Heart failure-related pleural effusion: determinants of pleural drainage and predictors of mortality in a cohort of 1372 patients

Graphical abstract

Heart failure (HF)-related pleural effusion: drainage requirements and prognosis. IPC: indwelling pleural catheter; NT-proBNP: N-terminal pro-brain natriuretic peptide.

Abstract

Background

Heart failure (HF) is one of the most common causes of pleural effusion, yet data on its management and prognosis in patients undergoing diagnostic thoracentesis are limited. The significance of invasive pleural interventions remains uncertain.

Methods

We conducted a retrospective analysis of a prospectively maintained pleural database, including consecutive patients with HF-related pleural effusion who underwent diagnostic thoracentesis. Clinical characteristics, pleural findings, need for invasive pleural management (therapeutic thoracentesis or indwelling pleural catheter (IPC)), and mortality were assessed. Fine–Gray competing-risk models and Cox regression were applied.

Results

A total of 1372 patients were included (median age 85 years). During follow-up, 726 (53%) patients required at least one therapeutic thoracentesis and 81 (6%) received an IPC. In the competing-risk analysis, effusion size equal to or more than half hemithorax (sudistribution hazard ratio (sHR) 2.33), bilateral effusion (sHR 1.37), pleural fluid albumin >1.2 g·dL⁻¹ (sHR 1.23), and male sex (sHR 1.20) were independently associated with invasive pleural management. Median survival was 14 months, with mortality rates of 50% at 1 year and 70% at 2 years. Mortality was independently associated with advanced age, elevated serum N-terminal pro-brain natriuretic peptide, renal dysfunction, and hypoalbuminemia, whereas pleural characteristics and invasive pleural management were not.

Conclusion

In patients with HF-related pleural effusion undergoing diagnostic thoracentesis, mortality is high and primarily driven by systemic markers of advanced HF. Invasive pleural interventions are frequently required in patients with large or bilateral effusions. In descriptive comparisons, survival was similar regardless of pleural intervention status, although causal inference is precluded by the observational design and time-dependent nature of intervention.

Shareable abstract

Heart failure-related pleural effusions requiring thoracentesis carry high mortality driven mainly by systemic disease severity. Large or bilateral effusions frequently require drainage, but survival was similar irrespective of pleural intervention status. https://bit.ly/4uyE947

Introduction

Heart failure (HF) is a leading cause of pleural effusion. Population-based studies indicate that nonmalignant pleural effusions account for approximately two-thirds of pleural disease related hospitalisations, with HF being the most frequent underlying aetiology [1]. In large contemporary clinical series of patients undergoing diagnostic thoracentesis, HF-related pleural effusion accounts for ∼30% of all cases [2], underscoring its central role in daily pleural practice.

HF-related pleural effusion is generally managed as a manifestation of haemodynamic congestion. Current cardiology-driven strategies focus on optimising pharmacological therapy, assuming that pleural fluid will resolve with diuretics and disease-modifying treatments. Consistent with this paradigm, the 2023 European Society of Cardiology guidelines for HF management do not provide specific recommendations on pleural procedures, such as therapeutic thoracenteses or indwelling pleural catheters (IPCs), or on referral pathways for patients with HF-related pleural effusion [3].

However, from a respiratory medicine perspective, HF-related pleural effusion poses practical management challenges. A relevant proportion of patients remain symptomatic despite optimised medical therapy, develop recurrent effusions, or require repeated pleural interventions for symptom control. Pleural-focused expert reviews acknowledge that iterative therapeutic thoracentesis is commonly used in refractory cardiac effusions and that IPCs may be considered in selected patients to alleviate symptoms and reduce healthcare utilisation [4]. Nevertheless, these approaches are supported mainly by small series and expert opinions, and robust real-world data on their use and determinants remain limited [5].

Concurrently, the long-standing perception of HF-related pleural effusion as a benign condition has been increasingly questioned. A recent European Respiratory Society statement on benign pleural effusions emphasised that several effusions traditionally classified as benign are associated with substantial morbidity, frequent hospitalisations, and poor survival, reflecting advanced underlying disease rather than a favourable clinical course [6].

Against this background, we conducted a study to characterise the clinical features of HF-related pleural effusion in real-world practice, identify baseline determinants of invasive pleural management, and describe survival outcomes according to pleural intervention status.

Methods

Study design and setting

We performed a retrospective cohort study based on a prospectively maintained database of patients who underwent diagnostic thoracentesis at our institution. The database captures consecutive patients evaluated by the pleural medicine unit, irrespective of the clinical service of origin. In the present study, we identified all patients with pleural effusion attributed to HF between January 2016 and June 2025. The study protocol was approved by the institutional ethics committee (reference number CEIC 1960).

Study population

Patients were eligible if they met the following criteria: 1) underwent diagnostic thoracentesis with pleural fluid analyses; and 2) HF was adjudicated as the primary cause of pleural effusion by an attending physician based on clinical grounds (typical symptoms and signs, chest radiographs, ECG, echocardiogram, pleural fluid chemistry, natriuretic peptide levels, response to diuretics, and exclusion of other primary causes of pleural effusion) [4, 6, 7]. Although a cardiologist was not directly involved in every individual case, patients were managed within a multidisciplinary heart failure programme involving both internists and cardiologists.

Patients with suspected concomitant causes of pleural effusion (e.g. malignancy, parapneumonic effusion, pulmonary embolism, hepatic hydrothorax, or renal failure related effusion) were excluded unless HF was determined to be the dominant aetiology. For patients undergoing multiple thoracenteses during the study period, only the first episode was included for baseline characterisation, whereas subsequent procedures were descriptively analysed as part of pleural management outcomes.

All pleural procedures (therapeutic thoracentesis and IPC insertions) were ultrasound-guided, with the puncture site selected using pleural ultrasonography immediately before the procedure.

Data collection

Data were extracted from the pleural database and electronic medical records of the patients. The variables collected at baseline included demographic characteristics (age and sex), body mass index, main causes of HF, New York Heart Association (NYHA) functional class, left ventricular ejection fraction (LVEF), serum N-terminal pro-brain natriuretic peptide (NT-proBNP) levels, renal and biochemical indices (glomerular filtration rate and serum albumin), pleural fluid characteristics (biochemistry, cell counts, Light's criteria classification, and NT-proBNP levels), imaging findings (ultrasound and radiographic assessment of pleural effusion laterality and estimated size, categorised as less than one-half hemithorax or equal to or more than one-half hemithorax), pleural procedures performed during follow-up (therapeutic thoracentesis and IPC placement), hospital length of stay and readmissions, and mortality.

Outcomes

Two primary outcomes were defined: 1) the need for invasive pleural management, operationalised as the receipt of therapeutic thoracentesis and/or IPC, and 2) all-cause mortality during follow-up. Secondary outcomes included the need for multiple therapeutic thoracentesis procedures, IPC removal rates, and IPC-related complications.

IPCs were removed when spontaneous pleurodesis was considered to have occurred, defined as a drainage volume <50 mL on three consecutive drainages performed ≥1 week apart, together with ultrasound confirmation of absent or minimal residual pleural fluid.

Statistical analysis

Continuous variables are summarised as mean±sd or median (interquartile range (IQR)), as appropriate, and categorical variables as counts and percentages. Group comparisons were performed using the Chi-squared or Fisher's exact test for categorical variables and the t-test or Mann–Whitney U-test for continuous variables.

To identify baseline predictors of invasive pleural management, a competing-risk regression model using Fine–Gray subdistribution hazards was fitted, treating all-cause mortality as a competing event. Candidate variables were selected a priori based on clinical relevance and pathophysiological plausibility, supported by univariable analyses. Continuous variables included in the multivariable models were dichotomised using clinically meaningful thresholds selected a priori (age ≥85 years to identify the “oldest-old” population, NT-proBNP >2500 pg·mL⁻¹ as a commonly used diagnostic threshold in pleural effusions of cardiac origin, and glomerular filtration rate <30 mL·min⁻¹·1.73 m⁻² to identify severe renal impairment). Pleural fluid albumin was dichotomised at 1.2 g·dL⁻¹, a pragmatic threshold close to the cohort median, to facilitate clinical interpretability. To avoid temporal bias, only baseline variables available at the time of thoracentesis were included. Subdistribution hazard ratios (sHRs) with 95% confidence intervals were reported.

To evaluate the predictors of mortality, Cox proportional hazards models were constructed, including baseline systemic variables associated with HF severity. Adjusted hazard ratios (95% CI) were reported, and proportional hazards assumptions were verified. The number of outcome events was substantially higher than the number of variables included in the multivariable models, minimising the risk of model overfitting.

For the multivariate models, multicollinearity was assessed and addressed by excluding redundant variables when necessary. Missing data were handled using complete-case analysis because missingness was minimal for the key variables included in the multivariable models and unlikely to materially influence the results. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test for descriptive purposes.

Analyses were performed using the SPSS statistical software (version 30.0). Statistical significance was set at p<0.05.

Results

Baseline characteristics

1372 patients with HF-related pleural effusion were included in this study. The median (IQR) age was 85 (78–90) years, 71% were aged ≥80 years, and 42% were women. Most patients had an advanced HF functional status (NYHA class III–IV in 75%). The leading aetiologies of HF were valvular disease (21%), ischaemic heart disease (20%), and arterial hypertension (17%). Renal dysfunction was common (glomerular filtration rate <30 mL·min⁻¹·1.73 m⁻² in 24%). The median serum NT-proBNP level was 6152 pg·mL⁻¹, with 81% of patients having values >2500 pg·mL⁻¹. LVEF was preserved (>50%) in 71% of cases. The median hospital stay was 7 days, and 42% of the patients were readmitted during follow-up (table 1). Most patients were receiving guideline-directed HF therapy at the time of thoracentesis, including loop diuretics in 75%, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers in 57%, β-blockers in 58%, mineralocorticoid receptor antagonists in 14%, and sodium-glucose cotransporter-2 inhibitors in 10%.

TABLE 1.

Baseline characteristics of the study population

Population	1372
Demographics and baseline clinical status
Age, years	85 (78–90)
Female	582 (42)
Body mass index	26 (23–29)
Glomerular filtration rate, mL·min−1·1.73 m−2	49 (31–69)
Hospital length of stay days	7 (4–11)
Readmission during follow-up	572 (42)
HF characteristics
HF with preserved ejection fraction (≥50%)	761/1066 (71)
HF with mildly reduced ejection fraction (41–49%)	66/1066 (6)
HF with reduced ejection fraction (≤40%)	239/1066 (22)
NYHA functional class I–II	337/1369 (25)
NYHA functional class III–IV	1032/1369 (75)
Serum NT-proBNP, pg·mL−1	6152 (3194–13 341)
Pleural effusion characteristics
Bilateral effusion on chest ultrasound	912/1355 (67)
Effusion size equal to or more than half hemithorax on chest radiograph	208/1366 (15)
Exudate by Light's criteria	394/1251 (32)
Exudate by pleural fluid-only three-test combination rule#	238/1286 (19)
Pleural fluid NT-proBNP, pg·mL−1	6993 (3888–15 843)
Pleural procedures
≥1 therapeutic thoracentesis	726/1372 (53)
≥2 therapeutic thoracenteses	242/1372 (18)
IPC inserted	81/1372 (6)

Data are presented as n, median (interquartile range) or n/N (%). Denominators vary because some variables were not available in all patients. HF: heart failure; NYHA: New York Heart Association; NT-proBNP: N-terminal pro-brain natriuretic peptide; IPC: indwelling pleural catheter. ^#: pleural fluid protein >3 g·dL⁻¹ or lactate dehydrogenase >250 IU·L⁻¹ or cholesterol >55 mg·dL⁻¹.

Pleural effusion characteristics

Pleural effusions were bilateral on ultrasound in 67% of the cases, predominantly showing a simple or anechoic pattern (87%). The effusion occupied less than half of the hemithorax on chest radiography in 85% of patients. Light's criteria classified 32% of effusions as exudates, despite pleural fluid NT-proBNP levels >2500 pg·mL⁻¹ in 86% (table 1).

Pleural procedures

A total of 726 (53%) patients required at least one therapeutic thoracentesis, and 242 (18%) underwent two or more procedures (table 1). IPCs were inserted in 81 (6%) cases, usually after two or more previous therapeutic thoracenteses (72%). IPC removal was achieved in 63% of patients after a median of 81 (20–161) days, and IPC-related infection occurred in four (5%) cases.

Complications related to IPC included four (5%) pleural infections and the use of intrapleural urokinase in seven (9%) patients because of pleural loculations impairing drainage. One accidental IPC removal and one case of post-procedural neuropathic pain were also recorded. Major complications following therapeutic thoracentesis were rare. One haemothorax occurred and was managed with two repeat therapeutic thoracenteses to evacuate the blood, and one pneumothorax resolved after needle aspiration.

Predictors of invasive pleural management

In a Fine–Gray competing-risk model treating death as a competing event, effusion size at least half hemithorax (sHR 2.33), bilateral effusion (sHR 1.37), absolute pleural fluid albumin >1.2 g·dL⁻¹ (sHR 1.23), and male sex (sHR 1.20) were independently associated with invasive pleural management (table 2).

TABLE 2.

Predictors of invasive pleural management in Fine–Gray competing risk regression analysis

	Subdistribution HR (95% CI)	p-value
Effusion size at least half hemithorax	2.33 (2.10–2.59)	<0.001
Bilateral effusion	1.37 (1.21–1.55)	<0.001
Pleural fluid albumin >1.2 g·dL−1	1.23 (1.08–1.40)	0.001
Male sex	1.20 (1.06–1.35)	0.003

HR: hazard ratio.

Survival and prognostic factors

The median follow-up was 10 (2–29) months. The median survival after thoracentesis was 14 (12–16) months. Mortality was 30%, 50%, 65%, 70% and 80% at 6, 12, 18, 24 and 36 months, respectively.

Multivariable Cox modelling identified five baseline factors independently associated with mortality: age ≥85 years (HR 1.82), serum NT-proBNP >2500 pg·mL⁻¹ (HR 1.65), serum albumin <3 g·dL⁻¹ (HR 1.34), glomerular filtration rate <30 mL·min⁻¹·1.73 m⁻² (HR 1.36), and effusion size at least half hemithorax (HR 1.49) (table 3). To assess whether pleural characteristics influenced prognosis independently of HF severity, a sensitivity Cox model excluding effusion size retained the systemic predictors with similar hazard ratios, supporting that mortality risk was primarily determined by HF severity rather than pleural features (table 4).

TABLE 3.

Multivariate Cox regression predictors of mortality

	Adjusted HR (95% CI)	p-value
Age ≥85 years	1.82 (1.56–2.19)	<0.001
Serum NT-proBNP >2500 pg·mL−1	1.65 (1.31–2.07)	<0.001
Effusion size equal to or more than half hemithorax	1.49 (1.20–1.85)	<0.001
Glomerular filtration rate <30 mL·min−1·1.73 m−2	1.36 (1.13–1.65)	0.001
Serum albumin <3 g·dL−1	1.34 (1.12–1.61)	0.002

HR: hazard ratio; NT-proBNP: N-terminal pro-brain natriuretic peptide.

TABLE 4.

Sensitivity Cox regression predictors of mortality excluding pleural effusion size

	Adjusted HR (95% CI)	p-value
Age ≥85 years	1.87 (1.58–2.21)	<0.001
Serum NT-proBNP >2500 pg·mL−1	1.67 (1.33–2.09)	<0.001
Serum albumin <3 g·dL−1	1.40 (1.17–1.68)	<0.001
Glomerular filtration rate <30 mL·min−1·1.73 m−2	1.38 (1.14–1.66)	0.001

HR: hazard ratio; NT-proBNP: N-terminal pro-brain natriuretic peptide.

In descriptive Kaplan–Meier comparisons, survival was similar between patients who ultimately underwent invasive pleural management and those who did not (median 13 (IQR 11–15) versus 15 (12–18) months; log-rank p=0.197) (figure 1).

FIGURE 1.

Kaplan–Meier survival curves according to invasive pleural management in patients with heart failure related pleural effusion. Note: invasive pleural management included therapeutic thoracentesis and/or indwelling pleural catheter placement. Differences were assessed using the log-rank test. The dotted horizontal and vertical lines indicate the 50% survival level and median survival time, respectively.

Discussion

In this large cohort of patients with HF-related pleural effusion undergoing diagnostic thoracentesis, we found that invasive pleural interventions were frequently required and that overall mortality was high, with prognosis driven predominantly by systemic markers of advanced HF.

Previous studies reporting mortality in HF-related pleural effusions have been limited by small sample sizes and heterogeneous populations. DeBiasi et al. [8] described a 1-year mortality of 53% in 32 patients undergoing thoracentesis, while two prospective cohorts including 86 and 158 HF cases, respectively, observed ∼50% 1-year mortality [9, 10]. A nationwide cohort in the United States of America demonstrated markedly reduced survival in 5156 HF patients with pleural effusion compared with 213 227 HF patients without pleural effusion (median 1.51 versus 3.23 years) [11]. In this context, our study represents the largest clinically adjudicated thoracentesis cohort to date (n=1372) and provides precise estimates of 1- and 2-year mortality (50% and 70%, respectively), reinforcing that pleural effusion in HF reflects an advanced systemic disease.

In our cohort, mortality was driven predominantly by systemic variables (very advanced age, markedly elevated serum NT-proBNP level, renal dysfunction, and hypoalbuminemia) rather than pleural characteristics. This pattern is consistent with evidence highlighting frailty as a key determinant of prognosis in pleural diseases. In a national Welsh cohort of nonmalignant pleural diseases, Short et al. [12] observed a stepwise increase in mortality with worsening frailty, independent of age, comorbidities, and pleural diagnosis. Barton et al. [13] similarly identified frailty as a strong predictor of 12-month mortality in a mixed pleural cohort, although only a minority (n=134) had HF. Although frailty was not formally assessed in our study, the very advanced age of our population and the high prevalence of renal dysfunction and hypoalbuminemia suggest substantial functional vulnerability, probably explaining both the poor survival observed and the limited incremental prognostic value of pleural features once HF severity is considered.

Therapeutic pleural intervention was common in our cohort, reflecting a recurrent or refractory pleural congestion phenotype. More than half of the patients required at least one therapeutic thoracentesis, and 6% ultimately received an IPC. The need for invasive pleural management was primarily determined by anatomical burden (large and bilateral effusions) and elevated pleural fluid albumin concentration (probably a reflection of intense diuretic therapy), identifying a subgroup in which pleural involvement is clinically consequential rather than incidental.

The role of pleural drainage as a therapeutic strategy for HF remains controversial. The only randomised trial to date (Thoracentesis to Alleviate Cardiac Pleural Effusion-Interventional Trial (TAP-IT)) did not support routine upfront thoracentesis in acute HF with reduced LVEF, as drainage of ∼1 L failed to improve days alive out of hospital, 90-day mortality, length of stay, or global health status [14]. However, TAP-IT was a pragmatic, unblinded trial conducted in a small, predominantly elderly and frail population and was not designed or powered to detect short-term symptomatic benefits, which provide the usual clinical rationale for thoracentesis. Importantly, observational studies have indicated that the persistence of pleural effusion at hospital discharge has adverse prognostic implications [15, 16]. In a Japanese registry of 3080 patients with HF and pleural effusion at admission [15], residual pleural effusion on chest radiographs at discharge was independently associated with higher 1-year mortality (HR 1.64) and increased risk of death or HF rehospitalisation (HR 1.38). Similar short-term risks have been reported when residual effusion is assessed by ultrasound [16]. An observational series of carefully selected diuretic-refractory patients further suggested that ultrasound-guided therapeutic thoracentesis may provide rapid symptomatic relief [17], although such data cannot establish durability or net clinical benefit. In the REDUCE randomised trial of refractory transudative pleural effusions, which included a substantial proportion of patients with HF, IPCs did not provide superior control of breathlessness compared with repeated therapeutic thoracentesis over 12 weeks, although both strategies were used to palliate symptoms [18]. Overall, current evidence supports a selective, symptom-oriented approach to pleural drainage over routine interventions.

This study has several limitations. Although data were collected prospectively in a dedicated pleural database, the analysis was retrospective and conducted at a single centre, limiting generalisability. The cohort was restricted to patients who underwent diagnostic thoracentesis; therefore, it represents a selected subset of patients with HF and clinically significant or atypical pleural effusions. Frailty was not formally assessed, although the very advanced age of the cohort and the high prevalence of renal dysfunction and hypoalbuminaemia suggested substantial functional vulnerability. Decisions regarding therapeutic thoracentesis and IPC placement were made based on clinical judgement rather than a pre-defined protocol, introducing potential confounding by indication. In addition, pleural interventions occurred during follow-up and therefore represent time-dependent exposures, which precludes causal inference regarding their effect on survival. Moreover, symptom burden and short-term dyspnoea relief were not systematically measured, precluding direct assessment of patient-reported benefits of pleural interventions. Finally, although most patients were receiving diuretic therapy at the time of thoracentesis, the adequacy of HF medical optimisation could not be formally assessed because of the retrospective design.

Conclusions

In patients with HF-related pleural effusion undergoing diagnostic thoracentesis, mortality is high and primarily driven by systemic markers of advanced HF rather than by pleural characteristics. Therapeutic pleural intervention is frequently required in patients with large or bilateral effusions. In descriptive comparisons, survival was similar between patients who did and did not undergo pleural intervention, although causal inference cannot be established because of the observational design. These findings support a pragmatic, selective approach to pleural drainage in HF, reserving intervention for patients with substantial pleural burden or refractory symptoms.