Risk Score Charts for Indian Population of Traumatic Brain Injury: Development and Validation
Abstract
Traumatic brain injury (TBI) poses a leading cause of mortality, morbidity, disability, and socioeconomic losses in all regions of the Globe. In Asia, many low and middle income countries including India are facing an... [ view full abstract ]
Traumatic brain injury (TBI) poses a leading cause of mortality, morbidity, disability, and socioeconomic losses in all regions of the Globe. In Asia, many low and middle income countries including India are facing an increased risk for TBI due to a rapid surge in urbanization, motorization and economic liberalization. Like diagnosis and treatment, the prognosis is a fundamental responsibility of all clinicians after a TBI. Our aim was to develop and validate the risk score charts for patients with moderate or severe traumatic brain injury using admission characteristics to predict in-hospital mortality and unfavourable functional outcome at 6-months post admission for clinical usefulness. For it, we used trauma database (n=1782 patients) of India’s largest level 1, tertiary care Jai prakash Narayn Apex Trauma Center, New Delhi, India. We developed three different models (n=1466) using stepwise logistic regression method and score charts based on regression coefficients to estimate probability for both the outcomes. We validated these models externally (n=316) and internally (bootstrap method), as well. For both the outcomes, model-1 included age, sex, motor score, pupillary reactivity, limb movement, cause of injury, and major extracranial injury as independent predictors; model-2 included CT features as independent predictors in addition to predictors of model-1 and model-3 included laboratory variables in addition to independent predictors of model-2. The discriminative ability of the three prognostic models was excellent in the development data set [AUC(95%CI):0.853(0.831-0.874)-0.891(0.872-0.911)] and in external validation data set [AUC(95%CI):0.819(0.769-0.869)-0.853(0.804-0.901)]. Calibration in development and validation data set for all the models was very good (H-L test p-value>0.05) except model-1 and model-3 for in-hospital mortality. On the basis of performance, these models are generalizable for predicting outcomes in new patients. We recommend to use the scores based on these models in predicting outcomes in such patients with brain injury in India and other similar countries.
Authors
-
Vineet Kumar Kamal
(All India Institute of Medical Sciences (AIIMS), New Delhi)
Topic Areas
V. Healthcare Service 5.1 Accessibility of healthcare services and its optimization 5.2 He , VI. Methodologies and technologies 6.1 Methodological issues in health research (e.g., MAU , I. Urbanization AND Health: what interactions? 1.1 New paradigms, concepts, methods, and t , II. Urban Health at the intersection of urban environment, social determinants and places , III. Urban Environments: what specificities? 3.1 Urban Environments as places of demograph , VI. Research and action 6.1 Collaboration; interaction of researchers; stakeholders 6.2 S , Topic #15
Session
PS-1 » POSTER SESSION 1 (12:10 - Friday, 1st April, TBA)
Paper
Development_and_Validation_of_the_Score_Charts_for_Indian_Head_Injury_Patients.docx
Presentation Files
The presenter has not uploaded any presentation files.