A Compass for the Heart That Hides Its Failure

Key Takeaway: Developed with data from over 20,000 patients and validated in more than 28,000, a new 14-variable risk calculator called LIFE-Preserved can predict both short-term and lifetime risk of hospitalization or cardiovascular death in patients with heart failure with preserved ejection fraction—a condition long considered too complex for reliable prognosis.

The Paradox of a Well-Pumping but Failing Heart

Imagine sitting across from a patient whose echocardiogram findings seem reassuringly normal: the heart contracts with adequate force, and the ejection fraction reads 55%. Yet, the patient cannot climb a single flight of stairs without becoming breathless. Their ankles are swollen. They have been hospitalized twice this year. This is the paradox of heart failure with preserved ejection fraction (HFpEF), a condition that now accounts for roughly half of all heart failure cases worldwide^[2]. Unlike heart failure with reduced ejection fraction, where the heart’s pumping power is visibly weakened, HFpEF hides in plain sight. The heart contracts normally, but it becomes stiff, struggles to relax, and fills with blood inefficiently. For decades, clinicians have lacked a reliable method to tell these patients what their future holds. A new risk model aims to change that.

Behind the Score: How the Model Was Developed

An international team of researchers developed the LIFE-Preserved risk model using data from 20,332 patients diagnosed with HFpEF in the Swedish Heart Failure Registry, one of the world’s most comprehensive national heart failure databases. The model was designed to predict a patient’s risk of two critical outcomes: hospitalization for heart failure or death from cardiovascular causes. The model’s most crucial feature is that it not only provides short-term predictions, such as the probability of an event within one year, but also delivers lifetime risk estimates. This capability has the potential to transform how clinicians and patients discuss the prognosis of a chronic disease over its long course^[1].

The score incorporates 14 markers, all of which are routinely obtained in standard clinical practice; it requires neither an unusual biomarker nor an advanced imaging technique. These variables include demographic information, comorbidities, and basic laboratory values. The model’s deliberate reliance on accessible data means that any clinic with electronic health records can, in principle, calculate this score at the bedside.

The 14 clinical parameters forming the backbone of LIFE-Preserved are: age, sex, diabetes, active smoking, ischemic heart disease, chronic obstructive pulmonary disease (COPD), atrial fibrillation, NYHA functional class, a history of prior hospitalization for heart failure, body mass index (BMI), heart rate, hemoglobin, NT-proBNP, and eGFR levels calculated from serum creatinine. However, the algorithm’s true elegance lies in its seamless integration of ‘competing risks’—an unavoidable reality in the HFpEF population. When non-cardiovascular deaths are ignored in these elderly, frail patients, risk ratios become artificially inflated, leading to an unnecessary treatment burden. The model prevents this bias by using ’cause-specific Cox models’ derived separately for men and women and an age-based time scale. Furthermore, a special time-sensitive risk coefficient is added for those who have had a decompensation episode within the last 6 months, as clinical data clearly show that a recent hospitalization increases the risk of an event within the first year by 2.06-fold in women and 1.97-fold in men.

What truly sets LIFE-Preserved apart from previous studies is the rigor of its external validation. The researchers tested the model in over 28,000 additional patients from major international clinical trials like EMPEROR-Preserved and TOPCAT, as well as real-world registries from the UK’s National Health Service and the US Department of Veterans Affairs system. Across these diverse populations, the model demonstrated robust discriminatory power, with pooled C-statistics of 0.714 in the trial cohorts and 0.658 in the registry cohorts. It also showed good calibration, meaning that the predicted risks closely matched what patients actually experienced over time.

Why HFpEF Has Been So Elusive

To grasp the importance of this score, one must understand why HFpEF has frustrated cardiologists for so long. In heart failure with reduced ejection fraction, the pathology is relatively straightforward: damaged heart muscle contracts weakly, cardiac output falls, and a well-defined neurohormonal activation cascade involving the renin-angiotensin-aldosterone system and the sympathetic nervous system drives disease progression^[3]. Therapies that block these pathways, such as ACE inhibitors, beta-blockers, and mineralocorticoid receptor antagonists, have dramatically improved survival.

HFpEF follows a different script. The left ventricle contracts adequately, but it becomes stiff and resistant to filling during diastole, the relaxation phase between heartbeats. This diastolic dysfunction increases pressure inside the heart chambers, which backs up into the lungs and the body, leading to the same symptoms of congestion and shortness of breath. However, the underlying drivers are far more heterogeneous. Obesity, diabetes, hypertension, atrial fibrillation, chronic kidney disease, and systemic inflammation converge through overlapping yet distinct mechanisms to create this syndrome^[4]. This heterogeneity has made it exceedingly difficult to find one-size-fits-all treatments and equally challenging to predict which patients will deteriorate rapidly and which will remain stable for years.

Until recently, treatment options for HFpEF were severely limited. The EMPEROR-Preserved trial, published in 2021, was a landmark study showing that the SGLT2 inhibitor empagliflozin reduced the risk of cardiovascular death and hospitalization for heart failure in patients with HFpEF^[5]. With effective therapies finally emerging, the need for accurate risk stratification has become urgent; clinicians need to know which patients will benefit most from aggressive intervention.

A recent wave of evidence has established that HFpEF is not a single clinical entity but a complex mosaic of metabolic, renal, and cardiovascular phenotypes. Indeed, following the path forged by SGLT2 inhibitors, the non-steroidal mineralocorticoid receptor antagonist finerenone (FINEARTS-HF) and the groundbreaking GLP-1 and dual GLP-1/GIP receptor agonists semaglutide and tirzepatide in cases arising from obesity have uniquely empowered clinicians to break the cycle of hospitalization and cardiovascular death. However, this enriched therapeutic arsenal brings a new dilemma: initiating aggressive multi-drug regimens in a patient with a very low baseline risk may offer limited absolute risk reduction while unnecessarily increasing the number needed to treat (NNT). It is at this juncture that the model lends mathematical precision to the phenotype-specific targeted therapy that guidelines have long emphasized.

Implications for Clinical Practice: What It Means for Patients

The practical implications of LIFE-Preserved are significant. First, it offers clinicians a structured, evidence-based framework for a conversation that has historically been ambiguous. Instead of saying, “Your heart failure might get worse, or it might not,” a physician can now provide personalized predictions for both the next year and the patient’s remaining lifetime. This level of detail is essential for shared decision-making, helping patients weigh the benefits and burdens of medications, lifestyle changes, and monitoring intensity.

Second, the score can guide resource allocation. Patients identified as high-risk can be directed to specialized heart failure programs, enrolled in remote monitoring protocols, or prioritized for newer therapies like SGLT2 inhibitors. Lower-risk patients, conversely, might be safely managed with less frequent follow-up in primary care.

Third, the lifetime risk estimate adds a dimension that short-term scores cannot offer. A 60-year-old patient with a modest one-year risk might still carry a strikingly high lifetime burden of heart failure events, a realization that can motivate earlier and more sustained adherence to treatment.

The tangible power of this predictive architecture in clinical practice can be understood through two different cases: Consider a 75-year-old female patient with atrial fibrillation, significant renal dysfunction (eGFR 35 mL/min/1.73 m²), a high NT-proBNP level (2120 pg/mL), and a history of a past decompensation event. On the other hand, consider a 55-year-old male patient with new-onset shortness of breath against a background of obesity (BMI 35 kg/m²), diabetes, and ischemic heart disease. When a standard SGLT2 inhibitor is initiated, the 75-year-old patient’s 2-year absolute event risk drops from 24.2% to 20.1% (a 4.1% absolute risk reduction), providing strong short-term protection and gaining her 0.6 event-free years of life. In contrast, while the 55-year-old patient’s 2-year event risk is a relatively modest 12.4%, the same treatment adds a full 1.7 additional healthy years to his life, free from a heart failure attack or cardiovascular death. The lifetime risk perspective debunks the misconception that ‘only old and severely ill patients benefit from medication,’ revealing the massive cumulative gain that early intervention promises in younger and middle-aged phenotypes.

Important Caveats

No study, however large, permanently settles a question. While the validation cohorts are impressively diverse, the development cohort was from Sweden, a population that is less racially and ethnically diverse than many others. The model’s performance in underrepresented populations warrants further investigation. The registry-based C-statistics, while acceptable, were lower than those from clinical trial cohorts, reflecting the more complex reality of everyday clinical practice. And like any prediction model, LIFE-Preserved estimates probabilities at a population level; it cannot say exactly what will happen to a single patient. It is a compass, not a crystal ball, and its greatest value will emerge when it is integrated into clinical workflows and tested prospectively in routine care.

From the perspective of practical integration, several methodological details should be highlighted. LIFE-Preserved excluded heart failure with mildly reduced ejection fraction (HFmrEF, EF 41-49%), as this intermediate group is pathophysiologically closer to HFrEF, and the sister LIFE-HF score is intended for them. Second, missing data for routine variables like NT-proBNP and functional class in the registry data were modeled using multiple imputation techniques to preserve the power of the parameters. Finally, the study was designed to allow for calibration via an expected/observed (E/O) ratio to capture local risk levels, due to a slight risk overestimation observed in the EMPEROR-Preserved and UK National Health Service (NHSE SDE) cohorts. This interactive calculator, which will soon be integrated into the CE-marked clinical decision support interface U-prevent, is flexible enough to dynamically absorb the relative risk reductions of new molecules as they enter the guidelines.

Scientific Sources

1. Reitsma TH, et al. Risk prediction in patients with heart failure with preserved ejection fraction: the LIFE-Preserved model. European heart journal. 2026;47(34):4773-4788. PubMed: https://pubmed.ncbi.nlm.nih.gov/41810940/
2. Dunlay SM, et al. Epidemiology of heart failure with preserved ejection fraction. Nat Rev Cardiol. 2017. DOI: 10.1038/nrcardio.2017.65
3. Packer M. The neurohormonal hypothesis: a theory to explain the mechanism of disease progression in heart failure. J Am Coll Cardiol. 1992. DOI: 10.1016/0735-1097(92)90167-l
4. Shah SJ, et al. Phenomapping for novel classification of heart failure with preserved ejection fraction. Circulation. 2015. DOI: 10.1161/CIRCULATIONAHA.114.010637
5. Anker SD, et al. Empagliflozin in heart failure with a preserved ejection fraction. N Engl J Med. 2021. DOI: 10.1056/NEJMoa2107038