OBJECTIVES: Describe a methodological proposal and five value-based ratio estimations based in real-world Spanish hospital care episodes, to serve as a reference platform to discuss and achieve value-based healthcare benchmarking.

METHODS: Patient Reported Experience and Outcome Measures (PREM+PROM) data was collected (January-June 2023) 6 months after Knee and Hip Prostheses surgery (set-KP, set-HP), as well as Heart Failure (set-HF) in-hospital admissions in 31 Spanish hospitals. Similar data was collected (February-April 2024) 4 months after programmed Laparoscopic Cholecystectomies (set-LCh) and Oncological Prostatectomies (set-OPr) in 23 Spanish hospitals (including retrospective questions on basal pre-surgery quality of life). PRE-data was based on CAHPS (2022) and Picker (2024) questionnaires, and PROs main source were ICHOM recommendations. All questionnaires scorings were re-scaled into 0-to-100 and averaged to get “value-based-units (VbU)” (each questionnaire counted equal). Costs data was calculated considering each patients’ episode profile with Machine-Learning and the Spanish Net of Hospital Costs database, based on 5 millions of hospital-per-episode costs and 10 partial-cost vectors. Crude PROMs scorings were compared across centers in each set. Costs were divided by PRM values for independent reference benchmark ratios splitting by size and public/private hospital status.

RESULTS: The ratios were based on 804 patients (set-KP), 729 (set-HP), 717 (set-HF), 952 (set-LCh) and 388 (set-OPr) patients. Cost per Value-Units were: 101€/VbU big public-84 medium public-89 private hospitals (set-KP), 110€/VbU-97-86 (set-HP), 102€/VU-46-59 (set-HF), 12€/VU-16-13 (set-LCh) and 68€/VU-56-62 (set-OPr). The higher variability (set-HF) was not associated to outcomes but to differential costs.

CONCLUSIONS: Value-based care invites to work for excellence care analyzing the variability in costs and relevant-to-patient self-reported measures. Methods and consensus evolved and many are still to overcome such like working with final-process outcomes without always considering basal. Nevertheless, the presented results prove the feasibility to approach practical reference benchmarking values.