Liberal versus restricted fluid prescription in stabilised patients with chronic heart failure: Result of a randomised cross-over study of the effects on health-related quality of life, physical capacity, thirst and morbidity

Abstract

Objective. To compare the effects of a restrictive versus a liberal fluid prescription, on quality of life, physical capacity, thirst and hospital admissions, in patients who had improved from NYHA class (III-)IV CHF to a stable condition without clinical signs of significant fluid overload. Design. The present study is a randomised cross-over study. Seventy-four patients (mean age 70±10 years, 16% women) -- with mild-moderate CHF -- were randomised 1:1 to either of two 16-week interventions. Intervention 1 prescribed a maximum fluid intake of 1.5 L/day. Intervention 2 prescribed a maximum fluid intake of 30–35 ml/kg body weight/day. Sixty-five patients completed the study. Results. There were no significant between-intervention differences in end-of-intervention quality of life, physical capacity or hospitalisation. However, there was a significant favourable effect on thirst and less difficulties to adhere to the fluid prescription during the liberal fluid prescription intervention. Conclusion. The results from this study indicate that it may be beneficial and safe to recommend a liberal fluid prescription, based on body weight, in stabilised CHF patients. These results warrant further investigation of the effects of fluid advice in CHF.

• Chronic heart failure
• thirst
• fluid prescription

Chronic heart failure (CHF) is associated with substantially shortened life expectancy and high hospital admission rates, especially among the elderly CHF patients 1. Besides the dominating clinical symptoms, such as shortness of breath, fatigue, exercise intolerance and peripheral oedema, thirst is a troublesome symptom for CHF patients. Thirst can be caused by several factors or a combination of these: 1 increased activation of neurohormonal systems stimulates the thirst centre in the hypothalamus, 2 xerostomia induced by diuretic therapy intensifies sense of thirst 2 and 3 a recommendation to be restrictive with fluid intake can increase the perceived thirst 3.

A recommendation of restrictive fluid intake is often given by international guidelines, and a fluid intake of 1.5–2 l/day is commonly proposed, especially to patients in New York Heart Association (NYHA) class III or IV 3, 4. However, this recommendation is not based on scientific evidence 5. Furthermore, there are no studies assessing the consequences of restricted versus a more liberal fluid prescription, if and when the condition improves and the patient is stable, and guidelines offer no advice in this regard. For a healthy person the physiological fluid requirement is 30–35 ml/kg body weight/day 2. Based on this knowledge, it may seem strange that a person with a body weight of 50 kg has the same fluid intake prescription as someone weighing 80 kg. Nevertheless, for most patients with CHF the recommended fluid intake is 1.5–2 l/day, regardless of body weight, and once the recommendation has been given, it is usually not changed. In healthy persons, fluid deficits of around 2% of the body weight or more, decline mental function and reduce physical exercise capability 6. Therefore, a strict fluid prescription to CHF patients could, hypothetically, lead to weakness and dysfunction of muscles and organs. In clinical practice, physicians and nurses experience other problems connected to a strict fluid restriction. Xerostomia causes, besides a constant feeling of thirst, also altered taste, impaired dental health and, in the worst cases, difficulties to speak. Dry skin and itching are other conditions that could be bothersome to patients due to a strict fluid restriction. Due to these nuisances, it may be difficult to comply with a strict fluid prescription and the patients are often tempted to not adhere 7. The guilt and agony this non-adherence may cause can have an opposite effect on the intended behaviour, and the patient can experience resignation towards the fluid prescription 8.

The aim of the present study was to compare the effects of a restrictive to those of a liberal fluid prescription, on quality of life, physical capacity, thirst and hospital admissions, in CHF patients who had improved from NYHA class (III-)IV CHF to a stable condition, without clinical signs of significant fluid overload.

Methods

Design and setting

The study design has been described in detail previously 5. This was a randomised cross-over study, carried out at the Heart Failure Unit, Department of Cardiology, Malmö University Hospital and at the Heart Failure Unit, Department of Cardiology, Linköping University Hospital, Sweden. Patients were recruited from September 2002 to January 2005. The study conforms with the principles outlined in the Declaration of Helsinki and was approved by the Regional Ethics Committees for Human Research at the Universities of Lund (LU 166-02) and Linköping (Li 03-150), Sweden.

Subjects

Eligible patients received information about the study at discharge from hospital or in connection with a visit to the out-patient heart failure clinic. The inclusion criteria were diagnosed CHF with left ventricular ejection fraction 45% or less and stable condition without clinical signs of significant fluid overload. Patients were not eligible who: a) had malignant or other severe diseases expected to substantially affect physical capacity, quality of life or morbidity; b) had dementia or other significant mental disorders expected to affect study outcome; c) had difficulties to understand or read the Swedish language; d) were unlikely to comply with the study interventions; e) had body weight below 65 kg (in which case the liberal fluid prescription would not differ much from the restricted one); f) were participating in another clinical trial; and g) needed fluid prescriptions due to other severe illnesses. Prior to study start, all patients had a fluid restriction of 1.5 l/day.

Randomization and intervention

At the end of the inclusion visit, a 1:1 telephone randomization was carried out by a person not otherwise involved in the study. Intervention 1. Patients were recommended to adhere to a maximum fluid intake of 1.5 l/day. Intervention 2. Patients were recommended a fluid intake based on 30 ml/kg body weight/day and were allowed to increase the fluid intake to a maximum of 35 ml/kg body weight/day, e.g. 2.4–2.8 l/day for a person weighing 80 kg. No recommendations were made regarding salt intake in this study. The entire study duration was 32 weeks and each intervention period was 16 weeks long. A nurse specially trained in the treatment and pathophysiology of CHF was responsible for the 13 study visits, and a cardiologist was always available for consultation. Diuretic therapy was adjusted according to the patients’ clinical need.

Measurements

EuroQol (EQ-5D) was used to measure self-perceived health status. EQ-5D is a generic single index defining health in 5 dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) 9. The scores for EQ-5D range from −0.59 to 1 10 where 1 is the best possible health. EQ-5D has been tested for responsiveness, validity and reliability in different patient populations 11, including a Swedish population 12. Besides the five statements, the EQ-5D consists of a visual analogue scale (VAS) ranging from 0 to 100. Zero denotes the worst possible health and 100 the best possible health.

The Minnesota Living with Heart Failure Questionnaire (MLWHFQ) was used to measure disease-specific quality of life. MLWHFQ consists of 21 items, each addressing a different aspect of life, such as symptoms and signs relevant to CHF, physical activity, social interaction, sexual activity, work and emotions 13. The possible score range is from 0 to 105 and lower scores indicate a better health-related quality of life 13. Validity of the questionnaire has been tested by content, construct and criterion-related validity. Reliability has been tested with Cronbach's alpha with satisfying results (0.90–0.94) 14. MLWHFQ appears to be reasonably sensitive to detect changes in quality of life and is recommended for use in non-pharmacological interventions 15. A linguistically tested Swedish version was used in the present study.

Submaximal physical performance was measured by the six-minute walk test 16. The walk test was conducted in a secluded corridor and the length of the course was 30 meters. The six-minute walk test has shown concordance with changes in symptoms, which is relevant in the present study 17.

Thirst and difficulties to adhere to the recommended fluid intake were measured by VAS. VAS is commonly used in clinical practice and research for evaluating pain 18, 19. Thirst is like pain a subjective experience and VAS has been used to evaluate thirst in patients with cancer and renal failure 20, 21, and was therefore chosen to evaluate thirst. The same type of scale was used to estimate difficulties to adhere to the recommended fluid intake. Face validity has been established by an expert panel of nurses and cardiologists specialized in CHF care and a test on 15 patients, and some small adjustments were made based on the recommendation from the patients. The scale was then re-tested on some of the same patients 22.

Numbers of hospitalisations, days in hospital and deaths during follow-up, as well as reasons for hospitalisation, were assessed from the hospital charts. To assure that no deterioration occurred, NYHA class 23, signs and symptoms of fluid retention (pulmonary rales, oedema, etc.), heart rate, blood pressure and body weight were assessed at each follow-up visit. The patients were provided with a protocol for recording of daily fluid intake and the instruction was to carefully measure and account for all fluid intake during the entire study period. All medication was assessed at each visit.

Statistical power and sample size calculation

The main study aim was to compare the effects of the two prescriptions on quality of life. The primary variable for estimating statistical power was self-perceived health status according to the EQ-5D VAS. A relative improvement of 10% of the EQ-5D VAS score was considered to be a clinically significant improvement. With 59 patients in each intervention, the study would have an 80% power at the 5% significance level to detect a 15% absolute between-intervention difference in the percentage of patients improving in EQ-5D VAS, e.g. 30% of patients improving in the 30 ml/kg/day intervention versus 15% of the patients improving in the 1.5 l/day intervention, or conversely. Thus, 118 participants were required. However, with the cross-over study design used, each patient was participating in both study interventions and, thereby, this number could be halved to 59 patients. In order to compensate for an expected 20% drop out rate, 74 patients were included. There was also an 80% power at the 5% significance level to detect a reduction in total hospitalisation from around 30% among patients on the 1.5 l/day intervention to around 10% among patients with the 30 ml/kg/day intervention, or conversely.

Statistical analysis

All study data are presented by intervention. To detect carry-over effects, analysis was performed according to the recommendations by Altman for cross-over trials 24. Descriptive analyses were used to describe the sample and the responses in the study variables. Due to the scale level, MLWHFQ, EQ-5D, sense of thirst, difficulties to adhere to the recommended fluid intake, hospital admissions and days spent in hospital, were analyzed using the Wilcoxon matched pair test for comparisons. For analysis of the six-minute walk test, Student′s paired t-test was used. The between-intervention comparisons are calculated from the measurements made at the end of each treatment period. Results are expressed as mean±standard deviation or median [inter-quartile range]. P < 0.05 denotes statistical significance.

Results

Study patients

Due to the cross-over design, patients who did not complete the study (n = 9) are not included in the analysis of the outcome variables. Table I depicts the baseline demographics and clinical characteristics for all the 74 randomised patients, the 9 patients dropping out and for the 65 patients completing the study. There were no significant differences between the whole study population and the drop-out patients, except that a significantly greater proportion of the drop-out subjects were in NYHA III (p = 0.01). Of the 9 drop-out patients, 6 patients died, 2 patients were excluded due to inability to comply with the study protocol, and 1 patient withdrew consent. Of the deceased patients, 4 died during the 1.5 l/day intervention and the primary cause of death was sudden death in 3 of these patients, whereas 1 died of deteriorating CHF. Two patients died during the 30 ml/kg/day intervention, 1 of myocardial infarction and 1 of deteriorating CHF. At baseline, 61 of 65 completing patients were treated with furosemide and the mean dose was 83 mg±75 mg. One patient was treated with 2.5 mg of bendroflumethiazid and 3 had no diuretic therapy. Diuretic doses were similar at baseline and study end. Few changes were made in the diuretic therapy during the study and they were generally temporary. During the 1.5 l/day intervention, the diuretic therapy was changed in 8 patients. It was increased at 13 occasions and decreased at 9 occasions. During the 30 ml/kg/day intervention, the diuretic therapy was changed in 6 patients. There were 9 increases and 9 decreases. All other cardiovascular medication was virtually unchanged during the study and there were no differences between the two interventions with regard to baseline or changes in cardiovascular medication.

Table I.  Demographics and clinical characteristics of the study sample at study start and at the beginning of each intervention.

	Total study sample (N = 74)	Drop-outs (n = 9)	1.5L/day intervention (n = 65)	30ml/kg/day intervention (n = 65)
Age (years), mean±SD	70±10	72±11	70±10	70±10
Gender
Men, n (%)	62 (84%)	8 (89%)	54 (83%)	54 (83%)
Women, n (%)	12 (16%)	1 (11%)	11 (17%)	11 (17%)
Vital signs and symptoms
Bodyweight (kg), mean±SD	86±15	79±10	87±15	87±15
Heart rate, mean±SD	70±10	73±9	70±10	70±11
Systolic blood pressure (mmHg), mean±SD	118±17	109±20	119±17	118±15
Diastolic blood pressure (mmHg), mean±SD	70±10	64±12	69±9	71±10
Blood samples
Haemoglobin g/L, mean±SD	135±14	130±12	137±15	135±16
Na^+ mmol/L, mean±SD	140±3	138±4	140±3	140±2
K^+ mmol/L, mean±SD	4.2±0.35	4.1± 0.5	4.2±0.4	4.3±0.3
Creatinine µmo/L, mean±SD	117±40	129±35	121±50	120±43
NYHA, n (%)
I	3 (4%)	0	4 (6%)	4 (6%)
II	65 (88%)	6 (67%)	56 (86%)	55 (85%)
III	6 (8%)	3 (33%)	5 (8%)	6 (9%)
Co-morbidities, n (%)
Ischemic heart disease	54 (73%)	7 (78%)	48 (74%)	48 (74%)
Myocardial infarction	40 (54%)	7 (78%)	33 (51%)	33 (51%)
Hypertension	17 (23%)	1 (11%)	17 (23%)	17 (23%)
Diabetes mellitus	19 (26%)	3 (33%)	16 (25%)	16 (25%)
Pharmacotherapy, n (%)
Diuretics	71 (96%)	9 (100%)	62 (95%)	62 (95%)
ACE-inhibitors	58 (78%)	7 (78%)	51 (78%)	51 (78%)
Beta-blockers	65 (88%)	9 (100%)	55 (85%)	56 (86%)
Spironolactone	32 (43%)	4 (45%)	28 (43%)	28 (43%)
Digitalis	21 (28%)	2 (22%)	19 (29%)	19 (29%)

NYHA = New York Heart Association classification, ACE, angiotensin converting enzyme.

Quality of life, physical capacity, signs and symptoms

No carry-over effects were found in the analysis of EQ-5D, MLWHFQ or six-minute walk test. These variables showed no significant within- or between-intervention differences (Table II), except for MLWHFQ that improved during both interventions, although the improvement was significant only during the first period. Body weight, serum-creatinine, electrolytes and haemoglobin were unchanged regardless of intervention (Table II). NYHA class, heart rate and blood pressure did also not differ significantly between interventions.

Table II.  Results at the end of each intervention and comparisons between-interventions.

	End of the 1.5 L/day intervention	End of the 30ml/kg /day intervention,	P-value
EQ-5D, median [IQR]	0.80 [0.66–0.92] (n = 63)	0.80 [0.70–92] (n = 63)	0.75
EQ-5D VAS, median [IQR]	70 [55–85] (n = 63)	70 [57–82] (n = 63)	0.88
MLWHFQ, median [IQR]	20 [11–40] (n = 64)	17 [9–44] (n = 64)	0.95
Six minute walk test, mean±SD	407±123 (n = 58)	410±119 (n = 58)	0.59
Sense of thirst, median [IQR]	51 [16–69] (n = 64)	23 [6–53] (n = 64)	<0.001
Difficulties to adhere to the fluid prescription, median [IQR]	23 [5–56] (n = 64)	6 [1–24] (n = 64)	<0.001
Bodyweight (kg), mean±SD	87,1±17 (n = 65)	87,5±18 (n = 65)	0.718
Haemoglobin g/L, mean±SD	136±16 (n = 60)	137±17 (n = 60)	0.85
Na^+ mmol/L, mean±SD	140±3 (64)	140±3 (64)	0.15
K^+ mmol/L, mean±SD	4.3±0.5 (=63)	4.2±0.4 (n = 63)	0.41
Creatinine µmo/L, mean±SD	127±55 (=64)	130±54 (n = 64)	0.30

IQR, interquartile range; MLWHFQ, Minnesota Living with Heart Failure Questionnaire.

Fluid intake, sense of thirst and difficulties to adhere to the fluid prescription

The mean fluid intake during the 1.5 l/day intervention was 17 ml/kg/day, irrespective of 16-week period. During the 30 ml/kg/day intervention, the mean fluid intake was 23 ml/kg/day in both periods. The difference between the interventions was significant (p < 0.0001) in both 16-weeks periods. Carry-over effects with regard to the order of the intervention were found in sense of thirst and difficulties to adhere to the fluid prescription. The between-intervention comparison in sense of thirst showed a significantly favourable result at the end of the 30 ml/kg/day intervention, as compared with the end of the 1.5 l/day intervention (Table II). Difficulties to adhere to the fluid prescription decreased during the 30 ml/kg/day intervention in both periods, and increased during the 1.5 l/day intervention in both periods. As for thirst, the between-intervention comparison in difficulties to adhere to the fluid prescription showed a significantly better result at the end of the 30 ml/kg/day intervention, as compared with the end of the 1.5 l/day intervention (Table II).

Morbidity

During the study there were twelve readmissions due to deterioration of CHF, 6 during the 1.5 l/day intervention and 6 during the 30 ml/kg/day intervention. One of the patients was readmitted 2 times during each intervention, in all 4 admissions. In Table III the readmissions are described. No statistically significant difference was found between the two interventions regarding readmissions (p = 1.00). Patients spent 42 days in hospital during the 1.5 l/day intervention and 56 during the 30 ml/kg/day intervention (p = 0.30). In regard to the order of the interventions, no carry-over effects were observed for readmissions.

Table III.  Description of the readmissions during the study (N = 65).

Number of patients not readmitted	Number of patients readmitted during both interventions	Number of patients readmitted only during the 30 ml/kg/day intervention	Number of patients readmitted only during the 1.5 L/day intervention
58	3	2	2

Discussion

This is the first study to investigate if a liberal fluid prescription affects quality of life, physical capacity, sense of thirst, difficulties to adhere to the fluid prescription, and morbidity, as compared with a traditional strict fluid intake prescription of 1.5 l/day, in patients who have improved from (moderate-)severe CHF to stable CHF with predominantly mild symptoms. The main findings are that sense of thirst and difficulties to adhere to the fluid prescription were significantly reduced by a less strict fluid prescription, without any significant differences in body weight, use of diuretics or other cardiovascular medication, quality of life, physical capacity or hospitalisation.

The mean daily fluid consumption was 1 479 ml/day during the 1.5 l/day intervention and 1 955 ml/day during the 30 ml/kg/day intervention. The patients in our study had a higher fluid intake than in the study by Travers et al. 25, but the results point in the same direction: fluid intake may not be as important to recovery and maintaining stability in patients with CHF as previously considered. Prior to the initiation of the present study, one might have expected the number of readmissions to be higher during the 30 ml/kg/day intervention, due to deterioration of CHF because of fluid overload. However, although the number of days spent in hospital was slightly higher during this intervention, the difference was far from statistically significant, and the number of patients with a hospital admission due to worsened CHF did not differ between the interventions. Based on the original assumption about hospitalisation rates, the study had 80% power to detect an approximately 20% absolute between-intervention difference. However, hospitalisation rates were unexpectedly low during both interventions, although 71 of the 74 patients in this study had been hospitalised due to CHF prior to study start. Due to this low hospitalisation rate we cannot preclude a clinically important between-intervention difference in hospitalisation.

Considering the differences observed in terms of thirst, it may seem odd that no significant between-intervention difference was found in quality of life. However, this may be due to methodological issues. Many of the instruments used to measure quality of life have a low sensitivity and are aimed at measuring overall quality of life. Alternatively, improvement in thirst might theoretically have been offset by worsening of symptoms of CHF, thus resulting in a neutral effect on quality of life. However, besides the few hospitalisations accounted for, very few patients reported worsened signs/symptoms of CHF, and cardiovascular medication (including diuretics) was virtually unchanged throughout the study, without any between-intervention differences. Despite a lack of a between-intervention difference in quality of life, the results of this study are potentially important, because they suggest reduced thirst without any negative effects on physical capacity of a change to a less strict, liberal fluid prescription in stabilized CHF patients with mild symptoms.

Although the randomisation procedure may diminish the confounding consequences of the carry-over effects on the study results, such effects are difficult to preclude, especially since statistical testing for carry-over effects has an inherited low power 26. Indeed, we detected carry-over effects on thirst and difficulties to adhere to the fluid prescription. We had no washout period, which can be used to distinguish the effects of the first intervention from the second. However, without the information obtained from the present study, we felt that it was impossible to decide how long a washout period would have to be.

In conclusion, this is the first study to evaluate the effects of a changed prescription of fluid intake in patients with CHF who have improved from NYHA class (III-)IV to stable CHF with mild symptoms. The end-of-intervention comparisons between a prescribed fluid intake of 1.5 l/day and one of 30 ml/kg body weight/day showed no significant differences in quality of life, physical capacity or morbidity. However, the 30 ml/kg body weight/day intervention was superior with regard to thirst and difficulties to adhere to the fluid prescription. This supports the idea that it may be beneficial and safe to recommend a liberal fluid prescription, based on body weight, in stabilized CHF patients. This may make life easier for many patients with CHF. This field of everyday clinical practice in CHF is very important and difficult. Therefore, this pilot study should be followed by a study with sufficient power to assess mortality and morbidity.

Acknowledgements

We thank the nurses Åsa Pettersson and Åsa Danielsson at the Department of Cardiology, Linköping University, Sweden for assistance with data collection in the study. There are no funding and competing interests.