Modelling Prognostic Factors on Traumatic Care Pathways: An Application of Multistate Models

Javanmardi, Fatemeh; Shayan, Zahra; Safarinejadian, Parisa; Paydar, Shahram; Shayan, Leila

doi:10.30491/tm.2025.469561.1751

Modelling Prognostic Factors on Traumatic Care Pathways: An Application of Multistate Models

Document Type : Original Article

Authors

Fatemeh Javanmardi ¹

Zahra Shayan ²

Parisa Safarinejadian ¹

Shahram Paydar ³

Leila Shayan ³

¹ Department of Biostatistics, School of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran

² Department of Biostatistics, Trauma Research Center, School of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran.

³ Trauma Research Center, Rajaee (Emtiaz) Trauma Hospital, Shiraz University of Medical Sciences, Shiraz, Iran.

10.30491/tm.2025.469561.1751

Abstract

Introduction: Traumatic experiences are commonplace worldwide and have been proven to have detrimental effects on health. Unlike traditional survival analysis, which focuses on a single endpoint, multistate models are beneficial for evaluating the progression of multiple health conditions over time. The primary goal of this paper is to employ a multistate model to ascertain the patterns in hospitalizations among traumatic patients.
Method: In this longitudinal study, 502 eligible trauma patients who were referred to Shahid Rajaee Hospital in Shiraz, Iran, were chosen and followed from July 2018 to March 2019. A Semi-Markov Multistate model was utilized in the current study. Patients were assumed to transition between five states. Transition times (triage → general ward, triage → death, general ward → Intensive Care Unit [ICU], general ward → death, and surgical ward → death) were assumed to follow an exponential distribution. The hazard ratio (HR) for each covariate was estimated for each transition.
Result: Based on in-hospital triage evaluations, some patients needed surgery. Injury Severity Score (ISS) more than 15 (HR= 1.41), blunt trauma with brain injury (HR=2.15), hypotension (HR=1.41), and low pulse rate (HR=1.40) increased the probable requirements of the surgery. Following surgical treatments, patients with moderate Glasgow Coma Scale (GCS) were more likely to die (HR=1.42). Further, those who had experienced blunt trauma and brain damage had a lower chance of death following surgery (HR=0.62). The need for intensive care directly after triage is more likely in cases with a severely low GCS score.
Conclusion: This study confirmed that elderly patients are at lower risk of surgical interventions after ICU. Cases with more injuries were more likely to require surgery after triage. Identifying specific prognostic factors that significantly impact the progression and outcomes of traumatic care can help healthcare providers prioritize interventions and allocate resources more effectively, ultimately enhancing patient outcomes.

Keywords

Trauma Care

Prognostic factors

Pathway

Models

Subjects