We developed and validated a predictive risk model for CTEPH using health insurance claims that can be used by payers/quality-of-care organizations to screen patients post-PE.

METHODS

Adult patients newly diagnosed with acute PE (index date) were identified from the Optum De-identified Clinformatics Extended DataMart (January 2007-March 2018; development set) and IBM MarketScan (January 2008-June 2019; validation set) databases. Predictors were identified 12 months before or on the index PE. Risk of "likely CTEPH" was assessed post-PE based on CTEPH-related diagnoses and procedures since the CTEPH diagnosis code (ICD-10-CM: I27.24) was not available until 1 October 2017. Stepwise variable selection was used to build the model using the development set; model validation was subsequently conducted using the validation set.

RESULTS

The development set included 93,428 patients, of whom 11,878 (12.7%) developed likely CTEPH. Older age (odds ratios [OR] = 1.16-1.49), female (OR = 1.09), unprovoked PE (i.e. without thrombotic factors; OR = 1.14), hypertension (OR = 1.07), osteoarthritis (OR = 1.08), diabetes (OR = 1.07), chronic obstructive pulmonary disease (OR = 1.11), obesity (OR = 1.21) were associated with higher odds of likely CTEPH, and oral anticoagulants with lower odds (OR= 0.50, all p < .01). C-statistic was 0.77 in the development and validation sets.

CONCLUSION

A claims-based risk model reliably predicted the risk of CTEPH post-PE and could be used to identify high-risk patients who may benefit from focused monitoring.

Predictive models for cardiovascular and kidney outcomes in patients with type 2 diabetes: systematic review and meta-analyses

OBJECTIVE

To inform a clinical practice guideline (BMJ Rapid Recommendations) considering sodium glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists for treatment of adults with type 2 diabetes, we summarised the available evidence regarding the performance of validated risk models on cardiovascular and kidney outcomes in these patients.

METHODS

We systematically searched bibliographic databases in January 2020 to identify observational studies evaluating risk models for all-cause and cardiovascular mortality, heart failure (HF) hospitalisations, end-stage kidney disease (ESKD), myocardial infarction (MI) and ischaemic stroke in ambulatory adults with type 2 diabetes. Using a random effects model, we pooled discrimination measures for each model and outcome, separately, and descriptively summarised calibration plots, when available. We used the Prediction Model Risk of Bias Assessment Tool to assess risk of bias of each included study and the Grading of Recommendations, Assessment, Development, and Evaluation approach to evaluate our certainty in the evidence.

RESULTS

Of 22 589 publications identified, 15 observational studies reporting on seven risk models proved eligible. Among the seven models with >1 validation cohort, the Risk Equations for Complications of Type 2 Diabetes (RECODe) had the best calibration in primary studies and the highest pooled discrimination measures for the following outcomes: all-cause mortality (C-statistics 0.75, 95% CI 0.70 to 0.80; high certainty), cardiovascular mortality (0.79, 95% CI 0.75 to 0.84; low certainty), ESKD (0.73, 95% CI 0.52 to 0.94; low certainty), MI (0.72, 95% CI 0.69 to 0.74; moderate certainty) and stroke (0.71, 95% CI 0.68 to 0.74; moderate certainty). This model does not, however, predict risk of HF hospitalisations.

CONCLUSION

Of available risk models, RECODe proved to have satisfactory calibration in primary validation studies and acceptable discrimination superior to other models, though with high risk of bias in most primary studies.

TRIAL REGISTRATION NUMBER

CRD42020168351.

A disease state approach to the pharmacological management of Type 2 diabetes in primary care: A position statement by Primary Care Diabetes Europe

Type 2 diabetes and its associated comorbidities are growing more prevalent, and the complexity of optimising glycaemic control is increasing, especially on the frontlines of patient care. In many countries, most patients with type 2 diabetes are managed in a primary care setting. However, primary healthcare professionals face the challenge of the growing plethora of available treatment options for managing hyperglycaemia, leading to difficultly in making treatment decisions and contributing to therapeutic inertia. This position statement offers a simple and patient-centred clinical decision-making model with practical treatment recommendations that can be widely implemented by primary care clinicians worldwide through shared-decision conversations with their patients. It highlights the importance of managing cardiovascular disease and elevated cardiovascular risk in people with type 2 diabetes and aims to provide innovative risk stratification and treatment strategies that connect patients with the most effective care.

Electronic Medical Record-Based Case Phenotyping for the Charlson Conditions: Scoping Review

Background

Electronic medical records (EMRs) contain large amounts of rich clinical information. Developing EMR-based case definitions, also known as EMR phenotyping, is an active area of research that has implications for epidemiology, clinical care, and health services research.

Objective

This review aims to describe and assess the present landscape of EMR-based case phenotyping for the Charlson conditions.

Methods

A scoping review of EMR-based algorithms for defining the Charlson comorbidity index conditions was completed. This study covered articles published between January 2000 and April 2020, both inclusive. Embase (Excerpta Medica database) and MEDLINE (Medical Literature Analysis and Retrieval System Online) were searched using keywords developed in the following 3 domains: terms related to EMR, terms related to case finding, and disease-specific terms. The manuscript follows the Preferred Reporting Items for Systematic reviews and Meta-analyses extension for Scoping Reviews (PRISMA) guidelines.

Results

A total of 274 articles representing 299 algorithms were assessed and summarized. Most studies were undertaken in the United States (181/299, 60.5%), followed by the United Kingdom (42/299, 14.0%) and Canada (15/299, 5.0%). These algorithms were mostly developed either in primary care (103/299, 34.4%) or inpatient (168/299, 56.2%) settings. Diabetes, congestive heart failure, myocardial infarction, and rheumatology had the highest number of developed algorithms. Data-driven and clinical rule–based approaches have been identified. EMR-based phenotype and algorithm development reflect the data access allowed by respective health systems, and algorithms vary in their performance.

Conclusions

Recognizing similarities and differences in health systems, data collection strategies, extraction, data release protocols, and existing clinical pathways is critical to algorithm development strategies. Several strategies to assist with phenotype-based case definitions have been proposed.

Association of ambient air pollution with cardiovascular disease risks in people with type 2 diabetes: a Bayesian spatial survival analysis

BACKGROUND

Evidence is limited on excess risks of cardiovascular diseases (CVDs) associated with ambient air pollution in diabetic populations. Survival analyses without considering the spatial structure and possible spatial correlations in health and environmental data may affect the precision of estimation of adverse environmental pollution effects. We assessed the association between air pollution and CVDs in type 2 diabetes through a Bayesian spatial survival approach.

METHODS

Taiwan's national-level health claims and air pollution databases were utilized. Fine individual-level latitude and longitude were used to determine pollution exposure. The exponential spatial correlation between air pollution and CVDs was analyzed in our Bayesian model compared to traditional Weibull and Cox models.

RESULTS

There were 2072 diabetic patients included in analyses. PM_2.5 and SO₂ were significant CVD risk factors in our Bayesian model, but such associations were attenuated or underestimated in traditional models; adjusted hazard ratio (HR) and 95% credible interval (CrI) or confidence interval (CI) of CVDs for a 1 μg/m³ increase in the monthly PM_2.5 concentration for our model, the Weibull and Cox models was 1.040 (1.004-1.073), 0.994 (0.984-1.004), and 0.994 (0.984-1.004), respectively. With a 1 ppb increase in the monthly SO₂ concentration, adjusted HR (95% CrI or CI) was 1.886 (1.642-2.113), 1.092 (1.022-1.168), and 1.091 (1.021-1.166) for these models, respectively.

CONCLUSIONS

Against traditional non-spatial analyses, our Bayesian spatial survival model enhances the assessment precision for environmental research with spatial survival data to reveal significant adverse cardiovascular effects of air pollution among vulnerable diabetic patients.

Impact on guidelines: The general practitioner point of view

Primary care physicians are uniquely placed to offer holistic, patient-centred care to patients with T2DM. While the recent FDA-mandated cardiovascular outcome trials offer a wealth of data to inform treatment discussions, they have also contributed to increasing complexity in treatment decisions, and in the guidelines that seek to assist in making these decisions. To assist physicians in avoiding treatment inertia, Primary Care Diabetes Europe has formulated a position statement that summarises our current understanding of the available T2DM treatment options in various patient populations. New data from recent outcomes trials is contextualised and summarised for the primary care physician. This consensus paper also proposes a unique and simple tool to stratify patients into 'very high' and 'high' cardiovascular risk categories and outlines treatment recommendations for patients with atherosclerotic cardiovascular disease, heart failure and chronic kidney disease. Special consideration is given to elderly/frail patients and those with obesity. A visual patient assessment tool is provided, and a comprehensive set of prescribing tips is presented for all available classes of glucose-lowering therapies. This position statement will complement the already available, often specialist-focused, T2DM treatment guidelines and provide greater direction in how the wealth of outcome trial data can be applied to everyday practice.

Development of risk models for major adverse chronic renal outcomes among patients with type 2 diabetes mellitus using insurance claims: a retrospective observational study

Objective: To develop and validate models allowing the prediction of major adverse chronic renal outcomes (MACRO) in patients with type 2 diabetes mellitus (T2DM) using insurance claims data.Methods: The Optum Integrated Real World Evidence Electronic Health Records and Claims de-identified database (10/01/2006-09/30/2016) was used to identify T2DM patients ≥50 years old. Risk factors were assessed over a 12-month baseline period, and MACRO were subsequently assessed until the end of data availability, continuous enrollment, or death. Separate models were built for moderate-to-severe diabetic kidney disease (DKD), end-stage renal disease (ESRD), and renal death. A random split-sample approach was employed, where 70% of the sample served for model development (training set) and the remaining 30% served for validation (testing set). C-statistics were used to assess model performance.Results: A total of 160,031 patients were included. Risk factors associated with MACRO for all models included adapted diabetes complications severity index, heart failure, anemia, diabetic nephropathy, and CKD. C-statistics ranged between 0.70 (moderate-to-severe DKD) and 0.84 (renal death) in the testing set. A substantial proportion (e.g. 88.7% for moderate-to-severe DKD) of patients predicted to be at high-risk of MACRO did not have diabetic nephropathy, proteinuria, or CKD at baseline.Conclusions: The models developed using insurance claims data could reliably predict the risk of MACRO in patients with T2DM and enabled patients at higher-risk of DKD to be identified in the absence of baseline diabetic nephropathy, CKD, or proteinuria. These models could help establish strategies to reduce the risk of MACRO in T2DM patients.

Risk of early mortality and cardiovascular disease in type 1 diabetes: a comparison with type 2 diabetes, a nationwide study

Background

Both type 1 and type 2 diabetes are well-established risk factors for cardiovascular disease and early mortality. However, few studies have directly compared the hazards of cardiovascular outcomes and premature death among people with type 1 diabetes to those among people with type 2 diabetes and subjects without diabetes. Furthermore, information about the hazard of cardiovascular disease and early mortality among Asians with type 1 diabetes is sparse, although the clinical and epidemiological characteristics of Asians with type 1 diabetes are unlike those of Europeans. We estimated the hazard of myocardial infarction (MI), hospitalization for heart failure (HF), atrial fibrillation (AF), and mortality during follow-up in Korean adults with type 1 diabetes compared with those without diabetes and those with type 2 diabetes.

Methods

We used Korean National Health Insurance Service datasets of preventive health check-ups from 2009 to 2016 in this retrospective longitudinal study. The hazard ratios of MI, HF, AF, and mortality during follow-up were analyzed using the Cox regression analyses according to the presence and type of diabetes in ≥ 20-year-old individuals without baseline cardiovascular disease (N = 20,423,051). The presence and type of diabetes was determined based on the presence of type 1 or type 2 diabetes at baseline.

Results

During more than 93,300,000 person-years of follow-up, there were 116,649 MIs, 135,532 AF cases, 125,997 hospitalizations for HF, and 344,516 deaths. The fully-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for incident MI, hospitalized HF, AF, and all-cause death within the mean follow-up of 4.6 years were higher in the type 1 diabetes group than the type 2 diabetes [HR (95% CI) 1.679 (1.490–1.893) for MI; 2.105 (1.901–2.330) for HF; 1.608 (1.411–1.833) for AF; 1.884 (1.762–2.013) for death] and non-diabetes groups [HR (95% CI) 2.411 (2.138–2.718) for MI; 3.024 (2.730–3.350) for HF; 1.748 (1.534–1.993) for AF; 2.874 (2.689–3.073) for death].

Conclusions

In Korea, the presence of diabetes was associated with a higher hazard of cardiovascular disease and all-cause death. Specifically, people with type 1 diabetes had a higher hazard of cardiovascular disease and all-cause mortality compared to people with type 2 diabetes.

Descriptive Analysis of Long- and Intermediate-Acting Insulin and Key Safety Outcomes in Adults with Type 2 Diabetes Mellitus

BACKGROUND

As new biosimilar and follow-on insulins enter the market, more data are needed on safety, effectiveness, and patterns of use for these products to inform prescriber and patient decision-making regarding treatment. Additionally, data are needed regarding real-world patterns of use to inform future studies comparing the safety and effectiveness of bio-similars to already approved agents for diabetes treatment.

OBJECTIVE

To analyze the medication use patterns, adverse events, and availability of glycated hemoglobin (A1c) values for adult patients with type 2 diabetes mellitus (T2DM) who use long-acting insulin (LAI) or neutral protamine Hagedorn (NPH), an intermediate-acting insulin.

METHODS

We used the Biologics and Biosimilars Collective Intelligence Consortium's (BBCIC) distributed research network (DRN) for this descriptive analysis. The analysis time frame was January 1, 2011, to September 30, 2015, and included patients continuously insured for at least 183 days before the first date of a filled prescription for LAI or NPH insulin alone or with rapid- or short-acting insulin or sulfonylureas, whether newly starting insulin or switching to a different product. Insulin exposure episodes were the unit of analysis, and patients were classified in cohorts according to treatment. We followed patients until end of health plan enrollment or the end of the study period. We used occurrence of a study outcome, switch to another medication regimen, discontinuation of the current medication, or study end date to mark the end of an insulin episode. We describe demographics and availability of A1c values for analysis. Study outcomes included severe hypoglycemic events and major adverse cardiac events (MACE).

RESULTS

We identified 103,951 patients with T2DM from a database of 39.1 million patients with commercial or Medicare Advantage pharmacy and medical benefits, who contributed 279,533 unique insulin exposure episodes. Most episodes (89%) included patients using LAI, and 52% of patients contributed data to 2 or more exposure cohorts. Insulin episodes lasted an average of 3.5 months, and patients had an average follow-up of 8.6 months. The unadjusted rate of severe hypoglycemic events requiring medical attention was 96.9 per 10,000 patient-years at risk (10kPYR). The unadjusted incident MACE rate was 676.9 events per 10kPYR. 38,330 T2DM patients in the BBCIC DRN had a baseline A1c available, and of those, less than 50% had a follow-up A1c result.

CONCLUSIONS

Among patients with T2DM, our observed insulin patterns of use and rates of severe hypoglycemic outcomes and MACE are consistent with other studies. We noted a paucity of A1c results available, which implies that additional data sources may be needed to augment the BBCIC DRN.

DISCLOSURES

This study was coordinated and funded by the Biologics and Biosimilars Collective Intelligence Consortium (BBCIC) and represents the independent findings of the BBCIC Insulins Principal Investigator and the BBCIC Insulins Research Team. Lockhart is employed by the BBCIC and the Academy of Managed Care Pharmacy (AMCP). Eichelberger was employed by the BBCIC and AMCP at the time of this study. McMahill-Walraven is employed by Aetna, a CVS Health business. Panozzo, Marshall, and Brown are employed by Harvard Pilgrim Healthcare Institute. Aetna was reimbursed for data and analytic support from Harvard Pilgrim Healthcare Institute and the Reagan Udall Foundation for the U.S. Food and Drug Administration. Aetna receives external funding through research grants and subcontracts with Harvard Pilgrim Healthcare Institute, which are funded by the FDA, NIH, PCORI, BBCIC, Pfizer, and GSK; the Reagan-Udall Foundation for IMEDS; and PCORI for the ADAPTABLE Study. This work was previously presented as a poster at AMCP Nexus 2018; October 22-25, 2018; in Orlando, FL.

Cholesterol levels and development of cardiovascular disease in Koreans with type 2 diabetes mellitus and without pre-existing cardiovascular disease

Background

The aim of the present study was to identify a threshold for the cholesterol level at which the risk of cardiovascular disease (CVD) begins to increase in people with type 2 diabetes mellitus (DM).

Methods

Using the Korean National Health Insurance Service database, 2,077,135 people aged ≥ 40 years with type 2 DM who underwent regular health checks between 2009 and 2012 were included. Subjects with previous CVD were excluded. Cox regression analyses were performed to estimate the risk of CVD for each low-density lipoprotein cholesterol (LDL-C) group using the < 70 mg/dL as the reference group.

Results

There were 78,560 cases of stroke (3.91%), and 50,791 myocardial infarction (MI, 2.53%) during a median follow-up of 7.1 years. Among participants not taking statins, LDL-C levels of 130–159 mg/dL and ≥ 160 mg/dL were significantly associated with the risk of MI: the hazard ratios (HRs) (95% confidence interval) were 1.19 (1.14–1.25) and 1.53 (1.46–1.62), respectively. Among participants taking statins, all categories of LDL-C level ≥ 70 mg/dL were significantly associated with increased risk of stroke and MI.

Conclusions

We identified an increased risk of CVD in people with an LDL-C level ≥ 130 mg/dL among individuals with type 2 DM not taking statins. The risk of CVD was significantly higher in those taking statins with an LDL-C level ≥ 70 mg/dL.

Predicting the risk of stroke among patients with type 2 diabetes: a systematic review and meta-analysis of C-statistics

OBJECTIVE

Stroke is a major cause of disability and death worldwide. People with diabetes are at a twofold to fivefold increased risk for stroke compared with people without diabetes. This study systematically reviews the literature on available stroke prediction models specifically developed or validated in patients with diabetes and assesses their predictive performance through meta-analysis.

DESIGN

Systematic review and meta-analysis.

DATA SOURCES

A detailed search was performed in MEDLINE, PubMed and EMBASE (from inception to 22 April 2019) to identify studies describing stroke prediction models.

ELIGIBILITY CRITERIA

All studies that developed stroke prediction models in populations with diabetes were included.

DATA EXTRACTION AND SYNTHESIS

Two reviewers independently identified eligible articles and extracted data. Random effects meta-analysis was used to obtain a pooled C-statistic.

RESULTS

Our search retrieved 26 202 relevant papers and finally yielded 38 stroke prediction models, of which 34 were specifically developed for patients with diabetes and 4 were developed in general populations but validated in patients with diabetes. Among the models developed in those with diabetes, 9 reported their outcome as stroke, 23 reported their outcome as composite cardiovascular disease (CVD) where stroke was a component of the outcome and 2 did not report stroke initially as their outcome but later were validated for stroke as the outcome in other studies. C-statistics varied from 0.60 to 0.92 with a median C-statistic of 0.71 (for stroke as the outcome) and 0.70 (for stroke as part of a composite CVD outcome). Seventeen models were externally validated in diabetes populations with a pooled C-statistic of 0.68.

CONCLUSIONS

Overall, the performance of these diabetes-specific stroke prediction models was not satisfactory. Research is needed to identify and incorporate new risk factors into the model to improve models' predictive ability and further external validation of the existing models in diverse population to improve generalisability.