The study followed consecutive adults who presented to the emergency department with new or worsening signs and symptoms compatible with acute heart failure between March 2022 and March 2026. Acute heart failure was confirmed retrospectively using clinical findings plus at least one objective indicator available by the index assessment, such as an age-adjusted NT-proBNP rule-in threshold, compatible echocardiographic abnormalities, or imaging showing pulmonary vascular congestion or pulmonary edema. The analysis included both new-onset disease and acute decompensation of chronic heart failure and did not require a particular left ventricular ejection fraction. Patients were excluded if, among other reasons, severe infection, active malignancy, major hepatic failure, a relevant blood disorder, albumin infusion, or immunosuppressive therapy could substantially distort the nutrition-related laboratory values.

PNI was calculated as serum albumin in grams per liter plus five times the peripheral lymphocyte count in units of 10⁹ per liter, using the first eligible measurements obtained within 24 hours of admission. The index time was the later collection time of the two specimens. Investigators used a prespecified literature-derived threshold to divide the cohort into low-PNI and high-PNI groups. This approach exposed an important limitation of interpreting the score as a direct nutritional assessment: serum albumin and lymphocytes can change with inflammation, congestion, fluid expansion, physiological stress, organ dysfunction, and medication exposure as well as with nutritional reserve. The low-PNI group was therefore already more vulnerable across several dimensions at the time of assessment.

Those differences were substantial. Low-PNI patients were older, had a lower body mass index, and were more likely to have chronic kidney disease or previous hospitalization for heart failure. At the index assessment, they had lower systolic blood pressure and oxygen saturation, faster heart and respiratory rates, more advanced Killip and New York Heart Association classes, and more frequent hypotension, hypoperfusion, and ongoing advanced support. Their laboratory profile also suggested greater inflammatory, cardiorenal, and cardiac stress: median C-reactive protein was 22.6 mg/L versus 9.8 mg/L, median NT-proBNP was 7,912 pg/mL versus 5,305 pg/mL, mean estimated glomerular filtration rate was 54.7 versus 68.1 mL/min/1.73 m², and mean high-sensitivity cardiac troponin I was 57 versus 35 ng/L. Echocardiography and chest imaging likewise showed lower mean ejection fraction, more right ventricular dysfunction, and more pulmonary congestion, edema, and pleural effusion in the low-PNI group.

Events were counted only if they occurred after the index PNI assessment and before discharge or in-hospital death. The primary composite comprised all-cause death, cardiogenic shock, sustained ventricular tachycardia or ventricular fibrillation, cardiac arrest requiring cardiopulmonary resuscitation, acute myocardial infarction, or worsening acute heart failure requiring treatment escalation. Overall, 112 patients, or 30.5%, experienced at least one qualifying event. The low-PNI group had higher rates of all-cause death (14.7% versus 5.7%), cardiogenic shock (18.6% versus 8.5%), sustained ventricular tachycardia or fibrillation (9.6% versus 3.8%), cardiac arrest (7.7% versus 2.8%), and worsening acute heart failure requiring escalation (32.7% versus 15.2%). Acute myocardial infarction did not differ significantly between groups, at 6.4% versus 3.3%. Low-PNI patients were also more often admitted to intensive care, ventilated, or given vasoactive drugs after assessment, and their median hospital stay was 10 days rather than 8.

The association also showed a graded pattern when PNI was treated as a continuous exposure. In the fully adjusted model, each five-point increase was associated with 30% lower odds of the composite endpoint, with an odds ratio of 0.70 and 95% confidence interval of 0.56–0.90; P = 0.003. The estimate for a 10-point increase was 0.49, with a 95% confidence interval of 0.31–0.82. Patients in the lowest PNI quartile had 2.57 times the adjusted odds of the primary outcome compared with those in the highest quartile, and each quartile increase was associated with lower odds, with an odds ratio of 0.72; P for trend = 0.004. The relationship remained in a complete-case analysis and when patients with a component event already present at assessment were excluded.

PNI alone had moderate discrimination, producing an apparent area under the receiver operating characteristic curve of 0.714, with 95% confidence interval 0.658–0.771. At the exploratory cohort-derived cutoff of 44.8, sensitivity was 60.7% and specificity 65.5%, showing that the score could not accurately classify individuals on its own. Its value was more apparent when added to a conventional model containing established clinical information. The conventional model achieved an AUC of 0.776, while the model with continuous PNI reached 0.829, an increase of 0.053 with a 95% confidence interval of 0.021–0.085; P = 0.001. Bootstrap correction reduced the corresponding AUCs to 0.752 and 0.803. The combined model also had a lower corrected Brier score, 0.151 versus 0.174, while its corrected calibration slope changed modestly from 0.86 to 0.88.

These results do not show that PNI causes deterioration or that changing the index would prevent cardiovascular events. The design was observational, based at one center, and vulnerable to selection bias, residual confounding, and inaccuracies in medical records. PNI was calculated from a single measurement, formal nutritional and frailty assessments were unavailable, and serial changes could not be examined. Multiple imputation addressed missing covariates under an unverifiable missing-at-random assumption. The composite also combined events of different severity, and treatment escalation can be influenced by clinical judgment and institutional practice. Nevertheless, a sensitivity analysis using a “hard” composite that excluded worsening heart failure and resource-use outcomes produced a similar adjusted association. After excluding patients with C-reactive protein of at least 50 mg/L, the estimate remained directionally consistent but became less precise, with an adjusted odds ratio of 1.79 and a 95% confidence interval of 0.98–3.26; P = 0.058.

The authors describe PNI as a potentially complementary early marker rather than a stand-alone triage rule or therapeutic target. Its ingredients are available from routine laboratory testing, making the calculation practical in emergency and inpatient care, but the observed model improvement does not itself demonstrate better clinical decisions or improved patient outcomes. Exploratory subgroup analyses found no statistically detectable variation in the association by age, sex, ejection-fraction category, diabetes, chronic kidney disease, NT-proBNP level, or atrial fibrillation; the interaction P-values were all at least 0.742. Some subgroup confidence intervals included 1.00, however, so the analysis indicates no detected effect modification rather than proving identical relationships across phenotypes. Prospective multicenter studies with external validation, longer follow-up, serial measurements, and formal nutritional, inflammatory, congestion, and frailty assessments are needed before PNI-guided management can be recommended.

Subject of Research: Association of early prognostic nutritional index with in-hospital adverse cardiovascular events in adults with acute heart failure

Article Title: Association of the prognostic nutritional index with in-hospital adverse cardiovascular events among emergency department patients with acute heart failure

Article References: Zhang, X.-P., Guo, W.-W., & Li, B. (2026). Association of the prognostic nutritional index with in-hospital adverse cardiovascular events among emergency department patients with acute heart failure. Frontiers in Nutrition, 13, Article 1949478. https://doi.org/10.3389/fnut.2026.1949478

Image Credits: AI Generated

DOI: 10.3389/fnut.2026.1949478

Keywords: acute heart failure, prognostic nutritional index, emergency medicine, in-hospital outcomes, nutritional status, risk stratification, albumin, lymphocyte count, cardiovascular events, retrospective cohort, Association, prognostic