Does an International Version of ICISS Predict Mortality for Patients Admitted With Trauma in Four Public University Hospitals in Urban India?
This trial has been completed.
|Condition||wounds and injuries|
|Collaborator||Lokmanya Tilak Municipal General Hospital|
|Start date||January 2016|
|End date||January 2017|
|Trial size||16047 participants|
|Trial identifier||NCT02724007, alice-claeson-201603271932|
The aim of this study is to validate international versions of the International Classification of Disease Injury Severity Score (ICISS) in adult trauma patients admitted to four public university hospitals in urban India.
time frame: Within 30 days of patient arrival to participating centre
time frame: Within 24 hours of patient arrival to participating centre
All participants of any age.
Inclusion Criteria: - History of trauma or death between arrival and admission Exclusion Criteria: - Isolated limb fractures without vascular injury
|Official title||Does an International Version of ICISS Predict Mortality for Patients Admitted With Trauma in Four Public University Hospitals in Urban India?|
|Description||Introduction Trauma causes 10% of the world's annual deaths. Trauma-related deaths are most common in low- and middle-income countries, which accounts for over 90% of global trauma mortality. In 2013 over 1.2 million trauma deaths occurred in India alone. Hence, India accounted for 25% of trauma-related deaths worldwide. Experiences from high income countries show that the establishment of trauma registers have been of importance for the improvement of trauma care and research, allowing for comparisons between different hospitals as well as over time. To be useful such comparisons should be risk-adjusted to account for differences in care and patient case mix. In trauma one of the most important patient case mix characteristics to adjust for is injury severity. Several different injury severity scores exist for this purpose. One of these scores is the International Classification of Disease Injury Severity Score (ICISS). In contrast to other established scores, such as the Injury Severity Score (ISS), ICISS can be calculated based on injuries coded using ICD for administrative purposes, making it cost effective. The validity of ICISS has primarily been studied in high-income countries ICISS requires an empirically estimated Survival Risk Ratio (SRR) for each ICD code. Some studies from high income countries has shown similar predictive performance of ICISS when using SRRs derived from other high-income countries. However, the generalisability of such SRRs to low- and middle-income countries such as India has not been researched. Therefore, the aim of this study was to validate an international version of ICISS in patients presenting with trauma to four public university hospitals in urban India. Study design Data from a multi-centre cohort study will be used. Setting Data from the Towards Improved Trauma Care Outcomes (TITCO) in India project will be used, which was conducted at four public university hospitals in urban India. The four centres were Lokmanya Tilak Municipal General Hospital in Mumbai, King Edward Memorial Hospital in Mumbai, Jai Prakash Narayan Apex Trauma Center in Delhi, and the Institute of Post-Graduate Medical Education and Research and Seth Sukhlal Karnani Memorial Hospital in Kolkata. Data was collected between October 2013 and September 2015. Patient data was collected by one project officer in every study center. The project officer worked eight hour shifts per day, with a rotating schedule between day and night shifts. All project officers had at least a health science master degree, and were continuously trained and supervised by project management. Details on the data collection process has been published elsewhere. Source and method of participant selection Eligible patients were identified by project officers through direct observation in the emergency room and/or through extraction from patient records. Data from patients admitted outside of the shifts was collected retrospectively within days. Patients were followed up until discharge, death, or 30 days, whichever occurred first. Patients transferred to other hospitals were considered as discharged. Covariates The explanatory variable will be ICISS. Other covariates include age, sex, mechanism of injury, mode of transportation and transfer status. These covariates will be used to characterise the study sample. Data sources/measurements Patient outcomes was followed up at 30 days, death in hospital or discharge, whichever occurred first. Patients discharged alive before 30 days were considered to be alive at 30 days after arrival to the hospital. ICISS will be assigned based on ICD-coding of the free text injury descriptions, using the 10th revision of the World Health Organization's International Classification of Diseases (ICD-10). Survival Risk Ratios from two published data sets will be used. One of the data sets had SRRs calculated from an Australian and New Zealand population (14). The other had SRRs created by pooling data from Australia, Argentina, Austria, Canada, Denmark, New Zealand and Sweden (12). Final ICISS for patients with multiple injuries will be calculated by multiplying the SRRs for all ICD codes assigned to the patient. Bias All patients matching the selection criteria will be included in the study. All data was collected from patient records and examinations done by hospital personnel with no association to the study. The data collectors worked independently form the project managers. The free text injury descriptions will be extracted from the data set and assigned ICD-codes with no other data available, including data on patient outcome. Study size The sample size required is calculated to give an 80% power to detect substantial differences in model performance such as 1.5 times too high or low predicted survival probability. The calculations were based on published recommendations on the sample size needed to detect such differences in the external validity of prediction models with a binary outcome. Based on these recommendations, the sample size needed to include 100 consecutive events, i.e. in-hospital deaths within 24 hours and all non-events enrolled during the same period, was calculated. In-hospital mortality within 24 hours was used for the sample size calculation to power the study for secondary outcomes as well. Quantitative variables All quantitative variables will be analysed as continuous. Statistical methods and analyses R will be used for all statistical analyses. A confidence level of 95% and a significance level of 5% will be used. For each patient one ICISS will be calculated for each of the two sets of SRR referenced above. Henceforth, the ICISS based on SRR from Australia and New Zealand will be denoted ICISS-ANZ, and the ICISS based on the international SRR derived by Gedeborg et al. will be denoted ICISS-I. The performance of ICISS-ANZ and ICISS-I in predicting death within 30 days and death within 24 hours will be estimated in terms of discrimination and calibration. The area under the receiver operating characteristics curve (AUROCC) will be used as a measure of discrimination. Calibration will be assessed visually by plotting observed versus predicted outcomes and will be quantified by calculating a calibration slope. The main analysis will be a complete case analysis in which only patients with complete data on outcomes and covariates will be included. Both ICISS-ANZ and ICISS-I will be based only on SRR calculated from ICD-10 codes that occurred in at least 20 patients in respective derivation dataset. Patients without any injury reported will be assigned an ICISS of 1. Sensitivity analyses Four sensitivity analyses will be conducted. The first sensitivity analysis will include patients with missing data on covariates but with complete outcome data. The second sensitivity analysis will exclude patients without any reported injury. The third sensitivity analysis will include all SRR regardless of the number of occurrences in the original derivation datasets. In the final sensitivity analysis the ICISS for each patient will be based only on unique ICD-10 codes, in other words, each ICD-10 code was only allowed to contribute one SRR to ICISS even if it occurred more than once in the same patient.|
Call for more information