The economic burden of disease of epithelial ovarian cancer in Spain: the OvarCost study

Incident patients were estimated from years 2017 to 2026, using a linear model between 2015, 2020, and 2025 as per GLOBOCAN predictions [1]. Accordingly, 3497 women diagnosed with ovarian cancer were estimated for the first year, and as per epidemiology data, the majority of cases (90%) are of epithelial origin [3]. Those were distributed by the four disease stages [20], as shown in Table 1. New cases diagnosed in the following years until 2026 were added each year assuming no changes in the distribution of disease stages over time. Incident patients’ distribution by stage comes from population-based cancer registries [20].

Transition probabilities depend on the disease stage assigned at diagnosis and the health state as patients enter the model. Death probability in the stable state (Pdss) was the mortality rate in the general Spanish female population [21] (Pnd, natural death probability). This mortality rate was estimated taking into account age at diagnosis and at each stage of the disease [15], as stated in Table 1, and its evolution over time; finally, it was transformed to probabilities as 1 − exp^{(rate at age)}.

Transition probabilities from stable state to post-progression state (Ppss) (Eq. 1) were assessed by progression-free survival (PFS) curves for each disease stage, which were built based on an exponential distribution of median PFS (Table 1; Fig. 2a) [9].

Equation 1. Probability of transitioning from stable to post-progression state:

Transition probabilities of patients who remain stable (Prss) were calculated as the inverse of the probability of dying plus the transition probability of progressing from stable to post-progression state. Transition probabilities from post-progression to death (Pdps) were estimated as the difference between overall survival (OS) and PFS, plus the probability of natural death (Eq. 2) (Fig. 2b). OS curves were built using an exponential distribution, taking into account the median OS at each stage published by Heintz et al. [9] (Table 1; Fig. 2b).

Equation 2. Probability of dying for patients at post-progression state (Pdps):

Transition probabilities of the patients who remain at the post-progression state (Prps) were estimated as the inverse of the probability of dying for patients at post-progression state.

Costs were expressed in 2016 Euros (Tables 1, 2 in Online Resource). Unit healthcare costs were the median value of the unit costs for each Autonomous Community in Spain [20‐37]. Since the costs come from different years, they were updated to € 2016 using the corresponding inflation rate: a medicine consumer price index (CPI) of 0.77% for direct healthcare costs (DHC) [40] (except for pharmaceutical costs and tariffs from Autonomous Communities already actualised) and a general CPI increase of 1.97% for direct non-healthcare costs (DNHC) [40]. In the Markov model, the costs were discounted at an annual rate of 3%, according to Spanish health technology assessment recommendations [41]. An annual growth rate of 1% was considered for labour productivity loss [42]. The average annual cost per patient was assessed dividing the total cost by the number of patients (those who were alive at the beginning of the year plus the incident patients in that year). Model costs were categorised as DHC, DNHC, and indirect costs (IC).

DHC included diagnosis and follow-up tests, treatments, and palliative care. Testing required at diagnosis according to the Spanish Society of Gynaecology and Obstetrics [12] include ovarian biopsy, biochemical analysis and vaginal ultrasonography, among others. Regarding selection criteria to identify BRCA mutation, it was agreed by the panel group that 20% of patients with EOC are referred to BRCA1/2 genetic test and genetic counselling, accounting two of these visits, before and after the test [43]. Of these patients, 5% are identified with the genetic mutation and an average of five family members are derived to genetic counselling [44], a transvaginal ultrasound and a blood test every 6 months to detect the tumour marker CA125 [43]. The patient’s follow-up depends on their disease state. The frequency for follow-up testing may be lower in stable patients, depending on the period of time that they remain at this health estate [12]. The percentage of patients hospitalized and those who attend to the emergency department due to EOC were also considered by disease stage (Table 1) [14]. Treatment management depends on many factors, such as the spread of the tumour and the patient’s clinical situation, being surgery or/and chemotherapy the standard of care [3, 9]. Treatment usually starts with the surgical excision of the tumour mass. Nevertheless, this procedure is not always possible and an interval debulking surgery is performed. This intervention is a surgical excision that takes place after patients have taken neoadjuvant chemotherapy [12]. Most patients receive adjuvant chemotherapy. However, patients in stages Ia and Ib do not need chemotherapy after surgery and only remain under clinical observation [12]. The type of surgery depends on the size and the spread of the tumour, and on whether or not the woman is planning to get pregnant in the future [12, 13].

Clinical experts panel classified chemotherapy as: (1) neoadjuvant: patients who receive 3 cycles of paclitaxel in combination with carboplatin before surgery and complete their treatment with other 3 cycles of chemotherapy [12]; (2) adjuvant: chemotherapy administered after surgery in the stable state (stages I, II, and III); (3) post-progression: chemotherapy administered at the post-progression state (stages I and II); and (4) advanced: chemotherapy administered at the post-progression states at stage III and at both states at stage IV (Table 1).

The recommended drugs used are based on the EOC treatment recommended by SEGO guidelines [12]: paclitaxel, carboplatin, doxorubicin, bevacizumab, cisplatin, gemcitabine, topotecan, trabectedin, and docetaxel. Its usage was accounted based on its market share [45]. Doses were calculated according to the usual clinical practice and product labels [46‐54] (Table 4 in Online Resource). Dose of carboplatin [47] was determined using the Calvert formula [55]. Du Bois et Du Bois formula was used to calculate the body surface area when necessary [56] (Table 3 in Online Resource). Patients’ height and weight were consulted in Spanish National Health Survey according to the mean age at diagnosis of each stage disease [15, 57] (Table 1). Drug costs were calculated using the list price (LP) [58], including Royal Decree Law 8/2010 deduction rate, when necessary, and a 4% of the value-added tax (VAT) entitled for Spain [58‐61]. For intravenous drugs, the model also considered non-vials optimization and the cost of administration for each drug (€0.32 per minute [62]): time of administration required for each one [46‐53] in the day hospital plus the cost of the 30-min preparation (Table 3 in Online Resource).

Palliative care is given to the patients in their last 48 days of life [63]. Up to 93.3% of the patients receive follow-up care at outpatient hospitals, while the remaining (6.7%) are assisted by palliative home care team [64]. Patients need a mean of 9.5 home visits of palliative-care services, while those who receive follow-up at the outpatient hospital are seen by a nurse [63]. In both cases, patients also pay four visits, on average, to the primary-care doctor [63] and they spent their terminal phase of their illness at home (59.6%) or at the hospital (40.4%) [65]. The last 3 days of this terminal phase, patients stay at home [66], and they are visited twice a day by a nurse [67]. Of these, 14% receive sedation [66]. The costs of visits [22‐39] and drugs used in the palliative-care phase were also considered [58‐61, 68‐71] (Tables 1, 4 in Online Resource).

DNHC considered were formal care costs (i.e., professional care financed by private or public funds) and informal care costs given at home (non-remunerated care from relatives or friends). Based on the literature, it was assumed that 17.4% of the patients received private care, 9.5% public care [63], and 93.4% received informal care [72] throughout their last 48 days of life [63]. On average, it was considered that public caregivers spent 1.5 h providing care [63], private caregivers 8 h [63], and informal caregivers 10.3 h [17]. According to the proxy good method [73], hourly wage for formal and informal caregivers was equally valued, €13.56 [74].

Lost labour productivity due to temporary or permanent leave and premature death were included as indirect costs (IC), using the human-capital method [75‐78]. At some point in the progress of the disease, patients with EOC become unable to develop their labour activities [79, 80]. Overall, 30% of stable patients lose 60 days due to temporary leave, while the remaining 70% lose over 70 days [45]. It was assumed that at advanced stages, patients are on sick leave, since progression starts until age of retirement (65 years). In case the patient’s death occurs before 65 years, sick leave period is assumed to last between the beginnings of the disease progression until patient dies. Patients on sick leave for more than 1 year were considered to be in permanent leave [81]. Labour productivity loss caused by premature death included lost productivity of patients who die before 65 years of age [21, 79, 82, 83]. The percentage of women employed and their respective salaries are used to estimate labour productivity losses (Table 5 in Online Resource).

Deterministic and univariate sensitivity analyses, including ten different scenarios, were conducted to examine the model’s robustness. According to the OvarCost Expert Panel, different scenarios were built based on the possible variation of the most sensitive parameters: percentage of patients who receive genetic counselling (from 35 to 70%), patients weight (± 10%), growth productivity discount rate on (from 0 to 2%), manufacturer’s drug price (− 10%), discount drug rate (from 0 to 6%), tests and medical visits cost (maximum and minimum prices in the Autonomous Communities), age at time of diagnosis (± 10%), bevacizumab dose recommendation (7.5 mg/kg), caregiver’s salary/informal care assessment per hour (€7.5) [84], and informal care hours received (± 30%).