All participants were recruited from the Thousand&2 Study with T2D patients followed at two secondary diabetes clinics in The Capital Region of Denmark: Steno Diabetes Center Copenhagen and the Diabetes Clinic at Herlev and Gentofte Hospital, University of Copenhagen. The study has previously been described in detail (11). In brief, a total of 2.158 T2D patients were invited to participate whereof 1.030 accepted. At study enrolment, patients were asked to fill out a standardised questionnaire with self-reported medical history including current medication, coronary heart disease (myocardial infarction, percutaneous coronary intervention, coronary artery bypass grafting), congestive HF, atrial fibrillation, self-reported dyspnoea, and cardiovascular risk factors (prior stroke, peripheral artery disease, family history of coronary artery disease, and smoking status). Laboratory values (cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglyceride, haemoglobin A1c, and creatinine) were obtained from the electronic health records. NT-pro brain-natriuretic peptide (NT-proBNP) was measured by use of MAGLUMITM 800 Chemiluminescence Immunoassay by Snibe Diagnostics, Shenzhen, China. Albuminuria was defined as urine albumin/creatinine ratio above 30 mg or urine albumin above 30 mg/day on at least 2 consecutive measurements. BMI was calculated from height- and weight measurements. Blood pressure was measured in supine position after 15 minutes rest with a validated device. Hypertension was defined as either a systolic blood pressure >140 mmHg or the ingestion of antihypertensive medication (e.g. beta blockers, calcium antagonists, angiotensin II receptor blocker, or angiotensin-converting enzyme inhibitor). These variables (haemoglobin A1c, systolic blood pressure, BMI, albuminuria, and cholesterols) are all simple clinical risk factors for the development and progression of cardiac function in patients with T2D, measured as part of routine follow-up (12).
Clinical examination
A 12-lead ECG (Cardiosoft version 6.61, GE Healthcare) was recorded at the visit and interpreted by two independent investigators (MCG and PGJ), any disagreements were resolved with discussion. A normal ECG was defined as the absence of all the following: (1) left or right bundle branch block defined as QRS-complex duration >120 ms; (2) abnormal Q-waves defined as a Q-wave duration of ≥0.02 s in V2-3 and a duration and depth of ≥0.03 s and ≥1 mm respectively in any other leads; (3) left ventricular hypertrophy (LVH) according to the Sokolow-Lyon criteria; (4) ST deviations defined as ST-depression ≥0.5 mm or inverted T-waves ≥1 mm. Echocardiographic recordings were performed and analysed by a single investigator (PGJ) in accordance with recommendations of the European Association of Echocardiography and the American Association of Echocardiography. Further details with respect to the echocardiographic recordings are published elsewhere (13). Hence, both the ECG and echocardiography were performed on the same day. HFpEF was defined as: Patients with dyspnoea, with a NT-pro-BNP >125 pg/mL and either one of the following four echocardiographic criteria fulfilled: (1) increased left ventricular mass index; (2) Septal E/e´ >15; (3) left atrial size >34 ml/m2; or (4) left ventricular ejection fraction between 41-50%. HFrEF was defined as a left ventricular ejection fraction ≤40%. In the present study, 308 patients were excluded (more than moderate heart valve disease or previous heart valve replacement n=34, missing NT-proBNP measurement n=74, missing ECGs n=68, missing echocardiographic measurement n=69, atrial fibrillation n=62, or missing follow-up data n=1). Patients with atrial fibrillation were excluded as they usually get a routine echocardiographic examination in Denmark. Further, E/e´ is unusable in patients with ongoing atrial fibrillation and their NT-proBNP is usually elevated making the HF and especially the diagnosis of HFpEF difficult.
Follow-up
Follow-up was performed through national registers and the end-point was the composite of incident CV events (defined as coronary revascularization, myocardial infarction (ICD-10 codes I21-I25), heart failure (ICD-10 codes I11, I13, I42, I43 and I50), cardiac arrest (I46), cerebrovascular disease (I60-I69 and peripheral artery disease (I70-I79)) or CV death.
Statistics
Parametric and non-parametric analyses were used to compare groups were applicable. Sensitivity, specificity, positive predictive values (PPV), and negative predictive values (NPV) for the diagnostic value of ECG were all calculated in different subgroups of patients with HF (either HFrEF + HFpEF, HFrEF alone, or HFpEF alone). Receiver Operating Characteristic curves were applied to determine the ability of an ECG to diagnose HF when added to a multivariable logistic regression model adjusted for age, sex, albuminuria, haemoglobin A1c, BMI and systolic blood pressure. Cumulative incidence curves were used to examine the association with the composite endpoint with non-cardiovascular death as competing risk. Association with prognosis was examined using uni- and multivariable Cox proportionate hazard regression models including the beforementioned covariates. All statistics were calculated using R for Mac, version 2.15.3 (R Project for Statistical Computing, Vienna University of Economics and Business Administration, Wien, Austria).
Approval
The study was conducted in accordance with the Helsinki Declaration and approved by the Danish National Committee on Biomedical Research Ethics (amendment to protocol no. H-3-2009-139)(14). All participants gave written informed consent to participate in the study.
Patient and public involvement
No patients were involved in the design, recruitment, or interpretation of the results in this study.