Use of electronic health record data mining for heart failure subtyping.

OBJECTIVE: To assess whether electronic health record (EHR) data text mining can be used to improve register-based heart failure (HF) subtyping. EHR data of 43,405 individuals from two Finnish hospital biobanks were mined for unstructured text mentions of ejection fraction (EF) and validated against clinical assessment in two sets of 100 randomly selected individuals. Structured laboratory data was then incorporated for a categorization by HF subtype (HF with mildly reduced EF, HFmrEF; HF with preserved EF, HFpEF; HF with reduced EF, HFrEF; and no HF). RESULTS: In 86% of the cases, the algorithm-identified EF belonged to the correct HF subtype range. Sensitivity, specificity, PPV and NPV of the algorithm were 94-100% for HFrEF, 85-100% for HFmrEF, and 96%, 67%, 53% and 98% for HFpEF. Survival analyses using the traditional diagnosis of HF were in concordance with the algorithm-based ones. Compared to healthy individuals, mortality increased from HFmrEF (hazard ratio [HR], 1.91; 95% confidence interval [CI], 1.24-2.95) to HFpEF (2.28; 1.80-2.88) to HFrEF group (2.63; 1.97-3.50) over a follow-up of 1.5 years. We conclude that quantitative EF data can be efficiently extracted from EHRs and used with laboratory data to subtype HF with reasonable accuracy, especially for HFrEF.

Diabetes status-related differences in risk factors and mediators of heart failure in the general population: results from the MORGAM/BiomarCaRE consortium.

BACKGROUND: The risk of heart failure among diabetic individuals is high, even under tight glycemic control. The correlates and mediators of heart failure risk in individuals with diabetes need more elucidation in large population-based cohorts with long follow-up times and a wide panel of biologically relevant biomarkers. METHODS: In a population-based sample of 3834 diabetic and 90,177 non-diabetic individuals, proportional hazards models and mediation analysis were used to assess the relation of conventional heart failure risk factors and biomarkers with incident heart failure. RESULTS: Over a median follow-up of 13.8 years, a total of 652 (17.0%) and 5524 (6.1%) cases of incident heart failure were observed in participants with and without diabetes, respectively. 51.4% were women and the mean age at baseline was 48.7 (standard deviation [SD] 12.5) years. The multivariable-adjusted hazard ratio (HR) for heart failure among diabetic individuals was 2.70 (95% confidence interval, 2.49-2.93) compared to non-diabetic participants. In the multivariable-adjusted Cox models, conventional cardiovascular disease risk factors, such as smoking (diabetes: HR 2.07 [1.59-2.69]; non-diabetes: HR 1.85 [1.68-2.02]), BMI (diabetes: HR 1.30 [1.18-1.42]; non-diabetes: HR 1.40 [1.35-1.47]), baseline myocardial infarction (diabetes: HR 2.06 [1.55-2.75]; non-diabetes: HR 2.86 [2.50-3.28]), and baseline atrial fibrillation (diabetes: HR 1.51 [0.82-2.80]; non-diabetes: HR 2.97 [2.21-4.00]) had the strongest associations with incident heart failure. In addition, biomarkers for cardiac strain (represented by nT-proBNP, diabetes: HR 1.26 [1.19-1.34]; non-diabetes: HR 1.43 [1.39-1.47]), myocardial injury (hs-TnI, diabetes: HR 1.10 [1.04-1.16]; non-diabetes: HR 1.13 [1.10-1.16]), and inflammation (hs-CRP, diabetes: HR 1.13 [1.03-1.24]; non-diabetes: HR 1.29 [1.25-1.34]) were also associated with incident heart failure. In general, all these associations were equally strong in non-diabetic and diabetic individuals. However, the strongest mediators of heart failure in diabetes were the direct effect of diabetes status itself (relative effect share 43.1% [33.9-52.3] and indirect effects (effect share 56.9% [47.7-66.1]) mediated by obesity (BMI, 13.2% [10.3-16.2]), cardiac strain/volume overload (nT-proBNP, 8.4% [-0.7-17.4]), and hyperglycemia (glucose, 12.0% [4.2-19.9]). CONCLUSIONS: The findings suggest that the main mediators of heart failure in diabetes are obesity, hyperglycemia, and cardiac strain/volume overload. Conventional cardiovascular risk factors are strongly related to incident heart failure, but these associations are not stronger in diabetic than in non-diabetic individuals. Active measurement of relevant biomarkers could potentially be used to improve prevention and prediction of heart failure in high-risk diabetic patients.

24-h urinary sodium excretion and the risk of adverse outcomes.

AIMS: The objective was to evaluate whether sodium intake, assessed with the gold standard 24-h urinary collections, was related to long-term incidence of death, cardiovascular disease (CVD) and diabetes mellitus (DM). METHODS: A cohort of 4630 individuals aged 25-64 years collected 24-h urine samples in 1979-2002 and were followed up to 14 years for the incidence of any CVD, coronary heart disease (CHD), stroke, heart failure (HF) and DM event, and death. Cox proportional hazards models were used to estimate the association between the baseline salt intake and incident events and adjusted for baseline age, body mass index, serum cholesterol, prevalent DM, and stratified by sex and cohort baseline year. RESULTS: During the follow-up, we observed 423 deaths, 424 CVD events (288 CHD events, 142 strokes, 139 HF events) and 161 DM events. Compared with the highest quartile of salt intake, persons in the lowest quartile had a lower incidence of CVD (hazard ratio [HR] 0.70; 95% confidence interval [CI], 0.51-0.95, p = .02), CHD (HR 0.63 [95% CI 0.42-0.94], p = .02) and DM (HR 0.52 [95% CI 0.31-0.87], p = .01). The results were non-significant for mortality, HF, and stroke. CONCLUSION: High sodium intake is associated with an increased incidence of CVD and DM.

The validity of heart failure diagnoses in the Finnish Hospital Discharge Register.

Background: Contemporary validation studies of register-based heart failure diagnoses based on current guidelines and complete clinical data are lacking in Finland and internationally. Our objective was to assess the positive and negative predictive values of heart failure diagnoses in a nationwide hospital discharge register. Methods: Using Finnish Hospital Discharge Register data from 2013-2015, we obtained the medical records for 120 patients with a register-based diagnosis for heart failure (cases) and for 120 patients with a predisposing condition for heart failure, but without a heart failure diagnosis (controls). The medical records of all patients were assessed by a physician who categorized each individual as having heart failure (with reduced or preserved ejection fraction) or no heart failure, based on the definition of current European Society of Cardiology heart failure guidelines. Unclear cases were assessed by a panel of three physicians. This classification was considered as the clinical gold standard, against which the registers were assessed. Results: Register-based heart failure diagnoses had a positive predictive value of 0.85 (95% CI 0.77-0.91) and a negative predictive value of 0.83 (95% CI 0.75-0.90). The positive predictive value decreased when we classified patients with transient heart failure (duration <6 months), dialysis/lung disease or heart failure with preserved ejection fraction as not having heart failure. Conclusions: Heart failure diagnoses of the Finnish Hospital Discharge Register have good positive predictive value and negative predictive value, even when patients with pre-existing heart conditions are used as healthy controls. Our results suggest that heart failure diagnoses based on register data can be reliably used for research purposes.