Application of Principal Component Analysis in the estimation of potential factors associated with the prevalence of deaths in cases of spinal cord trauma

Abstract

Background: In Brazil, the mortality indicators for spinal cord trauma (SCT) are poorly known. Therefore, identifyng potential risk factors for death due to SCI is fundamental for preventive actions which suits each country region’s specificities. Design and setting: Cross-sectional study (Parecer no. 1.451.493), with the analysis of 786 medical records of SCT cases hospitalized between 2010 and 2015 at a reference hospital in Belém, Pará, Brazil. Methods: We used Principal Component Analysis (PCA) to estimate potential factors that contribute to the SCT, the following factors were analyzed: intervention (conservative/surgical), hospitalization period ( one week), vertebral segment affected (cervical or not), treatment cost ( 1 minimum wage), management (multiprofessional or not), distance to care, sequels, age and gender. Data analysis was performed using R. Results: The type of management and intervention and patients’ age were factors that most influenced (~20%) the outcome (death/survival). However, this indicates that ~ 80% of the outcome’s variability is due to other factors. Conclusions: PCA is useful to indicate potential factors to be considered as influencing the prevalence of death in SCT cases. The use of these data in discriminatory analyzes, hereafter, can help to clarify this topic.