nature medicine

Article https://doi.org/10.1038/s41591-025-03628-4

Liberal fluid intake versus fluid restriction

in chronic heart failure: a randomized
clinical trial

Received: 13 February 2025 Job J. Herrmann 1,2, Hans-Peter Brunner-La Rocca3,4,

Lisette E. H. J. M. Baltussen1, Fabienne Beckers-Wesche3,
Accepted: 4 March 2025
Sebastiaan C. A. M. Bekkers5, Louise Bellersen1, J. W. Martijn van Eck6,
Published online: xx xx xxxx H. Carlijne Hassing1, Tiny Jaarsma7,8, Gerard C. M. Linssen 9, Ron Pisters10,
Sandra Sanders-van Wijk11, Marjolein H. I. Verdijk1, M. Louis Handoko 8,
Check for updates
Peter van der Meer 12, Frederik H. Verbrugge 13,14, James L. Januzzi Jr15,
Antoni Bayés-Genís16, Robby Nieuwlaat17, Laura Rodwell18,
D. H. Frank Gommans1,19 & Roland R. J. van Kimmenade 1

Fluid restriction is frequently recommended to patients with chronic

heart failure, but randomized clinical trials assessing the effects of fluid
restriction remain scarce. In this multicenter open-label trial, outpatients
with chronic heart failure were randomized to receiving advice for liberal
fluid intake versus receiving advice for fluid restriction, up to 1,500 ml per
day of fluid intake. The primary outcome of the trial was health status after
3 months, as assessed by the Kansas City Cardiomyopathy Questionnaire
Overall Summary Score (KCCQ-OSS). Secondary outcomes included thirst
distress and safety events. Among 504 randomized patients (67.3% male),
the KCCQ-OSS after 3 months was 74.0 in the liberal fluid intake group versus
72.2 in the fluid restriction group, with a mean difference after adjustment
for baseline scores of 2.17 (95% confidence interval −0.06 to 4.39; P = 0.06),
indicating that the primary outcome was not met. Thirst distress was
higher in the fluid restriction group and no differences were observed
for safety events between the two groups. These findings question the
benefit of fluid restriction in chronic heart failure. ClinicalTrials.gov
registration: NCT04551729.

Fluid restriction is frequently advised to patients with heart failure causing distress1,2,12–14. These factors may contribute to the observed low
(HF), on the intuitive assumption that a restrictive fluid regimen may adherence to fluid restriction3,9,15. In the light of these limited data on the
prevent episodes of congestion1–5. Clinical trial data on the effect of possible clinical benefit or harm, global clinical practice guidelines for
fluid restriction in HF are limited, with no study demonstrating a clear HF recognize the effect of fluid intake as a ‘gap in evidence’ and highlight
beneficial effect of limiting fluid intake on either HF hospitalizations or the need for additional research on the effects of fluid restriction16,17.
mortality6–11. By contrast, the available data suggest that fluid restric- Accordingly, the Fluid Restriction in Heart Failure versus Liberal
tion might even have a detrimental impact on patients’ quality of life, Uptake (FRESH-UP) study was designed to assess the effects of a liberal
because fluid restriction might lead to an increased sensation of thirst, versus a restrictive fluid regimen in patients with chronic HF in an

A full list of affiliations appears at the end of the paper. e-mail: [email protected]

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Screening/baseline visit
n = 524
Retracted informed consent: 1
Screen failures: 19
–1 –2
- GFR <30 ml min 1.73 m : 7
- NYHA class I: 7
- Change in HF medical therapy <14 days: 2
- Device implantation <3 months: 1
- HF diagnosis <6 months: 1
–1
- Sodium <130 mmol l : 1

Liberal fluid intake Fluid restriction

n = 254 n = 250

Visit month 3 not completed: 12 Visit month 3 not completed: 17

a
- No questionnaire: 9 - Nonadherence to fluid intake protocol: 6
- No show: 2 - No questionnaire: 5
- Participation stopped due to new - No show: 5
malignancy: 1 - Death: 1

Month 3 Month 3
Complete primary endpoint: Complete primary endpoint:
n = 242 n = 233

a
Retracted informed consent: 1

Month 6 Month 6
Complete secondary outcomes: Complete secondary outcomes:
n = 254 n = 249

Fig. 1 | Patient enrollment and follow-up. The primary analysis was performed in all patients with complete KCCQ at baseline and month 3. aSix patients in the fluid
restriction group did not complete the questionnaires at month 3 in relation to a preference to not continue with fluid restriction. Of these six patients, one withdrew
informed consent.

outpatient setting. The primary outcome measures of this randomized 139 (55.8%) were adherent to therapy based on their reported intake
trial were patient-reported outcomes, while safety was also evaluated. (fluid intake below 1,500 ml on at least 5 of 7 days), but 219 (87.6%)
patients self-reported that they were adherent during the entire
Results restricted period. Complete results for therapy adherence and mean
Patient characteristics fluid intake are shown in Extended Data Table 1.
Between 17 May 2021 and 13 June 2024, a total of 504 patients were
randomly assigned to liberal fluid intake (n = 254) or fluid restric- Outcomes
tion up to 1,500 ml per day (n = 250) (Fig. 1). The Kansas City Cardio- The KCCQ-OSS after 3 months was 74.0 in the liberal fluid intake group
myopathy Questionnaire (KCCQ) was completed by 242 (95.3%) and versus 72.2 in the fluid restriction group, with a mean difference after
233 (93.2%) patients at 3 months in the liberal fluid intake and fluid adjustment for baseline scores of 2.17 (95% confidence interval (CI)
restriction groups, respectively. Six patients in the fluid restriction −0.06 to 4.39; P = 0.06) (Fig. 2). Missing data analysis with multiple
group (2.4%) did not complete the questionnaires at month 3 in rela- imputation provided similar results (2.18, 95% CI −0.02 to 4.39; P = 0.05)
tion to a preference to not continue with the fluid restriction. Of these, (Extended Data Table 2). Perceived thirst distress was significant
one patient withdrew informed consent. lower in patients randomized to liberal fluid intake (TDS-HF: 16.9 versus
The baseline characteristics of the patients were balanced between 18.6, with a mean difference after adjustment for baseline scores of
both study arms (Table 1), except for the use of angiotensin recep- −2.29 (95% CI −1.09 to −3.49; P < 0.001)). Full results for the primary
tor–neprilysin inhibitor (P = 0.030), the use of any renin–angioten- and (key-)secondary outcomes are summarized in Table 2.
sin–aldosterone system inhibitors (P = 0.012) and medical history In those randomized to liberal fluid intake, significantly higher
with atrial fibrillation or flutter (P = 0.041). At baseline, the mean KCCQ Clinical Summary Score (KCCQ-CSS) and KCCQ Total Symp-
age was 69.2 ± 10.7 years, 67.3% of patients were male and most had tom Score (KCCQ-TSS) values were observed at 3-month follow-up
New York Heart Association (NYHA) class II symptoms (87.1%). Half of compared with those randomized to fluid restriction, with mean dif-
the patients had HF with reduced ejection fraction (51.6%). Just over ferences after adjustment for baseline scores of 2.39 (95% CI 0.20–
half (53.4%) adhered to some form of fluid restriction before study 4.57; P = 0.03) and 3.30 (95% CI 0.53–6.07; P = 0.02), respectively. The
participation. Almost all were receiving guideline-recommended results for the other KCCQ subdomains are shown in Extended
HF medical therapy. The median baseline KCCQ Overall Summary Data Table 3.
Score (KCCQ-OSS) and Thirst Distress Scale for patients with HF The proportion of patients with clinically meaningful changes
(TDS-HF) scores were 77.0 [interquartile range (IQR) 60.4–89.1] and in KCCQ-OSS was not significantly different for both randomization
15.5 [IQR 10.0–21.0], respectively. arms (increase or decrease of 5 points or more: 76 (31.4%) and 65 (26.9%)
Following randomization, the median patient-reported fluid versus 59 (25.3%) and 78 (33.5%) (P = 0.19) for liberal fluid intake com-
intake was significantly higher in patients with liberal fluid intake pared with fluid restriction, respectively). European Quality of Life Five
(1,764 ml [IQR 1,488–2,156]) compared with patients randomized Dimensions Five Levels questionnaire (EQ-5D-5L) scores were compa-
to fluid restriction (1,480 ml [IQR 1,357–1,561], difference of 284 ml; rable between both groups (liberal fluid intake: 0.83 [IQR 0.72–0.92]
P < 0.001). Of the patients randomized to liberal fluid intake, 178 (74.2%) versus fluid restriction: 0.81 [IQR 0.70–0.89]; P = 0.45).
and 82 (34.2%) had a mean intake above 1,500 ml and 2,000 ml per
day, respectively, compared with 98 (42.6%) and 11 (4.8%) patients in Safety
the fluid restriction arm, respectively. In 47 (20.4%) of the restricted No between-group significant differences were observed in the
patients, intake was above 1,600 ml. Of the restricted patients, composite of death, HF and all-cause hospitalizations or intravenous

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Table 1 | Characteristics of patients at baseline Table 1 (continued) | Characteristics of patients at baseline

Liberal fluid intake Fluid restriction Liberal fluid intake Fluid restriction
(n = 254) (n = 250) (n = 254) (n = 250)
Age, years 69.4 ± 10.6 69.0 ± 10.8 BMI, kg m−2 28.4 ± 5.2 27.9 ± 4.6
Male 170 (66.9) 169 (67.6) Weight, kg 86.2 ± 17.6 84.5 ± 16.6
White a
247 (97.2) 245 (98.0) Laboratory results
Quality of life Hemoglobin, mmol l−1 8.9 ± 0.9 9.0 ± 1.1
KCCQ-OSS 76.0 [59.9–90.2] 77.7 [61.7–88.5] Sodium, mmol l −1
139.6 ± 2.4 139.7 ± 2.5
TDS-HF 15.0 [10.0–22.0] 16.0 [11.0–21.0] Potassium, mmol l−1 4.6 ± 0.4 4.5 ± 0.4

EQ-5D-5L 0.85 [0.71–0.92] 0.81 [0.70–0.91] BUN, mmol l−1 8.0 ± 3.1 8.0 ± 3.6

NYHA functional class eGFR, ml min−1 1.73m−2 62.0 ± 17.2 62.6 ± 17.4

II 218 (85.8) 221 (88.4) NT-proBNP, ng l−1 430.0 507.4

[194.9–1,100.0] [193.5–1,300.0]
III 36 (14.2) 29 (11.6)
 Calculated osmolalityc, 293.7 ± 5.7 293.2 ± 5.7
LVEF mosm kg−1
% 40.3 ± 10.9 40.2 ± 10.8 Values are given as n (%), mean ± s.d. or median [IQR]. Analysis was conducted using
chi-squared or Fisher’s exact test and unpaired t-test or Mann–Whitney U-test, whichever
HFrEF 136 (53.5) 124 (49.6)
is appropriate. ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor
HFmrEF 60 (23.6) 70 (28.0) blocker; ARNI, angiotensin receptor–neprilysin inhibitor; BMI, body mass index; COPD,
chronic obstructive pulmonary disease; CRT, cardiac resynchronization therapy; DM,
HFpEF 58 (22.8) 56 (22.4) diabetes mellitus; eGFR, estimated GFR; GDMT, guideline-directed medical therapy; HFrEF,
heart failure with reduced ejection fraction; HFmrEF, heart failure with mildly reduced
Cause of HF
ejection fraction; HFpEF, heart failure with preserved ejection fraction; ICD, implantable
Ischemic 108 (42.5) 113 (45.2) cardioverter defibrillator; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid
receptor antagonists; NT-proBNP, N-terminal pro-B-type natriuretic peptide; RAASi, renin–
Nonischemic 146 (57.5) 137 (54.8) angiotensin–aldosterone system inhibitors; SGLT2i, sodium–glucose cotransporter-2
HF duration, years 5.0 [2.0–10.0] 4.0 [2.0–10.0] inhibitor. aSelf-reported. bCalculated only for patients with HFrEF, HFmrEF and heart
failure with improved ejection fraction. cCalculated as [sodium concentration] × 2 + [BUN
Fluid regimen pre-study participation concentration] + [glucose concentration].

Fluid restriction up to 1,000 ml 3 (1.2) 0 (0.0)

Fluid restriction up to 1,500 ml 43 (16.9) 46 (18.4) loop diuretic use or in the individual components of the outcome
 Fluid restriction up to 1,500 to 70 (27.6) 77 (30.8) (Table 3). The unmatched win ratio analysis resulted in 7.7% wins for the
2,000 ml liberal fluid intake and 7.8% wins for the fluid restriction (win ratio, 0.99;
95% CI 0.53–1.84; P = 0.97), resulting in a win odds of 1.00 (0.91–1.10;
Fluid restriction up to 2,000ml 14 (5.5) 16 (6.4)
P = 0.97) (Extended Data Fig. 1).
Liberal fluid intake 124 (48.8) 111 (44.4) Acute kidney injury occurred in 3 (1.2%) and 4 (1.6%) patients in
HF treatment the liberal fluid intake group and fluid restriction group, respectively.
RAASi 231 (90.9) 241 (96.4) No between-group differences were observed after 3 or 6 months
regarding changes in N-terminal pro–B-type natriuretic peptide con-
ACEi 43 (16.9) 34 (13.6)
centrations (−7.0 [IQR −90.0 to 100.0] versus −1.5 [IQR −98.3 to 133.8]
ARB 40 (15.7) 38 (15.2) and 13.2 [IQR −84.6 to 129.6] versus 0.0 [IQR −100.0 to 137.0], respec-
ARNI 149 (58.7) 170 (68.0) tively) or weight (0.0 [IQR −1.0 to 1.2] versus 0.0 [IQR −1.1 to 1.0] and
0.0 [IQR −2.2 to 1.0] versus 0.0 [IQR −1.3 to 1.0], respectively) (Table 3).
β-blocker 228 (89.8) 222 (88.8)
There was no significant difference in the observed changes
MRA 199 (78.3) 204 (81.6) (initiation, increases, decreases or termination) in oral loop diuretics
SGLT2i 144 (56.7) 162 (64.8) or other pharmacological HF therapy during 6 months of follow-up
Loop diuretics 133 (52.4) 125 (50.0) (Table 3).
The per-protocol analysis revealed similar results (Extended Data
 Furosemide equivalent 40 [20–60] 40 [20–60]
doses per day, mg
Tables 4–6).

Thiazides 6 (2.4) 7 (2.8)

Subgroup analyses
Digoxin 26 (10.2) 36 (14.4) The results of the pre-specified subgroup analyses demonstrate
GDMT score b
7.0 [5.0–8.0] 8.0 [6.0–9.0] a consistent treatment effect on the primary outcome with the
exception for baseline blood urea nitrogen (BUN) concentration
ICD 80 (31.5) 92 (36.8)
(Fig. 3 and Extended Data Fig. 2).
CRT 61 (24.0) 59 (23.6)
Medical history Discussion
The FRESH-UP study is a multicenter, randomized, open-label clinical
Atrial fibrillation or flutter 115 (45.3) 136 (54.4)
trial investigating the effect of a lifestyle advice of liberal fluid intake
COPD 32 (12.6) 33 (13.2) versus a lifestyle advice of fluid restriction up to 1,500 ml per day
DM 57 (22.4) 54 (21.6) on patient-reported outcomes and safety in 504 outpatients with
Hypertension 119 (46.9) 131 (52.4) chronic HF (NYHA II and III). No significant difference in KCCQ-OSS
was observed after 3 months, while thirst distress was significantly
Currently smoking 31 (12.2) 23 (9.2)
lower in the group of patients with liberal intake. There was no sign

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Adjusted mean difference 2.17 points (95% CI –0.06 to 4.39); P = 0.06 Table 3 | Safety outcomes and changes in medication
100
Liberal fluid Fluid restriction P value
intake (n = 254) (n = 249)

Safety events
90
Death 1 (0.4) 2 (0.8) 0.62
74.0 72.2
Mean KCCQ-OSS

All-cause hospitalization 20 (7.9) 15 (6.0) 0.42

(95% CI 71.5–76.6) (95% CI 69.6–74.7)
80
Hospitalization for HF 4 (1.6) 4 (1.6) 1.00

Intravenous loop diuretics usage 5 (2.0) 7 (2.8) 0.54

70 Acute kidney injurya 3 (1.2) 4 (1.6) 0.72

Any changes in loop diuretics 46 (18.1) 43 (17.3) 0.81

60  Furosemide equivalent doses 40 [20 to 40] 40 [20 to 40] 0.43

per day, mg
Liberal fluid intake Fluid restriction
0  Composite of death, any hospitalization 21 (8.3) 19 (7.6) 0.79
and intravenous loop diuretics
Baseline Month 3 Baseline Month 3
 Composite of death, HF hospitalization 6 (2.4) 9 (3.6) 0.41
Fig. 2 | Primary outcome: changes in KCCQ-OSS at 3 months. The primary and intravenous loop diuretics
outcome, KCCQ-OSS after 3 months, was 74.0 (95% CI 71.5 to 76.6) in the liberal
Win ratio 0.99 (0.53 to 1.84) 0.97
fluid intake group (n = 242) versus 72.2 (95% CI 69.6 to 74.7) in the fluid restriction
group (n = 233), with a mean difference after adjustment for baseline scores of Win odds 1.00 (0.91 to 1.10) 0.97

2.17 (95% CI −0.06 to 4.39; P = 0.06). KCCQ-OSS values are unadjusted means Net benefit −0.001 (−0.049 to 0.047) 0.97
(95% CI). NT-proBNP, ng l−1b

Month 3 422.0 [173.0 to 527.2 [196.4 to 0.19

1,200.0] 1,411.5]
Table 2 | Clinical outcomes Δ Baseline to month 3c −7.0 [−90.0 to −1.5 [−98.3 to 0.52
100.0] 133.8]
Liberal fluid intake Fluid restriction P value Month 6 530.5 [185.8 to 553.8 [179.8 to 0.65
(n = 242) (n = 233) 1,300.5] 1,425.0]
Primary outcome Δ Month 3 to month 6c 13.2 [−84.6 to 0.0 [−100.0 to 0.51
129.6] 137.0]
KCCQ-OSS 74.0 (71.5–76.6) 72.2 (69.6–74.7) 0.056 a

Weight, kgd
Key secondary outcome
Δ Baseline to month 3c 0.0 [−1.0 to 1.2] 0.0 [−1.1 to 1.0] 0.39
TDS-HF 16.9 (15.8–18.0) 18.6 (17.5–19.6) <0.001a
Δ Month 3 to month 6c 0.0 [−2.2 to 1.0] 0.0 [−1.3 to 1.0] 0.46
Other secondary outcomes
Medication changes
KCCQ-CSS 75.9 (73.4–78.4) 74.5 (71.9–77.1) 0.032a
Between baseline and month 3
KCCQ-TSS 78.5 (75.9–81.1) 77.2 (74.5–79.9) 0.020a
Loop diuretics any changes 21 (8.3) 23 (9.2) 0.70
KCCQ-OSS (−5 to +5)b 101 (41.7) 96 (41.2) 0.19  Loop diuretics dose increased or 10 (3.9) 14 (5.6) 0.38
initiation
 KCCQ-OSS 65 (26.9) 78 (33.5)
(−5 or less)b  Loop diuretics dose decreased or 11 (4.3) 9 (3.6) 0.68
termination
 KCCQ-OSS 76 (31.4) 59 (25.3)
(+5 or more)b Any HF medication changese 48 (18.9) 49 (19.7) 0.82

EQ-5D-5L 0.83 [0.72–0.92] 0.81 [0.70–0.89] 0.45 a Between month 3 and month 6

(n = 240) (n = 230) Any loop diuretics changes 28 (11.0) 27 (10.8) 0.95

 Loop diuretics dose increased or 19 (7.5) 19 (7.6) 0.95

Reported fluid intake, ml 1,764 [1,488–2,156] 1,480 [1,357–1,561] <0.001
initiation
Reported fluid intake, 21.6 [17.4–26.2] 17.9 [15.4–20.6] <0.001
 Loop diuretics dose decreased or 11 (4.3) 13 (5.2) 0.64
ml kg−1 termination
Clinical outcomes were assessed after 3 months. Values are given as n (%), unadjusted mean
Any HF medication changese 67 (26.4) 53 (21.3) 0.18
(95% CI) or unadjusted median [IQR]. ANCOVA, chi-squared and Mann–Whitney U-test were
used for analysis. aBased on the adjusted mean difference tested two-sided with ANCOVA During study follow-up
using baseline value as the covariate. bA change of 5 points is considered to be clinically
Loop diuretics any changes 46 (18.1) 43 (17.3) 0.81
important.
 Loop diuretics dose increased or 29 (11.4) 29 (11.6) 0.94
initiation
of compromised safety when a liberal fluid intake was prescribed,  Loop diuretics dose decreased or 20 (7.9) 22 (8.8) 0.70
as indicated by the absence of differences in safety events including termination

mortality, (HF) hospitalizations or change in HF pharmacological Any HF medication changese 98 (38.6) 82 (32.9) 0.19
therapy (for example, initiation or dose increase of loop diuretics). Safety was assessed during the 6 months follow-up. Values are given as n (%), median
Restriction of fluid intake is a common recommendation for [IQR] or ratio (95% CI). Analyses were conducted using chi-squared or Fisher’s exact test,
patients with HF, but with a low level of evidence regarding both effi- whichever is appropriate, and Mann–Whitney U-test. aAcute kidney injury is defined as a 50%
decline in eGFR relative to baseline, or decrease of >30 ml min−1 1.73 m−2 and to a value below
cacy and safety. Four previous studies have reported on the effects
60 ml min−1 1.73 m−2. bAvailable values at month 3 (n = 225 versus n = 224) and month 6 (n = 203
of solely liberal intake or fluid restriction in small study populations versus n = 193). cDifferences between two timepoints in individual patients with complete
(52–87 patients), in heterogeneous settings (for example, inpatient, data. dAvailable values at month 3 (n = 145 versus n = 139) and month 6 (n = 58 versus n = 66).
recently discharged, outpatient) with a variety of intervention methods
e
Changes in HF medication refers to changes in RAASi, beta-blockers, MRA, SGLT2i, loop and
thiazide diuretics.
(for example, fluid restriction up to 1,000, 1,500 or 2,500 ml or

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Liberal fluid intake Fluid restriction Adjusted mean difference P value

Subgroup % (n/N) % (n/N) Mean (95% CI) (interaction)
Sex at birth 0.16
Male 66.1% (160/242) 67.0% (156/233) 1.06 (−1.67 to 3.79)
Female 33.9% (82/242) 33.0% (77/233) 4.47 (0.58 to 8.36)
Median age 0.62
<71 years 46.3% (112/242) 48.1% (112/233) 2.83 (−0.21 to 5.87)
≥71 years 53.7% (130/242) 51.9% (121/233) 1.71 (−1.52 to 4.95)
HF type 0.42
HFrEF 52.9% (128/242) 48.1% (112/233) 0.77 (−2.37 to 3.91)
HFmrEF 23.1% (56/242) 28.8% (67/233) 4.33 (−0.30 to 8.96)
HFpEF 24.0% (58/242) 23.2% (54/233) 2.87 (−1.58 to 7.32)
NYHA class 0.18
II 85.1% (206/242) 89.3% (208/233) 1.75 (−0.52 to 4.02)
III 14.9% (36/242) 10.7% (25/233) 5.40 (−2.57 to 13.38)
Median NT−proBNP 0.66
–1
<451.5 ng l 53.7% (130/242) 48.1% (112/233) 2.50 (−0.39 to 5.38)
–1
≥451.5 ng l 46.3% (112/242) 51.9% (121/233) 1.54 (−1.86 to 4.94)
Median BMI 0.94
2
<27.68 kg m 48.5% (117/241) 51.3% (118/230) 1.99 (−1.16 to 5.13)
2
≥27.68 kg m 51.5% (124/241) 48.7% (112/230) 2.14 (−1.06 to 5.34)
Diabetes mellitus 0.07
Yes 22.3% (54/242) 22.3% (52/233) −1.68 (−6.93 to 3.56)
No 77.7% (188/242) 77.7% (181/233) 3.25 (0.81 to 5.70)
Median sodium 0.15
–1
<140.0 mmol l 45.9% (111/242) 47.2% (110/233) 0.44 (−2.68 to 3.56)
–1
≥140.0 mmol l 54.1% (131/242) 52.8% (123/233) 3.68 (0.53 to 6.84)
Median BUN 0.006
–1
<7.4 mmol l 49.6% (120/242) 46.8% (109/233) 5.37 (2.40 to 8.34)
–1
≥7.4 mmol l 50.4% (122/242) 53.2% (124/233) −0.82 (−4.08 to 2.45)
Median creatinine 0.82
–1
<97.0 mmol l 43.0% (104/242) 49.4% (115/233) 1.91 (−1.25 to 5.06)
–1
≥97.0 mmol l 57.0% (138/242) 50.6% (118/233) 2.43 (−0.73 to 5.59)
Loop diuretics usage 0.53
Yes 53.3% (129/242) 49.8% (116/233) 1.54 (−1.62 to 4.70)
No 46.7% (113/242) 50.2% (117/233) 2.97 (−0.19 to 6.13)
Overall 100% (242/242) 100% (233/233) 2.17 (−0.06 to 4.39)

−7 −6 −5 −4 −3 −2 −1 0 1 2 3 4 5 6 7 8
Adjusted mean difference

Fluid restriction better Liberal fluid intake better

Fig. 3 | Primary outcome, according to pre-specified subgroups. The primary outcome of the trial, adjusted mean differences in the KCCQ-OSS, tested two-sided
with ANCOVA using baseline KCCQ-OSS as covariate, according to subgroups that were pre-specified in the protocol. The pre-specified subgroup analysis for each
study site is provided in Extended Data Fig. 2.

based on body weight) and study design (parallel or cross-over)8–11. With regard to the primary outcome, a nonsignificant difference
These earlier studies provided mixed results that generally suggested between groups in KCCQ-OSS at follow-up after adjustment for base­
that liberal fluid intake would not be associated with a deleterious line scores of 2.17 (95% CI −0.06 to 4.39; P = 0.06) was demonstrated.
impact on health status or outcome8–11. Although this difference is not significant, it implies that according
Accordingly, before this study, the evidence to support fluid to the 95% CI, the actual true mean difference in KCCQ-OSS between
restriction in any setting was low, which was also recently underscored groups may be somewhere between 0.06 points lower and 4.39 points
by the Heart Failure Association of the European Society of Cardiol- higher for those who received advice for liberal fluid intake. On a
ogy18. This created equipoise for a well-powered randomized trial to patient-level, a change of 5 points in KCCQ-OSS follow-up scores
explore the question. is considered a clinically relevant change19. But importantly, smaller
The FRESH-UP study demonstrated that different lifestyle advice mean differences between groups may already be of clinical rele­
regarding fluid intake affected the behavior of the study population. vance, as recently demonstrated 20. For example, a mean differ-
A difference in self-reported daily fluid intake was observed, with ence of about 2.5 points between groups may seem small, but can
higher fluid intake of about 280 ml per day in the group with liberal actually relate to 50% of patients with a clinically relevant change
fluid intake. In this group, mean fluid intake remained in the recently of 5 points.
proposed normal range18. Fluid intake was reported for a relatively Because the result is nonsignificant, it would be inappropriate to
short period of 1 week, because it was considered that a longer period state that liberal fluid intake leads to a higher health status as assessed
of self-reporting on fluid intake might interfere with one of the sup- by the KCCQ-OSS. However, the observed difference between both
posed advantages of liberal fluid intake: patients are not required to randomization arms does make it very unlikely that fluid restriction
think about or report their fluid intake in their daily life. It remains, in a low-risk HF population will result in an improved health status
however, speculative what led to the nonsignificant difference in as assessed by the KCCQ-OSS. Moreover, no signal of harm related to
health status: for example, the actual difference in fluid intake, the liberal fluid intake in terms of safety events, serum biomarkers, weight
freedom of constant monitoring for those in the liberal fluid intake or change in HF medication was observed; although liberal fluid
group or other mechanisms. intake did result in lower thirst distress. This should be interpreted as

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

reassuring that there are no signals of harm when adhering to liberal 2. van der Wal, M. H. L., Waldreus, N., Jaarsma, T. & Kato, N. P. Thirst
fluid intake and should not be interpreted as definite equivalence in patients with heart failure in Sweden, the Netherlands, and
between the two strategies. Accordingly, our findings suggest no del- Japan. J. Cardiovasc. Nurs. 35, 19–25 (2020).
eterious impact of liberal fluid intake and question the need for fluid 3. van der Wal, M. H. L., Jaarsma, T., Jenneboer, L. C. & Linssen, G. C. M.
restriction among patients with stable, symptomatic chronic HF. Thirst in stable heart failure patients; time to reconsider fluid
Importantly, these results cannot be extrapolated to patients restriction and prescribed diuretics. ESC Heart Fail. 9, 2181–2188
who have more unstable HF, have been recently hospitalized or (2022).
have hyponatremia8, because these patients were excluded from 4. Vazir, A. et al. Decongestion strategies in patients presenting with
the FRESH-UP study. Future studies should also explore the effects acutely decompensated heart failure: a worldwide survey among
of fluid intake in patients with more advanced HF. Although this is physicians. Eur. J. Heart Fail. 25, 1555–1570 (2023).
a well-powered study evaluating the effect of liberal fluid intake in 5. Hoog, L., Stromberg, A., Waldreus, N. & Nymark, C. A national
chronic HF and focusing on important patient-reported outcomes as survey of healthcare professional’s clinical practice on fluid intake
a key outcome, it should be acknowledged that this trial has several in patients with heart failure. Abstract. Eur. J. Heart Fail. 26 (S2),
limitations. First, the potential biases inherent to open-label studies 3–643 (2024).
apply to the FRESH-UP study (for example, a possible reporting bias 6. De Vecchis, R., Baldi, C., Cioppa, C., Giasi, A. & Fusco, A. Effects
for thirst distress and/or fluid intake cannot be excluded)21. To limit of limiting fluid intake on clinical and laboratory outcomes
bias, the clinical events and safety analyses were evaluated blinded to in patients with heart failure. Results of a meta-analysis of
treatment allocation (PROBE design). Second, despite the study being randomized controlled trials. Herz 41, 63–75 (2016).
the largest in this field of research, it still concerns a relatively small 7. Li, Y., Fu, B. & Qian, X. Liberal versus restricted fluid
population with a short duration of follow-up. Although there is no administration in heart failure patients. A systematic review
evidence that safety was compromised, the study was not sufficiently and meta-analysis of randomized trials. Int. Heart J. 56, 192–195
powered for clinical events. Third, it is conceivable that the way the (2015).
lifestyle advice was presented to the participants has an effect on the 8. Albert, N. M., Nutter, B., Forney, J., Slifcak, E. & Tang, W. H.
outcome measures. To limit variation between sites, a standardized A randomized controlled pilot study of outcomes of strict
script was provided to each participating site on how to provide the allowance of fluid therapy in hyponatremic heart failure
lifestyle advice. The wording regarding the level of uncertainty about (SALT-HF). J. Card. Fail. 19, 1–9 (2013).
the efficacy and safety of both fluid regimens can be a matter of debate, 9. Holst, M., Stromberg, A., Lindholm, M. & Willenheimer, R.
yet it was based on the limited available data at the time of the study Liberal versus restricted fluid prescription in stabilised patients
design. Fourth, interpretation of the TDS-HF warrants caution because with chronic heart failure: result of a randomised cross-over
it has not yet been correlated with quality of life and further validation study of the effects on health-related quality of life, physical
is needed. Also, most of the study population was white, male and of capacity, thirst and morbidity. Scand. Cardiovasc. J. 42, 316–322
European descent, which does not reflect the global HF population, (2008).
making these results less generalizable to female patients or patients 10. Reilly, C. M., Higgins, M. & Dunbar, S. B. Clinical, symptom,
from other racial groups. In addition, no systematic screening log was functional, and QOL outcomes in a trial of prescribed fluid
recorded. Furthermore, the patient-reported fluid intake at week 6 may restrictions in persons with heart failure. Circulation 136, A19846
not be fully representative of the fluid intake during the entire study (2017).
period. The level of adherence to the different lifestyle advice may have 11. Travers, B. et al. Fluid restriction in the management of
attenuated power to detect between-group differences; however, this decompensated heart failure: no impact on time to clinical
is reflective of daily clinical practice. Lastly, other dietary data were not stability. J. Card. Fail. 13, 128–132 (2007).
collected (for example, sodium intake). 12. Holst, M., Stromberg, A., Lindholm, M. & Willenheimer, R.
In summary, the FRESH-UP is the largest randomized clinical trial Description of self-reported fluid intake and its effects on body
to investigate the effect of lifestyle advice of liberal fluid intake versus weight, symptoms, quality of life and physical capacity in patients
lifestyle advice of fluid restriction in patients with chronic HF. In the with stable chronic heart failure. J. Clin. Nurs. 17, 2318–2326
FRESH-UP study, the advice of a liberal fluid intake did not result in a (2008).
significant difference in health status compared with advice of a fluid 13. Waldreus, N., Hahn, R. G. & Jaarsma, T. Thirst in heart failure:
restriction up to 1,500 ml per day, but liberal fluid intake did result in a a systematic literature review. Eur. J. Heart Fail. 15, 141–149
lower perceived thirst distress. No signals of compromised safety were (2013).
observed when a liberal fluid intake was prescribed, as indicated by the 14. Waldreus, N., Chung, M. L., van der Wal, M. H. & Jaarsma, T.
absence of significant differences in safety events including mortality, Trajectory of thirst intensity and distress from admission to
(HF) hospitalizations or change in HF medications. These findings 4-weeks follow up at home in patients with heart failure. Patient
suggest no deleterious impact of liberal fluid intake and question the Prefer. Adherence 12, 2223–2231 (2018).
need of fluid restriction among patients with stable, symptomatic 15. van der Wal, M. H. et al. Compliance in heart failure patients: the
chronic HF. importance of knowledge and beliefs. Eur. Heart J. 27, 434–440
(2006).
Online content 16. Heidenreich, P. A. et al. 2022 AHA/ACC/HFSA Guideline for the
Any methods, additional references, Nature Portfolio reporting sum- Management of Heart Failure: a report of the American College
maries, source data, extended data, supplementary information, of Cardiology/American Heart Association Joint Committee on
acknowledgements, peer review information; details of author contri- Clinical Practice Guidelines. J. Am. Coll. Cardiol. 79, e263–e421
butions and competing interests; and statements of data and code avail- (2022).
ability are available at https://doi.org/10.1038/s41591-025-03628-4. 17. McDonagh, T. A. et al. 2021 ESC Guidelines for the diagnosis and
treatment of acute and chronic heart failure. Eur. Heart J. 42,
References 3599–3726 (2021).
1. Eng, S. H. et al. Thirst and factors associated with frequent thirst 18. Mullens, W. et al. Dietary sodium and fluid intake in heart failure.
in patients with heart failure in Spain. Heart Lung 50, 86–91 A clinical consensus statement of the Heart Failure Association of
(2021). the ESC. Eur. J. Heart Fail. 26, 730–741 (2024).

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

19. Spertus, J. A., Jones, P. G., Sandhu, A. T. & Arnold, S. V. Publisher’s note Springer Nature remains neutral with regard to
Interpreting the Kansas City Cardiomyopathy Questionnaire in jurisdictional claims in published maps and institutional affiliations.
clinical trials and clinical care: JACC State-of-the-Art Review.
J. Am. Coll. Cardiol. 76, 2379–2390 (2020). Springer Nature or its licensor (e.g. a society or other partner) holds
20. Abdel Jawad, M. et al. Interpreting population mean treatment exclusive rights to this article under a publishing agreement with
effects in the Kansas City Cardiomyopathy Questionnaire: the author(s) or other rightsholder(s); author self-archiving of the
a patient-level meta-analysis. JAMA Cardiol. 10, 32–40 accepted manuscript version of this article is solely governed by the
(2025). terms of such publishing agreement and applicable law.
21. Merlo, A., Mezue, K. & Ambrosy, A. P. Fluid restriction recommenda­
tions in heart failure: dry as a bone or quench your thirst? J. Card. © The Author(s), under exclusive licence to Springer Nature America,
Fail. 28, 1531–1533 (2022). Inc. 2025

1
Department of Cardiology, Radboud University Medical Center, Nijmegen, the Netherlands. 2Netherlands Heart Institute, Utrecht, the Netherlands.
3
Department of Cardiology, Maastricht University Medical Center, Maastricht, the Netherlands. 4Cardiovascular Research Institute Maastricht (CARIM),
Maastricht University, Maastricht, the Netherlands. 5Department of Cardiology, Bernhoven Hospital, Uden, the Netherlands. 6Department of Cardiology,
Jeroen Bosch Hospital, ‘s-Hertogenbosch, the Netherlands. 7Department of Health, Medicine and Caring Sciences, Linköping University, Linköping,
Sweden. 8Department of Cardiology, University Medical Center Utrecht, Utrecht, the Netherlands. 9Department of Cardiology, Hospital Group Twente,
Almelo/Hengelo, the Netherlands. 10Department of Cardiology, Rijnstate Hospital, Arnhem, the Netherlands. 11Department of Cardiology, Zuyderland
Medical Center, Heerlen/Sittard-Geleen, the Netherlands. 12Department of Cardiology, University Medical Centre Groningen, University of Groningen,
Groningen, the Netherlands. 13Centre for Cardiovascular Diseases, University Hospital Brussels, Jette, Belgium. 14Faculty of Medicine and Pharmacy, Vrije
Universiteit Brussel, Brussels, Belgium. 15Baim Institute for Clinical Research, Cardiology Division, Massachusetts General Hospital and Harvard Medical
School, Boston, MA, USA. 16Cardiology Service, Hospital Universitari Germans Trias i Pujol, CIBERVC, Badalona, Spain. 17Department of Health Research
Methods, Evidence, and Impact, McMaster University, Hamilton, Ontario, Canada. 18IQ Health Department, Radboud University Medical Center, Nijmegen,
the Netherlands. 19Department of Cardiology, Máxima Medical Center, Veldhoven, the Netherlands. e-mail: [email protected]

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Methods outcome measure for overall health status in patients with HF, has high
Study design predictive value for future events and is widely used in clinical trials
The FRESH-UP study is an investigator-initiated, multicenter, rando­ as the primary outcome19,20,24. KCCQ-OSS has a range from 0 to 100, in
mized, open-label clinical trial conducted at seven sites in the which a higher score indicates better health status. On a patient-level,
Netherlands between May 2021 and June 2024. The rationale and design a 5-point change (improvement or deterioration) in the KCCQ-OSS is
of the trial have been published previously22. The trial was performed a small but clinically important change and is associated with cardio-
in accordance with the previously published protocol and statisti- vascular death and hospitalization19,20. These within-patient thresh-
cal analysis plan (SAP) (Supplementary Information). The trial was olds cannot be applied to mean differences in KCCQ-OSS on a group
approved by the Medical Research Ethics Committee of Radboud level because smaller changes at this group level are associated with
University Medical Center and all institutional review boards of the differences in the proportion of patients improving by 5 points or
participating sites (Maastricht University Medical Center, Zuyderland more, reflecting treatment benefit for the individual patient20.
Medical Center, Rijnstate Hospital, Hospital Group Twente, Bernhoven The key secondary outcome of the trial was thirst distress as
Hospital and Jeroen Bosch Hospital). assessed by the TDS-HF at the same timepoint. The TDS-HF is a 5-point
The first author was the only investigator with full access to all Likert scale that consists of eight statements about thirst (Supplemen-
source data during the conduct of the study and was not involved tary Information). Patients were asked to rate these from 1 (strongly
in any clinical decision-making throughout the trial. The lead inves- disagree) to 5 (strongly agree)23. The total TDS-HF score ranges from
tigators (the last and second-to-last authors) and an independent 8 to 40. Higher scores indicate higher thirst distress and can be divided
event adjudication committee (the latter reviewed and adjudicated all into five categories: no distress (8 points), mild distress (9–16 points),
serious adverse events), were blinded for randomization and had no moderate distress (17–24 points), high distress (25–32 points) and
direct access to the source data. The members of the Data Safety and severe distress (33–40 points)25.
Monitoring Board, who performed two interim safety analyses, were Other secondary and exploratory outcomes included compari-
blinded for the follow-up health status data. The investigators did not sons of the KCCQ-CSS and KCCQ-TSS, and the proportion of patients
perform analyses on either efficacy or safety before database lock. with differences of 5 points or more in KCCQ-OSS compared with
baseline and EQ-5D-5L, patient-reported fluid intake at week 6, safety
Patients outcomes ((composite of) death, all-cause and HF hospitalization,
Patients treated for chronic HF for more than 6 months who had the requirement of intravenous loop diuretics and HF hospitalization,
NYHA class II or III symptoms were included. Patients were eligible acute kidney injury during the total 6 months of clinical follow-up),
regardless of left ventricular ejection fraction. Major exclusion crite- serum biomarkers, weight and medication use. The event adjudication
ria were recent (3 months before randomization) HF hospitalization, committee independently adjudicated the cause of death as HF-related,
coronary intervention or implantation of a pacemaker device, changes not HF-related but cardiovascular or not cardiovascular, and any hos-
in medical HF therapy in the 14 days before randomization, a glome­ pitalization as either primary HF-related or not.
rular filtration rate (GFR) < 30 ml min−1 1.73 m−2 and hyponatremia
(sodium concentration <130 mmol l−1). The complete inclusion and Statistical analysis
exclusion criteria are listed in the Supplementary Information. All The trial was designed to detect a difference of 2.5 points in KCCQ-OSS
patients provided written informed consent before participation. follow-up scores after 3 months between both randomization groups,
based on corresponding results of randomized HF trials19,24,26–29. For the
Randomization and study procedures liberal fluid intake group, a KCCQ-OSS of 66.25 points with a standard
Patients were randomly assigned in a 1:1 ratio to a standardized deviation of 20 points at follow-up was assumed, adjusted for baseline
lifestyle advice (Supplementary Information) of either liberal fluid scores with an estimated correlation of 0.88 between baseline and
intake (intake as desired without a maximum) or fluid restriction up follow-up scores19,24,26–29. To test this difference, 454 evaluable patients
to 1,500 ml per day for a period of 3 months. Randomization was per- were required. Anticipating a drop-out rate of 10%, the aim was to
formed using a web-based system (Castor EDC) with a random block include 506 patients.
randomization algorithm with block sizes of six, eight and ten and was Statistical analyses were performed according to the published
stratified by site. pre-specified SAP. Analyses of the primary, secondary and exploratory
At baseline, before randomization, the KCCQ, TDS-HF and outcomes were executed according to the intention-to-treat princi-
EQ-5D-5L were completed19,23,24. At 6 weeks following randomization, ple and after the final database lock. The primary and key secondary
during the intervention period, patients reported their daily fluid intake outcomes were tested hierarchically. For other secondary outcomes,
during seven consecutive days in a fluid intake diary. Three months no corrections for multiple testing were executed. For the primary
after randomization, patients visited the outpatient clinic to complete outcome, all patients with complete KCCQ-OSS at baseline and at
the follow-up KCCQ, TDS-HF and EQ-5D-5L questionnaires, or were 3-month follow-up were included, no data were imputed. The difference
contacted by phone and received the questionnaires on paper and were between the two randomization arms in KCCQ-OSS after 3 months
asked to return them by mail, or were asked to complete their question- was tested with the analysis of covariance (ANCOVA), using baseline
naires via a digital survey. Patients were also clinically evaluated accord- KCCQ-OSS as a covariate and testing for ANCOVA assumptions. For
ing to standard clinical practice, and adverse events were assessed. After the primary outcome, a sensitivity analysis for missing data was per-
this visit, patients were followed up until 6 months post-randomization formed with multiple imputation, as described in the SAP. The con-
for assessment of adverse events. Fluid management after month 3 sistency of the treatment effect was assessed among 12 pre-specified
was at the discretion of the treating physician and the patient. For subgroups. An additional per-protocol analysis was performed based
the final study visit, 6 months after randomization, patients visited on patient-reported fluid intake at week 6. For the fluid restriction
the outpatient clinic or were contacted by phone for clinical and arm, therapy adherence was defined as a daily fluid intake ≤1,500 ml
safety assessment. Study participation ended after this visit. Further on at least 5 of 7 days. If the patient-reported intake was not available,
details regarding follow-up are available in Supplementary Table 1. nonadherence was assumed. Patients randomized to the liberal fluid
intake arm were, by definition, considered adherent.
Outcomes The analyses of the secondary ordinal quality of life outcomes
The primary outcome was the result of the KCCQ-OSS at 3 months followed a similar approach to the primary analysis. Between-group dif-
after randomization. The KCCQ is a validated 23-item patient-reported ferences in proportions (for example, patients with clinically meaningful

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

changes in KCCQ or safety outcomes) and patient-reported fluid intake M.H.I.V., D.H.F.G. and R.R.J.v.K. acquired funding for the project. L.R.
were analyzed with chi-squared or Fisher’s exact test, and Mann–Whitney and D.H.F.G. planned the statistical analysis, which was executed by
U-test, respectively. J.J.H. and L.R. J.J.H. and L.R. accessed and verified the underlying
The exploratory safety outcomes were analyzed using an study data. J.J.H., H.-P.B.-L.R., L.E.H.J.M.B., F.B.-W., S.C.A.M.B., L.B.,
unmatched win ratio in which each comparison of two patients fol- J.W.M.v.E., H.C.H., G.C.M.L., R.P., S.S.-v.W. and M.H.I.V. participated in
lowed the hierarchy of mortality, HF hospitalization, requirement of patient enrollment and carried out the trial. J.J.H. drafted the initial
intravenous loop diuretics and all-cause hospitalization within the version of the manuscript with input from D.H.F.G. and R.R.J.v.K. All
total 6 months of follow-up. Analyses were performed with SPSS v.29 authors reviewed the data analyses, data interpretation and writing of
(IBM), R v.4.1.3 (R foundation for Statistical Computing) and Stata v.18.0 the report. All authors had full access to all the data in the study and
(StataCorp). The trial was prospectively registered at ClinicalTrials. had final responsibility for the decision to submit for publication. All
gov (NCT04551729). authors have seen and approved of the manuscript before submission.

Reporting summary Competing interests

Further information on research design is available in the Nature H.-P.B.-L.R. has received research grants from Innovative Health
Portfolio Reporting Summary linked to this article. Initiative Joint Undertaking and Roche Diagnostics, received
consulting fees from Astra Zeneca, Boehringer Ingelheim, Novartis,
Data availability Roche Diagnostics and Vifor Pharma, received honoraria from Roche
Anonymized participant data can be made available upon requests Diagnostics and participated on an Advisory Board of CeleCor
directed to the corresponding author. Data requests will be reviewed Therapeutics. S.S.-v.W. has received research grants from Boehringer
on the basis of scientific merit, ethical review, available resources and Ingelheim, CRL funds and ZonMW, received honoraria from Astra
regulatory requirements, and a response can be expected within two Zeneca, Bayer, Boehringer Ingelheim, Novartis, Novo Nordisk,
weeks. Following approvals, unidentified patient data and a data dic- Roche Diagnostics and Pfizer, and is a member of the board of the
tionary will be provided. The steering committee will have the right to Werkgroep Cardiologisch Centra Nederland and Stichting Perfusie.
review and comment on any draft papers based on these data before P.v.d.M. has received research grants from Astra Zeneca, Pfizer,
publication. Pharma Nord, Novo Nordisk, Ionis and Vifor Pharma, honoraria from
Astra Zeneca, Boehringer Ingelheim, BridgeBio, Daiichi Sankyo, Ionis,
References Novartis, Novo Nordisk, Pfizer, Pharma Nord, Pharmacosmos and Vifor
22. Herrmann, J. J. et al. Fluid REStriction in heart failure vs liberal Pharma, and participated on an Advisory Board for CorHeart. J.L.J.
fluid UPtake: rationale and design of the randomized FRESH-UP has received research grants from Abbott, Applied Therapeutics,
study. J. Card. Fail. 28, 1522–1530 (2022). Astra Zeneca, HeartFlow and Novartis, received consulting fees from
23. Waldreus, N., Jaarsma, T., van der Wal, M. H. & Kato, N. P. Abbott, Applied Therapeutics, Astra Zeneca, Beckman, Boehringer
Development and psychometric evaluation of the Thirst Distress Ingelheim, Bristol Myers, Intellia, Jana Care, Novartis, Pfizer, Merck,
Scale for patients with heart failure. Eur. J. Cardiovasc. Nurs. 17, Roche Diagnostics and Siemens, participated on an Advisory Board
226–234 (2018). for Abbott, AbbVie, Bayer, CVRx, Roche Diagnostics and Takeda, is a
24. Green, C. P., Porter, C. B., Bresnahan, D. R. & Spertus, J. A. member of the board of American College of Cardiology and Imbria
Development and evaluation of the Kansas City Cardiomyopathy Pharmaceuticals, and has stock (options) in Imbria Pharmaceuticals
Questionnaire: a new health status measure for heart failure. and Jana Care. A.B.-G. has lectured and/or participated in advisory
J. Am. Coll. Cardiol. 35, 1245–1255 (2000). boards for Abbott, Astra Zeneca, Bayer, Boehringer Ingelheim,
25. Eng, S. H. et al. Thirst distress in outpatients with heart failure in a Medtronic, Novartis, Novo Nordisk, Roche Diagnostics and Vifor
Mediterranean zone of Spain. ESC Heart Fail. 8, 2492–2501 (2021). Pharma. D.H.F.G. received a speaker’s fee from Novartis and travel
26. McMurray, J. J. et al. Angiotensin–neprilysin inhibition versus support from Astra Zeneca. R.R.J.v.K. received consulting fees from
enalapril in heart failure. N. Engl. J. Med. 371, 993–1004 (2014). Novo Nordisk and Roche Diagnostics, honoraria from Hippocrates
27. Veazie, P. J. et al. Cardiac resynchronization and quality of life in Academy and is a fiduciary member of the ESC working group on
patients with minimally symptomatic heart failure. J. Am. Coll. Peripheral Artery and Aortic Disease. The other authors declare no
Cardiol. 60, 1940–1944 (2012). competing interests.
28. Lewis, E. F. et al. Health-related quality of life outcomes in
PARADIGM-HF. Circ. Heart Fail. 10, e003430 (2017). Additional information
29. Ekman, I. et al. Heart rate reduction with ivabradine and health Extended data is available for this paper at
related quality of life in patients with chronic heart failure: results https://doi.org/10.1038/s41591-025-03628-4.
from the SHIFT study. Eur. Heart J. 32, 2395–2404 (2011).
Supplementary information The online version
Acknowledgements contains supplementary material available at
We thank the Netherlands Heart Institute for their logistical support, https://doi.org/10.1038/s41591-025-03628-4.
and both the Dutch Heart Foundation (grant no. 2019T100 to R.R.J.v.K.)
as well as the Academic Alliance Fund of the Radboud University Correspondence and requests for materials should be addressed to
Medical Center, Nijmegen and Maastricht University Medical Center Roland R. J. van Kimmenade.
(grant R0005106) for their financial support (H.-P.B.-L.R., F.B.-W.,
D.H.F.G., R.R.J.v.K.). The funder of the study had no role in study design, Peer review information Nature Medicine thanks Justin Ezekowitz,
data collection, data analysis, data interpretation or writing of the Wilfried Mullens and the other, anonymous, reviewer(s) for their
report. contribution to the peer review of this work. Primary Handling Editor:
Michael Basson, in collaboration with the Nature Medicine team.
Author contributions
H.-P.B.-L.R., T.J., F.B.-W., L.B., D.H.F.G. and R.R.J.v.K. wrote the study Reprints and permissions information is available at
protocol. All authors approved the protocol. L.E.H.J.M.B., F.B.-W., www.nature.com/reprints.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Fig. 1 | Win ratio. The win ratio for safety outcomes during the six months follow-up.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Fig. 2 | Primary outcome, according to pre-specified subgroup: study site. The primary outcome of the trial, adjusted mean difference in the
KCCQ Overall Summary, tested two-sided with ANCOVA analysis using baseline KCCQ Overall Summary Score as covariate, according to the pre-specified subgroup:
study site.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 1 | Therapy adherence and mean fluid intake

Therapy adherence and mean fluid intake based on patient-reported intake at week 6. Values are N (%). Chi-square test. †For the fluid restriction arm, therapy adherence was defined as a daily
fluid intake ≤1500 ml in at least five of seven days. If the patient-reported intake was not available, non-adherence was assumed. Patients randomized to the liberal fluid intake arm were, by
definition, considered adherent.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 2 | Missing data analysis

Primary outcome at 3 months after randomization. Missing data were imputed according to the pre-specified statistical analysis plan. The imputation models included baseline values of
the respective outcomes, all variables intended for the subgroup analyses (variables dichotomized for the subgroup analyses were included as continuous in the imputation model) and
the advised fluid regimen prior to randomization. Imputation was performed separately by treatment arm. Unadjusted mean (95% confidence interval) or adjusted mean difference (95%
confidence interval). KCCQ-OSS = Kansas City Cardiomyopathy Questionnaire Overall Summary Score; TDS-HF = Thirst Distress Scale for patients with HF. †Based on the adjusted mean
difference tested two-sided with ANCOVA analysis using baseline value as covariate.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 3 | KCCQ subdomains

KCCQ subdomains after 3 months. Total Symptom Score = mean of Symptom Burden Score and Symptom Frequency Score. Clinical Summary Score = mean of Total Symptom Score and
Physical Limitation Score. Overall Summary Score = mean of Total Symptom Score, Physical Limitation Score, Quality of Life Score and Social Limitation Score. Values are unadjusted mean
(95% confidence interval). †Based on the adjusted mean difference tested two-sided with ANCOVA analysis using baseline value as covariate.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 4 | Per Protocol Analysis: Characteristics of the patients at baseline

Values are N (%), mean ± SD or median [interquartile range]. There were no significant differences between both groups except for the use of any RAASi (P = 0.009), medical history with atrial
fibrillation or flutter (P = 0.014) and BMI (P = 0.024). Chi-square or Fisher exact test, and unpaired T-test or Mann-Whitney U test, whichever appropriate, without adjusting for multiple testing.
ACEi = Angiotensin Converting Enzyme inhibitor; ARB = Angiotensin Receptor Blocker; ARNI = Angiotensin Receptor/Neprilysin Inhibitor; BMI = Body Mass Index; BUN = blood urea nitrogen;
COPD = Chronic Obstructive Pulmonary Disease; CRT = Cardiac Resynchronization Therapy; DM = Diabetes mellitus; eGFR = Estimated Glomerular Filtration Rate; EQ-5D-5L = European Quality
of Life Five Dimensions Five Levels questionnaire; GDMT = Guideline directed medical therapy; HF = Heart Failure; - rEF = with reduced ejection fraction; - mrEF = with mildly reduced ejection
fraction; - pEF = with preserved ejection fraction; ICD = Implantable Cardioverter Defibrillator; KCCQ-OSS = Kansas City Cardiomyopathy Questionnaire Overall Summary Score; LVEF = Left
Ventricular Ejection Fraction; MRA = Mineralocorticoid Receptor Antagonists; NT-proBNP = N-terminal pro–B-type Natriuretic Peptide; NYHA = New York Heart Association; RAASi = Renin–
angiotensin–aldosterone system inhibitors; TDS-HF = Thirst Distress Scale for patients with HF; SGLT2i = Sodium-glucose Cotransporter-2 inhibitor. †Self-reported. ‡Calculated only for HFrEF,
HFmrEF and HFimpEF patients. §Calculated as [sodium concentration]*2 + [BUN concentration] + [glucose concentration].

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 5 | Per Protocol Analysis: Clinical outcomes

Values are N (%), unadjusted mean (95% confidence interval) or unadjusted median [interquartile range]. EQ-5D-5L = European Quality of Life Five Dimensions Five Levels questionnaire; KCCQ
= Kansas City Cardiomyopathy Questionnaire; -CSS = Clinical Summary Score; -OSS = Overall Summary Score; -TSS = Total Symptom Score; TDS-HF = Thirst Distress Scale for patients with HF.
†
Based on the adjusted mean difference tested two-sided with ANCOVA analysis using baseline value as covariate.

Nature Medicine
Article https://doi.org/10.1038/s41591-025-03628-4

Extended Data Table 6 | Per Protocol Analysis: Safety outcomes

Safety was assessed during the 6 months follow-up. Values are N (%), mean ± SD or median [interquartile range]. Chi-square or Fisher exact test, whichever appropriate, and Mann-Whitney U
test. HF = Heart failure; IV = intravenous; NT-proBNP = N-terminal pro–B-type natriuretic peptide. †Acute kidney injury is defined as a 50% decline in estimated glomerular filtration rate relative
to baseline, or decrease of >30 ml/min/1.73m2 and to a value below 60 ml/min/1.73m2. ‡Available values at month 3 (N = 231 vs N = 136) and month 6 (N = 214 vs N = 113). §Available values at month
3 (N = 145 vs N = 80) and month 6 (N = 58 vs N = 38). ¶Differences between two timepoints within individual patients with complete data.

Nature Medicine