The nursing process and total health cost variability: an analysis using machine learning.

Area of Science:

• Healthcare economics
• Nursing informatics
• Health services research

Background:

• The Spanish healthcare system does not currently integrate nursing process data into expenditure calculations.
• This omission may lead to inefficient resource allocation and suboptimal healthcare management.

Purpose of the Study:

• To determine if nursing process information, including functional patterns and NANDA-NIC-NOC taxonomy from clinical histories, influences predictions of total healthcare costs.
• To assess the predictive value of nursing data in healthcare expenditure models.

Main Methods:

• An analytical, retrospective observational study involving 1,691,075 individuals over 15 years old.
• Explanatory variables included age, sex, and nursing process data; total healthcare cost was the outcome.
• Bivariate analysis, multiple regression, and machine learning models (neural networks, XGBoost) were employed for multivariate analysis.

Main Results:

• 58% of the population had nursing process data. Individuals with assessed nursing patterns had higher average costs (€2304.17) than those without (€950.93).
• Those with a nursing diagnosis incurred higher average costs (€1,666) compared to those without (€840).
• Machine learning models achieved a coefficient of determination (R² ) of 21.45%, indicating significant predictive power.

Conclusions:

• The developed model, incorporating age, sex, and nursing process information, explains over 21% of the variability in total healthcare costs.
• Highlighting the impact of nursing care on patient costs can support its integration into management programs and risk adjustment models for better healthcare planning.