Is Pre-Injury Socioeconomic Status Associated With Outcomes in Patients With Traumatic Brain Injury? A Systematic Review

Abstract

While socioeconomic status (SES) is associated with a variety of health outcomes, the literature on the association between SES and traumatic brain injury (TBI) outcomes has not been formally summarized. This study aims to review existing literature to ascertain whether patients with low SES pre-injury have worse clinical outcomes after TBI compared with those with high SES, in high-income countries. A systematic search was conducted using the MEDLINE, Embase, and PsychINFO databases. Observational studies addressing the association between SES and TBI outcomes (mortality, functional, cognitive, and vocational outcomes) were included (published from 2000, written in English). Both pediatric and adult TBI groups were included. Thirty-two studies met the inclusion criteria. Measures of SES varied across studies. Mortality was assessed in seven studies; five reported an association between low SES and higher mortality. Five of eight studies showed an association between low SES and worse functional outcomes; results for cognitive (n = 13) and vocational outcomes (n = 10) were mixed. The results of this review suggest that SES is a variable of interest in the context of TBI outcomes and should be assessed at time of admission to assist in social work discharge planning and early mobilization of available community resources. Further work is required to better understand the impact of SES on TBI outcomes.

Introduction

Traumatic brain injury (TBI) is a leading cause of morbidity and mortality worldwide, with an estimated annual incidence of 69 millions globally.¹ Multiple factors can influence long-term outcomes after TBI, including injury-related factors (such as injury severity and additional systemic injuries) as well as demographic and contextual factors. The latter group contains several variables including age, comorbidities, and socioeconomic status (SES).

The SES can be defined as the combined economic and social status of an individual or a group unit such as a household. This is commonly measured in terms of education, income, and occupation, but additional metrics such as place of residence can also be used as proxy measures.^2,3 The SES has a bidirectional association with health, with evidence suggesting that SES influences one's ability to access healthcare resources and treatment, and develop and maintain healthy habits. Reciprocally, health can also influence SES in that health status has an impact on a person's ability to attend and succeed in schooling or work.²

Evidence has shown that SES mediates one's ability to access relevant health services—for example, because of services being inaccessible both geographically (reduced ability to travel) and financially (unaffordable in private healthcare settings).^4,5 Acceptability of health services to patients from different SES brackets also varies with evidence that lower SES is associated with reduced engagement with health screening and preventative health initiatives, potentially because of reduced awareness and health beliefs.⁶ All of these factors can influence a patient's ability to recover after TBI.

SES and TBI

Low SES has been identified as a risk factor for increased incidence of trauma and injuries, including TBI.^7,8 United Kingdom (UK)-wide data from the Trauma Audit and Research Network (TARN) showed that low SES was associated with increased 30-day mortality after minor but not major trauma.⁹ A study based in Scotland investigating mortality after TBI concluded that there is no association between lower SES and increased mortality, suggesting equal access to medical treatment, or that unequal access does not influence survival outcomes in this situation.¹⁰

On the contrary, a study based in the United States used lack of medical insurance as a proxy for assessing SES, showing that this is associated with mortality and length of hospital stay after TBI.¹¹ Additional studies have demonstrated links between pre-injury SES and outcomes other than mortality, such as cognitive outcomes, with mixed results.^12
–14

The SES is likely to affect different groups of patients differently depending on their wider context, healthcare systems, and other factors. Therefore, whether SES is an independent factor impacting outcomes after TBI is an important question to answer to provide better future strategies to protect patients from health inequalities or reduce health inequalities themselves. There is currently no systematic review investigating the relationship between SES and clinical outcomes after TBI.

Therefore, the purpose of this review is to synthesize the existing evidence around the association between SES and outcomes after TBI.

Methods

A systematic review of the literature was conducted using the following databases: MEDLINE, EMBASE, and PsychINFO via the Ovid platform. The PRISMA guidelines were followed when reporting the results.¹⁵

Original research articles meeting the following criteria were included:

Patients of all ages suffering from TBI, diagnosed clinically or radiographically.

Report of socioeconomic status of the patients with a definition of the measurement used (such as a deprivation index or other proxy measure but excluding education of the patient alone as this is often a precursor to downstream SES metrics).

Assessing outcomes after TBI, including mortality and clinical outcomes (e.g., functional neurological status (GOSE [Glasgow Outcome Scale-Extended]), cognitive or vocational outcomes).

Publications in the English language with full text available.

Observational research study design.

Reports from high-income countries (HIC) only, defined by HIC status set out by the World Bank, based on gross national income (GNI) per capita.¹⁶ To assess whether a country was a HIC at the time of the study, the GNI per capita threshold of the year of publication of the article was used. This was to limit the role of confounding factors such as different health system resources that could impact the results.

Published from 2000 onward. This was chosen to account for the developments in management of TBI within the last 20–30 years.

The final search was performed on 16th March 2021. The full list of search terms used, and number of articles screened is available in Appendix 1 of the Supplementary Materials.

Screening process

A two-stage review process was adopted with two-reviewer (SV, MS) abstract screening followed by a single-reviewer (SV) full-text review. Data extraction was performed using a custom-made data extraction table (Appendix 2 of the Supplementary Material). The PRISMA Flow Diagram, 2020 version, describes this process.¹⁷

Article exclusion criteria were grouped as follows:

Reason 1. Article is a published conference abstract only.

Reason 2. Article is a narrative or systematic review article, not original research.

Reason 3. Article investigates the wrong patient population. For example, if it includes all types of acquired brain injury including strokes without differentiating outcome data into a clear TBI subgroup.

Reason 4. Article assesses the wrong outcome measure. For example, investigates the incidence of TBI rather than outcomes following TBI.

Reason 5. Article does not use a defined exposure metric for socioeconomic status.

Reason 6. Article is a technical article describing a scoring algorithm without focusing on the relevant outcomes.

Reason 7. Article compares across multiple countries that are both HICs and low- and middle-income countries (LMICs).

Methodological quality assessment and level of evidence

The National Institutes of Health (NIH) Study Quality Assessment Tool for Observational Cohort and Cross-sectional studies, a validated assessment tool for observational studies, was used to evaluate study quality.¹⁸ The GRADE system was used to assess the level of evidence pertaining to the key outcomes of interest.^19,20

TBI outcomes

Because several outcomes are of interest after TBI, the following four outcome groups were investigated separately: mortality, functional, cognitive, and vocational outcomes. The rationale for these groups is described more in details in the results. The results of this review are reported as a narrative synthesis.

Results

There were 1576 studies identified by the pre-defined search strategy, and 32 studies met the inclusion criteria (Fig. 1). Most studies reported data from an individual country, although the study by Arango-Lasprilla and associates²¹ included participants recruited from 18 European countries as part of the CENTER-TBI project. The United States was the most represented country in 15 studies, followed by Australia (four studies), Norway (three studies), and The Netherlands, the United Kingdom, Canada, Israel represented in two studies each. Fourteen studies were single-center studies and 18 were multi-center studies.

FIG. 1.

Preferred Reporting Items for Systematic reviews and Meta-Analyses ( PRISMA) diagram for this systematic review.

Twenty studies collected data prospectively, and the remainder were retrospective cohort studies. Follow-up time varied greatly between studies with some following patients only through acute hospitalization (particularly those focused on mortality), and others with follow-up intervals ranging from 3–6 months to 40 years.

The Glasgow Coma Scale (GCS) was used to assess TBI severity in all studies. Nineteen (60%) studies included participants with any TBI severity, five studies included patients with mild TBI only,^22

–26 five studies included patients with moderate or severe TBI,^27

–31 two studies included severe TBI alone,^32,33 and one study included patients with complicated mild (GCS 13–15 with CT head imaging findings) to severe TBI.³⁴

Sample size varied widely between studies, ranging from 34 TBI participants to 187,354 individuals.^35,36 Overall, the median population size was 163.5 (interquartile range 87 to 1724.5). Nineteen studies included adult patients alone, nine included only pediatric patients, and four studies included both groups.

Measuring SES

The most common measure of SES used in the included studies was pre-injury employment or occupation, noted in 12 of the 32 studies.^{25

–28,30,31,33,37

–41} Eight studies investigating pediatric TBI populations used parental occupation as their measure of SES^{13,22,23,35,42

–45} and one used the highest education level of the primary caregiver as measure of SES.³⁴ Four studies, all based in the United States, used medical insurance as a proxy measure for SES.^11,36,46,47

Income was used by a small number of studies: Nuno and colleagues⁴⁸ used the median family income in the zip code of residence, classified into income brackets.⁴⁸ Heffernan and coworkers⁴⁶ also used median income in the zip code based on US census data as a continuous scale by $1000 increments. Huang and associates³⁸ used family income status, defined as low, middle, or high-income family, as one of their SES measures, alongside employment.

The SES can also be measured through indices or composite scores incorporating multiple variables. Such indices were used in six articles included in this review and include the Carstairs and Morris Index of Deprivation, English Indices of Multiple Deprivation (IMD), New Zealand Deprivation Index 2006, and the Socioeconomic Composite Index. Two studies created their own context-specific composite scores.^29,32

TBI outcomes based on SES

The included articles reported a range of outcome measures. To allow for comparison across studies and pooling results of studies addressing similar outcomes together, outcome measures were grouped into one of four categories: mortality, functional outcomes, cognitive outcomes, and vocational outcomes. Specifically, the functional outcomes category included several measures of pertaining to global functioning, such as assessment of overall health and disability, independence with activities of daily living, GOSE, and self-reported measures of persisting symptoms and quality of life. The cognitive category included measures of cognition and neuropsychiatric symptoms, and vocational outcomes pertained to return to work.

From a methodological perspective, it is important to note that all studies used multi-variable analysis when assessing the association between variables and their outcomes.

Mortality

Seven of the 32 studies investigated mortality as one of their primary outcomes (Table 1), all of which had populations of more than 1000 individuals per study. All but one study concluded that lower SES was associated with a statistically significant increase in mortality compared with higher SES. Five studies investigated in-hospital mortality, of which three showed a significant association with SES, and one showed a significant association between SES and mortality for severe TBI only. Dunn and colleagues⁴⁹ was the only study showing no difference in the odds of in-hospital mortality based on baseline SES: in this study, more deprived patients had higher survival rates than less deprived individuals (81.2% vs. 78%, p = 0.046) in univariate analysis, but this was not significant after multi-variable analysis, potentially suggesting confounding.

Table 1.

Summary of Studies With Mortality as Their Outcome

Study details	SES measures	Outcome	Results	Interpretation
Dunn et al. 2003⁴⁹ United Kingdom Participants: Adults Mean age 44.5 (SD not reported) N = 2932 Follow-up: not defined Design: Prospective cohort	Domicile post-code (deprivation category using the Carstairs and Morris Index of deprivation)	In hospital mortality	• Mortality rate after TBI in the emergency department was 3.1% in more deprived and 3.4% in less deprived groups. Overall mortality rate was 18.8% in the more deprived and 22% in the less deprived groups. • Univariate analysis showed lower mortality rate in the more deprived group (p = 0.046, 18.8% vs. 22%), but this was not significant in the final logistic regression model used to identify independent predictors of mortality after adjusting for other known predictors (p = 0.281).	No statistically significant difference.
Haines et al. 2019¹¹ United States Participants: Teenagers and adults Age range 15 - 89 N = 106708 Follow-up: hospital stay Design: retrospective	Medical insurance	In hospital mortality	• Uninsured and Medicare insured patients had higher odds of mortality (odds ratio OR = 1.052, p = 0.001, CI not reported) for head AIS = 5 compared with other insurance types. • No difference for lower AIS (2–4) (p > 0.05).	Statistically significant increase in mortality for severe injuries (AIS 5) only. No statistically significant difference for lower severity injuries.
Heffernan et al. 2011⁴⁶ United States Participants: Adults Mean age 53.8 N = 3101 Follow-up: hospital stay Design: retrospective	Medical insurance, median income in zip code	In hospital mortality	• Privately insured patients were less likely to die than Medicare/Medicaid patients (OR = 0.68, 95% CI = 0.51, 0.9) or uninsured patients (OR = 0.57, 95% CI = 0.41. 0.78) • For ISS 1–15, mortality increased in 10,000 income increments (trend).	Statistically significant increase in mortality for those without private insurance.
Kool et al. 2008⁵⁵ New Zealand Participants: Children and adults No age range N = 4347 Follow-up: not defined Design: retrospective	2006 New Zealand deprivation index	Overall mortality rate	• Rates of mortality showed an increasing trend with increasing deprivation with a linear relationship to deprivation index scores (values 1–10) (p < 0.001) for all age groups except the 10–14 age group. Increase in mortality rate of 2–7% in each age group.	Statistically significant increase in mortality with increasing deprivation.
McQuistion et al. 2016⁴⁷ United States Participants: Teenagers and adults Mean age 59 N = 187354 Follow-up: hospital stay Design: retrospective	Insurance status	In hospital mortality	• Uninsured patients, those with Medicare insurance and Medicaid insurance types had higher in hospital mortality compared with privately insured patients (private insurance is the reference). Uninsured OR = 1.37, p < 0.001, Medicare OR = 1.16, p < 0.001, Medicaid OR = 1.20, p < 0.001).	Statistically significant increase in mortality for those without private insurance.
Nuno et al. 2015⁴⁸ United States Participants: Children Age 0 - 23 months N = 5195 Follow-up: hospital stay Design: retrospective	Median income in zip code	In hospital mortality	• The highest median income group ($30,001–$35,000) had significant mortality reduction compared with the lowest income group (<$24,000) (OR = 0.46, 95% CI = 0.29, 0.72, p = 0.006	Statistically significant increase in mortality for the lowest income group compared with highest income group.
Selassie et al. 2014³⁶ United States Participants: Adults Mean age 42.8 N = 33695 Follow-up: up to 12 years (median 53 months) Design: retrospective	Insurance status	All-cause mortality rate	• Health insurance type was associated with relative risk of dying. Compared with the private insurance group, Medicare had a HR = 1.54 (1.40–1.69), Medicaid HR = 1.67 (1.48–1.87), uninsured HR = 1.27 (1.11–1.45), p < 0.001	Statistically significant increase in mortality rate for those without private insurance.

SES, socioeconomic status; TBI, traumatic brain injury; SD, standard deviation; AIS, Abbreviated Injury Scale, scores 1–5 with increasing severity; CI, confidence interval; HR, hazard ratio; ISS, Injury Severity Score.

Significance column: Standard text, low SES predicts worse outcomes; italic, low SES does not predict difference in outcomes. Scores 1 - 15 with increasing severity.

Studies based in the United States used medical insurance as a proxy measure for SES. Haines and coworkers¹¹ showed a significant increase in mortality with lower SES for severe head injuries, when comparing privately insured versus uninsured patients, although the effect was quite small (odds ratio [OR] = 1.052, p = 0.001, confidence interval [CI] not reported). Heffernan and associates,⁴⁶ McQuistion and colleagues,⁴⁷ and Selassie and coworkers³⁶ used insurance status as a proxy for SES. They all showed a significantly higher mortality risk for patients with Medicare/Medicaid insurance and uninsured individuals compared with those privately insured.

Heffernan and associates⁴⁶ showed that patients with private insurance were less likely to die compared with those with Medicare/Medicaid policies (OR = 0.68, 95% CI = 0.51 to 0.9) or uninsured (OR = 0.57, 95% CI = 0.41 to 0.78); McQuistion and colleauges⁴⁷ showed that uninsured patients were more likely to die in hospital compared with those with private insurance (OR = 1.37, 95% CI 1.27 to 1.48); and Selassie and coworkers³⁶ showed increased odds of mortality for patients without insurance (OR = 1.27, 95% CI 1.11 to 1.45) and indigent insurance (OR = 1.67, CI 1.48 to 1.87) compared with the privately insured group.

Functional outcomes

Eight studies addressed functional outcomes^{24,25,27,29,30,33,38,50} with a wide range of population sizes (from 51 to 8630) and follow-up duration (from three months to 40 years). The outcome measures varied: (Extended Glasgow Outcome Scale), which describes longer term functional outcomes, was used as a metric in three studies^25,33,50 while other social functioning and activities of daily living (ADL) scales were used by Hoofien and coworkers,³² Rassovsky and colleagues,²⁹ and Ulfarsson and associates.³³

The Patient Competency Rating Scale (PCRS) was used by van Walsem and colleagues³⁰ to assess competency with daily activities as reported by patients, and Yue and associates³⁵ used a satisfaction with life scale administered to patients at six months post-injury requiring individuals to rate items on a point-based scale. Huang and associates³⁸ assessed overall functioning based on being institutionalized or discharged home. Finally, Rabinowitz and coworkers²⁴ assessed persisting symptoms using a validated questionnaire (Rivermead Post Concussion Symptoms Questionnaire).

Results support the presence of an association between lower SES and worse functional outcomes. Five studies showed a significant link between lower SES and worse functional outcomes, including two large studies of 8630 patients by Huang and associates³⁸ and 1322 patients by Humphries and colleagues.⁵⁰ Huang and associates³⁸ showed a large increase in the odds of being institutionalized versus being discharged home for both SES exposure measures used (unemployment vs. employment: OR = 2.12, 95% CI = 1.34 to 3.34, and family income (average vs. low): OR = 2.20, CI = 1.66 to 2.92) in their multi-variable model. Humphries and colleagues⁵⁰ showed that 12-month GOSE scores worsened with increasing deprivation (increasing raw IMD scores) in their multi-variable ordinal regression model.

Two small-scale studies showed no association, with a total of 176 included patients across the two studies.^24,29 One study by Yue and coworkers²⁵ showed that patients who were unemployed pre-injury had lower GOSE scores at 6 months compared with those in employment (GOSE scores decreased by 0.50 points, CI = -0.88 to -0.11), but self-reported quality of life scores did not differ (Table 2).

Table 2.

Summary of Studies With Global Functioning Measures as Their Outcome

Study details	SES measures	Outcome	Results	Significance
Hoofien et al. 2002³² Israel Participants: adults Mean age 38.6 years N = 76 Follow-up: 14 years average Design: retrospective	Education, number of siblings, index of quality of services in the military^†	• Social integration: Extended ADL questionnaire. • Daily functioning	• Number of siblings was the most powerful predictor of functional and social outcomes (lower number of siblings, better outcomes), p < 0.01, (unclear effect size). • Quality of services in the military and education did not predict outcomes.	Statistically significant worse social and functional outcomes with increased sibling number.
Huang et al. 2019³⁸ Taiwan Participants: adults Age >18 years N = 8630 Follow-up: 6 months minimum Design: retrospective	Employment, family income	Functioning: institutionalization in long term facility vs. home	• Unemployment at time of injury and lower family income predict being institutionalized in long term facilities. • Univariate model: Unemployment odds ratio (OR) (reference = employed) OR = 5.96, 95% CI = 4.01-8.87. Middle-low and low family income OR (reference = average income) OR = 2.67, 95% CI = 2.08 to 3.41) • Multi-variable model: Unemployment OR = 2.12, 95% CI = 1.35-3.34; Family income OR = 2.20, 95% CI = 1.66 to 2.92	Statistically significant worse functional outcome with unemployment and lower income.
Humphries et al. 2020⁵⁰ United Kingdom Participants: adults Mean age 45.54 years N = 1322 Follow-up: 12 months Design: prospective cohort	English Indices of Multiple Deprivation (IMD) 2015 score	Global: GOSE at 12 months.	• Univariate analysis: increasing IMD raw scores (more deprived), predict worse GOSE at 12 months (p < 0.001) • Multi-ordinal regression: increasing raw IMD scores (worse deprivation) predict worse GOSE, OR = 0.99, 95% CI = 0.98-0.99, p = 0.002.	Statistically significantly worse GOSE scores for increased deprivation.
Rabinowitz et al. 2015²⁴ United States Participants: teenagers and adults Age range 12 -20 years N = 87 Follow-up: 3 months Design: prospective cohort	Socioeconomic Composite Index (SCI)	Persisting symptoms, assessed using the Rivermead post-concussion questionnaire and BSI global severity index.	• SCI does not predict persisting symptoms (p = 0.5033).	No statistically significant difference.
Rassovsky et al. 2015²⁹ Israel Participants: adults Mean age 26.06 years N = 89 Follow-up: 14 years average (1-53 range) Design: prospective cohort	SES factor: composite score of parental education, self-reported SES, salary, sibling number	• Daily functioning (ADL-home score and ADL-mobility). • Social functioning (social activity)	• SES factor does not predict any of the functional outcomes (ADL or social functioning), p > 0.05	No statistically significant difference.
van Walsem et al. 2020³⁰ Norway Participants: adults Mean age 30.5 years N = 94 Follow-up: 1, 2, 5, 10 years Design: prospective cohort	Employment at the time of injury, occupation type	Patient competency rating scale (self-reported functional competence: self-awareness, emotional and cognitive competencies, ADL).	• Employed individuals had significantly higher patient competency rating scale score trajectories (higher score means greater competency) than unemployed individuals in the hierarchical linear model. Effect size within the model: b-Weight = 18.01, 95% CI 6.95-29.07, p = 0.002). • Employment type (white collar vs blue collar) did not predict higher scores. b-Weight = 2.05, 95% CI -5.30 - 9.41, p = 0.58	Statistically significant difference - lower competency scores for unemployed individuals.
Ulfarsson et al. 2013³³ Sweden Participants: adults Mean age 37.9 years N = 51 Follow-up: 12 months Design: prospective cohort	Employment at time of injury	Functional scales, measures of HRQoL. 1) Short Form-36 Health Survey (SF-36) 2) GOSE 3) P-ADL	• Being on sick leave/unemployed (for any length of time) predicted worse P-ADL, SF-36 physical function (p = 0.04 and p = 0.02), and may be associated worse GOSE (p = 0.08). • Being on sick leave/unemployed for >12 months predicted significantly worse GOSE scores at 12 months (p = 0.01), worse P-ADL (p = 0.04), SF-36 (p = 0.03).	Statistically significant worse global scores for unemployed / sick leave individuals.
Yue et al. 2018²⁵ United States Participants: adults Mean age 39.8 years N = 162 Follow-up: 6 months Design: retrospective	Employment at time of injury	1) GOSE at 6 months 2) satisfaction with life scale (SWLS)	• Unemployed individuals also had lower education status. • Unemployed patients had lower GOSE scores at 6 months on multivariable analysis: decrease in 0.50 points, 95% CI -0.88, -0.11, p = 0.011). • QoL outcome (SWLS) was not significantly different on multivariate analysis: B = -1.79, CI -4.66, 1.08, p = 0.219.	Statistically significantly worse GOSE scores for unemployed individuals. No statistically significant difference for SWLS.

SES, socioeconomic status; ADL, acivities of daily living; CI confidence intrval; GOSE, Glasgow Outcome Scale Extended; SCI = socioeconomic composite index-

Significance column (significance based on p < 0.05): Standard text, low SES predicts worse outcomes; italic, low SES does not predict difference in outcomes.

†

SES indicator created by the authors because of the context in which the study was conducted.

Cognitive outcomes

Thirteen studies focused on cognitive and psychiatric outcomes.^{13,22

–25,29,32,34,35,42

–45} Variation existed between these studies in terms of the specific metrics and scales used to assess cognitive and psychiatric outcomes. These differences make it more challenging to compare results directly across the studies and increase heterogeneity of outcome data available. Nevertheless, most studies included measures of intellectual ability such as intelligent quotient (IQ) measures and other scales of intelligence, such as the Wechsler Intelligence Scale and learning tests. Scales of behavior were used in a small subset of studies.

Finally, two studies by Ryan and colleagues investigating psychiatric conditions used tailored interviews.^13,35 Results were mixed; five studies showed no association between SES and cognitive outcomes.^{13,23,29,42,43} Catroppa and colleagues⁴³ tested intellectual ability through a number of IQ measures, none of which showed significant differences across SES groups, both at 2- and 10-years post-injury. Five studies suggested a link between lower SES and worse outcomes,^24,32,34,44 and three showed that SES affected some cognitive domains but not others^25,35,45 (Table 3).

Table 3.

Summary of Studies Assessing Cognitive Outcomes

Study details	SES measures	Outcome	Results	Significance
Catroppa et al. 2003⁴² Australia Participants: Children Age range 8 – 12 years N = 70 Follow-up: 2 years Design: prospective cohort	Daniel's Scale of Occupational Prestige	Intellectual ability. IQ measures: verbal (VIQ), performance (PIQ), full scale intellectual quotient (FSIQ). Index scores: verbal comprehension (VC), perceptual organization (PO), processing speed (PS), freedom from distractibility (FFD).	• Overall, trends toward SES predicting outcome. • VIQ did not change based on SES groups (p = 0.09) • SES was the best predictor of PIQ and FSIQ, VC, PS, not statistically significant. • SES was overall the best predictor of intellectual ability but lacked statistical significance. • Severe TBI individuals are more affected overall (lower scores).	No statistically significant difference.
Catroppa et al. 2012⁴³ Australia Participants: Children Mean age 4.81 years (range 1 - 8) N = 40 Follow-up: 10 years Design: prospective cohort	Daniel's Scale of Occupational Prestige	Recovery at 10 years: 1) intellectual abilities, 2) educational skills (WRAT - wide range achievement test), 3) adaptive behaviors (ABAS)	• SES does not predict intellectual ability, educational skills outcomes at 10 years in any domains (ABAS; WRAT). 95% CI crosses 1 for all cognitive tests, p > 0.05	No statistically significant difference.
Cohen-Zimerman et al. 2019⁴⁴ United States Participants: Adults Mean age 58.24 years N = 54 Follow-up: 40 years Design: prospective cohort	Hollingshead occupational prestige scale (Parental occupation prestige score, as proxy of baseline SES)	Cognitive ability (Armed Forces Qualification Test - AFQT) collected pre-injury.	• Childhood SES accounts for a significant proportion of intelligence variance (p < 0.001) at baseline and all following time points for TBI. • Linear regression: SES predicts post-injury AFQT scores (lower SES, worse scores), score difference:-4.3, p < 0.001. • SES is a stronger predictor in left hemisphere injuries (compared with right or bilateral), p < 0.001.	Statistically significant difference - worse cognitive ability scores for lower SES.
Donders et al. 2004⁴⁵ United States Participants: Pediatric Mean age 14.5 years N = 100 Follow-up: 1 year Design: prospective cohort	Parental occupation	Neuropsychological tests: (1) Wechsler Intelligence Scale for Children WISC-III, (2) California Verbal Learning Test - Children CVLT-C, (3) Children's Category test CCT	• Lower parental occupation is associated with worse performance on verbal comprehension, freedom from distractibility and processing speed index on WISC-III (p < 0.01, p < 0.01 and p < 0.05 respectively), but not on other tests.	Statistically significant difference - worse performance on some domains but not all for lower SES.
Hoofien et al. 2002³² Israel Participants: Children Mean age 14.5 years N = 100 Follow-up: 14 years average Design: retrospective	Education, number of siblings, index of quality of services in the military	Cognitive abilities: WAIS-R, FSIQ, VIQ, PIQ, questionnaire. Psychiatric: SCL-90 GSI, PTDS	• Number of siblings is the stronger predictor of cognitive and psychiatric outcomes (lower n siblings, better cognitive outcome), p < 0.01 • Quality of family military service is the second strongest predictor (p < 0.01). • Education level does not predict cognitive outcomes p > 0.01	Statistically significant difference - worse cognitive outcomes for lower SES measures.
Max et al. 2013²² United States, Canada Participants: Children Mean age 9.81 years N = 79 Follow-up: 12 months Design: prospective cohort	Four-Factor Index.	Novel psychiatric disorders (NPD) development at 1 year after injury. Assessed using the K-SADS-PL semi structured interview, to establish DSM-IV psychiatric diagnoses	• NPS was significantly associated with SES (lower SES, more likely to have NPD). Average Four-Factor Index score in NPD group was 33.25, in No NPD group it was 40.56; SE = 0.63, p = 0.038.	Statistically significant increased likelihood of developing psychiatric disorders with low SES.
Max et al. 2015²³ United States, Canada Participants: Children Mean age 10.2 years N = 87 Follow-up: 24 months Design: prospective cohort	Four-Factor Index	NPD presence assessed at 24 months. DSM-IV psychiatric diagnosis.	• 87 children, 54 followed up at 2 years. 17 of 54 had NPD. • SES does not predict NPD. Four-Factor Index scores for NPD and No NPD groups were 35.7 and 39.0 respectively, p = 0.92	No statistically significant difference.
Rabinowitz et al. 2015²⁴ United States Participants: Teenagers and adults Age range 12 -20 years N = 87 Follow-up: 3 months Design: prospective cohort	Socioeconomic Composite Index (SCI)	Cognitive outcome, assessed using a series of cognitive tests (SDMT-O, SDMT-W, VSR Delay, VSR CLTR, BVMT total)	• SCI is the only predictor of poor cognitive outcome at 3 months (p = 0.007). • Mild TBI and orthopedic controls had comparable rates, higher than healthy group.	Statistically significant worse cognitive outcome according to SCI.
Rassovsky et al. 2015²⁹ Israel Participants: Adults Mean age 26.06 years N = 89 Follow-up: average 14 years (range 1 - 53) Design: prospective cohort	SES factor. Composite score of parental education, self-reported SES, salary, sibling number	Cognitive functioning (RAVLT, WCST, semantic fluency, phonemic fluency, ROCF delay). Mental status.	• SES factor does not predict any of the cognitive outcomes (p > 0.05 for all domains).	No statistically significant difference.
Ryan et al. 2014³⁵ Australia Participants: Children Age range 1-7 years N = 34 Follow-up: 16 years Design: prospective cohort	Daniel's Scale of Occupational Prestige	Emotion perception (EP) measured using the advanced clinical solutions (ACS) social perception subset. Domains: (1) affect naming, (2) prosody-face matching, 3) prosody-pair matching	• SES predicts prosody-pairs matching (mean difference 0.83, p = 0.38). High SES TBI group were significantly more accurate than the low SES TBI group (mean difference 2.96, p = 0.002). • No effect on the other 2 EP domains (p > 0.05).	Statistically significant difference - worse outcomes in one of three domains with low SES. No significant difference in other 2 domains.
Ryan et al. 2016¹³ Australia Participants: Children Age range 5-15 years N = 78 Follow-up: 12 and 24 months Design: prospective cohort	Australian and New Zealand Socioeconomic Classification of Occupations (parental)	Social problems, assessed with the Child Behaviour Checklist for 6–18 years (CBCL/6-18), specifically the Social Problems Scale (CBCL-SPS)	• SES did not significantly predict outcome at 24 months (p = 0.36).	No statistically significant difference.
Wade et al. 2020³⁴ United States Participants: Children Mean age 11.7 years N = 591 Follow-up: unclear Design: retrospective	Education level of primary caregiver	Behavior problems. Child Behavior Checklist (CBCL) - completed by parents, and Home and community social behavior scale (HCSBB).	• Parental education was correlated with lower levels of Total, Internalizing and Externalizing Problems, and was positively correlated with Social Competence (p < 0.001) • Children of parents with less than high school education had more than twice the odds of developing internalising (OR = 2.82, 1.27-6.28), externalising (OR = 2.31, 95% CI 1.02-5.24), and total problems (OR = 2.90, CI 1.20-6.44). • Children where parents had at least some post-secondary education had better social competence than those with HS/GED (diff = 4.6, 95% CI 2.69-6.6) and those with less than HS (diff = 7.1, 95% CI 3.68-10.52)	Statistically significant difference - worse outcomes for lower education level of caregiver.
Yue et al. 2018²⁵ United States Participants: Adults Mean age 39.8 years N = 162 Follow-up: 6 months Design: retrospective	Employment at time of injury	Neuropsychiatric: (1) PTSD checklist for civilians (PCL-C) (2) Brief Symptom Inventory 18 (BSI18) (3) Rivermead Post concussion symptoms questionnaire (RPQ). Neurocognitive: (1) Welchsler adult intelligence scale processing speed index (WAIS-PSI) (2) Trail Making Test (TMT) (3) CVLT-II	• Unemployed individuals had higher psychiatric distress: BS118 score 60.80 vs. 52.50, p < 0.001; RPQ13 scores 16.60 vs. 11.02, p = 0.001; PCL-C score increase of 5.99 for unemployed, 95% CI 0.76, 11.2, p = 0.025. • No differences in neurocognitive outcomes (p > 0.05)	Statistically significant difference - worse psychiatric outcomes with low SES. No statistically significant difference for cognitive outcomes.

SES, socioeconomic status; CI, confidence interval.

Significance column: Standard text, low SES predicts worse outcomes; italic, low SES does not predict difference in outcomes.

The study by Wade and coworkers³⁴ was the only study with a large population size of 591 (all other studies had populations of 34–164 patients). This study assessed child behavior problems and showed consistently over double odds of internalizing, externalizing, and total problems for children in the lower SES group after TBI (ORs, respectively for internalizing, externalizing, and total problems: OR = 2.82, 95% CI 1.27 to 6.28; OR 2.31, CI 1.02 to 5.24, and OR = 2.90, CI 1.20 to 6.44).

Vocational outcomes

Ten studies included vocational outcomes after TBI, all of which were investigated in adult patients only (Table 4).^{26

–29,31,32,37,39
–41} Vocational outcomes primarily consisted of return to work after TBI, assessed in all 10 studies. The follow-up time at which the final assessment was completed, however, varied from six months to 14 years. Six studies showed a significant association between lower SES pre-injury, often assessed as pre-injury employment and occupation type, and employment after TBI.

Table 4.

Summary of Studies With Vocational Outcomes

Study details	SES measures	Outcome	Results	Significance
Arango-Lasprilla et al. 2020³⁷ Multiple European countries Participants: Adults Mean age 44.2 years N = 1015 Follow-up: 1 year Design: retrospective	Employment at the time of injury	Employment at 1 year post- TBI	• Baseline employment: 72.02% full time, 15.96% part time, 12.02% unemployed. • At follow-up 61.38% employed, 38.62% unemployed. • Part-time employment or unemployment at injury predict lower probability of employment at 1 year, compared with full-time employment. Odds ratio (OR) 0.67, 95% CI 0.45, 1.00, p = 0.048 for part-time employment. OR 0.12, 95% CI 0.07, 0.19, p < 0.001 for unemployment	Statistically significant difference - lower odds of being unemployed with lower SES measure.
De Koning et al. 2017²⁶ The Netherlands Participants: Adults Age range 18 – 65 years N = 319 Follow-up: 6 to 12 months Design: prospective cohort	Occupational level pre-injury	Return to work (RTW) at 6 months and 12 months after injury. Groups: none (nRTW), partial (pRTW), complete (cRTW).	• Predictive models for 6 and 12 months. • In 6-month model, professional occupation predicts lower odds of unemployment (OR = 0.38, 95% CI 0.15-0.98) • Pre-injury unemployment had the lowest odds of returning to work. • In the 12-month model, SES does not predict RTW.	Statistically significant lower RTW at 6 months for low SES. No statistically significant difference at 12 months.
Forslund et al. 2014³¹ Norway Participants: Adults Mean age 30.9 years N = 105 Follow-up: 5 years Design: prospective cohort	Employment at the time of injury, occupational category	Employment at 1, 2, 5 years.	• Baseline: Blue collar: 50.5% (53), white collar 49.5% (52). Employed 81% (85). • Those employed at injury had a greater likelihood to be employed at 1, 2, 5 years (B = 0.52, CI 0.32, 0.72 p < 0.001). • White collar workers more likely to be employed than blue collar at 1, 2, 5 years (B = 0.21, CI 0.04, 0.38, p = 0.019). • Moderate TBI more likely to be employed than severe (p = 0.008)	Statistically significant difference - unemployment and lower occupation predict worse vocational outcomes.
Grauwmeijer et al. 2017²⁷ The Netherlands Participants: Adults Age range 18 – 67 years N = 109 Follow-up: 10 years Design: prospective cohort	Employment at the time of injury, occupational category	Employment at 10 years.	• Pre-injury employment, professional category are not predictors of employment at 10 years (p = 0.439)	No statistically significant difference.
Hoofien et al. 2002³² Israel Participants: Adults Mean age 38.6 years N = 76 Follow-up: 14 years average Design: retrospective	Education, number of siblings, index of quality of services in the military	Vocational: (1) index of employment, (2) index of level of employment, (3) index of stability of work	• Number of siblings was the strongest predictor of vocational outcomes (lower number of siblings, increased employment, and stability of work), p < 0.01. • Quality of services in the military and education did not predict outcomes.	Statistically significant difference in vocational outcome for number of siblings. No significant difference for military quality.
Howe et al. 2018²⁸ Norway Participants: Adults Mean age 30.3 years N = 97 Follow-up: 10 years Design: prospective cohort	Employment at the time of injury, occupational category	Employment at 1, 2, 5, 10 years	• Employment at time of injury predicts employment at all 4 time points. B = 0.447, 95% CI 0.254, 0.640, p < 0.001. • Higher occupational category also predicts employment. B = 0.243, 95% CI 0.074, 0.411, p = 0.005. • White collar workers at injury are more likely to be employed at 1, 2, 5, 10 years compared to blue collar.	Statistically significant difference - worse employment outcomes for unemployed and lower occupations.
Ketchum et al. 2012³⁹ United States Participants: Adults Mean age 29.9 years N = 418 Follow-up: 1 year Design: retrospective	Pre-injury employment	Employment at 1 year (in competitive employment vs unemployed)	• Pre-injury employment status was a significant predictor of 1 year employment (p < 0.001) • Odds of not being employed at 1 year were 2.8 for those not employed pre-injury, compared with those employed pre-injury.	Statistically significant difference - worse employment outcomes for unemployed pre-injury.
Keyser-Marcus et al. 2002⁴⁰ United States Participants: Adults Mean age 32.1 years N = 451 Follow-up: 5 years Design: prospective cohort	Pre-injury employment and productivity	Employment at 1, 2, 3, 4, 5 years	• Employed individuals at injury were 5.1 times more likely to return to work at 1 year. • Pre-injury employment predicted return to work at 1-, 2-, 3- and 5-years post-injury (p < 0.01 in each) • Unemployed pre-injury (compared with those employed) were less likely to be employed after injury. • Pre-injury employment did not predict return to work at 4 years only (p = 0.16).	Statistically significant difference in vocational outcome for pre-injury employment.
Machamer et al. 2005⁴¹ United States Participants: Adults Mean age 35.1 years N = 165 Follow-up: 3-5 years Design: prospective cohort	Pre-injury yearly earnings; pre-injury work stability	Work stability at 3-5 years. Defined as amount of time worked since injury: (1) never returned, (2) <50%, (3) 51-89%, (4) >90%	• Participants who did not return to work had higher rates of pre-injury unstable work (p < 0.001), compared with those who returned to work. • Pre-injury earnings predicted return to work, those working >90% since injury had highest income pre-injury (p = 0.003). • Higher income predicts maintenance of stable employment (p < 0.001).	Statistically significant difference - worse work stability for those with pre-injury unstable work and lower income.
Rassovsky et al. 2015²⁹ Israel Participants: Adults Mean age 26.06 N = 89 Follow-up: average 14 yr (range 1 - 53) Design: prospective cohort	SES factor.	Vocational status: occupation level	• SES factor does not predict vocational outcomes (p > 0.05).	No statistically significant difference.

SES, socioeconomic status; CI, confidence interval; RTW, return to work.

Significance column: Standard text, low SES predicts worse outcomes; italic, low SES does not predict difference in outcomes.

Two studies by Rassovsky and associates²⁹ and Grauwmeijer and colleagues²⁷ showed no association between SES and return to work, and both had long follow-up times of 10 years or more.^27,29 This could indicate that it takes longer for individuals with lower SES to return to work after TBI. De Koning and colleagues²⁶ showed a significant association between level of employment pre-injury and return-to-work at six months, but no significance at the later time point of 12 months.²⁶

Outcomes for pediatric patients with TBI

Most studies included in this review focused on adult TBI populations or included both older children and adults; therefore, the results up to this point present data from all included studies. This paragraph presents a closer look at those nine studies that focused solely on pediatric TBI participants. Of these, only one study had mortality as its primary outcome.⁴⁸ The remaining eight studies focused on cognitive outcomes.^{13,22,23,32,34,35,42,43} The evidence was mixed, with four studies showing that family SES is not associated with differential cognitive or psychiatric outcomes after TBI, three studies showing a link between lower SES and worse cognitive outcomes, and one study showing a mixed picture with lower SES predicting worse outcomes for only some of the cognitive domains but not others. Details are provided in Appendix 3 of the Supplementary Material.

Outcomes based on TBI severity

Five studies only recruited participants with mild TBI.^22

–26 Overall, no clear association between SES and outcomes was shown in the mild TBI groups; however, the outcomes studied in these five articles varied. A further five studies included participants with moderate and severe TBI but not mild TBI.^27

–31 They all assessed functional outcomes, largely defined as post-injury employment, and also showed mixed results with a trend toward a negative association between SES and worse outcomes (three studies: lower SES associated with worse functional outcomes, two studies: neutral). Two studies only recruited participants with severe TBI,^32,33 and both showed worse outcomes for lower SES in their populations.

Assessment of quality and level of evidence

The 14-item NIH Quality Assessment Tool for Observational Cohort and Cross-sectional Studies was used to assess methodological quality. All articles scored well for the domains assessed having a clear and defined study question, and the exposure being measured before the outcome: however, based on the scoring guidelines, most studies scored a “poor” quality rating overall, and 11 scored a “fair” rating. No studies scored a “good” rating overall. Having the exposure measured more than once in the study and blinding of the assessors of the outcomes to the exposure were particularly problematic.

In regard to the variable studied (SES), however, these are unlikely to have had an impact on the quality of data reported because SES is typically relatively fixed and unlikely to have changed significantly over a short period. In addition, none of the studies reported performing a priori calculation regarding required sample size, which is important to detect differences between groups.

The GRADE quality assessment was used to assess the level of the evidence surrounding each outcome of interest (Table 5). Inconsistency, referring to unexplained heterogeneity of the results, was identified as serious for the cognitive and vocational outcomes. Serious issues with indirectness or imprecision of the evidence were not identified for these outcomes. Overall, the quality of evidence was rated as moderate for mortality: this was the outcome where studies had variable risk of bias, with some studies only having moderate risk of bias (rated as fair), and where the results were most consistent. The other three outcomes had low and very low quality of evidence, owing to high risk of bias in individual studies, as well as high levels of heterogeneity in between studies.

Table 5.

GRADE Level of Evidence for Each Study Outcome and Recommendations

Outcome	Risk of bias	Quality of evidence	Recommendation
Mortality	Unclear	Moderate	Stronger evidence required. Bias may affect the interpretation of the results.
Cognitive	High	Low	Stronger evidence required. Bias likely to affect the interpretation of the results.
Functional	High	Very low	Stronger evidence required. Bias likely to affect the interpretation of the results.
Vocational	High	Very low	Stronger evidence required. Bias likely to affect the interpretation of the results.

Discussion

This review showed trends to support the hypothesis that lower SES is associated with worse outcomes after TBI. No studies showed better outcomes for lower SES groups. The mixed outcomes and heterogeneous data reported in this review, however, demonstrate that there are several challenges that should be considered when evaluating the SES of individuals with TBI that could impact study and patient outcomes.

All 32 studies included were observational. This is not surprising given the subject matter and inclusion criteria and has been shown to be the case for studies of SES in other clinical contexts.^5,51 It would not be possible to conduct randomized clinical trials to investigate SES; therefore, good quality evidence to inform clinical practice must be derived from robustly constructed, large scale observational studies.

SES measurement

The SES cannot be directly and objectively measured, but is a complex multi-dimensional context-specific construct, and measures vary according to the setting, country, or society of an individual.⁵² Universally acceptable and valid metrics are difficult to establish; therefore, proxy metrics are the best method established for estimating SES.

Several measures of SES were used across the studies included in this review, frequently including occupation and employment. These can be challenging to compare between different studies and preclude meta-analysis because occupation outcomes have no set definition for research purposes (e.g., criteria used to define full-time employment may vary between studies).

Vocational outcomes in the included studies were primarily return to employment after TBI. This outcome may be biased as a marker for post-TBI success because various employers have different tolerances for leaves of absence, medical insurance coverage, and physical and cognitive requirements on return to work, regardless of TBI or other types of injury. Several studies attempted to address this issue by evaluating pre-and post-injury productivity or relative change in employment level.

An additional limitation of investigating this outcome was the definition of employment itself. This varied across studies, with some authors including those in education in the employed category^28,40 and others including them in the unemployed category.³⁹ Impact of age and other demographic factors on return to work are also relevant. Research shows that older individuals with TBI (>40) are less likely to return to work post-injury compared with younger individuals.^53,54 This may be related to greater ability to retrain and long-term employment prospects making employers more willing to invest in younger workers, or simply a patient's personal choice to retire early considering their injury regardless of functional rehabilitation. In addition, longer rehabilitation after TBI may have more focus on return to work for younger individuals. Taken together, this indicates that there may be multiple confounders of the relationship between pre-injury SES and employment after TBI.

Medical insurance status was also used as a proxy for SES. ^11,36,46,47 This measure has the advantage of being readily available on medical records, but does not yield granular data and makes broad presumptions regarding a patient's employment or financial status based on presence or absence of insurance without regard for the vast variety of options for private and public funding available in many regions.

Results from the United States-based studies all demonstrated a correlation between lower SES, largely defined by presence or absence of insurance, and mortality risk. The difference in emergency care received after a trauma may not vary as much as outpatient or elective services, possibly confounding the association between insurance status and SES, thus biasing the interpretation of the results of these studies. On the other hand, insurance status likely has a large impact on rehabilitation services utilized after index hospitalization, thus very likely impacting some of the longer-term outcomes reported in included studies.

Several studies in entirely publicly funded medical systems also found associations between SES and outcomes, using measurements other than insurance status as proxies for SES, implying that insurance status alone is not a reliable measurement of SES, with variation existing within broad insurance groups.

Indices of deprivation were also used: Carstairs and Morris Index of Deprivation,⁴⁹ the English Index of Multiple Deprivation,⁵⁰ the New Zealand Deprivation Index,⁵⁵ and the Socioeconomic Composite Index.²⁴ The above indices have a key issue of providing area-level rather than individual-level deprivation scores. They also have the advantage of being context specific, because they can include tailored variables that apply to a local population or group, making them a very specific and likely more accurate measurement tool for SES. This is highly appropriate as SES is best understood within its own context; however, this can make wider comparison across populations more challenging.

Outcomes measurement

The outcomes evaluated in SES studies are also quite heterogeneous and susceptible to a variety of factors that could impact reported results. Most studies evaluating mortality reported in-hospital mortality without looking at 6- or 12-month relative risk of mortality, which could be impacted by several SES factors.

Functional outcomes can also be quite variable depending on the tool used and timeline studied. This review found that low SES appears to be correlated with poor functional outcomes, but specific outcomes measured can impact the data reported. For example, Huang and associates³⁸ demonstrated a significant relationship between family income and functional outcome that they defined as discharge destination to home or long-term facility. This study was conducted robustly, but a key limitation is the lack of information regarding caregiver status, outside assistance by home-care nursing staff, and level of independence at home, because this could vary from being completely independent to requiring assistance for most ADL.

This outcome is also limiting because families of higher SES would likely have a better ability to care for injured family members at home through the financial means to hire home-care nursing staff or potentially taking extended leave from work to care for patients during the recovery period, a luxury that is often not available to those of non-professional or hourly workers in many locations. Thus, discharge destination is likely a poor representation of true recovery and lacks detailed stratification regarding expected outcomes.

Other measurements of functional outcomes, such as level of independence and self-reported symptom and satisfaction scores have similar challenges. Analogous to the indices used as SES measurement, functional outcomes are a factor of their own societal context, linked to personal values, social dynamics, and cultural beliefs. This raises a challenge relating to the need to achieve a trade-off between using locally developed and meaningful metrics of functional outcomes and utilizing standardized measurements of outcomes such as GOSE that may be less specific but easier to compare between varied patient populations.

Cognitive outcomes, too, are highly susceptible to variations in measurement, reporting, and context-specific expectations. Cohen-Zimerman and colleagues⁴⁴ showed a significant association between childhood SES and intelligence scores years after injury (higher SES, higher intelligence scores). ^.The authors also noted, however, a similar significant association between SES and intelligence overall both in the controls and TBI groups, suggesting that the link between intelligence and SES exists even in the absence of a brain injury. They postulate that this may be explained by the theory of cognitive reserve, stating that pre-morbid cognitive enrichment (such as years of education, access to resources, and a stimulating environment) plays a neuroprotective role in modulating the effects of a subsequent brain injury.^56,57

Measures of intelligence are inherently flawed by biases and the context and cohort against which they are compared, making it challenging to conduct studies with varied patient populations. In addition, intellectual outcomes may not be the only measurement of consequence when evaluating cognitive impact of TBI; Yue and coworkers²⁵ showed that lower SES predicted higher psychological distress scores but no difference in neurocognitive domains. The authors considered whether there may be an element of confounding in that unemployment and more challenging social situations may themselves be a predictive factor of distress even in the absence of brain injury, drawing similar conclusions to the aforementioned complexity of this outcome paradigm. Psychological distress, however, is an important consideration for developing children, even in the absence of intellectual impairment.

Study designs

The size of the studies evaluated in this review varied and likely had an impact on reported findings. The studies evaluating the association between SES and mortality had a mean of 49,047 participants (range 2932 to 187,354), while the studies reporting on functional outcomes were smaller overall and findings were stratified by cohort size; the two studies that had large populations of greater than 1000 patients found there was an association between SES and poor functional outcomes^38,50 while those that were smaller demonstrated no link with outcomes.

Similar results were reported for cognitive outcomes. Both studies that evaluated the cognitive outcomes in children after TBI reported a relationship between SES and poor outcomes, but the estimates of the associations were imprecise, likely because of the small sample size of both studies (40 and 70 individuals, respectively).^42,43 Thus, a lack of association between SES and poor outcomes found in some studies could be because of underpowered studies; frequently these were the studies measuring complex psychosocial concepts that required multi-faceted evaluation and long-term follow up.

All studies used multi-variable analysis when studying the association between variables of interest and outcomes; a number of studies reported both univariable and multi-variable models, and in this case, the results included in this review represent the multi-variable analysis, because these adjust for other factors. This demonstrates that the authors attempted to account for other variables that may influence the results of their studies.

Implications for clinical practitioners

This work supports the hypothesis that SES is a variable of interest in the context of TBI outcomes. It flags that patients with TBI falling into the low SES category may require individualized and targeted interventions during their recovery to help them achieve similar outcomes to the higher-SES group, a concept that has been identified across diverse areas of clinical medicine.^58
–60

Societal changes are likely required to reduce socioeconomic inequalities, and healthcare professionals are a key professional group to initiate changes pertinent to the patient populations they treat.⁶¹ The SES sits within the domain of the social determinants of health, contributing to the social gradient of health, a concept describing how one's position within society influences health.⁶² A wider implication is therefore to advocate on behalf of vulnerable patient groups such as those with TBI.

The SES may be an upstream factor affecting an individual's ability to access health services because of lack of awareness of services available, physical challenges in reaching those (such as unavailable resources, distance, cost), weaker support networks to support one during the illness journey. These factors may therefore be part of the pathway linking SES to health outcomes. Therefore, addressing SES may be the correct strategy to reducing societal inequalities and could result in more equal access to care.

For instance, patients with TBI from lower SES could be offered targeted follow-up for early identification of complications and to facilitate an environment conducive to recovery, particularly regarding ability to participate in rehabilitation programs (such as childcare at physical therapy clinics or expanded paid time off from employment for doctors' appointments) and skills workshops for those required to change professions because of limitations from their injuries.

It is likely that those with prior unemployment, precarious employment, or low skilled jobs will require more targeted employment development training as part of their rehabilitation. Because TBI frequently affects young individuals, focusing on employment post-TBI is essential to reduce the overall loss of earning from a working-age population. Similarly, further resources could be allocated to community-based interventions in areas of greater need, to support patients locally, and strengthen their support networks.

Incorporating assessments of a patient's socioeconomic setting in early assessments could help to identify patients at risk of having worse outcomes during their initial hospital admission, allowing for resources and appropriate services to be mobilized to support them. In addition, a greater understanding of how SES affects outcomes could help establish more targeted prevention strategies in the community, such as more frequent teaching about and advocating for safety measures like bicycle helmets and seat belts in low SES areas.

At a practical level, this work suggests that taking time to ascertain a patient's sociodemographic circumstances could be crucial to facilitate targeted care in the rehabilitation after TBI. It could also be useful for designing robust prediction models for patient outcomes or strengthening existing models with SES information and instituting more specific preventative measures after injury.

Recommendations for future work

The findings of this review should be corroborated by large scale, longitudinal studies of SES and TBI in a range of healthcare settings. Understanding how the impact of SES on outcomes changes based on the health system it is studied in will be beneficial to identify protective factors as well as health system characteristics that are driving inequalities. Measuring SES in health research is challenging because of the heterogeneity of metrics available, and as suggested by Oakes and associates,⁵² new efforts should be made in conjunction with epidemiologists and experts to reach a consensus and identify widely accepted and reproducible measures. This would also facilitate quantitative pooling of results from multiple studies through meta-analysis.

While shorter follow-up duration may be suitable to investigate in-hospital mortality, longer follow-up times are required to evaluate cognitive and vocational sequelae of TBI, particularly in pediatric populations. Future studies should allocate resources to longer-term longitudinal follow-up to detect changes that may become apparent years after the initial injury.

Agreement should be sought to use validated and standardized outcome metrics for each domain of interest. For example, it would be useful to compare the same mortality time point, such as in-hospital mortality or 30-day mortality. Similarly, with regard to global functioning measures, using outcomes scales such as the GOSE, Disability Rating Scale (DRS), or the disease specific quality of life measure Quality of Life after Brain Injury (QOLIBRI) would facilitate reproducibility across individual studies and quantitative pooling of study results in meta-analysis.⁶³

Limitations

Studies were heterogeneous in terms of their designs and measures of SES and outcomes used, precluding the ability to perform a meta-analysis. Specifically, the following factors result in the inability to perform meta-analysis. First, there are limited groups of studies using the same outcome measure—for example, the GOSE (three studies); however, these use the GOSE at different time points (two studies use GOSE at 12 months, one at six months), making comparison in meta-analysis difficult.^25,33,50

In addition, and more importantly, these studies measure exposure (SES) differently, which complicates quantitative comparison across studies. For example, one study uses the English Indices of Multiple Deprivation as a SES metric,⁵⁰ which is a scale dividing the population into deciles from most to least deprived, the second study uses dichotomized employment versus unemployment groups,²⁵ and the third uses the employed versus unemployed categories again but further subdivided into duration of unemployment.³³ Because of these multiple elements of variation, we believe that a meta-analysis with the current data is not appropriate, due to comparing studies with multiple cohorts of patients. The same can be applied to other studies with the same outcome measure such as in-hospital mortality.

The methodology of the included studies must be scrutinized, specifically with regard to confounding and interactions between variables. The role of confounding variables and mediator variables on the causal pathway linking SES to TBI outcomes needs to be considered more deeply. For example, other behaviors such as risky drinking practices and unsafe driving behaviors, which have been associated with low SES, may also contribute to poor health outcomes before or after traumatic injury.^64,65

Similarly, lower SES is associated with a higher incidence of traumatic injuries in adult and pediatric populations.^66,67 Therefore, one must be careful to interpret mortality figures in proportion to the disease incidence, which can vary for different population groups.

A high proportion of articles reviewed originated from the United States. Careful consideration must be taken of the differing insurance paradigms, regarding public versus private insurance changes subsequent to ObamaCare that occurred during the inclusion period of this study (since 2000). Using privately insured versus non-insured status as a proxy for SES could be misleading because insurance coverage is optional and under the current paradigm, many young, healthy people may choose not to pay for insurance regardless of SES.

In addition, this combined public-private system results in complex insurance schemes with many patients partially covered by both public Medicare/Medicaid and private insurance, Veterans insurance, possible add-ons to the basic public insurance, and different levels of coverage between individuals in the same household (children covered by insurance when parents are not, married adults covered by different insurance methods through their jobs requiring various levels of pay-in, deductibles, and copayments). This could be a potential confounder for outcomes published in SES literature and reported in this review.

Finally, this systematic review focuses on the impact of SES on TBI outcomes in high-income countries but excludes low- and middle-income countries. This approach was chosen to narrow the scope and obtain a more homogeneous population. Including LMICs within the same review was not deemed appropriate because differences in outcomes could be attributable to several other factors including wider differences in medical care received. This would have resulted in difficulty in differentiating between confounding factors and true association between the exposure and the outcome. Therefore, the results of this review would be relevant to high income systems alone, limiting their applicability in lower income settings. A separate review would be required to make conclusions for those settings.

Conclusion

This is the first systematic review of the association between pre-injury SES and outcomes after TBI. Using a narrative synthesis of the available evidence, trends emerged suggesting that SES is a variable of interest in TBI. The link between SES and outcomes was most robust regarding mortality and functional outcomes and more uncertain for cognitive and vocational outcomes. Further large scale, prospective, longitudinal observational studies are required to further elucidate and quantify the impact that lower SES has on clinical outcomes in the TBI population.

Notably, this review raises the point that SES should be considered as one of the relevant demographic or background factors for patients with TBI. It is important for clinicians and healthcare professionals to be aware that SES may play an important role in the clinical outcomes of patients with TBI, and actively collecting data regarding SES and considering this within a patient's clinical journey may enable early mobilization of targeted interventions for specific patient groups.

Footnotes

Authors' Contributions

S.V. designed the study protocol, J.G provided input in the protocol design. S.V. and M.S. performed the screening, data extraction and manuscript writing processes. J.G. and P.J.A.H critically revised the manuscript. All authors have read and approved the final manuscript.

Funding Information

JG was supported by the Medical Research Council (Skills Development Fellowship). For the purpose of open access, the author has applied a Creative Commons Attribution (CC BY) licence (where permitted by UKRI, ‘Open Government Licence’ or ‘Creative Commons Attribution No-derivatives' (CC BY-ND) licence may be stated instead) to any Author Accepted Manuscript version arising. PJH is supported by the NIHR (Cambridge BRC, Medtech Co-operative on Brain Injury, Global Health Research Group on Acquired Brain and Spine Injury, Senior Investigator Award) and the Royal College of Surgeons of England.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Appendix 1

Appendix 2

Appendix 3

References

Maas

AIR

, Menon

, Adelson

, et al. Traumatic Brain Injury: Integrated Approaches to Improve Prevention, Clinical Care, and Research. Lancet Neurol, 2017; 16(12):987–1048; doi: 10.1016/S1474-4422(17)30371-X

Baker

EH.

Socioeconomic Status, Definition. In: The Wiley Blackwell Encyclopedia of Health, Illness, Behavior, and Society John Wiley & Sons, Ltd: Chichester, UK; 2014; pp. 2210–2214; doi: 10.1002/9781118410868.wbehibs395

Villalba

CMH.

Socioeconomic Status (SES). In: Encyclopedia of Quality of Life and Well-Being Research Springer: Netherlands; 2014; pp. 6210–6212; doi: 10.1007/978-94-007-0753-5_2805

McMaughan

, Oloruntoba O and Smith

. Socioeconomic Status and Access to Healthcare: Interrelated Drivers for Healthy Aging. Front Public Health, 2020; 8:231; doi: 10.3389/FPUBH.2020.00231

Stringhini

, Carmeli

, Jokela

, et al. Socioeconomic Status and the 25 × 25 Risk Factors as Determinants of Premature Mortality: A Multicohort Study and Meta-Analysis of 1 · 7 Million Men and Women. Lancet, 2017; 389(10075):1229–1237; doi: 10.1016/S0140-6736(16)32380-7

Nieman

, Marrone

, Szanton

, et al. Racial/Ethnic and Socioeconomic Disparities in Hearing Health Care Among Older Americans. J Aging Health, 2016; 28(1):68–94; doi: 10.1177/0898264315585505

Popal

, Berkeveld

, Ponsen

, et al. The Effect of Socioeconomic Status on Severe Traumatic Injury: A Statistical Analysis. Eur J Trauma Emerg Surg, 2019; 47(1):195–200

Heathcote

, Wullschleger

, Gardiner

, et al. The Importance of Place of Residence on Hospitalized Outcomes for Severely Injured Trauma Patients: A Trauma Registry Analysis. J Rural Health, 2020; 36(3):381–393; doi: 10.1111/jrh.12407

McHale

, Hungerford

, Taylor-Robinson

, et al. Socioeconomic Status and 30-Day Mortality after Minor and Major Trauma: A Retrospective Analysis of the Trauma Audit and Research Network (TARN) Dataset for England. Ramagopalan S V. ed. PLoS One, 2018; 13(12):e0210226; doi: 10.1371/journal.pone.0210226

10.

Liao

, Chang

, Yeh

, et al. Socioeconomic Deprivation and Associated Risk Factors of Traumatic Brain Injury in Children. J Trauma Acute Care Surg, 2012; 73(5):1327–1331; doi: 10.1097/TA.0b013e31826fc87f

11.

Haines

, Nguyen

, Vatsaas

, et al. Socioeconomic Status Affects Outcomes After Severity-Stratified Traumatic Brain Injury. J Surg Res, 2019; 235:131–140; doi: 10.1016/j.jss.2018.09.072

12.

Yeates

KO.

Race as a Moderator of Parent and Family Outcomes Following Pediatric Traumatic Brain Injury. J Pediatr Psychol, 2002; 27(4):393–403; doi: 10.1093/jpepsy/27.4.393

13.

Ryan

, van Bijnen

, Catroppa

, et al. Longitudinal Outcome and Recovery of Social Problems after Pediatric Traumatic Brain Injury (TBI): Contribution of Brain Insult and Family Environment. Int J Dev Neurosci, 2016; 49:23–30; doi: 10.1016/j.ijdevneu.2015.12.004

14.

Jonsson

, Catroppa

, Godfrey

, et al. Individual Profiles of Predictors and Their Relations to 10 Years Outcome after Childhood Traumatic Brain Injury. Brain Inj, 2013; 27(7–8):831–838; doi: 10.3109/02699052.2013.775493

15.

Anonymous. PRISMA Flow Diagram. n.d. Available from: https://prisma-statement.org/prismastatement/flowdiagram [Last accessed: 1/29/2021]

16.

Anonymous. WDI - The World by Income and Region. n.d. Available from: https://datatopics.worldbank.org/world-development-indicators/the-world-by-income-and-region.html [Last accessed: 7/14/2021]

17.

Moher

, Liberati

, Tetzlaff

, et al. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. Ann Inter Med, 2009; 151(4):264–269; doi: 10.7326/0003-4819-151-4-200908180-00135.

18.

Anonymous. Study Quality Assessment Tools | NHLBI, NIH. n.d. Available from: https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools [Last accessed: 7/5/2021]

19.

Ryan

How to GRADE the Quality of the Evidence. n.d.

20.

Guyatt

, Oxman

, Vist

, et al. GRADE: An Emerging Consensus on Rating Quality of Evidence and Strength of Recommendations. BMJ, 2008; 336(7650):924–926; doi: 10.1136/bmj.39489.470347.ad.

21.

Anonymous. Patient Recruitment // CENTER-TBI.Eu. n.d. Available from: https://www.center-tbi.eu/project/progression [Last accessed: 7/14/2021].

22.

Max

, Pardo

, Hanten

, et al. Psychiatric Disorders in Children and Adolescents Six-to-Twelve Months after Mild Traumatic Brain Injury. J Neuropsychiatry Clin Neurosci, 2013; 25(4):272–282; doi: 10.1176/appi.neuropsych.12040078

23.

Max

, Friedman

, Wilde

, et al. Psychiatric Disorders in Children and Adolescents 24 Months after Mild Traumatic Brain Injury. J Neuropsychiatry Clin Neurosci, 2015; 27(2):112–120; doi: 10.1176/appi.neuropsych.13080190

24.

Rabinowitz

, Li

, Mccauley

, et al. Prevalence and Predictors of Poor Recovery from Mild Traumatic Brain Injury. J Neurotrauma, 2015; 32(19):1488–1496; doi: 10.1089/neu.2014.3555

25.

Yue

, Rick

, Morrissey

, et al. Preinjury Employment Status as a Risk Factor for Symptomatology and Disability in Mild Traumatic Brain Injury: A TRACK-TBI Analysis. NeuroRehabilitation, 2018; 43(2):169–182; doi: 10.3233/NRE-172375

26.

De Koning

, Scheenen

, Van Der Horn

, et al. Prediction of Work Resumption and Sustainability up to 1 Year after Mild Traumatic Brain Injury. Neurology, 2017; 89(18):1908–1914; doi: 10.1212/WNL.0000000000004604

27.

Grauwmeijer

, Heijenbrok-Kal

, Haitsma

, et al. Employment Outcome Ten Years after Moderate to Severe Traumatic Brain Injury: A Prospective Cohort Study. J Neurotrauma, 2017; 34(17):2575–2581; doi: 10.1089/neu.2016.4846

28.

Howe

, Andelic

, Perrin

, et al. Employment Probability Trajectories up to 10 Years after Moderate-to-Severe Traumatic Brain Injury. Front Neurol, 2018; 9; doi: 10.3389/fneur.2018.01051

29.

Rassovsky

, Levi

, Agranov

, et al. Predicting Long-Term Outcome Following Traumatic Brain Injury (TBI). J Clin Exp Neuropsychol, 2015; 37(4):354–366; doi: 10.1080/13803395.2015.1015498

30.

van Walsem

, Howe

, Perrin

, et al. Trajectories of Self-Reported Competency up to 10 Years Following Moderate-to-Severe Traumatic Brain Injury. Brain Inj, 2020; 34(3):335–342; doi: 10.1080/02699052.2019.1704061

31.

Forslund

M V.

, Arango-Lasprilla

, Roe

, et al. Multi-Level Modelling of Employment Probability Trajectories and Employment Stability at 1, 2 and 5 Years after Traumatic Brain Injury. Brain Inj, 2014; 28(7):980–986; doi: 10.3109/02699052.2014.888770

32.

Hoofien

, Vakil

, Gilboa

, et al. Comparison of the Predictive Power of Socio-Economic Variables, Severity of Injury and Age on Long-Term Outcome of Traumatic Brain Injury: Sample-Specific Variables versus Factors as Predictors. Brain Inj, 2002; 16(1):9–27; doi: 10.1080/02699050110088227

33.

Ulfarsson

, Lundgren-Nilsson Å, Blomstrand

, et al. A History of Unemployment or Sick Leave Influences Long-Term Functioning and Health-Related Quality-of-Life after Severe Traumatic Brain Injury. Brain Inj, 2014; 28(3):328–335; doi: 10.3109/02699052.2013.865274

34.

Wade

, Kaizar

, Narad

, et al. Behavior Problems Following Childhood TBI: The Role of Sex, Age, and Time since Injury. J Head Trauma Rehabil, 2020; 35(5):E393–E404; doi: 10.1097/HTR.0000000000000567

35.

Ryan

, Anderson

, Godfrey

, et al. Predictors of Very-Long-Term Sociocognitive Function after Pediatric Traumatic Brain Injury: Evidence for the Vulnerability of the Immature “Social Brain.”. J Neurotrauma, 2014; 31(7):649–657; doi: 10.1089/neu.2013.3153

36.

Selassie

, Cao

, Church

, et al. Accelerated Death Rate in Population-Based Cohort of Persons with Traumatic Brain Injury. Journal of Head Trauma Rehabilitation, 2014; 29(3); doi: 10.1097/HTR.0b013e3182976ad3

37.

Arango-Lasprilla

, Zeldovich

, Olabarrieta-Landa

, et al. Early Predictors of Employment Status One Year Post Injury in Individuals with Traumatic Brain Injury in Europe. J Clin Med, 2020; 9(6):2007; doi: 10.3390/jcm9062007

38.

Huang

, Chang

, Escorpizo

, et al. WHODAS 2.0 Can Predict Institutionalization among Patients with Traumatic Brain Injury. Int J Environ Res Public Health, 2019; 16(9):1484; doi: 10.3390/ijerph16091484

39.

Ketchum

, Getachew

, Krch

, et al. Early Predictors of Employment Outcomes 1 Year Post Traumatic Brain Injury in a Population of Hispanic Individuals. NeuroRehabilitation, 2012; 30(1):13–22; doi: 10.3233/NRE-2011-0723

40.

Keyser-Marcus

, Bricout

, Wehman

, et al. Acute Predictors of Return to Employment after Traumatic Brain Injury: A Longitudinal Follow-Up. Arch Phys Med Rehabil, 2002; 83(5):635–641; doi: 10.1053/apmr.2002.31605

41.

Machamer

, Temkin

, Fraser

, et al. Stability of Employment after Traumatic Brain Injury. J Int Neuropsychol Soc, 2005; 11(7):807–816; doi: 10.1017/S135561770505099X

42.

Catroppa

and Anderson

. Recovery and Predictors of Intellectual Ability Two Years Following Paediatric Traumatic Brain Injury. Neuropsychol Rehabil, 2003; 13(5):517–536; doi: 10.1080/09602010343000084

43.

Catroppa

, Godfrey

, Rosenfeld J

, et al. Functional Recovery Ten Years after Pediatric Traumatic Brain Injury: Outcomes and Predictors. J Neurotrauma, 2012; 29(16):2539–2547; doi: 10.1089/neu.2012.2403

44.

Cohen-Zimerman

, Kachian

, Krueger

, et al. Childhood Socioeconomic Status Predicts Cognitive Outcomes across Adulthood Following Traumatic Brain Injury. Neuropsychologia, 2019; 124:1–8; doi: 10.1016/j.neuropsychologia.2019.01.001

45.

Donders

and Nesbit-Greene

. Predictors of Neuropsychological Test Performance after Pediatric Traumatic Brain Injury. Assessment, 2004; 11(4):275–284; doi: 10.1177/1073191104268914

46.

Heffernan

, Vera

, Monaghan

, et al. Impact of Socioethnic Factors on Outcomes Following Traumatic Brain Injury. J Trauma, 2011; 70(3):527–534; doi: 10.1097/TA.0b013e31820d0ed7

47.

McQuistion

, Zens

, Jung

, et al. Insurance Status and Race Affect Treatment and Outcome of Traumatic Brain Injury. J Surg Res, 2016; 205(2):261–271; doi: 10.1016/j.jss.2016.06.087

48.

Nuño

, Pelissier

, Varshneya

, et al. Outcomes and Factors Associated with Infant Abusive Head Trauma in the US. Journal of Neurosurgery: Pediatrics, 2015; 16(5):515–522; doi: 10.3171/2015.3.PEDS14544

49.

Dunn

, Henry J and Beard

. Social Deprivation and Adult Head Injury: A National Study. J Neurol Neurosurg Psychiatry, 2003; 74(8):1060–1064; doi: 10.1136/jnnp.74.8.1060

50.

Humphries

, Ingram

, Sinha

, et al. The Effect of Socioeconomic Deprivation on 12 Month Traumatic Brain Injury (TBI) Outcome. Brain Inj, 2020; 34(3):343–349; doi: 10.1080/02699052.2020.1715481

51.

Kivimäki

, Batty

, Pentti

, et al. Association between Socioeconomic Status and the Development of Mental and Physical Health Conditions in Adulthood: A Multi-Cohort Study. Lancet Public Health, 2020; 5(3):e140–e149; doi: 10.1016/S2468-2667(19)30248-8

52.

Oakes

and Rossi

. The Measurement of SES in Health Research: Current Practice and Steps toward a New Approach. Soc Sci Med, 2003; 56(4):769–784; doi: 10.1016/S0277-9536(02)00073-4

53.

Asikainen

, Kaste M and Sarna

. Patients with Traumatic Brain Injury Referred to a Rehabilitation and Re-Employment Programme: Social and Professional Outcome for 508 Finnish Patients 5 or More Years after Injury. Brain Inj, 1996; 10(12):883–900; doi: 10.1080/026990596123864

54.

Ponsford

, Olver

, Curran

, et al. Prediction of Employment Status 2 Years after Traumatic Brain Injury. Brain Inj, 1995; 9(1):11–20; doi: 10.3109/02699059509004566

55.

Kool

, Chelimo C and Ameratunga

. Head Injury Incidence and Mortality in New Zealand over 10 Years. Neuroepidemiology, 2013; 41(3–4):189–197; doi: 10.1159/000354782

56.

Steward

, Kennedy

, Novack

, et al. The Role of Cognitive Reserve in Recovery from Traumatic Brain Injury. J Head Trauma Rehabil, 2018; 33(1):E18–E27; doi: 10.1097/HTR.0000000000000325

57.

Donders

and Stout

. The Influence of Cognitive Reserve on Recovery from Traumatic Brain Injury. Arch Clin Neuropsychol, 2019; 34(2):206–213; doi: 10.1093/ARCLIN/ACY035.

58.

, Lane

, Pan

, et al. Racial and Socioeconomic Disparities in Hip Fracture Care. J Bone Joint Surg (American volume), 2016; 98(10):858; doi: 10.2106/JBJS.15.00676

59.

Van Der Waerden

JEB

, Hoefnagels C and Hosman

CMH

. Psychosocial Preventive Interventions to Reduce Depressive Symptoms in Low-SES Women at Risk: A Meta-Analysis. J Affect Disord, 2011; 128(1–2):10–23; doi: 10.1016/J.JAD.2010.02.137

60.

Fahlman

, McCaughtry

, Martin

, et al. Racial and Socioeconomic Disparities in Nutrition Behaviors: Targeted Interventions Needed. J Nutr Educ Behav, 2010; 42(1):10–16; doi: 10.1016/J.JNEB.2008.11.003

61.

Hocking

Occupational Therapists Driving Societal Change. Br J Occup Ther, 2014; 77(12):591–592

62.

Lucyk

and McLaren

. Taking Stock of the Social Determinants of Health: A Scoping Review. PLoS One, 2017; 12(5); doi: 10.1371/JOURNAL.PONE.0177306

63.

Shukla

, Devi BI and Agrawal

. Outcome Measures for Traumatic Brain Injury. Clin Neurol Neurosurg, 2011; 113(6):435–441; doi: 10.1016/J.CLINEURO.2011.02.013

64.

Girasek

and Taylor

. An Exploratory Study of the Relationship between Socioeconomic Status and Motor Vehicle Safety Features. Traffic Inj Prev, 2010; 11(2):151–155; doi: 10.1080/15389580903531598

65.

Voas

, Romano E and Peck

. Validity of Surrogate Measures of Alcohol Involvement When Applied to Nonfatal Crashes. Accid Anal Prev, 2009; 41(3):522–530; doi: 10.1016/J.AAP.2009.02.004

66.

Hale

, Bohnert

, Grekin

, et al. Traumatic Brain Injury in the General Population: Incidence, Mental Health Comorbidity, and Functional Impact. J Nerv Mental Dis, 2019; 207(1):38–42; doi: 10.1097/NMD.0000000000000915

67.

Pearce

, Li

, Abbas

, et al. Does the Home Environment Influence Inequalities in Unintentional Injury in Early Childhood? Findings from the UK Millennium Cohort Study. J Epidemiol Commun Health, 2012; 66(2):181–188; doi: 10.1136/JECH.2011.139626

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.10 MB

0.12 MB