Correction
2 Jun 2014: The PLOS ONE Staff (2014) Correction: Early Hospital Mortality among Adult Trauma Patients Significantly Declined between 1998-2011: Three Single-Centre Cohorts from Mumbai, India. PLOS ONE 9(6): e99694. https://doi.org/10.1371/journal.pone.0099694 View correction
Figures
Abstract
Background
Traumatic injury causes more than five million deaths each year of which about 90% occur in low- and middle-income countries (LMIC). Hospital trauma mortality has been significantly reduced in high-income countries, but to what extent similar results have been achieved in LMIC has not been studied in detail. Here, we assessed if early hospital mortality in patients with trauma has changed over time in an urban lower middle-income setting.
Methods
We conducted a retrospective study of patients admitted due to trauma in 1998, 2002, and 2011 to a large public hospital in Mumbai, India. Our outcome measure was early hospital mortality, defined as death between admission and 24-hours. We used multivariate logistic regression to assess the association between time and early hospital mortality, adjusting for patient case-mix. Injury severity was quantified using International Classification of Diseases-derived Injury Severity Score (ICISS). Major trauma was defined as ICISS<0.90.
Results
We analysed data on 4189 patients out of which 86.5% were males. A majority of patients were between 15 and 55 years old and 36.5% had major trauma. Overall early hospital mortality was 8.9% in 1998, 6.0% in 2002, and 8.1% in 2011. Among major trauma patients, early hospital mortality was 13.4%, in 1998, 11.3% in 2002, and 10.9% in 2011. Compared to trauma patients admitted in 1998, those admitted in 2011 had lower odds for early hospital mortality (OR = 0.56, 95% CI = 0.41–0.76) including those with major trauma (OR = 0.57, 95% CI = 0.41–0.78).
Conclusions
We observed a significant reduction in early hospital mortality among patients with major trauma between 1998 and 2011. Improved survival was evident only after we adjusted for patient case-mix. This finding highlights the importance of risk-adjustment when studying longitudinal mortality trends.
Citation: Gerdin M, Roy N, Dharap S, Kumar V, Khajanchi M, Tomson G, et al. (2014) Early Hospital Mortality among Adult Trauma Patients Significantly Declined between 1998-2011: Three Single-Centre Cohorts from Mumbai, India. PLoS ONE 9(3): e90064. https://doi.org/10.1371/journal.pone.0090064
Editor: Yu-Kang Tu, National Taiwan University, Taiwan
Received: September 2, 2013; Accepted: January 29, 2014; Published: March 3, 2014
Copyright: © 2014 Gerdin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was funded through grants from the Second Assist and the Swedish National Board of Health and Welfare. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Background
Traumatic injury is a major threat to global public health [1], [2]. More people die annually from traumatic injuries than from HIV/AIDS, tuberculosis, malaria, and obstetric conditions combined [1]. Over 90% of the five million annual deaths from traumatic injuries occur in low- and middle-income countries (LMIC) [3], [4]. A recent study estimated that almost two million lives could be saved each year, if hospital care for the injured, i.e. trauma care in LMIC can be improved and reach the same level as in high-income countries (HIC) [5]. However, it is currently unclear how such reduction can be achieved.
Although a large part of trauma mortality occurs at the injury-site and during pre-hospital transportation, around 30-50% of trauma mortality occurs in hospital [6], [7]. In HIC, hospital trauma mortality has been significantly reduced in recent years [8], [9]. Among the most important explanations for this reduction are implementation of trauma care systems and improved medical and surgical treatment [10], [11], including management of traumatic brain injury [12], haemorrhage control [13], and musculoskeletal injuries [14], [15]. These improvements in clinical care have been driven, in large part, by research progress, from the basic sciences to the systems and policy level.
Hospital trauma registers have played a key role in the advancement of patient-based research and trauma care in HIC [16], [17]. Trauma registers offer a unique opportunity to document patient characteristics and audit outcomes [18], thereby creating a platform for innovative clinical research. Moreover, trauma registers in HIC have been used to study longitudinal trends in hospital mortality [9]. Such studies are important as they form the basis for further research on factors underlying changes in hospital mortality. In contrast, few trauma registers exist in LMIC [19], [20], which limit the potential for studies of longitudinal trends in hospital mortality in these countries.
Studies from HIC have stressed the importance of risk-adjustment when comparing trauma hospital mortality rates between hospitals or over time [21], [22]. Risk-adjustment is commonly done by incorporating measures of patient case-mix including injury severity or mortality risk in the analysis [23]. Several different methods for assigning mortality risk exist but the evidence for which one to use is inconclusive [24]. Recent research from HIC has used the data-driven score International Classification of Disease (ICD) based Injury Severity Score (ICISS) to adjust for mortality risk [25], but there is no consensus regarding the appropriate approach in studies from LMIC. We adopted an explorative approach to risk-adjustment using ICISS, with the aim to assess if early hospital mortality in patients with trauma has changed over time in an urban lower middle-income setting.
Methods
Study design
We conducted a retrospective study of patients presenting to Lokmanya Tilak Municipal General Hospital (LTMGH), Mumbai, India (Figure 1).
Abbreviations: DV Dependent Variable, ICD International Classification of Disease, ICISS ICD-derived Injury Severity Score, IV Independent Variables, SRR Survival Risk Ratio.
Ethics statement
The institutional ethics committee of LTMGH approved the collation of the 2011 database, reference number IEC/22/10, and approved new analyses of all three dataset in an amendment to IEC/22/10. The ethics committee granted a waiver of consent.
Setting
India is the world's second most populous country, classified as a lower middle-income country by the World Bank [26]. In 2011, it was estimated that more than 500,000 people died from traumatic injuries in India [27]. In 2015, around 200,000 Indians are expected to die from road traffic injuries (RTI) [28]. Furthermore, in India, more than 50% of mortality from RTI occur in hospitals [29]. Mumbai is the most populous city in India, with more than18 million inhabitants [30]. Lokmanya Tilak Municipal General Hospital (Figure 2) is one of the four biggest public hospitals in Mumbai.
Abbreviations: ICU Intensive Care Unit.
Data
We retrospectively analysed three datasets of patient cohorts admitted to LTMGH trauma ward during 1998, 2002, and 2011 (Table 1). All three datasets were collated for previous research. At least one of the authors (NR, VK, or SD) of this study supervised collection of all three datasets. The three datasets were merged and duplicate entries were detected and removed using the automated approach suggested in STATA's FAQ [31].
Eligibility criteria
All non-duplicate observations were considered eligible. Observations falling outside of accepted value ranges were set to missing.
Variables
Our outcome of interest was early hospital mortality, defined as death between admission and 24 hours. According to Haider et al. [23], risk adjusted outcome studies should account for sex, age, mechanism of injury, and physiological- and anatomical injury severity. We were able to include all except physiological injury severity as covariates. Age was defined as a categorical variable, with the conventionally used cut-offs <15, 15–55, and >55 years. Mechanism of injury was defined as a nominal variable, categorised as fall, railway injury (RI), road traffic injury (RTI), assault, other, and unknown.
Anatomical injury severity
We used ICISS to quantify anatomical injury severity [32]. Our rationale was that ICISS has been shown to perform well compared to the more conventional Injury Severity Score (ISS) and other established injury severity measures such as Trauma and Injury Severity Score (TRISS) [33], [34], while also being easily computed using ICD-codes. To calculate ICISS, each ICD-code is assigned a survival risk ratio (SRR). Each SRR is equal to the proportion of patients who survived with a specific ICD-code in a reference population. In studies with large samples, the study sample is often used as the reference population [35], [36]. Smaller studies may use published SRRs from a larger, similar, reference population [37]. In literature searches, we found no published SRRs from a LMIC trauma population and therefore we decided to use our own sample as the reference population. Because our sample was small, this analysis should be considered explorative and we are well aware of the same-sample bias that potentially is introduced using this approach.
MG coded all patients' injuries using ICD-10: 2010 edition, down to the fourth level. We calculated SRRs by dividing the number of survivors with a specific ICD-code with the total number of patients with the same ICD-code. This was done separately for each cohort to account for potential differences in factors such as transport time and care that might have influenced outcomes in the three time periods. We used the ICISS1 version of ICISS, which means that for overall scoring of patients' injury severity we counted only the worst injury (lowest SRR). We chose ICISS1 because this version has been showed to have a higher predictive value compared to versions that take into account all of a patient's injuries [38]. We defined major trauma as ICISS<0.9 [39].
Statistical methods
We used STATA (STATA 12, StataCorp, Texas) for statistical analyses. Where applicable, a significance level of 5% and a confidence level of 95% were used. Sample characteristics are reported for each cohort. Continuous variables are reported using their median, inter-quartile range, and range because they were found to be non-normally distributed. Categorical variables are reported as proportions.
We assessed the univariate associations between covariates and early hospital mortality using logistic regression and report their unadjusted odd-ratios (OR). We used multivariate logistic regression to evaluate the association between time and early hospital mortality. Dummy variables were created for the three cohorts and the 2002 and 2011 cohorts were included in the logistic regression model and compared early hospital mortality with the early hospital mortality in 1998. Age, sex, mechanism of injury, and ICISS were included as covariates [23]. We conducted 13 subgroup analyses and assessed early hospital mortality trends separately in patients with and without major trauma, males, females, each age category, and in patients with history of fall, railway injury, and road traffic injury.
We used multiple imputation using chained equations to handle missing data. For each variable with missing data, we assessed the association between the probability of missing data and early hospital mortality using logistic regression. We found no significant associations, i.e. P-value>0.05 in all analyses, and hence deemed multiple imputation as appropriate. Each cohort was imputed separately. We specified our imputation model to impute sex using logistic regression, and age and mechanism of injury using multinominal logistic regression. We performed 20 imputations to achieve stable estimates.
Results
We analysed data on 4189 patients (Table 2), out of which 86.5% were male. Road traffic injury was the most common mechanism of injury across cohorts, followed by railway injury and falls. Out of all patients, 81.5% were between 15 and 55 years old. Early hospital mortality among all patients was 7.9% (Figure 3A), the median ICISS was 0.93, and 36.5% had major trauma (ICISS<0.9). The early hospital mortality rate among major trauma patients was 12.2% overall and 13.4%, 11.3%, and 10.9% in 1998, 2002, and 2011 respectively.
A. Unadjusted early hospital mortality rates across cohorts. B. Odds ratios for the 2002 and 2011 cohorts when treated as indicator variables in a multivariate logistic regression analysis with early hospital mortality as the dependent variable. The model was adjusted for sex, age, mechanism of injury and ICISS. Abbreviations: ICD International Classification of Disease, ICISS ICD-derived Injury Severity Score.
Univariate analysis
Our univariate logistic regression analyses (Table 3) showed that patients over 55 years of age had significantly higher odds of early hospital mortality compared to patients younger than 15 years of age (OR = 1.81, P-value = 0.026, 95% CI = 1.07–3.05). Patients with railway injury (OR = 3.21, P-value<0.001, 95% CI = 2.26–4.57), road traffic injury (OR = 1.78, P-value = 0.002, 95% CI = 1.24–2.55), or an unknown mechanism of injury (OR = 3.04, P-value = 0.004, 95% CI = 1.42–6.51) had significantly higher odds of early hospital mortality compared to patients with fall. Every 0.01 unit increase in ICISS was significantly associated with lower odds of early hospital mortality (OR = 0.95, P-value<0.001, 95% CI = 0.94–0.96), and having major trauma was significantly associated with higher odds of early hospital mortality (OR = 2.44, P-value<0.001, 95% CI = 1.93–3.09).
Multivariate analysis
Compared to 1998, the multivariate logistic regression model showed lower adjusted odds of early hospital mortality in 2002 (OR = 0.78, P-value = 0.118, 95% CI = 0.58–1.06) and significantly lower odds in 2011 (OR = 0.56, P-value<0.001, 95% CI = 0.41–0.75), (Table 4, Figure 3B). Males did not have significantly higher odds of early hospital mortality compared to females. There was no significant difference in odds of early hospital mortality difference between either older (>55 years) or younger (15–55 years) compared to the youngest (<15 years) patients. Railway injury (OR = 2.85, P-value<0.001, 95% CI = 1.98–4.10) and road traffic injury (OR = 1.54, P-value = 0.021, 95% CI = 1.07–2.21) retained largely the same effect sizes and remained significantly associated with early hospital mortality in multivariate analysis. Assault was significantly associated with lower odds of early mortality (OR = 0.46, P-value = 0.038, 95% CI = 0.21–0.95). Also, every 0.01 unit increase in ICISS remained significantly associated with lower odds of early hospital mortality (OR = 0.95, P-value<0.001, 95% CI = 0.94–0.96). Model estimates for each cohort analysed separately are available as Table S1-S3.
Subgroup analysis
Among patients with major trauma, the odds of early hospital mortality were significantly lower in 2011 (OR = 0.44, P-value<0.001, 95% CI = 0.30–0.66) compared to 1998 (Figure 4A). Males had significantly lower odds of early hospital mortality in 2011 (OR = 0.57, P-value<0.001, 95% CI = 0.41–0.78) compared to 1998 (Figure 4B). In patients between 15 and 55 years of age the odds of early hospital mortality were significantly lower in 2011 (OR = 0.52, P-value<0.001, 95% CI = 0.37–0.74) compared to 1998 (Figure 4C). For patients with railway injury the odds were significantly lower in 2011 (OR = 0.39, P-value<0.001, 95% CI = 0.24–0.64) compared to 1998 (Figure 4D). Patients with road traffic injury had significantly lower odds in 2011 (OR = 0.60, P-value = 0.047, 95% CI = 0.36–1.00) compared to 1998. Full model estimates for all subgroups, including non-significant results, are available as Table S4-13.
Odds ratios are early hospital mortality odds ratios. A. Major trauma was defined as ICISS<0.9. The model was adjusted for sex, age, mechanism of injury, and ICISS. B. The model was adjusted for age, mechanism of injury, and ICISS. C. The model was adjusted for sex, mechanism of injury, and ICISS. D. The model was adjusted for sex, age, and ICISS. Abbreviations: ICD International Classification of Disease, ICISS ICD-derived Injury Severity Score. *P-value<0.05, compared to the reference category.
Discussion
Our study showed that early hospital mortality in trauma patients decreased over a 13-year period in an urban lower middle-income setting. This trend was not observed in unadjusted analysis and our study therefore highlights the importance of risk-adjustment when comparing mortality rates over time. However, we used an explorative approach to compute ICISS as a measure of anatomical injury severity and this should be kept in mind when interpreting our results. The observed reduction in early hospital mortality was most pronounced in patients with major trauma, while no significant changes could be detected in patients with minor trauma. Similar trends as the one shown here have been observed for other major conditions in the same geographical area. For example, the infant mortality rate and under-five mortality rate in Maharashtra dropped by 25% and 34% respectively between 1992 and 2006 [40]. Similarly, Maharashtra's maternal mortality rate dropped by 23% between 1998 and 2009 [41], [42]. It is important to note that these figures are not risk-adjusted.
The implementation of trauma systems has been widely claimed as a major reason for reduced trauma mortality over time in HIC [11], [22], [43], [44]. The trauma system concept originates from the designation of dedicated trauma centres in the US some 40 years ago [45]. Today such systems ideally include streamlined preventive, prehospital, hospital, and rehabilitative measures [8]. Interestingly, the reduction in early hospital mortality observed in our study has occurred in the absence of a comprehensive trauma system. On the hospital level, LTMGH does employ a trauma team model, and one hypothetical explanation for the observed trend might be a maturation of trauma care efforts at institutional level, with improved equipment and enforced education of residents according to advanced trauma life support principles.
A functioning prehospital system is generally considered a vital component of a modern trauma system and improvements in prehospital triage and care may influence overall mortality rates. A recent systematic review found that the implementation of prehospital systems in LMIC has led to reduced overall mortality rates [46]. However, there is no organised prehospital system in Mumbai and most trauma patients arrive to hospital in taxis or are brought by the police [47]. Ambulances are used almost exclusively for inter-hospital transfers and are largely used as transport vehicles without any resuscitation equipment [47]. Therefore, the reduced odds of early hospital mortality can most likely not be attributed to improvements in prehospital care, but rather improved hospital trauma care.
A high proportion of the patients in this study were males. This finding is consistent with other research from LMIC [48]–[50]. Interestingly, the proportion of females seems to be higher in HIC trauma than in our study population [9], [51]. This discrepancy might be partly due to more women involved in work outside of the home in HIC compared to LMIC. Lastly, we found that RTI, falls, and RI were the most common mechanisms of injury. Although RTI and falls are major causes of trauma around the world, RI appears to be rather unique for Mumbai, and thus interesting to explore further as a subset of the Mumbai trauma population.
Methodological considerations and limitations
First, the difference in inclusion criteria between the three cohorts is a major limitation. To adjust for this limitation and to minimize bias, we adjusted our analyses for case-mix and also conducted several subgroup analyses. Second, we used ICISS to adjust for anatomical injury severity and used our own data to calculate SRRs and generate finals scores. As we highlighted in the methodology section this approach might have introduced a same-sample bias that in part can explain the good fit of ICISS. Our approach should therefore be considered explorative in nature. However, we argue that this was a more sensible approach compared to using SRRs derived from a completely different trauma population. Third, we analysed available datasets and were therefore not able to choose our time-points. Unfortunately, retrospective inquires into patient files spanning more than ten years back constitute a more or less impossible undertaking as files are paper based and stored under variable conditions over time. Finally, the data did not allow us to explore outcome measures beyond mortality between admission and 24 hours. We do acknowledge that later mortality and functional outcomes are important and should be the focus of prospective studies in the same setting.
Conclusions
In our study, we observed a significant reduction in early trauma mortality among patients with major trauma between 1998 and 2011. Improved survival was evident after adjusting for patient case-mix and this finding therefore highlights the importance of risk-adjustment in studies of mortality over time. Furthermore, as one of few studies from an urban lower-middle income setting spanning more than ten years, we show that survival has improved despite the absence of a trauma system and functioning prehospital organisation.
Supporting Information
Table S1.
Multivariate logistic regression model parameters, 1998 cohort analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s001
(PDF)
Table S2.
Multivariate logistic regression model parameters, 2002 cohort analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s002
(PDF)
Table S3.
Multivariate logistic regression model parameters, 2011 cohort analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s003
(PDF)
Table S4.
Multivariate logistic regression model parameters, patients with major trauma analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s004
(PDF)
Table S5.
Multivariate logistic regression model parameters, patients with minor trauma analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s005
(PDF)
Table S6.
Multivariate logistic regression model parameters, males analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s006
(PDF)
Table S7.
Multivariate logistic regression model parameters, females analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s007
(PDF)
Table S8.
Multivariate logistic regression model parameters, age<15 years analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s008
(PDF)
Table S9.
Multivariate logistic regression model parameters, age 15–55 years analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s009
(PDF)
Table S10.
Multivariate logistic regression model parameters, age>55 years analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s010
(PDF)
Table S11.
Multivariate logistic regression model parameters, patients with fall analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s011
(PDF)
Table S12.
Multivariate logistic regression model parameters, patients with railway injury analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s012
(PDF)
Table S13.
Multivariate logistic regression model parameters, patients with road traffic injury analysed separately.
https://doi.org/10.1371/journal.pone.0090064.s013
(PDF)
Acknowledgments
We would like to acknowledge the support by the Dean and Staff and Research Society of LTMGH. We would also like to thank scientific editor Angelika Hofmann for professional proofreading, Siddarth Daniels David, and our reviewers for insightful and helpful comments.
Author Contributions
Conceived and designed the experiments: MG NR JvS. Performed the experiments: NR SD VK. Analyzed the data: MG MP. Contributed reagents/materials/analysis tools: NR SD VK MK. Wrote the paper: MG NR GT JvS. Substantial contribution to interpretation of data: NR SD VK MK GT MP LFT JvS. Revised the article critically for important intellectual content: MG NR SD VK MK GT MP LFT JvS. Gave final approval of the version to be published: MG NR SD VK MK GT MP LFT JvS.
References
- 1. Lozano R, Naghavi M, Foreman K, Lim S, Shibuya K, et al. (2013) Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380: 2095–2128.
- 2.
World Health Organization (2004) World report on road traffic injury prevention. Geneva: World Health Organization.
- 3.
Mathers C, Boerma T, Ma Fat D (2008) The global burden of disease: 2004 update. Geneva: World Health Organization.
- 4.
World Health Organization (2010) Injuries and violence: the facts. Geneva: World Health Organization.
- 5. Mock C, Joshipura M, Arreola-Risa C, Quansah R (2012) An estimate of the number of lives that could be saved through improvements in trauma care globally. World J Surg 36: 959–963.
- 6. Pfeifer R, Tarkin IS, Rocos B, Pape HC (2009) Patterns of mortality and causes of death in polytrauma patients—has anything changed? Injury 40: 907–911.
- 7. Ker K, Kiriya J, Perel P, Edwards P, Shakur H, et al. (2012) Avoidable mortality from giving tranexamic acid to bleeding trauma patients: an estimation based on WHO mortality data, a systematic literature review and data from the CRASH-2 trial. BMC Emerg Med 12: 3.
- 8. Gruen RL, Gabbe BJ, Stelfox HT, Cameron PA (2012) Indicators of the quality of trauma care and the performance of trauma systems. Br J Surg 99 Suppl 197–104.
- 9. Glance LG, Osler TM, Mukamel DB, Dick AW (2012) Outcomes of adult trauma patients admitted to trauma centers in Pennsylvania, 2000-2009. Arch Surg 147: 732–737.
- 10. Nathens AB, Jurkovich GJ, Cummings P, Rivara FP, Maier RV (2000) The effect of organized systems of trauma care on motor vehicle crash mortality. JAMA 283: 1990–1994.
- 11. Celso B, Tepas J, Langland-Orban B, Pracht E, Papa L, et al. (2006) A systematic review and meta-analysis comparing outcome of severely injured patients treated in trauma centers following the establishment of trauma systems. J Trauma 60: 371–378 discussion 378.
- 12. Rosenfeld JV, Maas AI, Bragge P, Morganti-Kossmann MC, Manley GT, et al. (2012) Early management of severe traumatic brain injury. The Lancet 380: 1088–1098.
- 13. Gruen RL, Brohi K, Schreiber M, Balogh ZJ, Pitt V, et al. (2012) Haemorrhage control in severely injured patients. The Lancet 380: 1099–1108.
- 14. Balogh ZJ, Reumann MK, Gruen RL, Mayer-Kuckuk P, Schuetz MA, et al. (2012) Advances and future directions for management of trauma patients with musculoskeletal injuries. The Lancet 380: 1109–1119.
- 15. Gruen RL, Fitzgerald MC (2012) Towards a maturation of trauma research. The Lancet 380: 1033–1034.
- 16. Cameron PA, Gabbe BJ, Cooper DJ, Walker T, Judson R, et al. (2008) A statewide system of trauma care in Victoria: effect on patient survival. Med J Aust 189: 546–550.
- 17. Rutledge R (1995) The goals, development, and use of trauma registries and trauma data sources in decision making in injury. Surg Clin North Am 75: 305–326.
- 18.
Gliklich RE, Dreyer NA (2007) Registries for Evaluating Patient Outcomes: A User's Guide. Rockville, MD: Agency for Healthcare Research and Quality.
- 19. O'Reilly GM, Cameron PA, Joshipura M (2012) Global trauma registry mapping: a scoping review. Injury 43: 1148–1153.
- 20.
O'Reilly GM, Joshipura M, Cameron PA, Gruen R (2013) Trauma registries in developing countries: A review of the published experience. Injury.
- 21. Christensen MC, Parr M, Tortella BJ, Malmgren J, Morris S, et al. (2010) Global differences in causes, management, and survival after severe trauma: the recombinant activated factor VII phase 3 trauma trial. J Trauma 69: 344–352.
- 22. Moore L, Hanley JA, Turgeon AF, Lavoie A (2010) Evaluation of the long-term trend in mortality from injury in a mature inclusive trauma system. World J Surg 34: 2069–2075.
- 23. Haider AH, Saleem T, Leow JJ, Villegas CV, Kisat M, et al. (2012) Influence of the National Trauma Data Bank on the study of trauma outcomes: is it time to set research best practices to further enhance its impact? J Am Coll Surg 214: 756–768.
- 24. Tohira H, Jacobs I, Mountain D, Gibson N, Yeo A (2012) Systematic review of predictive performance of injury severity scoring tools. Scand J Trauma Resusc Emerg Med 20: 63.
- 25. Ciesla DJ, Tepas JJ 3rd, Pracht EE, Langland-Orban B, Cha JY, et al. (2013) Fifteen-year trauma system performance analysis demonstrates optimal coverage for most severely injured patients and identifies a vulnerable population. J Am Coll Surg 216: 687–695 discussion 695–688.
- 26.
The World Bank (2013) Country and Lending Groups. Geneva: The World Bank.
- 27.
National Crime Records Bureau (2011) Accidental deaths and suicides in India. New Delhi: Ministry of Home Affairs, Government of India.
- 28. Gururaj G (2008) Road traffic deaths, injuries and disabilities in India: current scenario. Natl Med J India 21: 14–20.
- 29.
Gururaj G (2005) Injuries in India: A national perspective. Background Papers: Burden of Disease in India Equitable Development - Healthy Future. New Delhi: National Commission on Macroeconomics and Health, Ministry of Health & Family Welfare, Government of India. pp. 325–347.
- 30.
Press Information Bureau (2011) INDIA STATS : Million plus cities in India as per Census 2011. Government of India.
- 31.
Gould W (1999) How do I identify duplicate observations in my data? Texas: StataCopr LP.
- 32. Osler T, Rutledge R, Deis J, Bedrick E (1996) ICISS: an international classification of disease-9 based injury severity score. J Trauma 41: 380–386 discussion 386–388.
- 33. Rutledge R, Osler T, Emery S, Kromhout-Schiro S (1998) The end of the Injury Severity Score (ISS) and the Trauma and Injury Severity Score (TRISS): ICISS, an International Classification of Diseases, ninth revision-based prediction tool, outperforms both ISS and TRISS as predictors of trauma patient survival, hospital charges, and hospital length of stay. J Trauma 44: 41–49.
- 34. Meredith JW, Evans G, Kilgo PD, MacKenzie E, Osler T, et al. (2002) A comparison of the abilities of nine scoring algorithms in predicting mortality. J Trauma 53: 621–628 discussion 628–629.
- 35.
Blomberg H, Svennblad B, Michaelsson K, Byberg L, Johansson J, et al. (2013) Prehospital trauma life support training of ambulance caregivers and the outcomes of traffic-injury victims in sweden. J Am Coll Surg 217: 1010–1019 e1012.
- 36. Gedeborg R, Thiblin I, Byberg L, Melhus H, Lindback J, et al. (2010) Population density and mortality among individuals in motor vehicle crashes. Inj Prev 16: 302–308.
- 37. Wong SS, Leung GK (2008) Injury Severity Score (ISS) vs. ICD-derived Injury Severity Score (ICISS) in a patient population treated in a designated Hong Kong trauma centre. Mcgill J Med 11: 9–13.
- 38. Kilgo PD, Osler TM, Meredith W (2003) The worst injury predicts mortality outcome the best: rethinking the role of multiple injuries in trauma outcome scoring. J Trauma 55: 599–606 discussion 606–597.
- 39. Diggs BS, Mullins RJ, Hedges JR, Arthur M, Newgard CD (2008) Proportion of seriously injured patients admitted to hospitals in the US with a high annual injured patient volume: a metric of regionalized trauma care. J Am Coll Surg 206: 212–219.
- 40.
International Institute for Population Sciences and Macro International (2008) National Family Health Survey (NFHS-3), India, 2005-06: Maharashtra. Mumbai: International Institute for Population Sciences.
- 41.
Sample Registration System (2011) Special Bulletin on Maternal Mortality in India 2007-2009. New Delhi: Office of Registrar General, India.
- 42.
Sample Registration System (2000) SRS Bulletin. Volume 33. Issue 1. New Delhi: Office of Registrar General, India.
- 43. Gabbe BJ, Simpson PM, Sutherland AM, Wolfe R, Fitzgerald MC, et al. (2012) Improved functional outcomes for major trauma patients in a regionalized, inclusive trauma system. Ann Surg 255: 1009–1015.
- 44. Shackford SR, Hollingworth-Fridlund P, Cooper GF, Eastman AB (1986) The effect of regionalization upon the quality of trauma care as assessed by concurrent audit before and after institution of a trauma system: a preliminary report. J Trauma 26: 812–820.
- 45. West JG, Trunkey DD, Lim RC (1979) Systems of trauma care. A study of two counties. Arch Surg 114: 455–460.
- 46. Henry JA, Reingold AL (2012) Prehospital trauma systems reduce mortality in developing countries: a systematic review and meta-analysis. J Trauma Acute Care Surg 73: 261–268.
- 47. Roy N, Murlidhar V, Chowdhury R, Patil SB, Supe PA, et al. (2010) Where there are no emergency medical services-prehospital care for the injured in Mumbai, India. Prehosp Disaster Med 25: 145–151.
- 48. Marson AC, Thomson JC (2001) The influence of prehospital trauma care on motor vehicle crash mortality. J Trauma 50: 917–920 discussion 920–911.
- 49. Demyttenaere SV, Nansamba C, Nganwa A, Mutto M, Lett R, et al. (2009) Injury in Kampala, Uganda: 6 years later. Can J Surg 52: E146–150.
- 50. Ichikawa M, Chadbunchachai W, Marui E (2003) Effect of the helmet act for motorcyclists in Thailand. Accid Anal Prev 35: 183–189.
- 51. Brennan PW, Everest ER, Griggs WM, Slater A, Carter L, et al. (2002) Risk of death among cases attending South Australian major trauma services after severe trauma: the first 4 years of operation of a state trauma system. J Trauma 53: 333–339.