The administration of fluids is one of the most common therapies for patients in inten- sive care. However, excessive fluid administration increases patient mortality; therefore, proper management of administered fluids is advisable. In this context, fluid creep refers to the amount of intravenous fluids that critically ill patients receive as a diluent for med- ications and to maintain catheter patency, which significantly contributes to the overall quantity of administered fluids. We conducted a retrospective single-center analysis using the MargheritaTre database with the aim of identifying variables associated with high fluid creep administration and developed a predictive model that estimates the risk of intensive care patients receiving high fluid creep. This study considered two aspects: first, the fluid received in the first 24 hours of intensive care admission, and second, the total fluid received by patients during the first week of hospitalization. The study includes 4786 patients who received an average of 1.6 liters of fluids in the first 24 hours and an average of 2.4 liters during the first week of hospitalization. The results show a significant association with AKI developed within the first 24 hours (P-value 0.015), sodium (P-value 0.029), bicarbonate (P-value 0.036), respiratory failure (P-value < 0.001), absence of comorbidities (P-value < 0.001), metabolic balance disorder (P-value 0.002), tumor without metastasis (P-value 0.017), organic brain coma (P-value 0.02), trauma (P-value 0.01), antiplatelet therapy (P-value 0.02), vasculopathy (P-value 0.03), surgery (P-value 0.04), septic shock (P-value < 0.001), intensive treatment (P- value < 0.01), creep within the first 24 hours (P-value < 0.001), heart rate (P-value 0.028), respiratory abonrmalities (P-value < 0.01), mixed shock (P-value 0.003), intensive treatment (P-value 0.024), COPD (P-value 0.025), AKI with non-conservative diuresis (P- value 0.029), hypovolemic shock (P-value 0.027), mild chronic pulmonary disease (P-value 0.045), and cerebral vasculopathy (P-value 0.048). The best machine learning prediction results for the first day and the first week showed respectively an accuracy of 0.71 and 0.80 on the hidden test set in predicting patients at risk of receiving high fluid creep. In conclusion, this study investigates the role of various variables associated with high fluid creep both during the first 24 hours and during the first week and demonstrates an exemplary model for identifying patients at higher risk of receiving fluid creep.
La somministrazione di fluidi è una delle terapie più comuni per i pazienti in terapia inten- siva. Tuttavia, un’eccessiva somministrazione di fluidi aumenta la mortalità dei pazienti, pertanto è consigliabile una corretta gestione dei fluidi somministrati. In questo contesto, per fluid creep si intende la quantità di fluidi endovenosi che i pazienti gravemente malati ricevono come diluente per i farmaci e per mantenere la pervietà dei cateteri, che con- tribuisce in modo significativo alla quantità complessiva di fluidi somministrati. Abbiamo condotto un’analisi retrospettiva monocentrica utilizzando il database MargheritaTre con l’obiettivo di identificare le variabili associate a un elevata somministrazione di fluid creep e abbiamo sviluppato un modello predittivo che stima il rischio del paziente in terapia intensiva di ricevere un elevato scorrimento dei fluidi; questo studio è stato condotto con- siderando due aspetti: prima lo scorrimento ricevuto nelle prime 24 ore di ammissione in terapia intensiva e poi lo scorrimento totale ricevuto dai pazienti durante la prima settimana di ricovero. Lo studio comprende 4786 pazienti che hanno ricevuto una media di 1,6 litri di fluidi nelle prime 24 ore e una media di 2,4 litri durante la prima setti- mana di ricovero. I risultati mostrano un’associazione significativa tra AKI sviluppato nelle prime 24 ore(P-value 0.015), sodio (P-value 0.029 ), bicarbonato (P-value 0.036), insufficienza respiratoria (P-value < 0.001), assenza di comorbidità (P-value < 0.001), dis- tubi dell’equilibrio metabolico (P-value 0.002), tumore senza metastasi (P-value 0.017), coma cerebrale (P-value 0.02), trauma (P-value 0.01), terapia anticoagulante (P-value 0.02), vasculopatia (P-value 0.03), tipo di chirurgia (P-value 0.04), shock settico (P-value < 0.001), trattamento intensivo (P-value < 0.01), creep nelle prime 24 ore (P-value < 0.001), frequenza cardiaca (P-value 0.028), anomalie del tratta respiratorio (P-value < 0.01), schock misto (P-value 0.003), trattamento intensivo (P-value 0.024), BPCO (P- value 0.025), AKI a diuresi non conservativa (P-value 0.029), shock ipovolemico (P-value 0.027), malattia polmonare cronica lieve (P-value 0.045) and vasculopatia cerebrale (P- value 0.048). I migliori risultati della predizione di machine learning per il primo giorno e per la prima settimana hanno mostrato rispettivamente un’accuratezza di 0,71 e 0.80 sul set di test nascosto nel prevedere i pazienti a rischio di ricevere un elevato fluid creep. In conclusione, questo studio indaga il ruolo di diverse variabili nell’essere associate a un elevato fluid creep sia durantele prime 24 ore sia durante la prima settimana e mostra un modello promettente per identificare i pazienti a maggior rischio di ricevere elevata fluid creep.
Effect and impact of fluid creep in intensive care unit patients receiving intravenous fluid therapy
CARPANI, GIULIA
2022/2023
Abstract
The administration of fluids is one of the most common therapies for patients in inten- sive care. However, excessive fluid administration increases patient mortality; therefore, proper management of administered fluids is advisable. In this context, fluid creep refers to the amount of intravenous fluids that critically ill patients receive as a diluent for med- ications and to maintain catheter patency, which significantly contributes to the overall quantity of administered fluids. We conducted a retrospective single-center analysis using the MargheritaTre database with the aim of identifying variables associated with high fluid creep administration and developed a predictive model that estimates the risk of intensive care patients receiving high fluid creep. This study considered two aspects: first, the fluid received in the first 24 hours of intensive care admission, and second, the total fluid received by patients during the first week of hospitalization. The study includes 4786 patients who received an average of 1.6 liters of fluids in the first 24 hours and an average of 2.4 liters during the first week of hospitalization. The results show a significant association with AKI developed within the first 24 hours (P-value 0.015), sodium (P-value 0.029), bicarbonate (P-value 0.036), respiratory failure (P-value < 0.001), absence of comorbidities (P-value < 0.001), metabolic balance disorder (P-value 0.002), tumor without metastasis (P-value 0.017), organic brain coma (P-value 0.02), trauma (P-value 0.01), antiplatelet therapy (P-value 0.02), vasculopathy (P-value 0.03), surgery (P-value 0.04), septic shock (P-value < 0.001), intensive treatment (P- value < 0.01), creep within the first 24 hours (P-value < 0.001), heart rate (P-value 0.028), respiratory abonrmalities (P-value < 0.01), mixed shock (P-value 0.003), intensive treatment (P-value 0.024), COPD (P-value 0.025), AKI with non-conservative diuresis (P- value 0.029), hypovolemic shock (P-value 0.027), mild chronic pulmonary disease (P-value 0.045), and cerebral vasculopathy (P-value 0.048). The best machine learning prediction results for the first day and the first week showed respectively an accuracy of 0.71 and 0.80 on the hidden test set in predicting patients at risk of receiving high fluid creep. In conclusion, this study investigates the role of various variables associated with high fluid creep both during the first 24 hours and during the first week and demonstrates an exemplary model for identifying patients at higher risk of receiving fluid creep.File | Dimensione | Formato | |
---|---|---|---|
2024_04_Carpani_Giulia_Tesi.pdf
Open Access dal 20/03/2025
Descrizione: Tesi
Dimensione
1.57 MB
Formato
Adobe PDF
|
1.57 MB | Adobe PDF | Visualizza/Apri |
2024_04_Carpani_Giulia_Executive_Summary.pdf
Open Access dal 20/03/2025
Descrizione: Executive Summary
Dimensione
518.3 kB
Formato
Adobe PDF
|
518.3 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/219157