Early Weight Loss in Acute Heart Failure Fails as Reliable Stand-In for Survival in Trial Review

For decades, cardiologists treating patients hospitalized with acute heart failure have leaned on a simple metric: the scale. When a patient’s weight drops several kilograms in the first days of admission, clinicians have often assumed the decongestion is working and that the patient is on track for a better outcome after discharge. A new systematic review and trial-level meta-regression published in Frontiers in Cardiovascular Medicine challenges that assumption, finding that short-term weight loss shows only a directional, mechanism-dependent, and statistically non-robust association with fewer deaths or heart failure events in the weeks and months that follow.

The research team, led by investigators at the Quzhou Affiliated Hospital of Wenzhou Medical University in China, took an unusual analytical approach. Rather than pooling average treatment effects across studies, they paired two effects within each randomized comparison: the short-term effect on a decongestion marker such as body weight, and the subsequent effect on clinical outcomes such as mortality, heart failure hospitalization, or rehospitalization. This structure more closely mirrors what formal surrogate validation would require—demonstrating that a treatment’s effect on a candidate marker reliably predicts its effect on the true clinical endpoint.

The researchers searched PubMed/MEDLINE, OpenAlex, and ClinicalTrials.gov through 18 July 2026. From 9,326 records initially identified, they screened 7,300 titles and abstracts, assessed 32 full-text reports representing 29 distinct trials, and ultimately included 24 trials reported across 27 articles in a qualitative synthesis. For the primary quantitative analysis, only 12 comparisons from 10 independent trials could be paired—weight change measured at 24 to 96 hours against clinical follow-up ranging from 30 days to a median of 9.9 months. Trials spanned publication years 2007 through 2025 and enrolled between 59 and 4,133 randomized participants, covering loop diuretic strategies, vasopressin antagonists, SGLT2 inhibitors, acetazolamide, ultrafiltration, and guidance-based approaches.

The central statistical result was an errors-in-variables meta-regression slope of −0.302 on the log clinical effect scale. Each additional 1-kilogram between-group difference in weight loss corresponded to a relative clinical effect ratio of 0.74, with a percentile 95% interval of 0.56 to 1.26 and a trial-jackknife 95% interval of 0.49 to 1.11. Although the point estimate leaned toward fewer events with greater weight loss, the interval allowed clinically meaningful benefit, no association, or even an adverse relation. The model used 10,000 successful trial-cluster bootstrap fits to account for the fact that some trials, such as DOSE and ROSE-AHF, contributed more than one comparison with shared participants.

Sensitivity analyses exposed how fragile the association was. When restricted to pharmacological trials only, the estimate rose to 1.04—essentially null. The original 48-to-96-hour protocol window gave a similar 0.74 (0.48–1.36). Retaining one comparison per trial yielded 0.69. Excluding the small, open-label, retrospectively registered Early UF 2020 ultrafiltration study gave 0.78. The low-risk-of-bias-only set was identical to the pharmacological-only set, meaning intervention class and study quality could not be separated in this small sample. Leave-one-trial-out ratios ranged from 0.64 to 0.82 per kilogram, with the largest shifts following omission of ROSE-AHF and UNLOAD.

The mechanism behind this fragility is physiological. Body weight reflects multiple processes beyond extracellular fluid removal—baseline measurement conditions, intake, bowel movements, and changes unrelated to congestion. Tolvaptan, a vasopressin antagonist, promotes water excretion (aquaresis) without proportional sodium removal; in the EVEREST outcome trial, it improved early weight and symptoms but showed no reduction in mortality or heart failure hospitalization over a median 9.9-month follow-up. Ultrafiltration removes isotonic fluid directly, yet its net benefit depends on removal rate, vascular refilling, renal perfusion, catheter complications, and patient selection. UNLOAD reported greater weight loss and fewer heart failure rehospitalizations through 90 days, while CARRESS-HF found similar weight loss, worse renal function, more serious adverse events, and no clinical advantage through 60 days in patients with cardiorenal syndrome.

The review also examined non-weight markers in a structured qualitative synthesis. In ADVOR, acetazolamide increased successful decongestion, urine volume, and natriuresis but did not reduce death or heart failure hospitalization at 3 months. PUSH-AHF increased 24-hour natriuresis by 63 mmol with a guided algorithm yet showed no reduction in the 180-day death or heart failure hospitalization composite (hazard ratio 0.92; 95% CI 0.62–1.38). Small pilot studies offered mixed signals: EMPA-RESPONSE-AHF (80 participants) and CAVAL US-AHF (60 participants) showed directionally concordant marker and clinical effects but were underpowered, while EMPAG-HF accrued too few 30-day deaths to establish any clinical relation. The authors emphasize these categories are descriptive and lack the evidential status of the quantitative weight analysis.

The distinction between prognostic value and surrogacy is central to the paper’s argument. Patients with persistent congestion after treatment are sicker and more likely to die or be readmitted—that is a prognostic association. But it does not follow that an intervention which improves the congestion marker will proportionally improve outcomes. Formal surrogate validation requires the treatment effect on the candidate surrogate to reliably predict the treatment effect on the true endpoint, ideally across many trials and mechanisms, with both individual-level and trial-level evidence. With only 10 independent trials, heterogeneous endpoints and follow-up, mixed treatment mechanisms, and no individual participant data, the authors explicitly declined to estimate trial-level R² or interpret their slope as mediation or formal validation.

Several limitations temper the findings. Screening, extraction, and risk-of-bias assessment were performed by a single investigator with a same-investigator verification pass, not independently duplicated. Direct searching was limited to three databases; Embase and CENTRAL were unavailable. The review was not prospectively registered in PROSPERO. Some weight standard errors used approximate denominators or interval conversions, one clinical effect relied on approximate Kaplan–Meier risk intervals, and hazard ratios and risk ratios were combined as a pragmatic approximation on the log scale. Within-comparison covariance between marker and clinical effect estimates was unavailable and assumed to be zero. The residual between-comparison heterogeneity estimate sat near the boundary, and the profile likelihood upper limit of 0.130 indicates clinically relevant residual heterogeneity cannot be excluded.

The clinical message is not that decongestion markers are useless. Serial weight on the same scale, early urine output and urinary sodium after loop diuretic dosing, renal electrolyte safety, and discharge ultrasound evidence of residual congestion all remain valuable indicators of treatment response. But no isolated weight, sodium, or ultrasound value should determine discharge readiness. The authors propose that future acute heart failure trials adopt a core reporting set—fixed-time weight, 6- or 24-hour urine output, early spot urinary sodium and 24-hour total sodium excretion, a standardized congestion score, kidney and electrolyte safety, discharge residual congestion, and at least 90-day mortality and heart failure events—enabling subsequent individual-participant and trial-level validation of multidimensional profiles rather than reliance on any single marker.

Subject of Research: Trial-level association between short-term decongestion markers and post-discharge clinical outcomes in acute heart failure

Article Title: Short-term decongestion and clinical benefit in acute heart failure: a systematic review and trial-level meta-regression

Article References: Wang, H., Fang, H., Wu, M., Chen, Q., Fu, A., & Xu, M. (2026). Short-term decongestion and clinical benefit in acute heart failure: a systematic review and trial-level meta-regression. Frontiers in Cardiovascular Medicine, 13, Article 1942130. https://doi.org/10.3389/fcvm.2026.1942130

Image Credits: AI Generated

DOI: 10.3389/fcvm.2026.1942130

Keywords: acute heart failure, decongestion, surrogate endpoint, body weight, meta-regression, natriuresis, ultrafiltration, heart failure hospitalization, diuretics, trial-level association, Short-term, clinical