Introduction

Tolvaptan (TLV), an oral selective vasopressin type 2 receptor antagonist, was approved on 27 October 2010 in Japan and came on the market on 14 December the same year. TLV has been only used for hyponatremia in other countries; however, it is also used to improve the fluid retention by heart failure (HF) in Japan [1]. Especially, the usefulness of TLV for HF complicated with renal dysfunctions has been reported [2, 3], and also renal protective effect of TLV is known [4]. The reduction in cardiovascular mortality by the long-term use of TLV was denied in the EVELEST trial [5]. Some previous studies reported that long-term use of TLV reduced or increased death and/or HF hospitalization [6, 7], and therefore, there is no unified judgement regarding the long-term use of TLV. On the other hand, it has been reported that the early use of TLV from admission improved diuretic response [8] and preserved renal protection [9]. However, the relationship between the timing of TLV administration after hospitalization and the length of hospital stay is still unclear.

Materials and methods

Statement of ethics

All investigations were performed in accordance with the Declaration of Helsinki and were approved by the Toho University Omori Medical Center Ethical Committee (24-123). The present study was a single center, retrospective observational study.

Study subjects

The subjects were 369 consecutive acute decompensated HF (ADHF) patients who received TLV during hospitalization from February 2011 to June 2016. ADHF was diagnosed according to the Framingham criteria, the American Heart Association guideline or the European Society of Cardiology guideline [10]. The exclusion criteria were as follows: (1) patients who died in hospital; (2) patients who transferred hospital; and (3) clinical scenario 4 or 5. Finally, we analyzed 247 ADHF patients. We did not include age in the exclusion criteria, because the safety and effectiveness of TLV in elderly patients with ADHF had been already demonstrated [11].

Study outcomes

We investigated the relationship between the time until commencement of TLV from hospitalization and the length of hospital stay. Study subjects were divided into two groups: E (the early use of TLV) group and D (the delayed use of TLV) groups by the median of time until commencement of TLV from hospitalization. We compared the following findings on admission between two groups. Moreover, univariate analysis for longer hospital stay (hospital stay of median length or more) was performed. We used multivariate analysis and adjusted with factors found to be significant in univariate analysis.

Clinical profile

Age, gender, and New York Heart Association Classification (NYHA) were investigated. Systolic blood pressure (sBP), diastolic BP (dBP), and heart rate (HR) were evaluated. BPs were measured with aneroid sphygmomanometer on admission. HR was evaluated by standard 12-lead electrocardiography in the supine position. In addition, height and weight were investigated. We calculated body mass index (BMI), using the following formula: BMI = weight (kg)/height (m)2. In addition, we investigated the whole length of hospital stay and the length of hospital stay after initiation of TLV.

Concomitant medications

The time until commencement of TLV after hospitalization and the maximum dosage of TLV while hospitalized were investigated. At discharge, the continuous administration rate and the dosage of TLV were also evaluated. Moreover, we investigated the administration rates of renin–angiotensin–aldosterone system inhibitor (RAAS-I) and be-ta blocker as oral cardio-protective concomitant medications at discharge. RAAS-I was defined as angiotensin converting enzyme inhibitor, angiotensin II type 1a receptor blocker and mineral corticoid receptor antagonist. The dosage of furosemide transformed oral loop diuretics at discharge (furosemide 20 mg is equivalent to azosemide 30 mg) were also evaluated [12]. We also investigated the maximum dosage of daily loop diuretics while hospitalized. For loop diuretics, both intravenous infusion and oral preparation were evaluated, and expressed with oral furosemide equivalent amount (2 mg of oral furosemide are equivalent to 1 mg of intravenous furosemide). In addition, we evaluated the administration rates of continuous intravenous infusion of cardiotonic agents [dobutamine and/or phosphodiesterase (PDE) III inhibitor] and human atrial natriuretic peptide.

Laboratory data, chest X-ray, and transthoracic echocardiography

C-reactive protein (CRP), electrolytes (sodium, potassium), liver function (total bilirubin, aspartate aminotransferase, alanine aminotransferase, and lactate dehydrogenase), renal function [blood urea nitrogen (BUN), creatinine (Cre)], uric acid, hemoglobin, hematocrit, and brain natriuretic peptide (BNP) were measured. Moreover, BUN/Cre ratio was calculated.

Two physicians blinded for the examination calculated the cardiothoracic ratio from Chest X-ray film utilizing the maximal cardiac diameter and the intrathoracic diameter.

We evaluated cardiac size (left atrial dimension and left ventricular end-diastolic/end-systolic dimensions), systolic function [ejection fraction (EF)], and wall thickness (interventricular septal wall thickness and posterior left ventricular wall thickness) from transthoracic echocardiography (TTE). EF was calculated with the Teichholz method [13] by parasternal long-axis view or modified Simpson’s method [14] by apical two or four-chamber view.

Statistical analysis

Continuous variables were expressed as mean ± standard deviation. We compared the groups by Unpaired Student’s t test. Statistical significance was considered at P < 0.05 in all instances. Univariate and multivariate analyses were performed by applying Cox proportional hazard models. We used a Windows computer [Excel (Microsoft XP)] and EZR (Saitama Medical Center, Jichi Medical University), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria version 2.13.0) [15].

Results

TLV and the length of hospital stay

We analyzed a regression curve of the relationship between the time until commencement of TLV from hospitalization and the length of hospital stay (Fig. 1). The time until commencement of TLV from hospitalization and the length of hospital stay showed a significant positive correlation (P < 0.001, r 2 = 0.390). This result was maintained in the analysis restricted for patients with hospital stay within 50 days (Fig. 2, P < 0.001, r 2 = 0.235). In the present study, the median of time until commencement of TLV from hospitalization was 4 days. Thus, patients who started TLV within 4 days after admission were classified into group E (n = 128), and patients with 5 days or later were classified into group D (n = 119). The length of hospital stay in group D was significantly longer than group E (31.9 ± 20.4 and 21.0 ± 12.9 days, P < 0.001). However, there was no difference in the length of hospital stay after initiation of TLV in both groups (Table 2). The median length of hospital stay in this study subjects was 21 days. In univariate and multivariate analysis, we evaluated the factors related to the long hospital stay (22 days or more).

Fig. 1
figure 1

Regression curve of the relationship between time until commencement of TLV from hospitalization and the length of hospital stay. Time until commencement of TLV from hospitalization were strongly related with the length of hospital stay (P < 0.001, r 2 = 0.390)

Full size image
Fig. 2
figure 2

Regression curve of the relationship between time until commencement of TLV from hospitalization and the length of hospital stay in patients with hospital stay within 50 days. Time until commencement of TLV from hospitalization were strongly related with the length of hospital stay in patients with hospital stay within 50 days (P < 0.001, r 2 = 0.235)

Full size image

Baseline characteristics between both groups

Table 1 shows baseline characteristics of the groups. The severity of HF evaluated with NHYA and BNP was not significantly different between the groups. In addition, there were no significant differences in sBP, dBP, and HR. In laboratory findings, BUN in group D was significantly higher than in group E (33.2 ± 19.8 in group D and 28.4 ± 18.1 in group E, P = 0.023). There were no significant differences in the other laboratory data. Similarly, there were no significant differences in CTR and TTE findings between both groups.

Table 1 Baseline characteristics of the subjects
Full size table

The dosage of TLV and concomitant medications

Tolvaptan and concomitant medications are shown in Table 2. The maximum dosage of TLV while hospitalized in group D had no statically significant difference compared with group E. There were no differences in the continuous administration rate and the dosage of TLV at discharge. In addition, the continuous administration rate of loop diuretics at discharge and the dosage of loop diuretics while hospitalized and at discharge had also no significant differences in both groups. On the other hand, the administration rate of cardiotonic agents in group D was significantly higher than in group E (20.1 ± 42.3% in group D and 9.4 ± 29.3% in group E, P = 0.010).

Table 2 Tolvaptan and concomitant medications
Full size table

Univariate and multivariate analyses for prediction of longer hospital stay

In univariate analysis, 12 factors were predicted for longer hospital stay within the above findings (Table 3). Only time until commencement of TLV from hospitalization independently predicted the longer hospital stay in multivariate analysis (odds ratio 1.32, 95% confidence interval 1.13–1.53, P < 0.001).

Table 3 Univariate and multivariate analyses for prediction of longer hospital stay, 22 days or more
Full size table

Discussion

The condition of heart failure in this study populations

In the present study, the median and average length of hospital stay were 21 and 26.3 days, respectively. Attend registry showed that the median and average length of hospital stay by HF in Japanese real world data were 21 and 30 days, respectively [16]. In addition, average age in the present study was similar to the Japanese real world data [16]. Therefore, the results of the present study reflect the Japanese real world data.

The severity of heart failure between both groups

In the present study, at the time of hospitalization, no significant difference was seen in the levels of HF evidenced with NYHA and BNP between both groups. The dosage of diuretics (TLV and loop diuretics) while hospitalized and at discharge were also not statistically different, and the accumulated body fluid between both groups was equivalent. The usefulness of additional administration of TLV to the standard diuretic therapy has been reported [17]. On the other hand, the delayed initiation of TLV did not prolong the length of hospital stay after initiation of TLV in the present study. The effectiveness of TLV at early stage, such as decreasing mean pulmonary artery wedge pressure and pulmonary vascular resistance index, has been reported [18]. Therefore, the length until improvement of HF between both groups was equivalent due to the effectiveness of TLV at early stage. Consequently, it is important to evaluate the timing of TLV initiation to the patients hospitalized due to ADHF. In addition, in the present study, significantly more cardiotonic agents were administrated to the patients in group D, and patients with low efficacy in the conventional HF treatment had been administrated cardiotonic agents before TLV. Although TLV initiation was selected by the attending physician, since this is a retrospective study, early use of TLV is useful for patients who can not obtain sufficient effect by conventional HF treatment.

TLV and the length of hospital stay

In the present study, the time until commencement of TLV from hospitalization and the length of hospital stay showed a significant positive correlation, regardless of the length of hospital stay. In addition, only time until commencement of TLV from hospitalization independently predicted the longer hospital stay in multivariate analysis in the present study. The vasopressin concentration in patients with chronic HF is higher than that in healthy populations [19]. Moreover, it has also been reported that as NYHA rises, the vasopressin concentration in patients with ADHF also increased [20]. Therefore, it is considered useful to use V2 receptor antagonist, TLV, when the vasopressin concentration increase, such as the cases of ADHF. In the present study, probably, the V2 receptor was activated and contributed to these results. Previous study reported that early use of TLV was associated with reduction of in-hospital death in ADHF patients [21]. In addition, we also reported that the continuation of TLV prolongs the period until re-hospitalization [22]. V2 receptor antagonism may have played a role in these results.

Study limitation

The present study is a single-centered retrospective observational study. TLV initiation was decided by the attending physician; therefore, the baseline characteristics (including BUN) may have been influenced by the choice of medication. It has been reported that BUN may be a useful marker for the judgment of the effectiveness of TLV in patients hospitalized for ADHF [23]. Moreover, we could not evaluate the response of TLV and activation of V2 receptor. The prediction of TLV responders (like urine osmolality) has been reported in some previous studies [24, 25]. However, these factors could not be measured in the present study because of the retrospective study design. On the other hand, there is no responder regulation by daily urine volume. In addition, this study is not randomized prospective study, and hence, the group D may preferentially include the patients with poor response to the early treatment with furosemide. This potential bias may flaw the strength of the authors’ conclusion. Thus, it is necessary to conduct further prospective clinical studies with matched backgrounds to confirm these results.

Conclusions

The present study revealed that the time until commencement of TLV after hospitalization was related to the length of hospital stay. Further clinical studies are necessary to confirm these results, because the present study was a single-centered retrospective study.