Cost and burden analysis requires estimating the prevalence and distribution of TNBC cases across all cancer stages (I, II, III and IV). To obtain the most accurate picture of TNBC in Portugal, a cumulative incidence model (CIM) with annual cycles was developed to estimate its prevalence in 2019. The CIM allows the prediction of TNBC evolution in female patients whose diagnosis was made between 2010 and 2019, establishing distinct patient pathways according to cancer stage at diagnosis, namely, locoregional (stages I–III) and metastatic (stage IV) (Fig. 1).

Model structure of pathways according to triple-negative breast cancer locoregional (i) and metastatic (ii) stages.

The annual probability of death was calculated on the basis of published overall survival (OS) data from the Surveillance, Epidemiology, and End Results (SEER) Program from the United States of America, due to the lack of available Portuguese data [11]. For patients with locoregional disease at the time of diagnosis, the risk of recurrence estimation was based on progression-free survival (PFS) curves published by Urru et al. (2018) [12]. Recurrence estimates according to type (locoregional or metastatic recurrence) were based on Steward et al. (2014) [12, 13]. The published Kaplan‒Meier (KM) curves allow differentiation between disease stages (I, II, III and IV), which have been extrapolated to the period under analysis through exponential distribution parametric curves. Nonetheless, an adjustment for recurrence and mortality rates was made on the basis of evidence that the risk of recurrence and death is greater in the first 5 years after diagnosis [10]. Thus, the estimated annual probabilities of death and locoregional and metastatic recurrences, according to cancer stage, are shown in Table 1.

TNBC incidence estimates in Portugal were based on official data [2, 14, 15]. However, as only data from 2010, 2018 and 2019 were published for the period under analysis, the incidence rates for the remaining years were calculated by linear regression. As stated by the Portuguese Society of Oncology, the TNBC subtype accounts for 15% of all cases of breast cancer (BC) [16]. The expert panel (EP) (Table 2) validated the distribution of cases by each cancer stage.

TNBC-related mortality in 2019 was based on the aforementioned CIM, assuming the average life expectancy loss reported by the Global Burden of Disease (GBD) 2019 for each death occurrence [17]. Similarly, cancer stage distribution was also calculated with CIM, whose disability weight values were confirmed by the GBD 2019 (Table 3) [18]. To address the impact of each death in terms of years of life lost (YLLs) and to estimate indirect costs, a distribution of new patients diagnosed with TNBC stratified by age group was necessary. Data from van Maaren et al. (2019) were used to estimate the number of TNBC cases [10].

The burden of disease was assessed on the basis of impact in terms of years of healthy life lost due to disease or premature death (DALY), a metric commonly used by the World Health Organization (WHO) [19]. It includes two time-based indicators, namely, the YLL and the years lost due to disability (YLD). One DALY represents the loss of the equivalent of 1 year living with full health [20]. The present study adopted the Global Burden of Disease (GBD) methodology of 2010, which considers the following [21]:

$$DALY \left(c,s,a,t\right)=YLL \left(c,s,a,t\right)+YLD \left(c,s,a,t\right)$$

[c = cost; s = sex; a = age; t = time]

While YLL is calculated as the number of deaths multiplied by the standard life expectancy at the age at which death occurs, YLD is the number of female patients in each cancer stage multiplied by the corresponding disability weight (Table 3) [18, 22].

The disability weight metric is as follows: “diagnosis and primary phase of treatment” are applicable to new TNBC cases; “clinically disease-free” is valid for patients with local and locoregional stages of TNBC, who have undergone conservative surgery or mastectomy (according to EP inputs); “metastatic phase” accounts for the TNBC metastatic stage; and the “terminal phase” applies to female TNBC patients who died during the year of analysis (Table 3).

The cost of illness was estimated from a societal perspective, considering the prevalence of TNBC and the distribution of patients across all cancer stages (I, II, III and IV) in 2019, including all patients diagnosed over the last 10 years (2010–2019). These estimates were based on inputs from several data sources, such as publicly available data, data from the literature and the EP [23,24,25,26,27,28,29,30]. The EP was conducted to support the cost analysis, both direct and nondirect health care expenditures, and to fill in information gaps regarding standard of care (SoC) approaches. The average calculation was the chosen method to reach consensus in the EP.

Direct medical costs included outpatient and inpatient expenditures and were estimated according to the EP. The outpatient expenditures included outpatient planned and emergency consultations, complementary diagnostic exams (CDE), primary care consultation and pharmacologic therapeutics. Inpatient status considered hospitalizations and outpatient episodes that occurred at the National Healthcare Service (NHS) hospitals, coded according to the 10th Revision of International Classification of Diseases (ICD-10) available from the Hospital Morbidity Database (HMDB). Unit costs were based on official data and remained constant over time [23, 24, 26]. Patient transportation to and from the hospital to attend appointments and treatments/CDE were assessed using the EP input. The cost of each trip was established according to Barros et al. (2015) [25]. Table 1 (see electronic supplementary material [ESM]) summarizes the unit costs related to direct expenditures, as well as the respective data sources.

The indirect costs, such as early retirement, sick leave and work absences due to TNBC, were established following the human capital approach, according to which the indirect costs are estimated on the basis of the company´s expenditure on each employee lost to adverse health disorders [27]. Inputs from the EP supported these estimates.

Considering that women’s working age averages 52 years on the basis of the CIM, the average monthly gross wage was estimated, taking into account the wage per level of education completed and the proportion of women employed in Portugal who completed each of these levels [28,29,30]. On average, the impact of each employed woman on the company, including the social security contribution, adds up to €15,725 per year [28,29,30].

The impact of the TNBC condition on women’s participation in the job market was obtained from the EP and is summarized in Table 4. Expert estimates combined with the employer's wage payment per employed woman and estimates of the number of TNBC cases in employed women allowed estimating TNBC indirect costs related to absenteeism or early retirement from work.

In terms of absenteeism, (i) each set of CDEs, with the same nature, represented the loss of a quarter of a full working day, (ii) each consultation represented half a day of work, (iii) each day of a hospital visit or radiotherapy treatment represented a full day of absenteeism, and (iv) TNBC-related hospitalization accounted for the loss of working days equivalent to twice the average hospitalization period (i.e., 9.68 days). The cost of labour absenteeism was estimated as the number of working days lost to medical care services related to the diagnosis and treatment of TNBC multiplied by the mean cost of a full working day lost (i.e., €43).