Cost-effectiveness of tolvaptan for the treatment of hyponatraemia secondary to syndrome of inappropriate antidiuretic hormone secretion in Sweden

Hyponatraemia (HN) is the most common electrolyte disturbance in clinical practice, occurring in approximately 15–30 % of hospitalised patients [1, 2]. It is commonly defined as a serum sodium concentration ([Na⁺]) <135 mmol/L [2‐5], with severe HN defined as <125 mmol/L [2]. Hyponatraemia is associated with increased mortality, morbidity and length of hospital stay [2].

Hyponatraemia can be euvolaemic (i.e. without an associated change in body fluid volume), hypervolaemic (i.e. body fluid volume is increased), or hypovolaemic (body fluid volume is decreased). These three types of HN have different sets of possible aetiologies. The most common cause of euvolaemic HN is the syndrome of inappropriate antidiuretic hormone secretion (SIADH) [3, 6], which is the focus of our analysis. In SIADH water retention is caused by inappropriate increased release of, or responsiveness to, antidiuretic hormone (ADH) [7]. This can be due to malignancy, pneumonia, central nervous system disorders or certain drugs [7]. Although SIADH can be associated with many different types of malignancy, it is most commonly observed with small cell lung cancer (SCLC) and studies have shown a higher mortality rate in SCLC patients with HN than in those without [8, 9]. Drug-induced SIADH usually resolves upon discontinuation of the medication [7].

Swedish guidelines for HN, published by the Swedish Endocrine Society and the Swedish Association for Anaesthesiology and Intensive Care [1], recommend resolution of the underlying condition as the primary treatment strategy in patients with SIADH. If symptoms are mild, HN secondary to SIADH may be treated with fluid restriction. Diagnosis and treatment should be re-evaluated if the patient does not respond within 12–24 h. If the symptoms are more pronounced the patient will require treatment in intensive care [1]. In patients who do not respond to or cannot be treated with fluid restriction, symptoms may exacerbate and patient outcomes may be signficiantly affected. For patients in whom fluid restriction has been unsuccessful or is unsuitable, vasopressin receptor antagonists (i.e., tolvaptan) are recommended [1]. In clinical practice, subjects with HN, especially if it is associated with no or mild symptoms, or if initial treatment was unsuccessful, may often not receive treatment [10].

Tolvaptan, an oral vasopressin V₂ receptor antagonist, is the only vaptan approved by the European Medicines Agency (EMA) for the treatment of adult patients with HN secondary to SIADH in 2009 [11]. Tolvaptan blocks binding of ADH to the V₂ receptor which induces electrolyte-free water excretion. The result is a reduction in body water without loss of body electrolytes, which leads to an increase in serum [Na⁺] [4, 5].

The cost-effectiveness of tolvaptan for the treatment of HN secondary to SIADH has never been studied in a European setting. Worldwide, three studies have previously evaluated the potential cost impact of tolvaptan in HN [12‐14]; however, one of these was performed in a different indication (heart failure) (14). Of the remaining two studies, Dasta et al. [13] considered a US hospital population with HN secondary to SIADH and observed a total cost offset, including the acquisition cost of tolvaptan, of $694 per admission. The authors estimated that tolvaptan would be cost-neutral with duration of therapy of ≥6.78 days. In a South Korean analysis, Lee et al. [14] found that, in patients hospitalised for treatment and monitoring of euvolaemic and hypervolaemic HN, tolvaptan treatment was less costly and more efficacious compared with placebo.

We have performed an economic evaluation, from a Swedish societal perspective, to determine the cost-effectiveness of tolvaptan, compared with no active treatment (NAT) for HN secondary to SIADH in adult patients who have either failed to respond to fluid restriction or for whom the use of fluid restriction is not suitable.

Base case results for the three different populations, based on cohorts of 100,000 individually modelled patients, are presented in Table 3. Detailed results are provided in the Additional file 1. Tolvaptan was associated with reduced per patient costs (savings of SEK 2,678 to SEK 8,411[€289 to €908]) and increased incremental QALYs (0.0018 to 0.0028) versus NAT. Tolvaptan was therefore considered to be the dominant (i.e. both cost-saving and resulting in a QALY gain) strategy in all three populations.

The results from the scenario analyses are presented in the Additional file 1. The most important determinants of cost-effectiveness in our model were reduction in hospital LOS associated with tolvaptan (removal of this benefit resulting in ICERs of SEK >1.4 million), duration of treatment with tolvaptan (an increase in treatment duration resulting in a considerable ICER increase in the ‘all SIADH’, but not change in the SCLC and pneumonia populations) and long term treatment with tolvaptan in SCLC patients (SEK 2.9 million).

The results of univariate sensitivity analyses for the ‘general SIADH’ population are presented in Fig. 2. The most influential parameters were two of the coefficients used in the estimation of the duration of tolvaptan treatment; the constant term and baseline [Na⁺]. When these were varied between their upper and lower 95 % confidence intervals, they were able to generate NMB < SEK 0 at a willingness-to-pay threshold of SEK 1.2 million. A recent review of the literature around the value of a statistical life in Sweden suggests a willingness-to-pay threshold of SEK 1.2 million per QALY [26].

The present analysis assessed the cost-effectiveness of tolvaptan in three populations of patients with HN secondary to SIADH for whom the alternative treatment option is NAT. In the base case for the ‘general SIADH’ population, tolvaptan was the dominant strategy compared with NAT, as it was associated with a reduction in cost and an incremental QALY gain. Cost savings are predicted based on reductions in LOS, whilst QALY gains are a consequence of associated improvements in HRQL. Compared with NAT, tolvaptan was also the dominant strategy in the SCLC and pneumonia populations, associated with incremental cost savings and an incremental QALY gain. The results were sensitive to assumptions relating to the duration of tolvaptan treatment, LOS associated with NAT and the reduction in LOS associated with tolvaptan treatment.

The data used to calculate LOS and HRQL was obtained from the SALT I and SALT II clinical trials [4, 5]. The patients in the trials were treatment-naïve, whereas patients in the model had either failed to respond to fluid restriction or the use of fluid restriction was not suitable. Patients that fail to respond to fluid restriction may have more severe forms of SIADH and may be unable to excrete electrolyte-free water [27]. Therefore, they may respond differently to tolvaptan compared with the patients in SALT I and SALT II. Although the assumed positioning of tolvaptan in the model may be considered a limitation of the analysis, the positioning was based on Swedish guidelines for HN [1]. Furthermore, there are indications, from small case series [28, 29] that tolvaptan quickly restores serum [Na⁺] concentrations in a population consistent with that modelled (i.e. in patients with HN secondary to SIADH who have failed fluid restriction or for whom fluid restriction is unsuitable).

The model assumes a reduction in LOS of 20 % based on the overall reduction observed in patients receiving tolvaptan compared with patients receiving placebo in the SIADH population of the SALT trials [4]. This reduction was not statistically significant, and the studies were not powered to demonstrate a difference in reduction of LOS. However, a subgroup analysis of euvolaemic patients with HN in the investigator-diagnosed SIADH group, combined with patients classified as ‘other’ (all patients not meeting clinical criteria for diagnosis of heart failure or cirrhosis), demonstrated a significantly shortened LOS favouring tolvaptan (4.70 ± 3.89 versus 8.40 ± 9.67 days; p = 0.045; a reduction of 44 %) [4]. These preliminary findings should ideally be confirmed by clinical trials designed to evaluate this outcome. A reduction of LOS, as demonstrated in the SALT trials, is supported by preliminary data from the HN registry [30] showing that the median LOS from the start of treatment was shortest for tolvaptan (3 days) and longest for no treatment (6 days). Published data in HN relevant to the current analysis are limited, not least due to the complexity of the condition and the clinical pathway. Every effort was made to identify the most relevant data for the model, using real word data sources where possible. Sweden-specific data were used where available.

Basing model LOS data in the SCLC and pneumonia populations on evidence from the SALT trials may be considered a source of uncertainty, as the percentage of patients in the study population with SCLC or pneumonia is not known. The patients enrolled in these studies had either euvolaemic or hypervolaemic HN associated with chronic heart failure, cirrhosis or SIADH (with no condition affecting >50 % of patients) [5]. Our approach is also supported by other studies, which suggest that the mean LOS assumed for the NAT arm of the model (8 days) is reasonable for the pneumonia population [31‐33]. Data for the SCLC population are scarce; Salahudeen et al. [34] report mean (SD) LOS of 10.3 (10.0) for patients with HN in a mixed US oncology population.

Patient baseline characteristics were derived from the ART study, which identifies HN patients using ICD-9/10 codes. This method of identifying patients has high specificity, but low sensitivity [35]. However, the data are specific to Sweden and include patients from all three populations modelled. The ART study is therefore the most relevant source of patient characteristics data for this analysis.

It remains unclear to what extent [Na⁺] correction or improvement is associated with improved long-term outcomes. There is some evidence that HN in hospitalised patients is associated with increased mortality (5-year data are available), with mild HN showing similar mortality rates to severe HN [36]. The model assumes no mortality benefit associated with [Na⁺] correction and does not explicitly include the benefits of treatments which may be administered to achieve [Na⁺] correction. The model assumes no residual HRQL or cost benefits associated with tolvaptan following discontinuation, and other longer-term benefits, such as any benefits of [Na⁺] correction on readmissions, are ignored by the model, and therefore cost-effectivness may be underestimated.

Another possible factor not considered in our analysis, due to the lack of data, is the possibility of overcorrection, which may occur due to unexpected diuresis after resolution of the cause of the original water retention. Overcorrection of HN can have severe neurological consequences due to demyelination occurring in the brain [37].

In modelling this condition, there was dissociation between the treatment effect (e.g. changes in [Na⁺] and HRQL. HRQL is based on reported QoL at Day 30 of SALT I and II. Transient adverse events could therefore have been missed, which would lead to an underestimation of HRQL losses. Although the assumption that 100 % of the ‘general SIADH’ population resolve their underlying condition following discharge is an over-simplification, the effect on incremental cost-effectiveness is likely to be minimal (please see Section 1.1.6 of the Additional file 1 for the underlying reasoning). Modelling HN secondary to SIADH without building several disease-specific models required a high degree of generalisation.

Published data relevant to the current analysis are limited but the best available data have been utilised from the randomised SALT trials. The main costs and benefits associated with short-term tolvaptan use are included in the model, but data limitations could lead to additional benefits not being accounted for, resulting in a possible under-estimation of cost-effectiveness. Despite a possible underestimation of cost-effectiveness, due to data limitations, the current analysis shows that tolvaptan treatment is associated with cost savings. The results of our analysis will be helpful to clinicians in the field of HN in establishing appropriate treatment strategies for patients with euvolaemic HN in the hospital setting.

This is the first cost-utility analysis of tolvaptan for the treatment of HN secondary to SIADH in a European setting. Treatment with tolvaptan was associated with reduced costs and better health outcomes (measured as QALYs) compared with NAT in the ‘general SIADH’, SCLC and pneumonia populations considered in this economic evaluation. Based on the assumptions within this evaluation, tolvaptan therefore represents a cost-effective treatment option for patients with HN secondary to SIADH who have either failed to response to fluid restriction or for whom the use of fluid restriction is not suitable.

Not applicable. We conducted an economical assessment of data which is already in the public domain and are referenced to throughout the manuscript where relevant.

Not applicable. We conducted an economical assessment of data which is already in the public domain and are referenced to throughout the manuscript where relevant.

The datasets supporting the conclusions of this article are included within the article (and its additional file).