Background Prostate cancer is the second most frequent malignant tumour among men, with approximately 900,000 new cases per year.

However, the financial situation of our public health system and the tight budgets of hospitals demand the best use of healthcare resources. Therefore, physicians should use cost-effective strategies in clinical practice.

Purpose To analyse the incremental cost-effectiveness ratio (ICER) of cabazitaxel as a treatment for metastatic prostate cancer in our hospital, and to compare those results with the available literature.

Material and methods A retrospective observational study was carried out to estimate our population progression-free survival (PFS), measured by the Response Evaluation Criteria in Solid Tumours (RECIST). The study lasted 15 months and included all 0–1 ECOG performance status patients who received cabazitaxel for metastatic prostate cancer in our hospital.

Afterwards, costs were calculated using actual dosages and current prices including industry discounts. Administration or indirect costs were not considered.

The incremental cost-effectiveness ratio (ICER) of cabazitaxel was obtained.

Finally, the literature perspective and our estimation from real practice were compared.

Results Ten patients were treated with cabazitaxel for metastatic prostate cancer during the inclusion period. Median age was 68 years old (range 61–83). The median dose was 20 mg m^-2. Four patients required dose reductions (10–15 mg m^-2). The median PFS was 7.0 months.

Pharmacological costs were €15,923.5 per patient treated, and therefore our estimated ICER was €2,274.78/month of PFS.

While published ICERs were estimated at €12,546.32/month of PFS, we obtained a much lower ICER. Hence, a discrepancy of €10,271.54/month of PFS was found.

Conclusion The summary of products characteristics (SPC) recommended dose is higher than actual practice, and consequently costs are overestimated in studies based on SPC data.

On the whole, pharmacoeconomic analyses should be applied in specific settings since general analyses may not be congruous with actual scenarios.

References and/or acknowledgements No conflict of interest.