Article Access Statistics | | Viewed | 1033 | | Printed | 32 | | Emailed | 0 | | PDF Downloaded | 58 | | Comments | [Add] | |
|

 Click on image for details.
|
|
|
COMMENTARY |
|
|
|
Year : 2017 | Volume
: 65
| Issue : 1 | Page : 87-88 |
Outcome prediction in traumatic brain injury: Is it a “Holy Grail?”
Sivashanmugam Dhandapani, Kanchan K Mukherjee
Department of Neurosurgery, Postgraduate Institute of Medical Education and Research, Chandigarh, India
Date of Web Publication | 12-Jan-2017 |
Correspondence Address: Dr. Kanchan K Mukherjee Department of Neurosurgery, Postgraduate Institute of Medical Education and Research, Chandigarh India
 Source of Support: None, Conflict of Interest: None  | Check |
DOI: 10.4103/0028-3886.198243
How to cite this article: Dhandapani S, Mukherjee KK. Outcome prediction in traumatic brain injury: Is it a “Holy Grail?”. Neurol India 2017;65:87-8 |
Traumatic brain injury (TBI) is an increasingly common cause of death and disability among the young in the developing world, causing a colossal socioeconomic burden.[1],[2] Outcome after severe TBI does not merely depend upon the major prognostic factors but also on various other factors such as nutrition, economic status, and access to rehabilitation, which play an even greater role in low and middle income countries.[1],[3] Studies on outcome prediction are as much needed and relevant as the therapeutic trials, especially in countries where the outcome prediction models from developed countries might not fit well.[4],[5] Second, being a resource-constrained country, we require predictive models to optimally allocate our facilities in areas relevant to our needs.[4],[5]
Deepika et al.,[6] in their prospective study of 88 patients who were admitted in the intensive care unit with severe TBI, found the disability rating scale (DRS) at discharge to have a “fair-to-good” predictive value in determining Glasgow outcome scale extended (GOSE) at 6 months using a bivariate rank-order correlation coefficient and area under “receiver operating characteristics (ROC)” curve (AUC). The DRS at discharge had significant correlation with DRS and GOSE at 6 months; whereas the DRS and GOSE at 6 months had significant correlation with each other. While the conventional outcome prediction models relied on admission parameters, the suggested model highlights the utility of outcome prediction among patients alive at discharge. DRS was originally developed as a way to track a TBI patient from “coma to community,” immensely helping in identifying patients that are most likely to benefit from intensive rehabilitation care. While patients with lower DRS may not need much support, it also may not be cost effective to provide an aggressive support to those with DRS at discharge being greater than 20.6. Those with discharge DRS just below 20 would benefit the most from vigorous supportive management.
Though the present study is relevant for the focused rehabilitation of selected patients in a resource-constrained set-up, using DRS at discharge poses a few issues limiting its utility. The discharge protocols of different institutions may vary based on the interval elapsed since the admission, the surgical or conservative management administered, the hospitalization complications encountered such as infection, whether the hospital stay was in the general or the private ward, whether the patient was discharged to domiciliary care or to a peripheral centre, and finally, the willingness of guardians to actually get the patient discharged.[7],[8] As DRS is known to have a greater sensitivity than GOS in detecting and measuring clinical changes in patients with severe TBI, the rank order correlation of DRS at discharge with GOSE at 6 months may not come as a surprise. The impact of baseline prognostic factors on DRS omits patients who have died due to various reasons, further limiting its utility in detecting differences across various treatment groups.
The outcome prediction models in TBI either tend to become too simple to accurately predict the actual condition of the patient or too complex to be practical in the clinical practice.[5],[7] The frailty of outcome prediction has been due to the inclusion of select patients, and excluding factors which otherwise may play a key role in the real world.[7],[9] Because of these reasons, the International Mission for Prognosis and Analysis of Clinical Trials (IMPACT) project recommends broad criteria for the inclusion of patients based upon wider generalizability, detailed treatment protocols, extensive reporting of baseline confounding factors, and covariate adjusted statistical analysis to mitigate heterogeneity due to age, Glasgow coma Scale score (GCS), pupillary findings, computed tomography (CT) characteristics, and the simultaneous presence of the extracranial injury.[7],[9] Studies on TBI should incorporate these recommendations, so that the results are scientifically robust and applicable in a wide variety of patients without any bias. This is especially important among studies with a smaller number of patients, which can skew the results to a significant extent. For studies on outcome estimation, a comparative evaluation of simpler outcome scales with more sensitive tests is also recommended, such as the assessment of neuropsychological status and functional independence or of the composite outcome data combining multiple measures, having their inherent strengths and weaknesses.
As the list of prognostic factors keeps growing to encompass biomarkers, nutritional, and socioeconomic factors,[1],[3] larger and more extensive datasets of all possible factors from different centres across the world constituting “Big data,” such as the IMPACT (International Mission for Prognosis and Analysis of Clinical Trials in TBI) and the CRASH (corticosteroid randomisation after significant head injury) datasets, are needed to generate comprehensive and lucid evidence for outcome prediction.[4]
» References | |  |
1. | Dhandapani SS, Manju D, Sharma BS, Mahapatra AK. Prognostic significance of age in traumatic brain injury. J Neurosci Rural Pract 2012;3:131-5.  [ PUBMED] |
2. | Tripathi M, Tewari MK, Mukherjee KK, Mathuriya SN. Profile of patients with head injury among vehicular accidents: An experience from a tertiary care centre of India. Neurol India 2014;62:610-7.  [ PUBMED] |
3. | Dhandapani SS, Manju D, Mahapatra AK. The economic divide in outcome following severe head injury. Asian J Neurosurg 2012;7:17-20.  [ PUBMED] |
4. | Gao J, Zheng Z. Development of prognostic models for patients with traumatic brain injury: A systematic review. Int J Clin Exp Med 2015;8:19881-5. |
5. | Mukherjee KK, Sharma BS, Ramanathan SM, Khandelwal N, Kak VK. A mathematical outcome prediction model in severe head injury: A pilot study. Neurol India 2000;48:43-8.  [ PUBMED] |
6. | Deepika A, Devi I, Shukla D. Predictive validity of disability rating scale in determining functional outcome in patients with severe traumatic brain injury. Neurol India 2017;65:83-6. |
7. | Dhandapani S, Sarda AC, Kapoor A, Salunke P, Mathuriya SN, Mukherjee KK. Validation of a new clinico-radiological grading for compound head injury: Implications on the prognosis and the need for surgical intervention. World Neurosurg 2015;84:1244-50. |
8. | Tripathi M, Kapoor A, Bajaj A, Kaur R, Mukherjee KK. You need not operate every case of compound depressed skull fracture. J Yoga Phys Ther 2016;6:2. |
9. | Maas AI, Steyerberg EW, Marmarou A, McHugh GS, Lingsma HF, Butcher I, et al. IMPACT recommendations for improving the design and analysis of clinical trials in moderate to severe traumatic brain injury. Neurotherapeutics. 2010;7:127-34. |
|
 |
|
|
|
|