Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jan 23.
Published in final edited form as: Circulation. 2023 Nov 11;149(4):293–304. doi: 10.1161/CIRCULATIONAHA.123.067530

Optimal screening for predicting and preventing the risk of heart failure among adults with diabetes without atherosclerotic cardiovascular disease: a pooled cohort analysis

Kershaw V Patel 1,*, Matthew W Segar 2,*, David C Klonoff 3, Muhammad Shahzeb Khan 4, Muhammad Shariq Usman 5, Carolyn S P Lam 6, Subodh Verma 7, Andrew P DeFilippis 8, Khurram Nasir 1, Stephan J L Bakker 9, B Daan Westenbrink 10, Robin P F Dullaart 9, Javed Butler 11,12, Muthiah Vaduganathan 13, Ambarish Pandey 5
PMCID: PMC11257100  NIHMSID: NIHMS1950557  PMID: 37950893

Abstract

Background:

The optimal approach to identify individuals with diabetes who are at high-risk for developing heart failure (HF) to inform implementation of preventive therapies is unknown, especially in those without atherosclerotic cardiovascular disease (ASCVD).

Methods:

Adults with diabetes and no HF at baseline from 7 community-based cohorts were included. Participants without ASCVD who were at high-risk for developing HF were identified using 1-step screening strategies: risk score (WATCH-DM ≥12); N-terminal pro-B-type natriuretic peptide (NT-proBNP; ≥125 pg/mL); high-sensitivity cardiac troponin (hs-cTnT ≥14 ng/L, hs-cTnI ≥31 ng/L); echocardiography-based diabetic cardiomyopathy (echo-DbCM; LA enlargement, LV hypertrophy, or diastolic dysfunction). High-risk participants were also identified using 2-step screening strategies with a second test to additionally identify residual risk among those deemed low-risk by the first test: WATCH-DM/NT-proBNP, NT-proBNP/hs-cTn, NT-proBNP/echo-DbCM. Across screening strategies, the proportion of HF events identified, 5-year number needed to treat (NNT5) and number needed to screen (NNS5) to prevent 1 HF event with an SGLT2i among high-risk participants, and cost of screening were estimated.

Results:

The initial study cohort included 6,293 participants (48.2% women), of which 77.7% without prevalent ASCVD were evaluated with different HF screening strategies. At 5-year follow-up, 6.2% of participants without ASCVD developed incident HF. The NNT5 to prevent 1 HF event with an SGLT2i among participants without ASCVD was 43 (95% CI, 29–72). In the cohort without ASCVD, high-risk participants identified using 1-step screening strategies had low NNT5 (22 for NT-proBNP to 37 for echo-DbCM). However, a substantial proportion of HF events occurred among participants identified as low-risk using 1-step screening approaches (29% for echo-DbCM to 47% for hs-cTn). 2-step screening strategies captured most HF events (75–89%) in the high-risk subgroup with a comparable NNT5 as the 1-step screening approaches (30–32). The NNS5 to prevent 1 HF event was similar across 2-step screening strategies (45–61). However, the number of tests and associated costs were lowest for WATCH-DM/NT-proBNP ($1,061) compared with other 2-step screening strategies (NT-proBNP/hs-cTn: $2,894; NT-proBNP/echo-DbCM: $16,358).

Conclusion:

Selective NT-proBNP testing based on the WATCH-DM score efficiently identified a high-risk primary prevention population with diabetes expected to derive marked absolute benefits from SGLT2i to prevent HF.

Keywords: biomarkers, heart failure, risk, type 2 diabetes mellitus

INTRODUCTION

Heart failure (HF) is one of the most common initial manifestations of cardiovascular disease (CVD) in diabetes, and its incidence is rising.1,2 Evidence from randomized clinical trials has established the efficacy of sodium-glucose cotransporter 2 inhibitors (SGLT2i) in reducing the risk of HF in diabetes, with greater absolute risk reductions observed among those with (vs. without) prevalent atherosclerotic CVD (ASCVD).3,4 Accordingly, consensus statements for diabetes management recommend SGLT2i for HF prevention in adults with prevalent ASCVD as well as those free of ASCVD who have elevated risk.5,6 The optimal risk stratification approach to identify individuals with diabetes and no prevalent ASCVD who have an elevated risk of HF and may benefit from early initiation of SGLT2i is not well-established.

An optimal HF risk stratification tool should be able to identify high- vs. low-risk individuals with a meaningful gradient in risk between the two groups. It should also capture the majority of HF events in the group identified as high-risk to limit the missed residual risk. Finally, the approach should be efficient and cost-effective to allow for broad implementation in clinical settings. Several strategies, ranging from clinical risk scores to cardiac biomarkers to echocardiography, have demonstrated good performance in identifying individuals with diabetes who are at an increased risk of HF.7–10 However, a comparison of different screening approaches, when used alone or in combination, to evaluate HF risk and inform the allocation of therapies such as SGLT2i has not been performed. Thus, the optimal screening strategy for HF in a primary prevention population with diabetes is unknown.

In the present study, we evaluated the performance of different screening approaches to identify adults with diabetes but no prevalent ASCVD who are at high risk for developing HF and may benefit from SGLT2i. Based on prior work, we hypothesized that the yield of cardiac biomarker testing would be improved with a diabetes-specific HF risk score to identify adults with diabetes who may derive greater absolute benefits from SGLT2i.11

METHODS

Deidentified data from the Atherosclerosis Risk in Communities (ARIC) study, Cardiovascular Health Study (CHS), Framingham Heart Study (FHS) Offspring Study and Third Generation Cohort, and Multi-Ethnic Study of Atherosclerosis (MESA) were obtained from the National Institute of Health Biologic Specimen and Data Repository Coordinating Center (BioLINCC). Deidentified data from the Chronic Renal Insufficiency Cohort (CRIC) study were obtained from the National Institute of Diabetes and Digestive and Kidney Disease Central Repository. Prevention of Renal and Vascular End-Stage Disease (PREVEND) study deidentified data were obtained from the study coordinating center. Data used in the present study will not be made available by the authors for reproducing the study results. Deidentified data can be obtained from data repositories and coordinating centers by submitting research proposals and obtaining appropriate approvals.

Pooled cohorts and study population

The present study included individual-level participant data pooled from the 7 epidemiological cohort studies described above. The design and protocol of each study cohort have been published previously and are described briefly in the Supplemental Methods.12–19 ARIC visit 5, FHS Offspring Cohort exam 6, and the baseline visits of CHS, CRIC, FHS Third Generation Cohort, MESA, and PREVEND were considered the baseline visits for the present analysis. The cohort for the present analysis included participants with diabetes who were at least 40 years of age and were free of HF at baseline. Exclusion criteria included missing HF status, missing outcomes, or >20% missing baseline data for key covariates (see Table S1). Each participant provided written informed consent at the time of enrollment in the primary study. The present analysis was considered exempt from Institutional Review Board approval at the University of Texas Southwestern Medical Center in Dallas.

Definition of diabetes and prevalent ASCVD status

Diabetes was defined using established criteria as described previously.12–19 Specifically, diabetes was defined as either 1) fasting plasma glucose (FPG) ≥126 mg/dL, 2) self-reported diagnosis of diabetes, or 3) self-reported use of anti-hyperglycemic medication. Additionally, in the ARIC cohort, diabetes was also defined by the presence of random plasma glucose ≥200 mg/dL. Due to limited availability across study cohorts, hemoglobin A1c (HbA1c) was not included in the definition of diabetes for the primary analysis. A similar definition of diabetes was used in prior pooled cohort analyses.10,11 The presence of ASCVD was defined as a prior history of non-fatal myocardial infarction (MI), coronary revascularization, or stroke (ischemic or hemorrhagic). History of MI was defined based on self-report or hospitalization for MI. Coronary revascularization was defined as either percutaneous coronary intervention or coronary artery bypass graft surgery (CABG).

Clinical covariates

Participants from each of the 7 cohorts underwent detailed examinations using standardized protocols as described in the Supplemental Methods.12–19 Demographic data (age, sex, race/ethnicity) and smoking history were obtained from self-reported questionnaires. Study-specific protocols were used to measure systolic and diastolic blood pressure (BP), weight, and height. Body mass index (BMI) was calculated as the ratio of weight in kilograms and height in meters squared. Serum glucose, creatinine, and lipids were measured according to study-specific protocols. A standard 12-lead electrocardiogram (ECG) was obtained as part of each study cohort protocol. Missing data for covariates with <20% missingness were imputed using random forest imputation.20

HF risk screening strategies

Among participants without ASCVD, different screening strategies were applied to identify high-risk individuals, as shown below.

1-step screening strategies to assess HF risk

WATCH-DM risk score

The WATCH-DM score derivation and validation details have been previously published.11,21,22 Briefly, the WATCH-DM score was developed and validated to predict the 5-year risk of incident HF among adults with diabetes. The WATCH-DM score incorporates 10 readily available clinical, laboratory, and ECG parameters, including 7 continuous variables (age, BMI, systolic BP, diastolic BP, serum creatinine, high-density lipoprotein [HDL] cholesterol, and FPG) and 3 binary variables (QRS >120 ms, history of MI, history of CABG). An alternative WATCH-DM model without an ECG parameter and incorporating HbA1c rather than FPG has also been validated.22 A WATCH-DM score ≥12 was defined as elevated and identified individuals at high risk for developing HF based on the optimal level for predicting HF risk in the present study population assessed by Youden’s index.11 Individuals with a WATCH-DM score <12 were identified as low risk.

Natriuretic peptide assessment

B-type natriuretic peptide (BNP) and N-terminal pro-BNP (NT-proBNP) were measured using well-established assays at the baseline visit for the present study (Supplemental Methods). BNP was measured using standard assays in the CRIC study and FHS Offspring Study.23–25 NT-proBNP was measured in all other cohorts (ARIC, CHS, FHS Third Generation Cohort, MESA, PREVEND).26–30 To facilitate comparison across natriuretic peptide assays, BNP values were transformed into Z-score equivalent NT-proBNP values as previously described.31 Based on an established threshold, NT-proBNP ≥125 pg/mL (or Z-score equivalent) was used to define high-risk individuals, and NT-proBNP <125 pg/mL identified low-risk individuals.5,32

Troponin assessment

High-sensitivity assays were used to quantify cardiac troponin levels (hs-cTn) across cohorts. Hs-cTn was not available in the CHS data available from BioLINCC. Hs-cTnT was measured in all cohorts using commercially available assays26,29,30,33 except for FHS Offspring Cohort and Third Generation Cohort, which assessed hs-cTnI.25,34 Consistent with prior literature, individuals were considered high risk for developing HF based on a hs-cTnT ≥14 ng/L or hs-cTnI ≥31 ng/L.9,35,36

Echocardiography-based diabetic cardiomyopathy

Diabetic cardiomyopathy is characterized by the presence of subclinical abnormalities in cardiac structure or function.10 Consistent with prior literature, echocardiography-based diabetic cardiomyopathy (echo-DbCM) was assessed and defined based on the presence of one of the following: left atrial enlargement, left ventricular hypertrophy, or diastolic dysfunction.10 Additional details regarding the definition of diastolic dysfunction used in the present study are provided in the Supplemental Methods. The presence of echo-DbCM identified individuals at high risk for developing HF. Echocardiographic data to evaluate echo-DbCM were available for participants from ARIC, CHS, and CRIC cohorts.

2-step screening strategies to assess HF risk

The 2-step screening strategies were based on combining NT-proBNP testing with other screening strategies to identify high-risk individuals as described below. In the 2-step screening strategies, the assessments were performed sequentially such that all participants were screened by the first test and only individuals who were not identified as high-risk for HF based on the initial test underwent subsequent assessment with the second test (Figure 1).

Figure 1. Sequential testing in the 2-step screening strategies.

Figure 1.

Open in a new tab

The 2-step screening strategies were based on combining two tests as needed. Participants identified as high risk for developing heart failure based on the first test do not undergo further testing. A second test is performed only for participants who do not identify as high risk based on the first test.

WATCH-DM / NT-proBNP

Based on previously published literature, HF risk was assessed based on a 2-step approach incorporating the WATCH-DM score assessment in all participants, followed by consideration of NT-proBNP testing among those with a low WATCH-DM score.11 High-risk individuals were identified based on the presence of a high WATCH-DM score or elevated NT-proBNP among those with a low WATCH-DM risk score.

NT-proBNP / hs-cTn

This multi-marker approach included an assessment of NT-proBNP in all participants, followed by an assessment of hs-cTn among those with low NT-proBNP. High-risk individuals for HF were identified based on elevated NT-proBNP or elevated hs-cTn among those with low NT-proBNP.

NT-proBNP / echo-DbCM

In this multi-modality approach, NT-proBNP was measured in all participants, followed by echocardiography among those with low NT-proBNP. Individuals were identified as high risk for developing HF based on elevated NT-proBNP or the presence of echo-DbCM if NT-proBNP was low.

Outcome of interest

The primary outcome of interest for the present study was incident HF. Each cohort adjudicated HF events using previously described, well-adjudicated protocols as detailed in the Supplemental Methods. An incident HF event was defined as hospitalization for HF, outpatient HF, or HF-associated death. Most adjudicated incident HF events across included cohorts were based on a hospitalization for HF, which is consistent with adjudicated outcomes in contemporary SGLT2i trials.3 To allow for comparable follow-up cohorts, outcomes were censored at 5 years.

Statistical analysis

Individual-level participant data were pooled and harmonized across all 7 cohorts as described in the Supplemental Methods. Baseline characteristics were displayed as mean (standard deviation) and number (percent) for continuous and categorical variables, respectively.

The primary analysis evaluating HF risk screening strategies included participants without ASCVD. In the cohort without ASCVD at baseline, the 5-year cumulative incidence and Kaplan-Meier (KM) estimate for incident HF was assessed. The 5-year number needed to treat (NNT5) to prevent 1 HF event with an SGLT2i was estimated assuming a 37% (95% CI, 20–50%) relative risk reduction in HF based on the published meta-analysis by McGuire et al.3 For reference, the NNT5 to prevent 1 HF event with an SGLT2i among participants with ASCVD was estimated based on the corresponding HF risk and assuming a 30% (95% CI, 22–38%) relative risk reduction in HF.3 The NNT5 with an SGLT2i to prevent 1 HF event was calculated as the inverse of the absolute risk difference based on the relative risk reduction and 5-year KM estimates as previously described.37

In the subgroup without ASCVD, participants were categorized into high- vs. low-risk groups based on the different 1- and 2-step screening strategies discussed above. Across high- and low-risk groups, the characteristics of participants were compared using one-way ANOVA for continuous variables and a chi-square test for categorical variables. Cumulative incidence curves and log-rank tests were used to evaluate the 5-year unadjusted risk of HF across high- vs. low-risk groups. The NNT5 to prevent 1 HF event with an SGLT2i was calculated across high- vs. low-risk categories as described above. Bootstrapping with 2,000 replicates was performed to obtain the standard errors and 95% CI around the prevalence of high- and low-risk groups by each screening strategy as well as the estimated effect of an SGLT2i. The proportions (95% CI) of overall HF events that were captured by the high- and low-risk subset of each screening strategy were also calculated.

The 5-year number needed to screen (NNS5) to prevent 1 HF event was calculated for screening strategies by dividing the NNT5 by the prevalence of high-risk individuals for that specific strategy in the population without ASCVD.38 For 2-step screening strategies, the number of secondary tests (serum biomarkers or echocardiograms) needed to prevent 1 HF event by treatment of high-risk individuals with an SGLT2i was estimated based on the NNS5 and the prevalence of low-risk individuals identified by the initial screening tool (WATCH-DM or NT-proBNP, see Figure 1) using a bootstrapping approach. The cost of 1- and 2-step screening strategies incorporating biomarkers and/or echocardiograms was estimated based on the total number of tests needed to prevent 1 HF event by treatment of high-risk individuals with an SGLT2i as well as the Centers for Medicare and Medicaid Services Clinical Laboratory Fee Schedule39 and Hospital Outpatient Prospective Payment System (Table S2).40

To evaluate the utility of screening strategies across different comorbidity burden strata, subgroup analyses were performed to evaluate the performance of different screening strategies across strata of comorbidity burden (0–1 vs. ≥2 comorbid risk factors: hypertension, obesity [BMI ≥30 kg/m2], and chronic kidney disease [CKD, estimated glomerular filtration rate <60 mL/minute/1.73 m2 using the CKD-epi equation41]). Sensitivity analysis was also performed for select 1- and 2-step screening strategies using a BMI-specific high-risk threshold for NT-proBNP (BMI <30 kg/m2: NT-proBNP ≥125 pg/mL [or Z-score equivalent], BMI ≥30 kg/m2: NT-proBNP ≥100 pg/mL [or Z-score equivalent]) and an alternative, previously validated version of the WATCH-DM score without an ECG parameter and replacing FPG with HbA1c among participants with available HbA1c data .10,22 All statistical analyses were performed using R 4.0.3 (R Foundation, Vienna, Austria) with a two-sided p-value <0.05 indicating statistical significance.

RESULTS

Among 7,643 participants with diabetes, the present study included 6,293 participants (Figure S1). Compared with participants who were included, those who were excluded due to missing data >20% for key covariates (n = 511) were more commonly Black, had higher BMI, and had a greater prevalence of ASCVD (34.6% vs. 22.3%) (Table S3). Among included study participants, missingness in covariates of interest was most commonly observed for smoking (19%), natriuretic peptides (16%), and lipid parameters (11–13%) (Table S4).

In the present study, 4,889 (77.7%) participants had no ASCVD. The baseline characteristics of participants without ASCVD and those with prevalent ASCVD are shown in Table S5. Among participants without ASCVD, approximately one-half were women (51.5%), one-quarter were Black (28.3%), and 5.3% had atrial fibrillation reported at baseline.

Over 5-year follow-up, there was a total of 301 HF events (6.2%) among participants without ASCVD, which accounted for 62.8% of all HF events in the study. The NNT5 for prevention of 1 HF event with an SGLT2i was 43 (95% CI, 29–72) among participants without ASCVD. In the reference group with ASCVD, the 5-year cumulative incidence of HF was 12.7% and the NNT5 for prevention of 1 HF event with an SGLT2i was 24 (95% CI, 18–34).

Performance of 1-step screening strategies among individuals without ASCVD

Among individuals without ASCVD, 1-step screening with the WATCH-DM score identified 47.7% of participants as high risk for incident HF (Table S6). In contrast, a biomarker-based approach identified 34.9% of participants as high-risk using NT-proBNP and 27.1% of those as high-risk using hs-cTn. An echocardiographic assessment identified 60.9% of individuals as high risk by echo-DbCM criteria. Participants identified as high-risk using the WATCH-DM score or NT-proBNP levels were older and had a greater prevalence of HF risk factors than the corresponding low-risk groups (Table 1, Table S7).

Table 1.

Baseline characteristics of participants without a history of ASCVD stratified by WATCH-DM

Low WATCH-DM
(n = 2,557)		 High WATCH-DM
(n = 2,332)		 P-value
Age, years 64.7 (10.2) 70.7 (8.6) <0.001
Men 1,117 (43.7) 1,252 (53.7) <0.001
Black 628 (24.6) 757 (32.5) <0.001
Systolic BP, mm Hg 132 (19) 138 (22) <0.001
Diastolic BP, mm Hg 72 (11) 69 (12) <0.001
BMI, kg/m2 29.5 (5.4) 32.9 (7.4) <0.001
Current smoking 304 (11.9) 134 (5.7) <0.001
Hypertension 1,919 (75.0) 2,161 (92.7) <0.001
Anti-hypertensive medication 1,663 (65.0) 1,955 (83.8) <0.001
Statin 906 (35.4) 1,190 (51.0) <0.001
Serum creatinine, mg/dL 1.0 (0.4) 1.4 (0.7) <0.001
HDL cholesterol, mg/dL 49 (14) 45 (11) <0.001
LDL cholesterol, mg/dL 116 (38) 106 (35) <0.001
Total cholesterol, mg/dL 196 (45) 185 (43) <0.001
Triglycerides, mg/dL 161 (91) 168 (106) 0.01
Fasting plasma glucose, mg/dL 141 (51) 150 (59) <0.001
QRS, ms 87 (18) 99 (23) <0.001
High NT-proBNP 679 (26.6) 1,029 (44.1) <0.001
High hs-cTn* 331 (15.8) 753 (39.5) <0.001
Echo-DbCM(+)* 818 (55.7) 1,007 (65.9) <0.001
High WATCH-DM or high NT-proBNP 679 (26.6) 2,332 (100) <0.001
High NT-proBNP or high hs-cTn 900 (35.2) 1,477 (63.3) <0.001
High NT-proBNP or echo-DbCM(+)* 984 (67.0) 1,207 (78.9) <0.001
Cohort
 ARIC 583 (22.8) 622 (26.7) <0.001
 CHS 458 (17.9) 424 (18.2)
 CRIC 496 (19.4) 906 (38.9)
 FHS Third Generation Cohort 79 (3.1) 3 (0.1)
 FHS Offspring Study 109 (4.3) 50 (2.1)
 MESA 589 (23.0) 260 (11.1)
 PREVEND 243 (9.5) 67 (2.9)
Open in a new tab

Data are presented as mean (standard deviation) and number (percent) for continuous and categorical variables, respectively. Comparisons across groups were performed using one-way ANOVA and chi-square test for continuous and categorical variables, respectively.

*

Data were available to evaluate high hs-cTn and echo-DbCM among subsets of participants (n = 4,007 and n = 2,998, respectively).

Abbreviations: ARIC, Atherosclerosis Risk in Communities Study; ASCVD, atherosclerotic cardiovascular disease; BMI, body mass index; BP, blood pressure; CHS, Cardiovascular Health Study; CRIC, Chronic Renal Insufficiency Cohort Study; CVD, cardiovascular disease; Echo-DbCM(+), echocardiogram-based diabetic cardiomyopathy; FHS, Framingham Heart Study; HDL, high-density lipoprotein; hs-cTn, high-sensitivity cardiac troponin; LDL, low-density lipoprotein; MESA, Multi-Ethnic Study of Atherosclerosis; NT-proBNP, N-terminal pro-B-type natriuretic peptide; PREVEND, Prevention of Renal and Vascular End-Stage Disease

The high-risk subgroup identified by each 1-step screening strategy had a higher cumulative incidence of HF than the corresponding low-risk group (Figure S2). Among the 1-step screening strategies, the gradient in 5-year risk of HF among participants identified as high vs. low risk was largest for NT-proBNP (5-year KM estimate: 13.1% vs. 3.8%, respectively) and smallest for echo-DbCM (5-year KM estimate: 8.1% vs. 5.01%, respectively) (p-value <0.01 for all). The NNT5 to prevent 1 HF event using an SGLT2i was lowest among high-risk participants identified using NT-proBNP and was similar using other 1-step screening strategies and comparable to that observed among those with ASCVD (Figure 2A). The NNT5 for prevention of 1 HF event with an SGLT2i was at least two-fold higher in the low- versus high-risk groups based on 1-step screening strategies with a risk score (WATCH-DM) and biomarkers (NT-proBNP and hs-cTn). However, the high-risk groups identified using each of the 1-step screening strategies accounted for only 53–71% of HF events (Figure 3). Thus, approximately 30–50% of HF events in the population without prevalent ASCVD occurred in the low-risk groups.

Figure 2. Five-year number needed to treat for prevention of an incident HF event using the treatment effect of an SGLT2i among participants with no history of ASCVD stratified by 1- (Panel A) and 2-step screening strategies (Panel B).

Figure 2.

Open in a new tab

WATCH-DM: high risk = score ≥12, low risk = score <12; NT-proBNP: high risk ≥125 pg/mL, low risk <125 pg/mL; Hs-cTn: high-risk ≥14 ng/L for hs-cTnT or ≥31 ng/L for hs-cTnI, low-risk <14 ng/L for hs-cTnT or <31 ng/L for hs-cTnI; Echo-DbCM: high-risk = left atrial enlargement, left ventricular hypertrophy, or diastolic dysfunction, low-risk = no left atrial enlargement, left ventricular hypertrophy, and diastolic dysfunction; WATCH-DM / NT-proBNP: high risk = score ≥12 or NT-proBNP ≥125 pg/mL, low-risk = score <12 and NT-proBNP <125 pg/mL; NT-proBNP / hs-cTn: high-risk = NT-proBNP ≥125 pg/mL or hs-cTnT ≥14 ng/L or hs-cTnI ≥31 ng/L, low-risk = NT-proBNP <125 pg/mL and hs-cTn <14 ng/L and hs-cTnI <31 ng/L; NT-proBNP / echo-DbCM: high risk = NT-proBNP ≥125 pg/mL or left atrial enlargement, left ventricular hypertrophy, or diastolic dysfunction, low risk = NT-proBNP <125 pg/mL and no left atrial enlargement, left ventricular hypertrophy, and diastolic dysfunction.

The black dotted line represents the upper end of the 95th percentile of the NNT5 to prevent 1 HF event with an SGLT2i among participants with history of ASCVD (n = 34). The error bars represent the standard error calculated using the 95% confidence interval.

Abbreviations: ASCVD, atherosclerotic cardiovascular disease; echo-DbCM, echocardiogram-based diabetic cardiomyopathy; HF, heart failure; hs-cTn, high-sensitivity cardiac troponin; NNT5, number needed to treat for 5 years; NT-proBNP, N-terminal pro-B-type natriuretic peptide; SGLT2i, sodium-glucose cotransporter 2 inhibitor.

Figure 3. Proportion of HF events identified among high- and low-risk participants with no history of ASCVD stratified by 1- and 2-step screening strategies.

Figure 3.

Open in a new tab

See Figure 2 legend for description of groups and abbreviations. Data are presented as percentage (95% confidence interval).

Performance of 2-step screening strategies among individuals without ASCVD

The general approach to 2-step screening strategies for identifying high-risk participants based on two sequential tests is shown in Figure 1. High WATCH-DM score or elevated NT-proBNP among participants with low WATCH-DM scores identified 61.6% of participants as high risk (Table S6). Similarly, elevated NT-proBNP levels as the initial test followed by elevated hs-cTn or presence of echo-DbCM among those with low NT-proBNP levels identified 48.7% and 73.1% of individuals as high risk, respectively. The risk of incident HF was 3.0- to 3.6-fold higher in the high- vs. low-risk groups based on 2-step screening strategies (Figure S3). The NNT5 to prevent 1 HF event using an SGLT2i among high-risk individuals without ASCVD identified by each of the 2-step screening strategies was similar and comparable to that observed among individuals with ASCVD (Figure 2B). Furthermore, the high-risk groups identified using 2-step screening strategies accounted for most HF events (75–89%), with only ~1–2 in 10 HF events occurring among low-risk participants (Figure 3).

NNS5 and cost of screening to prevent 1 HF event using 1- and 2-step screening strategies.

The NNS5 to prevent 1 HF event by treatment of high-risk individuals with an SGLT2i was lower in 2- versus 1-step screening approaches (45–61 versus 60–93, respectively) (Figure 4, Figure S4). However, the number of screening tests (biomarkers or echocardiograms) needed to prevent 1 HF event varied substantially across different strategies. Among 2-step screening strategies, the risk score-based strategy (WATCH-DM / NT-proBNP) had the lowest number of diagnostic tests needed (27 NT-proBNP tests to prevent 1 HF event) followed by the multi-modality strategy (NT-proBNP / echo-DbCM: 74 tests [45 NT-proBNP tests and 29 echocardiograms among those without elevated NT-proBNP]) and multi-marker strategy (NT-proBNP / hs-cTn: 101 biomarker tests [61 NT-proBNP tests and 40 hs-cTn tests among those without elevated NT-proBNP]) (Table S8). The cost of screening to prevent 1 HF event was also lowest for the risk score-based strategy using initial WATCH-DM assessment followed by NT-proBNP testing among those with a low WATCH-DM score ($1,061 per 1 HF event prevented) (Figure 4). In contrast, the cost of screening to prevent 1 HF event was more than double for the multi-marker strategy ($2,894) and highest in the multi-modality strategy incorporating biomarker and echocardiography testing ($16,358).

Figure 4. Five-year number needed to screen for prevention of an incident HF event using the treatment effect of an SGLT2i and additional cost of testing stratified by 2-step screening strategies.

Figure 4.

Open in a new tab

See Figure 2 legend for description of groups and abbreviations. Test quantity and cost are shown in Tables S2, S8. The error bars represent the standard error calculated using the 95% confidence interval.

Subgroup and sensitivity analyses

In subgroup analysis, 2,013 participants (41.2%) had low (0–1) and 2,876 participants (58.8%) had high (≥2) burden of comorbid risk factors. Across both comorbidity burden strata, a large gradient was noted in NNT5 to prevent 1 HF event using an SGLT2i among individuals identified as high vs. low risk using the proposed screening strategies (Figure S5). For example, in the low comorbidity burden strata, there was a ~6-fold gradient in the NNT5 among individuals identified as high vs. low risk using the WATCH-DM / NT-proBNP 2-step screening strategy (32 vs. 180). Similarly, in the high comorbidity strata, the NNT5 was ~2.5-fold lower among individuals identified as high vs. low risk using the WATCH-DM / NT-proBNP 2-step screening strategy (31 vs. 78). The NNT5 for high-risk individuals using the WATCH-DM / NT-proBNP 2-step screening strategy was similar across both low vs. high-comorbidity strata.

In sensitivity analysis using BMI-specific NT-proBNP thresholds, the performance of the NT-proBNP-based screening strategies in identifying high vs. low HF risk individuals was similar to that observed using a single NT-proBNP cutoff (Figure S6, Figures 2 and 4). Similarly, in the sensitivity analysis, using an alternative, previously validated version of WATCH-DM without an ECG parameter and using HbA1c demonstrated comparable risk stratification performance as the original WATCH-DM score with QRS duration and FPG (Figure S7, Figures 2 and 4).

DISCUSSION

In this large, pooled analysis of individuals with diabetes, we observed several important findings. First, among adults with diabetes and no history of ASCVD, 1-step screening strategies demonstrated good risk stratification performance, with elevated NT-proBNP identifying the highest risk subgroup. The NNT5 for prevention of 1 HF event with an SGLT2i among high-risk individuals identified by each 1-step screening strategy was comparable to that observed among individuals with established ASCVD. However, despite adequate risk stratification, a large proportion of HF events occurred in low-risk individuals identified based on 1-step screening strategies. Second, a 2-step screening approach that combined NT-proBNP testing with other risk assessment tools was better able to identify high-risk individuals without ASCVD at baseline, accounting for ~85% of all HF events. Furthermore, the NNT5 and NNS5 to prevent 1 HF event with SGLT2i in high-risk individuals based on 2-step screening strategies were low. Finally, among the 2-step screening strategies, sequential use of the WATCH-DM score followed by selective NT-proBNP testing among individuals with a low WATCH DM score was the most efficient strategy, requiring the fewest number of tests and lowest screening cost to prevent 1 HF event.

Prior studies have demonstrated the utility of several risk screening strategies to identify individuals with diabetes who are at an increased risk for HF development.7–10 These strategies include HF risk scores, cardiac biomarkers, and echocardiographic assessment to identify those with subclinical cardiomyopathy.7–10 Furthermore, multi-specialty consensus reports recommend using NT-proBNP and hs-cTn to screen for high-risk individuals to identify those likely to develop HF.5,32 However, each strategy has its limitations. For example, clinical risk scores, while being generalizable and an inexpensive approach, may not capture all high-risk individuals as these do not account for the presence of subclinical CVD that is better captured by cardiac biomarkers and non-invasive imaging. Similarly, while models incorporating cardiac biomarkers have good risk discrimination and calibration, measuring biomarkers in all individuals with diabetes is logistically challenging and likely cost-prohibitive, especially echocardiography.8 Furthermore, elevated NT-proBNP, a currently recommended screening strategy, may not identify individuals with obesity who are at increased risk of developing HF due to lower NT-proBNP levels with increased BMI.5,10,32 The present study addresses these gaps in HF risk screening by evaluating the usefulness of different 1- and 2-step screening strategies to inform the allocation of SGLT2i for HF prevention.

We observed that while 1-step screening strategies were effective in identifying high-risk individuals for HF, a substantial proportion of HF events occurred among individuals classified as low-risk. Thus, a risk-based approach to prevention that allocates effective but costly preventive therapies to individuals identified as high-risk using these screening strategies may miss a large proportion of preventable HF events. These observations suggest that 1-step screening strategies may not be sufficient to inform HF prevention therapies at the population level. Thus, there is a need for a complementary screening strategy that can better capture HF risk among individuals with diabetes. To this end, we observed that a 2-step approach that incorporates complementary prognostic information from a clinical risk score and NT-proBNP can better identify high-risk individuals who account for nearly 85% of future HF events. Furthermore, such an approach is also efficient and potentially cost-effective and can be implemented in electronic health record (EHR) -based contemporary clinical practices to inform a risk-based approach to HF risk screening in diabetes.

The WATCH-DM score includes traditional risk factors such as BP, BMI, and glucose which are each associated with incident HF risk.21 However, these well-established clinical risk factors do not fully explain the development of all HF events as evidenced by elevated risk of HF among individuals with control of these parameters.42 The addition of NT-proBNP to the WATCH-DM score improves HF risk prediction and similar findings have been observed for this combination approach using other risk scores.11,43 The enhanced prognostic utility of NT-proBNP in low- versus high-risk individuals may be related to the identification of a distinct pathological mechanism not identified among individuals with a low WATCH-DM score. The present study highlights the utility of combining markers from multiple assessment tools that capture different pathophysiological processes to minimize the residual risk of HF.

A 2-step approach to HF risk assessment using the WATCH-DM score followed by selective NT-proBNP testing, if needed, results in lower costs than multi-marker and multi-modality 2-step screening strategies. The WATCH-DM score integrates clinical data that is readily available in the EHR as part of standard care for patients with diabetes, such as demographics, vitals, and recommended laboratory tests, including kidney function and lipid measurements.21 Leveraging a machine learning-based approach, the WATCH-DM score is less sensitive to missingness in specific variables and can be programmed to be used in different clinical settings. Using data from the EHR, the WATCH-DM score can be calculated at no additional cost beyond routine practice, was previously implemented and validated in an EHR, and is available online for public use (https://www.cvriskscores.com/about.html).22 Additionally, a high WATCH-DM score identified one-half of individuals as high-risk for developing HF, and this subgroup could likely forego NT-proBNP testing as this would unlikely meaningfully change risk estimates and, thus, management recommendations. Similar to WATCH-DM, other risk scores such as RECODE, BRAVO, TRS-HFDM, QRISK, and DM-CURE can also potentially be implemented in the EHR and combined with biomarker testing, as suggested by our 2-step screening strategy to identify individuals at high risk for developing HF.7,44

Our study findings have important implications. Over the past decades, HF prevention in diabetes has received considerably less attention than preventing ASCVD complications.32 A risk-based approach has been associated with greater use of atherosclerotic CVD preventive therapies, and a similar methodology can be translated to HF prevention with SGLT2i.45 Matching effective but expensive preventive therapies to the highest-risk individuals who are mostly likely to benefit would be an efficient and cost-effective strategy for HF prevention. Individuals without ASCVD identified as high risk for HF using screening strategies evaluated in the present study had comparable NNT5 for HF prevention as those with prevalent ASCVD. Based on the comparable NNT5 and a small proportion of missed HF events occurring in the low-risk group, the 2-step strategy is an effective approach for screening. Furthermore, these risk stratification approaches were also effective in de-risking by identifying low-risk individuals for developing HF who may derive less benefit from costly therapies such as SGLT2i. HF risk assessment incorporating a clinical risk score followed by selective biomarker testing may help inform shared decision-making discussions with individuals at high and low risk for developing HF. Regarding biomarker testing in a primary care setting, it is likely that natriuretic peptide testing in combination with HF risk scores, as demonstrated in the 2-step screening strategy, may be practical and cost-effective. Future prospective studies are needed to evaluate the clinical and cost-effectiveness of such 2-step strategies incorporating a diabetes-specific HF risk score with selective biomarker testing for prevention of HF in diabetes and potentially other populations such as individuals with a recent MI.

Limitations

Several limitations of our study should be considered. First, the study population included older adults with a high burden of comorbidities, and findings from the present study may not be generalizable to other populations. Second, the present study was a pooled cohort analysis, and there may be differences in cohort and participant characteristics. However, data were harmonized using a comprehensive strategy for categorical and continuous variables and adequate harmonization was confirmed. Third, the diabetes definition used in the primary analysis was based on FPG, which was consistently available across all pooled cohorts. HbA1c is commonly used to diagnose diabetes in contemporary clinical settings, and there is a possibility that some individuals with diabetes may have been missed using FPG. Fourth, the biomarker cutoffs for abnormal levels of hs-cTn and natriuretic peptides may vary across different cohorts. Future studies in external cohorts may consider cohort-specific biomarker cut-offs to identify high-risk individuals. Fifth, 1- and 2-step screening strategies evaluated in the present study do not consider other important prognostic factors in diabetes, such as diabetes duration and socioeconomic status.46,47 Sixth, HF events were defined primarily by hospitalization for HF, which may limit the generalizability of our findings for preventing the development of outpatient-only HF events. Finally, cost data presented in this analysis focused on the cost of screening to prevent 1 HF event and does not account for downstream costs of subsequent evaluation and management of individuals identified as high risk.

Conclusions

Among community-dwelling adults with diabetes who were free of CVD, screening with a clinical risk score, cardiac biomarkers, and echocardiography identified individuals who were at elevated risk for HF. However, substantial residual risk of HF persists among those classified as low risk using these 1-step screening approaches. A 2-step screening strategy combining a clinical risk score with selective NT-proBNP testing among those deemed low risk based on the risk score may be an effective and cost-efficient approach to identify high-risk individuals for HF development who are more likely to benefit from SGLT2i therapy.

Supplementary Material

Supplemental Publication Material

CLINICAL PERSPECTIVE.

What is new?

  • Among adults with diabetes and no prevalent atherosclerotic cardiovascular disease (ASCVD), the 5-year number needed to treat to prevent one heart failure (HF) event using an SGLT2i in those identified as high-risk with a risk score, cardiac biomarkers, or echocardiography was comparable to those with diabetes and prevalent ASCVD.

  • Sequential use of a risk score followed by natriuretic peptide testing among those deemed low-risk by the risk score identified 84% of incident HF events, had a low 5-year number needed to screen, fewest tests, and lowest screening cost to prevent one HF event among 2-step screening strategies.

What are the clinical implications?

  • A 2-step screening strategy to assess HF risk incorporating a clinical risk score, such as WATCH-DM, with selective natriuretic peptide testing may help efficiently identify a high-risk primary prevention population with diabetes.

  • Future studies are needed to evaluate the clinical and cost-effectiveness of 2-step screening strategies to assess HF risk that combine a clinical risk score with selective cardiac biomarker testing among those deemed low risk using the risk score.

ACKNOWLEDGEMENTS

The authors thank the study participants, staff, and investigators of the Atherosclerosis Risk in Communities (ARIC) study, Cardiovascular Health Study (CHS), Chronic Renal Insufficiency Cohort (CRIC), Framingham Heart Study (FHS) Offspring Cohort and Third Generation Cohort, Multi-Ethnic Study of Atherosclerosis (MESA), and Prevention of Renal and Vascular End-Stage Disease (PREVEND) study.

SOURCES OF FUNDING

Dr. Patel and Dr. Pandey have received research support from the National Heart, Lung, and Blood Institute (R21HL169708).

DISCLOSURES

Dr. Patel has served as a consultant to Novo Nordisk. Dr. Segar has received honoraria from Merck. Dr. Khan has received personal fees from Merck. Dr. Lam has received research support from AstraZeneca, Bayer, Boston Scientific, and Roche Diagnostics; has served as a consultant or on advisory boards/steering committees/executive committees for Actelion, Alleviant Medical, Allysta, Amgen, ANaCardio AB, Applied Therapeutics, AstraZeneca, Bayer, Boehringer Ingelheim, Boston Scientific, Cytokinetics, Darma, EchoNous, Impulse Dynamics, Ionis Pharmaceutical, Janssen Research and Development, Medscape, Merck, Novartis, Novo Nordisk, Radcliffe Group, Roche Diagnostics, Sanofi, Siemens Healthcare Diagnostics, Us2.ai and WebMD Global; and has served as cofounder and non-executive director of Us2.ai. Dr. Verma holds a Tier 1 Canada Research Chair in Cardiovascular Surgery; has received research grants and honoraria from Amarin, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, HLS Therapeutics, Janssen, Novartis, Novo Nordisk, PhaseBio, and Pfizer; has received honoraria from Sanofi, Sun Pharmaceuticals, and the Toronto Knowledge Translation Working Group; is a member of the scientific excellence committee of the EMPEROR-Reduced trial (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure With Reduced Ejection Fraction); has served as a national lead investigator of the DAPA-HF and EMPEROR-Reduced trials; and is the president of the Canadian Medical and Surgical Knowledge Translation Research Group, a federally incorporated not-for-profit physician organization. Dr. Nasir is on the advisory board of Amgen, Novartis, Novo Nordisk, and his research is partly supported by the Jerold B. Katz Academy of Translational Research. Dr. Butler has received consulting fees from Boehringer Ingelheim, Cardior, CVRx, Foundry, G3 Pharma, Imbria, Impulse Dynamics, Innolife, Janssen, LivaNova, Luitpold, Medtronic, Merck, Novartis, Novo Nordisk, Relypsa, Roche, Sanofi, Sequana Medical, V-Wave Ltd, and Vifor. Dr. Vaduganathan is supported by the KL2/Catalyst Medical Research Investigator Training award from Harvard Catalyst (National Institutes of Health/National Center for Advancing Translational Sciences Award UL 1TR002541); and has served on advisory boards or has received research grant support from American Regent, Amgen, AstraZeneca, Baxter Healthcare, Bayer AG, Boehringer Ingelheim, Cytokinetics, and Relypsa. Dr. Pandey has received research support from the National Institute on Aging GEMSSTAR Grant (1R03AG067960-01) and the National Institute on Minority Health and Disparities (R01MD017529). Dr. Pandey has received grant funding (to the institution) from Applied Therapeutics, Gilead Sciences, Ultromics, Myovista, and Roche; has served as a consultant for and/or received honoraria outside of the present study as an advisor/consultant for Tricog Health Inc, Lilly USA, Rivus, Cytokinetics, Roche Diagnostics, Sarfez Therapeutics, Edwards Lifesciences, Merck, Bayer, Novo Nordisk, Alleviant, Axon Therapies, and has received nonfinancial support from Pfizer and Merck. Dr. Pandey is also a consultant for Palomarin Inc. with stocks compensation.

NONSTANDARD ABBREVIATIONS AND ACRONYMS

Echo-DbCM

echocardiography-based diabetic cardiomyopathy

HF

heart failure

Hs-cTn

high sensitivity cardiac troponin

NT-proBNP

N-terminal pro-B-type natriuretic peptide

SGLT2i

Sodium-glucose cotransporter 2 inhibitor

Footnotes

REFERENCES

  • 1.Shah AD, Langenberg C, Rapsomaniki E, Denaxas S, Pujades-Rodriguez M, Gale CP, Deanfield J, Smeeth L, Timmis A, Hemingway H. Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1.9 million people. Lancet Diabetes Endocrinol. 2015;3:105–113. doi: 10.1016/S2213-8587(14)70219-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Honigberg MC, Patel RB, Pandey A, Fonarow GC, Butler J, McGuire DK, Vaduganathan M. Trends in Hospitalizations for Heart Failure and Ischemic Heart Disease Among US Adults With Diabetes. Jama Cardiol. 2021;6:354–357. doi: 10.1001/jamacardio.2020.5921 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.McGuire DK, Shih WJ, Cosentino F, Charbonnel B, Cherney DZI, Dagogo-Jack S, Pratley R, Greenberg M, Wang S, Huyck S, et al. Association of SGLT2 Inhibitors With Cardiovascular and Kidney Outcomes in Patients With Type 2 Diabetes: A Meta-analysis. Jama Cardiol. 2021;6:148–158. doi: 10.1001/jamacardio.2020.4511 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mahaffey KW, Neal B, Perkovic V, de Zeeuw D, Fulcher G, Erondu N, Shaw W, Fabbrini E, Sun T, Li Q, et al. Canagliflozin for Primary and Secondary Prevention of Cardiovascular Events: Results From the CANVAS Program (Canagliflozin Cardiovascular Assessment Study). Circulation. 2018;137:323–334. doi: 10.1161/CIRCULATIONAHA.117.032038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, Deswal A, Drazner MH, Dunlay SM, Evers LR, et al. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2022;145:e895–e1032. doi: 10.1161/CIR.0000000000001063 [DOI] [PubMed] [Google Scholar]
  • 6.ElSayed NA, Aleppo G, Aroda VR, Bannuru RR, Brown FM, Bruemmer D, Collins BS, Das SR, Hilliard ME, Isaacs D, et al. 10. Cardiovascular Disease and Risk Management: Standards of Care in Diabetes-2023. Diabetes Care. 2023;46:S158–S190. doi: 10.2337/dc23-S010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Razaghizad A, Oulousian E, Randhawa VK, Ferreira JP, Brophy JM, Greene SJ, Guida J, Felker GM, Fudim M, Tsoukas M, et al. Clinical Prediction Models for Heart Failure Hospitalization in Type 2 Diabetes: A Systematic Review and Meta-Analysis. J Am Heart Assoc. 2022;11:e024833. doi: 10.1161/JAHA.121.024833 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Pandey A, Vaduganathan M, Patel KV, Ayers C, Ballantyne CM, Kosiborod MN, Carnethon M, DeFilippi C, McGuire DK, Khan SS, et al. Biomarker-Based Risk Prediction of Incident Heart Failure in Pre-Diabetes and Diabetes. JACC Heart Fail. 2021;9:215–223. doi: 10.1016/j.jchf.2020.10.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Berg DD, Wiviott SD, Scirica BM, Zelniker TA, Goodrich EL, Jarolim P, Mosenzon O, Cahn A, Bhatt DL, Leiter LA, et al. A Biomarker-Based Score for Risk of Hospitalization for Heart Failure in Patients With Diabetes. Diabetes Care. 2021;44:2573–2581. doi: 10.2337/dc21-1170 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Segar MW, Khan MS, Patel KV, Butler J, Tang WHW, Vaduganathan M, Lam CSP, Verma S, McGuire DK, Pandey A. Prevalence and Prognostic Implications of Diabetes With Cardiomyopathy in Community-Dwelling Adults. J Am Coll Cardiol. 2021;78:1587–1598. doi: 10.1016/j.jacc.2021.08.020 [DOI] [PubMed] [Google Scholar]
  • 11.Segar MW, Khan MS, Patel KV, Vaduganathan M, Kannan V, Willett D, Peterson E, Tang WHW, Butler J, Everett BM, et al. Incorporation of natriuretic peptides with clinical risk scores to predict heart failure among individuals with dysglycaemia. Eur J Heart Fail. 2022;24:169–180. doi: 10.1002/ejhf.2375 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Investigators ARIC. The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. Am J Epidemiol. 1989;129:687–702. [PubMed] [Google Scholar]
  • 13.Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263–276. doi: 10.1016/1047-2797(91)90005-w [DOI] [PubMed] [Google Scholar]
  • 14.Feldman HI, Appel LJ, Chertow GM, Cifelli D, Cizman B, Daugirdas J, Fink JC, Franklin-Becker ED, Go AS, Hamm LL, et al. The Chronic Renal Insufficiency Cohort (CRIC) Study: Design and Methods. J Am Soc Nephrol. 2003;14:S148–153. doi: 10.1097/01.asn.0000070149.78399.ce [DOI] [PubMed] [Google Scholar]
  • 15.Feinleib M, Kannel WB, Garrison RJ, McNamara PM, Castelli WP. The framingham offspring study. Design and preliminary data. Preventive Medicine. 1975;4:518–525. doi: 10.1016/0091-7435(75)90037-7 [DOI] [PubMed] [Google Scholar]
  • 16.Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979;110:281–290. doi: 10.1093/oxfordjournals.aje.a112813 [DOI] [PubMed] [Google Scholar]
  • 17.Splansky GL, Corey D, Yang Q, Atwood LD, Cupples LA, Benjamin EJ, D’Agostino RB Sr., Fox CS, Larson MG, Murabito JM, et al. The Third Generation Cohort of the National Heart, Lung, and Blood Institute’s Framingham Heart Study: design, recruitment, and initial examination. Am J Epidemiol. 2007;165:1328–1335. doi: 10.1093/aje/kwm021 [DOI] [PubMed] [Google Scholar]
  • 18.Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, Greenland P, Jacob DR Jr., Kronmal R, Liu K, et al. Multi-Ethnic Study of Atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–881. [DOI] [PubMed] [Google Scholar]
  • 19.Hillege HL, Janssen WM, Bak AA, Diercks GF, Grobbee DE, Crijns HJ, Van Gilst WH, De Zeeuw D, De Jong PE, Prevend Study G. Microalbuminuria is common, also in a nondiabetic, nonhypertensive population, and an independent indicator of cardiovascular risk factors and cardiovascular morbidity. J Intern Med. 2001;249:519–526. doi: 10.1046/j.1365-2796.2001.00833.x [DOI] [PubMed] [Google Scholar]
  • 20.Stekhoven DJ, Buhlmann P. MissForest--non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012;28:112–118. doi: 10.1093/bioinformatics/btr597 [DOI] [PubMed] [Google Scholar]
  • 21.Segar MW, Vaduganathan M, Patel KV, McGuire DK, Butler J, Fonarow GC, Basit M, Kannan V, Grodin JL, Everett B, et al. Machine Learning to Predict the Risk of Incident Heart Failure Hospitalization Among Patients With Diabetes: The WATCH-DM Risk Score. Diabetes Care. 2019;42:2298–2306. doi: 10.2337/dc19-0587 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Segar MW, Patel KV, Hellkamp AS, Vaduganathan M, Lokhnygina Y, Green JB, Wan SH, Kolkailah AA, Holman RR, Peterson ED, et al. Validation of the WATCH-DM and TRS-HF(DM) Risk Scores to Predict the Risk of Incident Hospitalization for Heart Failure Among Adults With Type 2 Diabetes: A Multicohort Analysis. J Am Heart Assoc. 2022;11:e024094. doi: 10.1161/JAHA.121.024094 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Anderson AH, Yang W, Hsu CY, Joffe MM, Leonard MB, Xie D, Chen J, Greene T, Jaar BG, Kao P, et al. Estimating GFR among participants in the Chronic Renal Insufficiency Cohort (CRIC) Study. Am J Kidney Dis. 2012;60:250–261. doi: 10.1053/j.ajkd.2012.04.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Wang TJ, Larson MG, Levy D, Leip EP, Benjamin EJ, Wilson PW, Sutherland P, Omland T, Vasan RS. Impact of age and sex on plasma natriuretic peptide levels in healthy adults. Am J Cardiol. 2002;90:254–258. doi: 10.1016/s0002-9149(02)02464-5 [DOI] [PubMed] [Google Scholar]
  • 25.Wang TJ, Wollert KC, Larson MG, Coglianese E, McCabe EL, Cheng S, Ho JE, Fradley MG, Ghorbani A, Xanthakis V, et al. Prognostic utility of novel biomarkers of cardiovascular stress: the Framingham Heart Study. Circulation. 2012;126:1596–1604. doi: 10.1161/CIRCULATIONAHA.112.129437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Jia X, Al Rifai M, Ndumele CE, Virani SS, de Lemos JA, Lee E, Shah AM, Echouffo-Tcheugui JB, Bozkurt B, Hoogeveen R, et al. Reclassification of Pre-Heart Failure Stages Using Cardiac Biomarkers: The ARIC Study. JACC Heart Fail. 2023;11:440–450. doi: 10.1016/j.jchf.2022.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Patton KK, Ellinor PT, Heckbert SR, Christenson RH, DeFilippi C, Gottdiener JS, Kronmal RA. N-terminal pro-B-type natriuretic peptide is a major predictor of the development of atrial fibrillation: the Cardiovascular Health Study. Circulation. 2009;120:1768–1774. doi: 10.1161/CIRCULATIONAHA.109.873265 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Fradley MG, Larson MG, Cheng S, McCabe E, Coglianese E, Shah RV, Levy D, Vasan RS, Wang TJ. Reference limits for N-terminal-pro-B-type natriuretic peptide in healthy individuals (from the Framingham Heart Study). Am J Cardiol. 2011;108:1341–1345. doi: 10.1016/j.amjcard.2011.06.057 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Seliger SL, Hong SN, Christenson RH, Kronmal R, Daniels LB, Lima JAC, de Lemos JA, Bertoni A, deFilippi CR. High-Sensitive Cardiac Troponin T as an Early Biochemical Signature for Clinical and Subclinical Heart Failure: MESA (Multi-Ethnic Study of Atherosclerosis). Circulation. 2017;135:1494–1505. doi: 10.1161/CIRCULATIONAHA.116.025505 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Brouwers FP, de Boer RA, van der Harst P, Voors AA, Gansevoort RT, Bakker SJ, Hillege HL, van Veldhuisen DJ, van Gilst WH. Incidence and epidemiology of new onset heart failure with preserved vs. reduced ejection fraction in a community-based cohort: 11-year follow-up of PREVEND. Eur Heart J. 2013;34:1424–1431. doi: 10.1093/eurheartj/eht066 [DOI] [PubMed] [Google Scholar]
  • 31.Anand IS, Claggett B, Liu J, Shah AM, Rector TS, Shah SJ, Desai AS, O’Meara E, Fleg JL, Pfeffer MA, et al. Interaction Between Spironolactone and Natriuretic Peptides in Patients With Heart Failure and Preserved Ejection Fraction: From the TOPCAT Trial. JACC Heart Fail. 2017;5:241–252. doi: 10.1016/j.jchf.2016.11.015 [DOI] [PubMed] [Google Scholar]
  • 32.Pop-Busui R, Januzzi JL, Bruemmer D, Butalia S, Green JB, Horton WB, Knight C, Levi M, Rasouli N, Richardson CR. Heart Failure: An Underappreciated Complication of Diabetes. A Consensus Report of the American Diabetes Association. Diabetes Care. 2022;45:1670–1690. doi: 10.2337/dci22-0014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Lidgard B, Zelnick LR, Go A, O’Brien KD, Bansal N, * CSI. Framingham and American College of Cardiology/American Heart Association Pooled Cohort Equations, High-Sensitivity Troponin T, and N-Terminal Pro-Brain-Type Natriuretic Peptide for Predicting Atherosclerotic Cardiovascular Events Across the Spectrum of Kidney Dysfunction. J Am Heart Assoc. 2022;11:e024913. doi: 10.1161/JAHA.121.024913 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Xanthakis V, Enserro DM, Larson MG, Wollert KC, Januzzi JL, Levy D, Aragam J, Benjamin EJ, Cheng S, Wang TJ, et al. Prevalence, Neurohormonal Correlates, and Prognosis of Heart Failure Stages in the Community. JACC Heart Fail. 2016;4:808–815. doi: 10.1016/j.jchf.2016.05.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Bhatia PM, Daniels LB. Highly Sensitive Cardiac Troponins: The Evidence Behind Sex-Specific Cutoffs. J Am Heart Assoc. 2020;9:e015272. doi: 10.1161/JAHA.119.015272 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.IFCC Committee on Clinical Applications of Cardiac Bio‐Markers. High-Sensitivity* Cardiac Troponin I and T Assay Analytical Characteristics Designated by Manufacturer IFCC Committee on Clinical Applications of Cardiac Bio-Markers (C-CB) v052023. Accessed September 24, 2023. https://pub-180a8d00f517477ba49634e6b2b147e3.r2.dev/2023/09/High-Sensitivity-Cardiac-Troponin-I-and-T-Assay-Analytical-Characteristics-Designated-By-Manufacturer-v052023.pdf
  • 37.Suissa S Calculation of number needed to treat. N Engl J Med. 2009;361:424–425. doi: 10.1056/NEJMc0903274 [DOI] [PubMed] [Google Scholar]
  • 38.Rembold CM. Number needed to screen: development of a statistic for disease screening. BMJ. 1998;317:307–312. doi: 10.1136/bmj.317.7154.307 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Centers for Medicare & Medicaid Services. Clinical Laboratory Fee Schedule. CY 2023 Q3 Release. Accessed August 29, 2023. https://www.cms.gov/medicare/medicare-fee-service-payment/clinicallabfeesched/clinical-laboratory-fee-schedule-files/23clabq3 [Google Scholar]
  • 40.Centers for Medicare & Medicaid Services. Hospital Outpatient PPS. Addendum B Updates. Accessed August 29, 2023. https://www.cms.gov/medicare/medicare-fee-service-payment/hospitaloutpatientpps/addendum-and-addendum-b-updates/july-2023-0 [Google Scholar]
  • 41.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Rawshani A, Rawshani A, Franzen S, Sattar N, Eliasson B, Svensson AM, Zethelius B, Miftaraj M, McGuire DK, Rosengren A, et al. Risk Factors, Mortality, and Cardiovascular Outcomes in Patients with Type 2 Diabetes. N Engl J Med. 2018;379:633–644. doi: 10.1056/NEJMoa1800256 [DOI] [PubMed] [Google Scholar]
  • 43.Chahal H, Bluemke DA, Wu CO, McClelland R, Liu K, Shea SJ, Burke G, Balfour P, Herrington D, Shi P, et al. Heart failure risk prediction in the Multi-Ethnic Study of Atherosclerosis. Heart. 2015;101:58–64. doi: 10.1136/heartjnl-2014-305697 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Lin Y, Shao H, Shi L, Anderson AH, Fonseca V. Predicting incident heart failure among patients with type 2 diabetes mellitus: The DM-CURE risk score. Diabetes Obes Metab. 2022;24:2203–2211. doi: 10.1111/dom.14806 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Jacobs JA, Addo DK, Zheutlin AR, Derington CG, Essien UR, Navar AM, Hernandez I, Lloyd-Jones DM, King JB, Rao S, et al. Prevalence of Statin Use for Primary Prevention of Atherosclerotic Cardiovascular Disease by Race, Ethnicity, and 10-Year Disease Risk in the US: National Health and Nutrition Examination Surveys, 2013 to March 2020. Jama Cardiol. 2023;8:443–452. doi: 10.1001/jamacardio.2023.0228 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Echouffo-Tcheugui JB, Zhang S, Florido R, Hamo C, Pankow JS, Michos ED, Goldberg RB, Nambi V, Gerstenblith G, Post WS, et al. Duration of Diabetes and Incident Heart Failure: The ARIC (Atherosclerosis Risk In Communities) Study. JACC Heart Fail. 2021;9:594–603. doi: 10.1016/j.jchf.2021.06.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Rawshani A, Svensson AM, Zethelius B, Eliasson B, Rosengren A, Gudbjornsdottir S. Association Between Socioeconomic Status and Mortality, Cardiovascular Disease, and Cancer in Patients With Type 2 Diabetes. JAMA Intern Med. 2016;176:1146–1154. doi: 10.1001/jamainternmed.2016.2940 [DOI] [PubMed] [Google Scholar]
  • 48.Gottdiener JS, Arnold AM, Aurigemma GP, Polak JF, Tracy RP, Kitzman DW, Gardin JM, Rutledge JE, Boineau RC. Predictors of congestive heart failure in the elderly: the Cardiovascular Health Study. J Am Coll Cardiol. 2000;35:1628–1637. doi: 10.1016/s0735-1097(00)00582-9 [DOI] [PubMed] [Google Scholar]
  • 49.Fitzpatrick JK, Ambrosy AP, Parikh RV, Tan TC, Bansal N, Go AS, Investigators CS. Prognostic value of echocardiography for heart failure and death in adults with chronic kidney disease. Am Heart J. 2022;248:84–96. doi: 10.1016/j.ahj.2022.02.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Bansal N, Hyre Anderson A, Yang W, Christenson RH, deFilippi CR, Deo R, Dries DL, Go AS, He J, Kusek JW, et al. High-sensitivity troponin T and N-terminal pro-B-type natriuretic peptide (NT-proBNP) and risk of incident heart failure in patients with CKD: the Chronic Renal Insufficiency Cohort (CRIC) Study. J Am Soc Nephrol. 2015;26:946–956. doi: 10.1681/ASN.2014010108 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Publication Material
NIHMS1950557-supplement-Supplemental_Publication_Material.pdf (1.1MB, pdf)

RESOURCES