Trend and Geographic Analysis for Traumatic Brain Injury Mortality and Cost Based on MarketScan Database

Abstract

The objective of the current research was to examine the current epidemiology of traumatic brain injury (TBI); to determine the effects of geographic region, co-morbidities, year of injury, injury severity, and demographics on hospital costs, length of stay (LOS), and mortality. All subjects were drawn from the Thomason Reuters MarketScan^® database. Statistical methods used included descriptive analysis, bivariate analysis, logistic regression, and the Geographic Information System (GIS) software, ArcMap. We studied 76,313 patients with TBI from 2004 to 2009 (52,721 with commercial insurance and 23,592 with Medicare) from the MarketScan database. As age increased, mortality rate and median LOS increased. The median hospital costs for adults were the highest ($13,000 for ages 18–64) compared with children ($8000 for age 0–14) and elderly persons ($9000 for age ≥65). The mortality rate for the elderly population has decreased slightly (11.1% in 2004 to 9.9% in 2009 for men, and 7.0% to 6.9% for women); however, their hospital costs have increased significantly ($6899 in 2004 to $11,567 in 2009 for men; $6784 to $9782 for women). Concerning the impact of geography, the western United States (e.g., Washington and California) had lower mortality rates and higher median costs while the southeast United States had the highest mortality and mixed median costs. Both overall mortality and median LOS have remained relatively stable over the years. Hospital cost, however, has increased for the elderly population even after accounting for the inflation. There is significant geographic variation for both mortality and hospital costs.

Introduction

T raumatic brain injury (TBI) is a serious public health problem in the United States. Each year, approximately 1.7 million persons receive a diagnosis of TBI in the United States.¹ The cognitive and emotional impairments from TBI have profound effects on a person's career, aspirations, as well as their social and family lives long after the initial injury.² Currently, in the United States, approximately 5.3 million Americans live with a disability that resulted from TBI. In addition, the cost of treatment, rehabilitation, and care totals approximately 30 billion US dollars each year.¹

Past studies have shown that the mortality rate of persons with TBI is the highest in the young and elderly population. In addition, males are usually twice as likely to sustain a TBI as females³; however, in ages 0 to 15, females are more likely to die.⁴ Age and severity have been shown to have a powerful effect on hospital costs.⁵ Little has been done concerning regional impact on TBI mortality and cost. In addition, some studies have only focused on the pediatric population,⁶ others on adults,⁷ and a few on the elderly population.⁸ The Thomason Reuters MarketScan^® Research Database, a fully integrated patient level data source of commercial, Medicare, and Medicaid claims, allows investigation of outcome variables from children to elders to update the current epidemiology of TBI.

This article aims to determine the effects of demographics, comorbidity, injury severity, calendar year, and geographic region on the outcome variables related to TBI: mortality rate, length of stay (LOS), and hospital costs. The current study not only updates the general pattern of the outcome variables over different age and sex groups, but also reveals the trend of the outcome variables over time (2004–2009). In addition, the impact of geographic regions on the outcome variables was specifically examined. The results of this study may provide some insight for policy makers.

Methods

All subjects were drawn from the Thomason Reuters MarketScan Research Database. The MarketScan data contains records from employers, health plans, and government and public organizations. The databases contain person-specific clinic use and cost as well as enrollment information with more than 170 million unique patients since 1995. Geographic data were not available for Medicaid claims, so the data set used for this study was limited to Medicare and commercial claims from the years 2004 to 2009. Subjects were selected only if their first diagnosis was an International Classification of Diseases (ICD)-9 Code for TBI. These were 800.xx through 804.xx, 850.xx through 854.xx, and 959.xx.

Outcome variables and predictors

The primary outcomes of interest in this study included mortality, cost, and LOS. Mortality rate was defined as the in-hospital mortality rate; cost as the calculated cost of all the expenses to the hospital of the primary stay adjusted by the medical component of the inflation to US dollars in the year 2009⁹; and LOS as that of the primary stay in days.

The dependent variables include the ICD-9 Injury Severity Score (ICISS), the Charlson Co-morbidity Index (CCI), sex, age, and the calendar year of diagnosis. The ICISS is a survival score based on the ICD-9 classification of trauma injuries, and is defined as “the product of all survival risk ratios for an individual patient's traumatic ICD-9 codes.”¹⁰ The ICISS ranges from 0 to 1 with 1 being the least severe.¹⁰ The ICISS was further validated as being superior to the Injury Severity Score (ISS) and the Trauma and Injury Severity Score (TRISS) in predictive capabilities.¹¹ Since then, the ICISS has been extensively used in trauma research studies. The CCI is meant to predict the long-term prognosis for patients and is dependent on the range of co-morbid conditions such as cancer, heart failure, or head trauma. Different pre-existing conditions are assigned a score with 0 being no conditions and 6 being a malignant tumor or acquired immunodeficiency syndrome. A patient's CCI is the sum of the scores for each separate co-morbid score.¹² CCI was classified as 0, 1, 2, and ≥3 in this study.

Statistical methods

The patient characteristics were summarized by median and interquartile ranges (IQR) if the variables were continuous. Otherwise, they were summarized as incidences and percentage of incidences (Table 1). A chi-square test was applied to test whether mortality rates are the same for females versus males, commercial insurance versus Medicare, different levels of CCI (0–3), and different years (2004–2009). The p values based on the chi-square test and the adjusted p values based on the Bonferroni method were reported (Table 2). Significance was defined as an adjusted p value<0.05.

Table 1.

Characteristics of the Outcome Variables and Predictors for All Patients with Traumatic Brain Injury

Variables	Median (IQR) OR occurrence (percentage)
Number of records	76313
Age	49 (20, 72)
LOS	3 (1, 6)
Sex
Male (sex=1)	46284 (60.7%)
Female (sex=2)	30029 (39.3%)
Insurance
Commercial	52721 (69.1%)
Medicare	23592 (30.9%)
ICISS	0.920 (0.806, 0.967)
In-hospital death
Yes	4155 (5.44%)
No	72158 (94.56%)
Hospital costs in 2009 dollars	10694 (5561, 24606)
Charlson Comorbidity Index (CCI)
0	45163 (59.2%)
1	16819 (22.0%)
2	7903 (10.4%)
3	6428 (8.4%)

IQR, interquartile range; OR, odds ratio; LOS, length of stay; ICISS, ICD-9 Injury Severity Score.

Table 2.

Bivariate Analysis for the Variables Associated with Mortality (n=76313)

	Mortality
Variables	Died	Survived	p value	Adjusted p value
Overall	4155 (5.44%)	72158 (94.56%)
Sex
Male (sex=1)	2700 (5.83%)	43584 (94.17%)	<0.0001	<0.001
Female (sex=2)	1455 (4.85%)	28574 (95.15%)
Insurance
Commercial	2207 (4.19%)	50514 (95.81%)	<0.0001	<0.001
Medicare	1948 (8.23%)	21644 (91.77%)
Charlson Co-morbidity Index (CCI)
0	1804 (3.99%)	43359 (96.01%)	<0.0001	<0.001
1	1100 (6.54%)	15719 (93.46%)
2	604 (7.64%)	7299 (92.36%)
3	647 (10.07%)	5781 (89.93%)
Year
2004	511 (5.73%)	8414 (94.27%)	0.0458	0.1832
2005	592 (5.97%)	9326 (94.03%)
2006	544 (5.07%)	10180 (94.93%)
2007	619 (5.18%)	11342 (94.82%)
2008	922 (5.40%)	16147 (94.60%)
2009	967 (5.46%)	16749 (94.54%)

A simple plot of hospital cost versus LOS for all patients was implemented to examine the relationship between hospital cost and LOS. The incidence and mortality rate for different age groups were plotted for females and males separately so that the disparity in sex and age could be examined (Panels A1, A2 in Fig. 1). The boxplots of the median hospital costs and LOS for different age groups were also plotted (Panels A3 and A4 in Fig. 1). The trend of the outcome variables for females and males among children (0–17), adult (18–64), and elderly (≥65) were plotted over the years 2004–2009 (Fig. 2).

FIG. 1.

Incidence and mortality classified by age and sex, length of stay, and hospital costs by age.

FIG. 2.

Trends in mortality, median length of stay, and median costs classified by age and sex.

The joint effect of multiple factors on mortality rate was examined by using a multiple logistic regression, where the predictors included age, CCI, sex, and ICISS. The Akaike Information Criterion (AIC) was used to select a parsimonious model that has a tradeoff between goodness-of-fit and a smaller number of predictors.¹³

To examine the geographic effect on the mortality and hospital cost, the GIS software ArcMap was applied.¹⁴ Mortality rate and median cost for the primary stay was summarized by state. Those states with fewer than 25 incidences were exempt from the map, because the inferences based on the small sample size may not be reliable. The maps were coded by the 10th percentile, 50th percentile, and 90th percentile of the outcome variables (Fig. 3).

FIG. 3.

(A) Mortality rates by state for patients with traumatic brain injury aged 18–44 years. (B) Median costs by state for patients with traumatic brain injury aged 18–44 years.

Results

Patient characteristics

A total of 76,313 patients with TBI were included in this analysis. Of those, 52,721 subjects had commercial insurance and the remaining 23,592 had Medicare. Among the 76,313 patients with TBI, 60.7% were males and 59.2% patients had a CCI of 0. The median and IQR were 49 and (20, 72) years old for age at injury, 3 and (1, 6) days for LOS, and $10,694 (5,561, 24,606) for hospital costs (Table 1).

The overall mortality was 5.44% with a slightly higher rate in males (5.83%) than females (4.85%). Those who were commercially insured (4.19%) had lower mortality rates compared with Medicare subjects (8.23%); however, this is confounded by age because Medicare patients are older than 65 years. As the CCI increased, so did the mortality rate. The mortality rates across the 6 years were not significantly different (Table 2).

Sex-age effect and trends on outcome variables

The total number of female incidents was generally lower than male incidents with the exception of those older than 75 years (Fig. 1 Panel A1). Based on Figure 1 Panel A2, during the childhood ages of 0–14 years, females had a slightly higher mortality rate; during the adolescent ages of 15–19 years, there is a steep increase in mortality rate for males and females; and during the middle ages of 19–44 years, the overall mortality rate remains relatively stable. With ages older than 45 years, there is a significant sex-disparity in mortality rate. Based on Figure 1 Panels A3 and A4, during the ages 19–44, the LOS and hospital costs remained relatively the same—approximately 2 days and $13,000, respectively. The LOS displays an upward trend as ages increase; however, the median hospital costs for adults are the highest ($13,000 for ages 18–64 years) compared with children ($8000 for ages 0–14 years) and the elderly ($9000 for ages ≥65 years). The hospital cost was adjusted by the medical component of inflation to 2009 US dollars.

The trend plot in Figure 2 demonstrates that pediatric patients had the lowest mortality rates, hospital costs, and the shortest LOS among the three age groups, and there was no obvious trend over the years 2004–2009. Adults have the highest hospital costs among the three age groups, although the trend was not significant except for the latter years of 2008 to 2009. The elderly have the highest mortality rate and LOS among the three age groups. The mortality rate for elderly persons decreased slightly over the years 2004–2009 (11.1% in 2004 to 9.9% in 2009 for elderly men; 7.0% in 2004 to 6.9% in 2009 for elderly women), while the cost for both elderly women and men increased significantly over these years ($6899 in 2004 to $11,567 in 2009 for elderly men; $6784 in 2004 to $9782 in 2009 for elderly women). This increasing trend was not observed for children and adults.

Predictors for mortality

To examine the joint impact of the variables age, CCI, sex, and ICISS on mortality, a logistic regression was performed, and the odds ratio, 95% confidence interval (CI), and p values are reported in Table 3 under the column title “Full Model.” The findings for the effect of age and sex based on the logistic regression model are consistent with what was found from the plots in Figures 1 and 2. From Table 3, age, CCI, sex, and ICISS all have p values much smaller than 0.05, indicating that these variables are significantly associated with mortality. The odds ratio for young persons versus adults is 0.54 with the 95% CI (0.47, 0.62), indicating that young persons have lower mortality than adults; and the odds ratio for elders versus adults is 2.66 with the 95% CI (2.44, 2.90), indicating that elders have higher mortality than adults. The odds ratio increases as CCI increases from 1 to ≥3, indicating that mortality increases as CCI increases. The odds ratio for 0.1 unit increase of ICISS is 0.597 with 95% CI (0.591, 0.604), indicating an inverse association between ICISS and mortality rate.

Table 3.

Results Based on Logistic Regression for All Patients (n=76313)

	Full model			Model excluding ICISS		Model excluding CCI
	Odds ratio	95% CI	p value	Odds ratio	p value	Odds ratio	p value
Age
Young vs. adult	0.54	(0.47, 0.62)	<2e–16	0.43	<2e–16	0.49	<2e–16
Elder vs. adult	2.66	(2.44, 2.90)	<2e–16	1.53	<2e–16	3.40	<2e–16
CCI
1 vs. 0	1.37	(1.25, 1.50)	<3.9e–11	1.36	1.41e–13
2 vs. 0	1.68	(1.50, 1.89)	<2e–16	1.46	3.59e–13
3 vs. 0	2.89	(2.56, 3.25)	<2e–16	1.87	<2e–16
Sex (female vs. male)	0.82	(0.76, 0.88)	<3.1e–07	0.71	<2e–16		5.42e–08
ICISS (0.1 unit increase)	0.597	(0.591, 0.604)	<2e–16			0.81	<2e–16
AIC		22293			31263		22587

ICISS, ICD–9 Injury Severity Score; CCI, Charlson Co–Morbidity Index; AIC, Akaike Information Criterion.

Because both ICISS and CCI describe the patients overall severity of disease and comorbidity, additional two logistic regressions were performed to determine which one is a stronger predictor for mortality: one excluding ICISS and one excluding CCI. The result for the model excluding ICISS was reported under the column “Model excluding ICISS” in Table 3, while the result for the model excluding CCI was reported under the column “Model excluding CCI” in the same table. The model excluding CCI has a much smaller AIC value than the model excluding ICISS; that is, the goodness-of-fit of the model excluding CCI is much better than the model excluding ICISS. Therefore, excluding CCI is less important than excluding ICISS, indicating that ICISS is a stronger predictor of mortality than CCI.

Geographic effect on mortality rate and cost

Figure 1 and an additional logistic regression analysis (not shown) revealed that the patients in the 19–44 age group had a similar mortality rate and cost throughout the age range for males and females. To reduce the confounding effect from age and sex, only the population in the age range 19 through 44 was included in the evaluation of the geographic impact on mortality and cost. In addition, to make the inference robust, Puerto Rico, Hawaii, Alaska, Wyoming, North Dakota, Vermont, and Rhode Island were excluded because each had less than 25 incidences. The mortality rate and the median cost by state are shown in Figure 3. The geographic areas with the higher mortality rates tended to be in the southeastern and the midwest regions of the United States. The four states with the highest mortality rates were Idaho, Kansas, Mississippi, and North Carolina. The four states with the lowest mortality rates were Oregon, Alabama, Arkansas, and Connecticut. Two states in the northeast area had a high mortality rate—New York and Maine—while two states in the western United States had a high mortality rate—Idaho and Montana. The southwest region was the only region with all states under the 50th percentile of mortality rates.

The map of median costs per state is the near inverse of that of mortality rates. While the states with the lower mortality rates lied in the West, the states with the greater median costs were also concentrated in the West. The states with the four lowest median costs were Arkansas, New York, Maryland, and New Jersey. Those with the greatest median costs were Washington, California, Idaho, and Wisconsin.

Arkansas was notable because it is in the bottom 10th percentile for cost and also for mortality rates. Many states in the southeast United States have a high mortality rate and high cost, while some others have a high mortality rate and low cost. New York is in the bottom 10th percentile for cost but in the top 50th percentile in mortality rate. Idaho is in the top 10th percentile for both mortality rate and cost. States with high mortality rates and high costs include Maine, Virginia, West Virginia, South Carolina, North Carolina, Florida, Tennessee, Indiana, Mississippi, Texas, Nebraska, and Montana.

To examine whether the mortality and cost among different states could be caused by types of injuries, we classified the TBIs into three types according to the codes: (1) codes 800.xx–804.xx relate to skull fractures, (2) codes 850.xx–854.xx to intracranial injury, and (3) code 959.xx to other causes of brain injury. The mortality rate is 6.2% for skull fractures, 5.4% for intracranial injury, and 0.3% for other causes of brain injury. The percentages of occurrences of skull fractures, intracranial injury, and other causes of brain injury are 30%, 66%, and 4%, respectively. We examine the occurrence, mortality rate, and median cost for each type of injury at state level. We found that the percentages of occurrences of the three types of injuries in different states are similar. In almost all states, the median cost for skull fracture is higher than the cost by intracranial injury, and both are higher than the median cost for other causes of brain injury. Because of the similarity of the percentage of occurrences of different types of injuries among the different states, however, neither the differences of mortality nor differences of cost among different states could be explained by the types of injuries.

Discussion and Future Research

In this study, the following conclusions were made: (1) males had an overall higher mortality rate than females (Table 2); however, this difference was not observed in children but slight in adults and dramatically in elders. The elderly women had a much lower mortality rate than that for the elderly men (age ≥65) (Fig. 1); (2) the mortality rate increases as the CCI increases (3.99% for CCI being 0 vs. 10.1% for CCI being 3) (Table 2); (3) the hospital cost for the adults was higher than for both children and elders, and the hospital cost for children and adults did not have a significant trend for 2008 to 2009; however, the hospital cost for elderly persons, both female and male, had increased significantly over these years (Fig. 2); (4) the LOS increased as age increased; however, the median LOS for female children (age ≤17) was 1 day more than male children, while the median LOS for female adults and elders was 1 day less than males (Fig. 2); (5) the geographic variation on mortality and hospital cost was significant; the western region (e.g., Washington and California) had the greatest low mortality rates and also had almost the highest median costs, while in the southeastern region, almost all states had high mortality rates, but half had high median cost, and half had low median costs.

This study found that the age group with the highest mortality rate was the group of older than 65 years, which agreed with past studies.¹ In this study, however, mortality was lowest in the youngest age group. A previous study has shown that the mortality rate for children aged 15–19 was a peak²; however, the current study showed a decreased mortality rate for this group. This may be in part because of the Centers for Disease Control heads-up and sports concussion training programs (http://www.cdc.gov/concussion/HeadsUp/youth.html). All this may suggest that mortality from TBI has improved among young children due to public safety campaigns.

In examining costs of hospitalization for TBI, hospital costs generally increased as LOS increased. There is a high variation in hospital costs for the same LOS, however. For patients that only stay 1, 2, or 3 days, the cost ranges from a few hundred to several millions of US dollars. One study shows that LOS is related to clinical factors (e.g., ISS) as well as nonclinical factors (e.g., insurance status and discharge destination).¹⁴ The large range of hospital cost for patients who stay just 1 day, however, might point out a problem in the cost-efficiency of some hospitals.

It is worthwhile to note that the mortality rate for elderly men slightly decreased over the years 2004–2009, while hospital cost increased significantly over the years for both elderly women and men. The cost of the health care for TBI has increased significantly among the elderly population, although not for children and young adults. It is important to note the large difference in mortality rates between males and females. This impact of sex on mortality rate has been previously suggested to be related to a progesterone effect.¹⁶ The finding of the sex-disparity on mortality rate is consistent with that by Berry and colleagues,¹⁷ yet contrary to some other studies.^18
–20 A prospective study may be needed for definite conclusions. Females have a lower median cost and LOS than males each year, and the effect of sex on LOS and hospital cost has not been addressed in previous studies.

With regard to geographic variations, the western region, with the lowest mortality rates, also had higher median costs. While in the southeast region, almost all states had high mortality rates; half had high median cost, and half had low median costs. Future studies could examine whether high mortality rates correlate with high costs for the metropolitan statistical areas, where a metropolitan area is a large population nucleus with adjacent communities having a high degree of social and economic integration with that core. Thus, one may examine whether TBI mortality is related to a region's socioeconomic status.²¹

Although MarketScan data can provide valuable information, there are several limitations to this study. First, MarketScan data are administrative data and may be subject to potential errors in coding of information. Second, the selection of subjects was based on ICD-9 codes for TBI, where there is no differentiation possible as to injury severity; thus this study cannot differentiate the outcome variables according to the severity of injuries. Third, the Glasgow Coma Score (GCS) is an important variable for monitoring change in the level of consciousness of brain injured patients²²; however, GCS scores were not reported for patients with TBI in this study. The inclusion of ICISS may play a similar role as including the GCS. Fourth, this study did not have information on longer-term outcomes such as neurocognitive deficits, functional capabilities, or discharge disposition, which limit TBI cost estimates, and description of longer-term outcomes.

Fifth, race was not included in this database; therefore, our analyses were not adjusted for race. Sixth, the geographic variations were examined at the state level not by metropolitan areas, limiting the examination of impact of socioeconomic status. Seventh, the mortality rate is a hard end-point, but LOS and hospital costs are relatively soft: patients dying early will be the cheapest and have the shortest LOS; however, the current study did not examine the LOS and hospital costs between the survived patients with TBI and the deceased patients.

Conclusion

Using large nationwide claim data available in the MarketScan database, this study shows that the overall mortality and median LOS have remained relatively stable over the years for each age and sex group. Although the hospital cost for children and adults has remained relatively stable over the years, hospital cost has increased for the elderly population even after accounting for inflation, indicating increasing cost with Medicare but not with commercial insurance. Significant geographic variations for both mortality and hospital costs were also found. The western United States generally had lower mortality rates and higher median costs while the southeast United States had the highest mortality rate and mixed median costs. The current study updated the current epidemiology of TBI, and the results of this study may provide some insight for policy makers.

Footnotes

Acknowledgment

The content of this article does not represent the views of the Department of Veterans Affairs or the United States Government.

Author Disclosure Statement

No competing financial interests exist.

References

Faul

, Likang

, Wald

M.M.

, Coronado

V.G.

2010. Traumatic Brain Injury in the USRetrievedFebraury 13, 2010. http://www.cdc.gov/features/dsTBI_BrainInjury/. 2010 February 13.

Thurman

D.J.

, Alverson

, Dunn

K.A.

, Guerrero

, Sniezek

J.E.

1999. Traumatic brain injury in the United States: a public health perspective. J. Head Trauma Rehabil., 14:602–615.

Bruns

J.J.

, Hauser

A.W.

2003. The epidemiology of traumatic brain injury: a review. Epilepsia, 10,Suppl. 10:2–10.

Morrison

W.E.

, Arbalaez

J.J.

, Fackler

J.C.

, De Maio

, Paidas

C.N.

2004. Gender and age effects on outcome after pediatric traumatic brain injury. Pediatr. Crit. Care Med., 5:145–151.

High

, Hall

, Rosenthal

, Mann

, Zafonte

1996. Factors affecting hospital length of stay and charges following traumatic brain injury. J. Head Trauma Rehabil., 11:85–96.

Schneier

A.J.

, Shields

B.J.

, Hostetler

S.G.

, Xiang

, Smith

G.A.

2006. Incidence of pediatric traumatic brain injury and associated hospital resource utilization in the United States. Pediatrics, 118:483–492.

Maas

A.I.

, Stocchetti

, Bullock

2008. Moderate and severe traumatic brain injury in adults. Lancet Neurol., 7:728–741.

Thompson

H.J.

, McCormick

W.C.

, Kagan

S.H.

2006. Traumatic brain injury in older adults: epidemiology, outcomes, and future implications. J. Am. Geriatr. Soc., 54:1590–1595.

United States Department of Labor: CPI inflation calculator. http://www.bls.gov/data/inflation_calculator.htm. 2012 March 5.

10.

Osler

, Rutledge

, Deis

, Bedrick

1996. ICISS: an international classification of disease-9 based injury severity score. J. Trauma, 41:380–388.

11.

Rutledge

, Osler

, Emery

, Kromhout-Schiro

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.

12.

Charlson

, Pompei

, Ales

, MacKenzie

. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis., 40:373–383.

13.

Allison

P.D.

1999. Logistic Regression Using the SAS System: Theory and Application. SAS Institute Inc: Cary, NC.

14.

Kurland

K.S.

, Gorr

W.L.

2006. GIS Tutorial for Health, 3rd. ESRI Press: Redlands, CA.

15.

Brasel

K.J.

, Lim

H.J.

, Nirula

, Weigelt

J.A.

2007. Length of stay: an appropriate quality measure? Arch. Surg., 142:461–466.

16.

, Huang

, Qin

, You

2012. Progesterone for acute traumatic brain injury. Cochrane Database Syst. Rev., 10:CD008409.

17.

Berry

, Ley

E.J.

, Tillou

, Cryer

, Margulies

D.R

, Salim

2009. The effect of gender on patients with moderate to severe head injuries. J. Trauma, 67:950–953.

18.

Ottochian

, Salim

, Berry

, Chan

L.S.

, Wilson

M.T.

, Margulies

D.R.

2009. Severe traumatic brain injury: is there a gender difference in mortality? Am. J. Surg., 197:155–158.

19.

Leitgeb

, Mauritz

, Brazinova

, Janciak

, Majdan

, Wilbacher

, Rusnak

2011. Effects of gender on outcomes after traumatic brain injury. J. Trauma, 71:1620–1626.

20.

Kraus

J.F.

, Peek-Asa

, McArthur

2000. The independent effect of gender on outcomes following traumatic brain injury: a preliminary investigation. Neurosurg. Focus, 8:e5.

21.

Colantonio

, Moldofsky

, Escobar

, Vernich

, Chipman

, McLellan

2011. Using geographical information systems mapping to identify areas presenting high risk for traumatic brain injury. Emerg. Themes Epidemiol., 8:7.

22.

Gabbe

B.J.

, Cameron

P.A.

, Finch

C.F.

2003. The status of the Glasgow Coma Scale. Emerg. Med.(Fremantle)15:353–360.