The Italian public healthcare system is managed regionally, providing universal coverage at the point of delivery, which is largely free of charge and primarily supported by general taxation (18). Its policies are grounded on the fundamental values of universality, free access, freedom of choice, pluralism in provision, and equity.

Veneto is an Italian region located in the North-eastern area of the country, with a population of 4.9 million residents and a mean age of 54.4 years. Considering the latest available socio-economic indicators, the Veneto region has a per capita gross domestic product (GDP) of 33,800 euros, with an unemployment rate of 5.8%, and 29.3% of Veneto inhabitants aged 30-34 have a university degree (11, 19).

In 2015, the Veneto Oncology Network (ROV) detailed the complete procedures for the clinical management of CMM patients in a document based on the current national and international literature (20–23). It defines standardized clinical care pathways from diagnosis to end-of-life care, as well as a set of indicators to assess the quality of care by evaluating the consistency between recommendations and real-world clinical practice (24).

This population-based cohort study includes all 2015 and 2017 incident cases of cutaneous malignant melanoma recorded by the Veneto Cancer Registry (RTV) in the cutaneous malignant high-resolution regional cancer registry, which provides demographic coverage of the entire region on a two-year basis.

The RTV recording procedures rely on various informative sources such as pathology reports, clinical charts, death certificates, and health service administrative records. The following variables were extracted by high resolution registry for the present study: age, sex, 8th edition AJCC TNM stage at diagnosis, and CMM histological subtypes.

The cost analysis was conducted from a health system perspective. Data on visits to outpatient clinics, specialist services, drug prescriptions, hospital or hospice admissions, treatments at the emergency department, and the use of medical devices were obtained from the regional administrative subject-level databases (see below). The cost of any diagnostic or therapeutic (surgical or other) interventions was based on the reimbursement rates established by the Veneto Regional Authority. For cost assessment we considered the following sources:

These distinct databases and their cost categorizations contribute to a more comprehensive analysis of healthcare expenditure patterns in the studied context, which is the Italian healthcare system.

Each patient was linked via a unique and anonymous identification code to all administrative data (see the cost assessment sources list above). All costs sustained from two years before melanoma diagnosis up to four years after diagnosis were included. The average real-world costs per year per patient (total and per single item of expenditure) were calculated and stratified by the following clinical variables: site of primary tumour, histological phenotype, and TNM stage at initial cancer assessment.

Differences in total costs, item expense distribution (comparing < 64 vs. 65-year-old patients), TNM stage, CMM histology subtype (i.e., nodular versus non-nodular), and survival time were also assessed by stratifying cost data by these variables.

Costs are summarized as survival-weighted means and medians, 95% confidence intervals for mean and minimum–maximum intervals are also provided.

Survival-weighted costs were calculated (for each year and for the entire period observed) by dividing the sum of total incurred costs by the observed person-time, i.e., the sum of total time (days of the year) in which each patient was alive, and then transformed to person-years.

A log-linear multivariable analysis was performed to investigate potential cost drivers among the available clinic-pathological characteristics. The mean person-years melanoma-specific costs sustained from diagnosis to 4 years after diagnosis defined the output variables of four different models adapted to the data. Costs were transformed to the logarithmic in order to manage the skewed distribution and meet linear regression assumptions. The independent variables considered were age, sex, TNM stage at diagnosis, mitotic count, ulceration, regression, growth phase, TILs, together with tumour location, histology subtype, and time to death.

Results were deemed statistically significant at the p < 0.05 level. Data preparation, analyses and visualizations were conducted using R (25) and Python (26).

Overall, analysing all data together, CMM patients tend to have higher costs in the first year after diagnosis, which then gradually decrease in subsequent years. As shown in the stage-specific sub analysis, this is only true when considering all CMM patients together, regardless of the stage of the disease at diagnosis ( Figure 3 ).

In a previous publication analysing direct health care costs in patients with cutaneous malignant melanoma, Lyth and colleagues also reported a similar decrease in mean costs over time, both overall and depending on the clinical stage at diagnosis (15). Our data further reveal specific trends, by grouping costs into cost items (e.g., hospitalization, outpatient services, inpatient drugs, etc.). It should be noted here that under the Italian NHS, the reported costs of care are equal to the sum of the individual fees for each medical procedure provided to the patient, which are fixed amounts defined a priori by regional regulations. Thus, it can be observed that hospital-related costs peak in the first year after diagnosis and then gradually decrease over the years, suggesting that hospitalization is mainly considered at the beginning of treatment. In contrast, hospice-related spending shows the opposite trend, as patients with CMM could benefit from such care towards the end of their lives. The reason why hospital drug costs peak two years after diagnosis and then decline in subsequent years deserves to be investigated in future studies, as there is no obvious explanation either from the scientific literature or from analysis of the clinical care pathway established by regional guidelines (27). Finally, spending associated with outpatient services and emergency department visits does not seem to follow any obvious pattern that can be analysed.

As is known, advanced tumour stages are associated with higher costs of care (14, 15, 28, 29). Indeed, patients in our cohort that had stage IV CMMs were associated with costs that were more than 20 times higher than those with stage I (€80,726 vs. €3,807). Two other studies also reported melanoma-related costs of care according to tumour stage and time at diagnosis – these 2016 studies were carried out by Serra-Arbeloa et al. (14) and Lyth et al. (15). They drew data from the official Spanish national health system reports and the Swedish cancer registry (respectively). A comparison of first-year melanoma treatment costs shows that for stage I melanoma patients, we in Italy show the lowest first-year costs (€1,401), compared to the Spanish (€3,103) and the Swedes (€2,670). For stage IV patients, the Swedes reported the lowest expenditure (€29,291), followed by our present study (€30,762), with the Spanish reporting significantly higher expenditure than both (€88,268).

In most cases, the main cost driver is inpatient drugs, however this is not the case for stage I CMMs, which instead show higher costs related to outpatient services. The relative impact of hospitalization is greater for lower cancer stages because other cost items are relatively limited in these early stages of the disease. Even though it should be noted that hospitalizations are (in absolute value) costlier for more advanced melanomas, this drops slightly for stage IV melanomas.

As stated by Cullison C.S. and Bordeaux J.S. in a letter to the editor, excision of primary cutaneous melanoma is a low-risk operation usually associated with few complications, therefore hospitalization is not an expected consequence (30). While patients with stage I CMM often recover after the first tumour removal, more advanced melanoma patients face higher recurrence rates, which may influence the course of their disease in subsequent years and thus also increase costs (8, 31–33). Indeed, Leenaman and colleagues reported a 47% recurrence rate for stage III melanoma patients, compared to 29% for stage II and 8% for stage I (33). In stage IV advanced metastatic cutaneous malignant melanomas, surgical or other ablative treatments are limited to palliative care or as a complementary treatment in a combined approach to oligometastases (20, 23, 34–38).

Other costs four years from diagnosis differ widely among melanoma stages. Data highlight how the use of inpatient drugs heavily influences the cost of care for CMM, moving from an average cost of just over €6,000 for stage II melanomas to nearly €50,000 for stage IV melanomas. It should be noted that emergency department visits are always a minor factor in the total cost associated with CMM care, always accounting for less than 1% of the total expenditure.

Overall, data show that death increases healthcare costs associated with CMM, where patients who survive beyond the observation period show substantially lower annual expenditure than patients who succumb to their disease. Specifically, patients who died during the observation period had the highest overall costs in their last year of life. In a recent literature review, Pisu and colleagues (39) summarized healthcare expenditure related to cancer care for multiple tumour types, stating that the initial and end-of-life phases are associated with the highest expenditure (both direct and out-of-pocket costs), while costs over the middle phases are generally lower. In an older systematic review covering articles from 1990-2011, Guy and colleagues (28) analysed the cost of melanoma care, highlighting that healthcare costs tend to be highest in the terminal phase and lowest in the interim phase (largely according to two studies conducted in the US). Consistently our present study indicates that end-of-life costs are highest and increased by hospice care (peaking in the last year of the patients’ lives) as well as hospitalization-related costs (which are higher than those seen in intermediate phases).

Other factors can also significantly influence costs associated with CMM care (both overall and per year in the four years following diagnosis), such as tumour growth patterns. As hypothesized several decades ago, tumour progression in melanomas goes through successive stages of growth, from radial to vertical to metastatic disease (40–42). Invasion of the lower layers of the dermis has been associated with increased tumour aggressiveness, also linked by some authors to the expression of specific biological markers by the tumour (42). Consequently, it can be argued that the higher costs of vertically growing CMMs may be a consequence of the higher intensity of care required due to the more aggressive behaviour of these tumours (14, 43). However, this direct association between tumour growth pattern, treatment intensity and overall cost of care needs specific investigation to test its validity.

The present study comes with the typical limitations and strengths seen in cohort studies that are based on regional, linked administrative databases. Even if a defined clinical pathway is common between different healthcare facilities in a region, the simple fact of including subjects that were treated in different hospitals could have distorted the results, not to mention the effect of dataset inaccuracies or missing values. Furthermore, due to the structure of these databases, our study could not explore the indirect costs of cutaneous malignant melanoma, thus limiting the goal of the research. Further studies are also needed to analyse cost differences after recent changes to the CMM treatment paradigm and the introduction of immunotherapies.

In contrast, the main strength of this study lies in its population-based design, which enables it to provide information on costs retrieved from real-world clinical practice, while minimizing the risk of selection bias that can occur with independently collected data.