Abstract
Abstract
Objectives
To describe and compare injury severity, clinical management and mortality among haemorrhage, traumatic brain injury (TBI) and combined haemorrhage+TBI cohorts in a South African trauma population.
Design
Secondary analysis of data from the Epidemiology and Outcomes of Prolonged Trauma Care (EpiC) multicentre observational cohort study, April 2021 to December 2024.
Setting
Six government health facilities and affiliated government ambulances across the Western Cape, South Africa.
Participants
18 840 adult patients with blunt and/or penetrating traumatic injuries enrolled in the EpiC registry were eligible for this secondary analysis. Patients <18 years of age and prisoners are excluded from EpiC. Burn injuries, non-haemorrhage and non-TBI injuries were excluded from this analysis. Patients were assigned to one of three mutually exclusive cohorts (haemorrhage, TBI or haemorrhage+TBI) using anatomic and physiologic criteria.
Primary and secondary outcome measures
The primary outcome was crude, all-cause prehospital or in-hospital mortality. Mantel-Haenszel risk ratios (RRs) for mortality were adjusted for injury severity, injury force type and mode of prehospital transportation. Secondary outcomes included critical interventions, hospital admission, length of stay and time from injury to death.
Results
A total of 7614 patients met inclusion: 5146 (67.6%) haemorrhage, 1556 (20.4%) TBI and 912 (12.0%) haemorrhage+TBI. The majority were male (85.1%) with a median age of 31 years. Patients with haemorrhage+TBI received more intensive care, including airway interventions (38%), transfusions (41%) and intensive care unit admissions (36%). Crude mortality was highest in haemorrhage+TBI (31.3%), followed by TBI (7.7%) and haemorrhage (5.4%). After injury-pattern specific severity adjustment, mortality risk remained significantly higher in haemorrhage+TBI than in TBI (RR 1.87, 95% CI 1.60 to 2.19) and haemorrhage (RR 3.38, 95% CI 2.96 to 3.85). Estimates were similar after additional adjustment for injury force and prehospital transportation (RR 1.84 and 2.98).
Conclusions
In this South African population, haemorrhage+TBI was the least prevalent but most high-risk cohort with greater morbidity and mortality compared with haemorrhage or TBI alone, even after adjusting for injury severity. These findings may inform system level planning and resource allocation particularly in resource-limited settings.
Keywords: Africa South of the Sahara, Brain Injuries, EPIDEMIOLOGY, Mortality, TRAUMA MANAGEMENT
STRENGTHS AND LIMITATIONS OF THIS STUDY.
This study leverages a large, prospective multicentre injured patient cohort with standardised data collection by trained on-site research clinicians, strengthening the reliability and validity of the patient-level data.
The study is conducted in the Western Cape of South Africa, a setting with hybrid characteristics of both high-income and low-income health systems, offering findings with relevance to a range of contexts.
As a secondary analysis of prospectively collected observational data, the study is limited to variables available in clinical records, precluding capture of interventions and clinically relevant factors that are not routinely documented.
Operational definitions for haemorrhage, traumatic brain injury and combined haemorrhage+traumatic brain injury, as well as injury-pattern-specific severity stratifications, were systematically developed for the parent study using established trauma literature while incorporating physiologic and anatomic criteria applicable to resource-constrained settings where imaging and laboratory diagnostics may be limited.
Unmeasured survivor bias is an inherent limitation, as patients sustaining the most severe injuries may die prior to health system contact and are therefore not captured, potentially underestimating true injury burden and mortality.
Introduction
Trauma is a leading cause of death and disability worldwide, causing 4.4 million deaths annually of which 90% occur in low- and middle-income countries (LMICs).1 Haemorrhage and traumatic brain injury (TBI) are consistently identified as the top causes of trauma mortality.2
Haemorrhage is a dominant cause of early trauma deaths and is often preventable with timely interventions.2 3 In contrast, TBI deaths occur later and are influenced by access to diagnostic imaging and neurosurgical care, leading to mortality rates threefold to fourfold higher in LMICs than high-income countries.4 Patients with concomitant extracranial haemorrhage and TBI are at particularly high risk for poor outcomes given dual insults with competing physiology and management priorities. As an example, the accepted practice of permissive hypotension5 for patients with isolated traumatic haemorrhage must be reconsidered in patients with concurrent TBI who rely on a maintained cerebral perfusion pressure; reductions in systolic blood pressures and shock states are known to increase TBI mortality and worsen neuro-outcomes.6–11
While studies have established the global prevalence and acuity of haemorrhage2 3 and separately TBI,2 4 literature describing the epidemiology and outcomes of patients with concomitant haemorrhage+TBI is sparse, particularly in LMICs. This gap reflects both limited trauma registry infrastructure for surveillance and data collection in LMICs12 and the historical reliance on definitions, trauma scoring systems, and evidence derived from high-income settings, which may not be valid or contextually appropriate in resource-constrained settings.13 It is well-established that high-quality, standardised emergency care delivered in context-specific manners reduces morbidity and mortality across both traumatic and medical conditions.14 15 To effectively advance trauma care and strengthen LMIC health systems, locally grounded research is essential to clarify the burden of high-risk injuries, including haemorrhage+TBI. Such evidence can directly inform ambulance prioritisation, prognostication tools, resource allocation policies, and early management strategies for critically injured patients. Embedding local data into trauma guidelines and system design strengthens coordination across prehospital and facility-based care, ultimately improving patient outcomes.
South Africa, particularly the Western Cape, has one of the world’s highest incidences of trauma and post-injury mortality.16 The Western Cape health system has features that resemble higher-income systems with blood products and specialty care and lower-income systems characterised by tiered facilities with variable access to diagnostic tests, serving as a valuable setting to advance scientific knowledge regarding the optimal prioritisation and management of trauma patients. The Epidemiology and Outcomes of Prolonged Trauma Care (EpiC) study is a multicentre, prospective observational cohort study in this region that has enrolled >22 000 adult trauma patients since 2021.17 Findings from the EpiC study are actively driving emergency systems strengthening activities in the Western Cape, and regional experts have called for more data on haemorrhage and TBI to accelerate evidence-based reform.18 19
This study describes the definitions, prevalence, clinical characteristics, critical interventions and mortality across three key trauma cohorts—haemorrhage, TBI and haemorrhage+TBI—within the EpiC study population. Findings will inform trauma system enhancements and improve care delivery within South Africa and offer unique insights for other LMICs facing similar injury patterns and health system limitations.
Methods
This descriptive study used a secondary analysis of patient-level data collected from the EpiC study17 between April 2021 and December 2024. The EpiC study enrolls adult (age ≥18) patients with trauma, including haemorrhage, TBI, drownings, hangings, high-risk burns or wounds, complicated orthopaedic injuries and multisystem injuries. Adult patients are eligible for the EpiC study if they receive care from participating government ambulances and/or at least one of six government health facilities (Delft, Khayelitsha Site B, Ceres, Khayelitsha, Worcester, Tygerberg). Patients are enrolled if they present within 24 hours of an injury and have any one of the following criteria: Abbreviated Injury Scale20 (AIS) severity ≥2 in any body region, Triage Early Warning Score21 ≥5, need for damage control surgery or intensive care unit (ICU)-level resuscitation, hospital admission for traumatic injury or death. Patients not meeting these thresholds may still qualify based on clinical findings of TBI, haemorrhage or high-risk wounds or fractures. Patients are excluded from EpiC if they are <18 years of age, prisoners or have a simple isolated orthopaedic injury of the extremity or isolated soft tissue injury (figure 1).
Figure 1. Flow chart of patient eligibility, inclusion/exclusion and categorisation. EC, emergency centre; EMS, emergency medical services; HEMS, helicopter emergency medical services; SATS, South African Triage Scale; SATS, South African Triage Scale; TBI, traumatic brain injury; TEWS, Triage Early Warning Score.*Includes drownings, hangings, high-risk or complex wounds, complicated orthopedic injuries, or patients missing data for cohort assignment.
This secondary analysis included screened and eligible EpiC patients who experienced extracranial haemorrhage, TBI or concurrent haemorrhage+TBI from blunt and/or penetrating injury forces. Non-haemorrhage/TBI injuries (eg, burns, high-risk wounds, complicated orthopaedic injuries) were excluded from this analysis. Missingness for individual variables among both included and excluded is presented in online supplemental table 1.
Operational definitions for each injury type in our inclusion criteria were created for the EpiC study to guide research methodology and capture the predominant injury driving early management strategies. Definitions for haemorrhage and TBI were informed by trauma literature.22–24 Thresholds for AIS severity, diagnostic tests and/or clinical parameters were used to assign patients to one of three primary cohorts that were mutually exclusive—assignments were hierarchical, first assigning patients to haemorrhage+TBI if they met criteria for both injury types, then isolated TBI, then isolated haemorrhage. A combination of physiologic and anatomic criteria was employed to fit the epidemiological and clinical realities of the Western Cape, including limited access to early imaging and laboratory diagnostics. Severity subgroups of mild, moderate and severe for haemorrhage and TBI were defined using established criteria specific to each injury pattern.22–24 Patients meeting any single parameter for a severe injury were assigned severe, then evaluated for and assigned moderate if applicable. Patients without any moderate or severe indicators were assigned to the mild subgroup.
The haemorrhage cohort was defined as non-head/face AIS severity score ≥1 with at least one abnormality in physiologic or clinical parameters at initial presentation. Patients with a head AIS severity score ≥3 were excluded from this cohort. Haemorrhage severity was determined based on thresholds for vitals, labs, medications and transfusion volumes (table 1).
Table 1. Criteria for mild, moderate and severe haemorrhage subgroups.
| Haemorrhage cohort definition |
|---|
| Non-head/face AIS severity score ≥1 with at least one of the following on initial presentation: systolic blood pressure ≤100, shock index ≥1.0, lactate ≥4.0, base deficit ≥6, pH ≤7.2, Sequential Organ Failure Assessment cardiovascular component score ≥3, administration of medications for haemodynamic support (tranexamic acid, vasopressors, inotropes or blood products) and/or a haemorrhage-related diagnosis. |
| Haemorrhage severity stratification* | |||
|---|---|---|---|
| Criteria | Mild | Moderate | Severe |
| Systolic blood pressure | >90 mm Hg | >70 to ≤ 90 mm Hg | ≤70 mm Hg |
| Shock Index | ≥0.7 to <1.2 | ≥1.2 to <1.6 | ≥1.6 |
| Base deficit | <6 mmol/L | ≥6 to <10 mmol/L | ≥10 mmol/L |
| Lactate | --- | --- | ≥10 mmol/L |
| Haemoglobin | >70 g/L | >50 to ≤70 g/L | ≤50 g/L |
| Packed red blood cells | <3 units | ≥3 to <5 units | ≥5 units |
| Whole blood | 0 units | >0 to <3 units | ≥3 units |
| Fresh frozen plasma | 0 units | >0 to <3 units | ≥3 units |
| Freeze dried plasma | <2 units | ≥2 to <4 units | ≥4 units |
| Platelets | 0 units | ≤1 unit | >1 unit |
| Cryoprecipitate | 0 units | ≤1 unit | >1 unit |
| Vasopressors | --- | --- | ≥1 dose/24 hours |
| Inotropes | --- | --- | ≥1 dose/24 hours |
Severity assignment: Patients meeting at least one severe criterion across any of the parameters were classified as severe regardless of values on other parameters. Patients meeting at least one moderate criterion (and no severe criteria) were classified as moderate. The remainder of patients were classified as mild if they met neither moderate nor severe criteria.
AIS, Abbreviated Injury Scale.
The TBI cohort was defined as head AIS severity score ≥2. Patients with a non-head AIS severity score ≥3 were excluded from this cohort. TBI severity was determined using Glasgow Coma Scale (GCS) scores and CT (computed tomography) brain imaging findings (table 2).
Table 2. Criteria for mild, moderate and severe TBI subgroups.
| TBI cohort definition |
|---|
| Head AIS severity score ≥2 |
| TBI severity stratification* | |||
|---|---|---|---|
| Criteria* | Mild | Moderate | Severe |
| GCS Score | 13–15 | 9–12 | ≤8 |
| Abnormal CT brain imaging severity=severe | + | ||
| Abnormal CT brain imaging severity=m oderate | + | ± | |
| Abnormal CT brain imaging severity | ||
|---|---|---|
| Findings on CT brain imaging | Abnormal CT brain: moderate | Abnormal CT brain: severe |
| Penetrating brain parenchymal injury | X | |
| Intracranial haemorrhage (epidural, subdural, subarachnoid, intraparenchymal and/or intraventricular) | X | |
| Traumatic axonal injury/diffuse axonal injury | X | |
| Midline shift | ≤5 mm | >5 mm |
| Cerebral oedema | X | |
| Hydrocephalus/ventriculomegaly | X | |
| Basal cistern compression | X | |
| Herniation | X | |
| Hemispheres with traumatic injury | Unilateral or Bilateral | |
±Means may be present in the cohort but not used to exclusively define the cohort.
Severity assignment: Patients were classified as severe if they had a GCS ≤8 and/or a severe abnormal CT brain finding, regardless of the value of the other parameters. If neither severe criterion was met, patients were classified as moderate if they had a GCS 9–12 and/or a moderate abnormal CT brain finding. The remainder of patients were classified as mild if they met neither moderate nor severe criteria.
AIS, Abbreviated Injury Scale; CT, computed tomography; GCS, Glasgow Coma Scale; TBI, traumatic brain injury.
Patients in the haemorrhage+TBI cohort met inclusion criteria for both haemorrhage and TBI. Severity was defined by the combinations shown in table 3.
Table 3. Criteria for haemorrhage+TBI severity subgroups.
| Mild haemorrhage | Moderate haemorrhage | Severe haemorrhage | |
|---|---|---|---|
| Mild TBI | Mild | Mild | Moderate |
| Moderate TBI | Mild | Moderate | Severe |
| Severe TBI | Moderate | Severe | Severe |
TBI, traumatic brain injury.
Data included demographics (age, sex, geographic location), injury characteristics (force, mechanism), ambulance transports (primary transport, inter-facility transport to higher level of care), anatomic scores (AIS, New Injury Severity Score25 (NISS)), vitals, laboratory results, clinical interventions (critical procedures, medications, blood products, surgery), and outcomes (lengths of stay, Sequential Organ Failure Assessment (SOFA) scores, mortality, time to death). Data were abstracted from clinical records by trained, on-site research clinicians.
Descriptive statistics were generated for all variables: medians and interquartile ranges (IQRs) for continuous variables and frequencies and percentages for categorical variables. Analyses used variable-specific available data, accounting for missingness among physiologic/laboratory data. The primary outcome was crude, all-cause, prehospital or in-hospital mortality for each cohort. Characteristics and outcomes were reported for each cohort as a whole and for the higher-risk moderate-to-severe subgroups. Additional descriptive statistics were generated for each cohort by injury force type (any penetrating injury vs blunt-only injury).
Mantel-Haenszel methods were used to calculate severity-adjusted mortality risk ratios (RRs) comparing respective cohorts. We conducted three sets of adjusted analyses: First, we stratified our population by injury-pattern-specific definitions of severity (mild, moderate and severe as defined in tables 1 and 2) that account for both anatomic and physiologic criteria. Second, we stratified our population by the cross-classification of these injury-pattern-specific definitions of severity with injury force type, dichotomised into any penetrating injuries versus blunt only, and whether or not the patient received primary emergency medical services (EMS) transportation from the scene of injury. Third, we stratified our population by anatomic severity using NISS categorised as mild or moderate (NISS ≤15), severe (NISS 16–25) or profound (NISS ≥26). Analyses were performed using SAS V.9.4 (2023, SAS Institute) and R V.4.5.1 (2025-06-13).
Patient and public involvement
Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Results
Of 18 840 patients enrolled in EpiC during the study period, 7614 (40.4%) met inclusion for this analysis (figure 1). The majority were young (median age 31.3 (25.5, 38.3) years) males (6477, 85.1%) injured in urban areas (4555, 59.8%). 912 (12.0%) met criteria for haemorrhage+TBI, 1556 (20.4%) TBI and 5146 (67.6%) haemorrhage. Haemorrhage patients predominantly sustained penetrating injuries (3732, 72.5%) from stabbings (2906, 56.5%). TBI patients mostly experienced blunt trauma (1130, 72.6%) from strikes (880, 56.6%). More than half of the haemorrhage+TBI cohort (548, 60.1%) received EMS transport to the first health facility with a median time from injury to EMS arrival on scene of 37 min (23, 86) (table 4).
Table 4. Demographic, injury, ambulance, anatomical and physiological factors by EpiC cohort.
| Haemorrhage+TBI | TBI | Haemorrhage | ||||
|---|---|---|---|---|---|---|
| All N=912 |
Mod/Sev N=602 |
All N=1556 |
Mod/Sev N=878 |
All N=5146 |
Mod/Sev N=2743 |
|
| Demographics | ||||||
| Age, years | 32 (26.2, 38.1) | 31.9 (25.7, 37.5) | 32.3 (26.1, 40.2) | 32.3 (26.1, 40.6) | 30.8 (25.4, 37.9) | 31.2 (25.6, 38.4) |
| Sex (male) | 787 (86.3) | 519 (86.2) | 1368 (87.9) | 793 (90.3) | 4322 (84) | 2358 (86) |
| Urban injury location | 688 (75.4) | 480 (79.7) | 953 (61.2) | 659 (75.1) | 2914 (56.6) | 1653 (60.3) |
| Injury | ||||||
| Injury force | ||||||
| Blunt | 606 (66.4) | 426 (70.8) | 1130 (72.6) | 655 (74.6) | 1153 (22.4) | 537 (19.6) |
| Penetrating | 181 (19.8) | 112 (18.6) | 256 (16.5) | 145 (16.5) | 3732 (72.5) | 2063 (75.2) |
| Blunt+penetrating | 125 (13.7) | 64 (10.6) | 170 (10.9) | 78 (8.9) | 261 (5.1) | 143 (5.2) |
| Injury mechanism | ||||||
| Fall | 11 (1.2) | 10 (1.7) | 99 (6.4) | 59 (6.7) | 116 (2.3) | 39 (1.4) |
| Firearm | 84 (9.2) | 63 (10.5) | 98 (6.3) | 81 (9.2) | 917 (17.8) | 534 (19.5) |
| Stabbing or cut | 150 (16.4) | 72 (12.0) | 205 (13.2) | 83 (9.5) | 2906 (56.5) | 1582 (57.7) |
| Struck/hit | 345 (37.8) | 222 (36.98.2) | 880 (56.6) | 496 (56.5) | 649 (12.6) | 338 (12.3) |
| Vehicular injury | 316 (34.6) | 230 (38.2) | 257 (16.5) | 147 (16.7) | 504 (9.8) | 233 (8.5) |
| Other/unknown | 6 (0.7) | 5 (0.8) | 17 (1.1) | 12 (1.4) | 54 (1.0) | 17 (0.6) |
| Ambulance | ||||||
| Primary transport | 548 (60.1) | 398 (66.1) | 662 (42.5) | 381 (43.4) | 2213 (43.0) | 1184 (43.2) |
| Injury to EMS arrival* (mins) | 37 (23, 86) |
34 (21, 76) |
51 (26, 135) |
38 (24, 117) |
52 (28, 120) |
47 (27, 120) |
| Inter-facility transport to higher level | 468 (51.3) | 282 (46.8) | 818 (52.6) | 552 (62.9) | 1716 (33.3) | 984 (35.9) |
| Anatomical | ||||||
| NISS | 27 (17, 38) | 34 (22, 43) | 16 (8, 27) | 25 (17, 29) | 9 (3, 17) | 11 (3, 22) |
| Maximum AIS | ||||||
| Head/face | 3 (3, 4) | 3 (3, 5) | 3 (2, 3) | 3 (3, 4) | 1 (1, 1) | 1 (1, 1) |
| Neck/torso | 3 (1, 3) | 3 (2, 4) | 1 (1, 2) | 1 (1, 2) | 3 (1, 3) | 3 (2, 4) |
| Extremity | 2 (1, 3) | 2 (1, 3) | 1 (1, 2) | 1 (1, 2) | 1 (1, 3) | 1 (1, 3) |
| External | 1 (1, 1) | 1 (1, 1) | 1 (1, 1) | 1 (1, 1) | 1 (1, 1) | 1 (1, 1) |
| Physiological | ||||||
| SBP†, mm Hg | 92 (74, 108) | 84 (52, 101) | 118 (106,131) | 117 (105, 130) | 96 (83, 106) | 84 (71.5, 92) |
| Shock index† | 1 (0.8, 1.2) | 1 (0.8, 1.3) | 0.8 (0.6, 0.9) | 0.8 (0.6, 0.9) | 1 (0.8, 1.2) | 1.1 (0.9, 1.3) |
| GCS† | 11 (3, 15) | 3.5 (3, 13) | 15 (13, 15) | 14 (9, 15) | 15 (15, 15) | 15 (14, 15) |
| Haemoglobin‡, g/L | 113 (97, 129) | 109 (93, 125) | 126 (113, 140) | 125 (112, 138) | 120 (105, 134) | 115 (99, 130) |
| Lactate§, mmol/L | 4.5 (2.7, 7.7) | 5.3 (3.1, 9.2) | 2.6 (1.5, 4.1) | 2.6 (1.6, 4) | 4.2 (2.5, 6.5) | 4.9 (3, 8.1) |
| pH§ | 7.3 (7.2, 7.4) | 7.3 (7.2, 7.3) | 7.4 (7.3, 7.4) | 7.3 (7.3, 7.4) | 7.3 (7.3, 7.4) | 7.3 (7.2, 7.3) |
| Base deficit§, mmol/L | 6.3 (3.4, 10.9) | 7.8 (4.4, 12.9) | 3 (1.6, 6.1) | 3.2 (1.7, 6) | 5.1 (2.5, 8.7) | 6.8 (3.8, 10.3) |
| Circulatory diagnosis | 295 (32.3) | 248 (41.2) | 84 (5.4) | 60 (6.9) | 1394 (27.1) | 1172 (42.7) |
Median (Q1, Q3) is presented for numeric variables and n (%) for categorical variables.
‘Mod/Sev’ includes patients classified as ‘moderate’ or ‘severe’ within each respective cohort.
Injury to EMS arrival (mins) only considers patients who were transported by EMS (excludes patients who self-presented to a health facility).
Worst recorded among primary EMS and first facility.
Worst among the first 24 hours.
First lab values.
AIS, Abbreviated Injury Scale; EMS, emergency medical services; GCS, Glasgow Coma Scale; NISS, New Injury Severity Score; SBP, systolic blood pressure; TBI, traumatic brain injury.
Anatomically, haemorrhage patients exhibited more severe injuries in the neck/torso (median AIS=3 (1, 3)). Haemorrhage+TBI patients exhibited a combination of severe injuries across the head/face (median AIS=3 (3, 4)) and neck/torso (median AIS=3 (1, 3)). Median NISS was highest in haemorrhage+TBI (27 (17, 38)). Physiologically, haemorrhage+TBI patients had the lowest median SBP (92 mm Hg (74, 108)), highest median lactate (4.5 (2.7, 7.7)) and lowest median GCS (11 (3, 15)) (table 4).
Regarding interventions, approximately one in three haemorrhage+TBI patients received airway (3467, 37.9%), breathing (408, 44.7%), circulation (281, 30.8%) or blood product (378, 41.4%) interventions. Patients with haemorrhage had the most surgeries (799, 15.5%) (table 5).
Table 5. Critical interventions and outcomes by EpiC cohort.
| Haemorrhage+TBI | TBI | Haemorrhage | ||||
|---|---|---|---|---|---|---|
| All N=912 |
Mod/Sev N=602 |
All N=1556 |
Mod/Sev N=878 |
All N=5146 |
Mod/Sev N=2743 |
|
| Interventions within 24 hours | ||||||
| Airway | 346 (37.9) | 332 (55.1) | 177 (11.4) | 167 (19.0) | 335 (6.5) | 309 (11.3) |
| Breathing | 408 (44.7) | 345 (57.3) | 124 (8) | 109 (12.4) | 1768 (34.4) | 1148 (41.9) |
| Circulation | 281 (30.8) | 220 (36.5) | 166 (10.7) | 104 (11.8) | 1113 (21.6) | 779 (28.4) |
| Disability | 76 (8.3) | 58 (9.6) | 52 (3.3) | 23 (2.6) | 128 (2.5) | 65 (2.4) |
| Blood products | 378 (41.4) | 300 (49.8) | 141 (9.1) | 108 (12.3) | 1411 (27.4) | 916 (33.4) |
| Surgery | 71 (7.8) | 59 (9.8) | 65 (4.2) | 62 (7.1) | 799 (15.5) | 566 (20.6) |
| TXA | 238 (26.1) | 164 (27.2) | 85 (5.5) | 56 (6.4) | 1205 (23.4) | 735 (26.8) |
| Cardiovascular support | 157 (17.2) | 152 (25.2) | 25 (1.6) | 22 (2.5) | 294 (5.7) | 285 (10.4) |
| Outcomes | ||||||
| Hospital length of stay, days | 4.4 (1.2, 12.1) | 4.8 (0.9, 13.9) | 1.7 (0.9, 5) | 3.2 (1.4, 8.4) | 2.7 (0.7, 6.3) | 3.4 (0.9, 7.5) |
| ICU admission | 327 (35.9) | 296 (49.2) | 204 (13.1) | 192 (22.1) | 504 (9.8) | 436 (15.9) |
| ICU length of stay, days* | 2.6 (0.7, 7.7) | 2.5 (0.7, 7.7) | 1.8 (0.8, 5.1) | 1.7 (0.8, 4.9) | 2.4 (0.9, 5.1) | 2.6 (1, 5.2) |
| Severe organ dysfunction† | 437 (47.9) | 400 (66.4) | 219 (14.1) | 206 (23.7) | 699 (13.6) | 630 (23.0) |
| Died during EMS transport | 11 (1.2) | 11 (1.8) | 4 (0.2) | 4 (0.5) | 11 (0.2) | 11 (0.4) |
| Death (all causes) | 285 (31.3) | 268 (44.5) | 120 (7.7) | 115 (13.1) | 278 (5.4) | 255 (9.3) |
| Time: injury to death, hours | 20 (4.6, 99.9) | 18.8 (4.56, 88) | 34.6 (8.3, 92.2) | 34.4 (8.2, 90.7) | 12.3 (2.6,100.4) | 10 (2.3, 85.1) |
Median (Q1, Q3) is presented for numeric variables and n (%) for categorical variables.
‘Mod/Sev’ includes patients classified as ‘moderate’ or ‘severe’ within each respective cohort. Airway: chin lift/jaw thrust manoeuvre, suction, oropharyngeal/nasopharyngeal device, supraglottic airway, endotracheal tube and/or surgical airway.
Breathing: needle decompression, chest seal, thoracostomy, chest tube, positive ventilation and/or oxygen administration.
Circulation: CPR, cardiocentesis, pelvic binder, bleeding control, tourniquet and/or embolisation.
Disability: C-Collar for suspected spinal injury, C-Collar for any EMS GCS <13 and/or femur traction splint.
Cardiovascular support: SOFA cardiovascular ≥3, vasopressors, inotropes.
ICU length of stay (days) only considers patients that had an ICU visit.
Severe organ dysfunction defined as SOFA score ≥3 within the first 24 hours.
CPR, cardiopulmonary resuscitation; EMS, emergency medical services; GCS, Glasgow Coma Scale; ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment; TBI, traumatic brain injury; TXA, tranexamic acid.
Morbidity and mortality outcomes are presented in table 5. Haemorrhage+TBI had the highest proportion of ICU admissions (327, 35.9%) and longest ICU stays (median 2.6 days (0.7, 7.7)). Median duration from injury to death was longest for TBI (34.6 hours (8.3, 92.2)). Patients with haemorrhage+TBI experienced the highest crude mortality (285, 31.3%) compared with TBI (120, 7.7%) and haemorrhage (278, 5.4%). Regarding subgroups, moderate-to-severe haemorrhage+TBI patients experienced the most severe injuries (median NISS=34, (22, 43)), lowest median haemoglobin levels (109 (93, 125)) and lowest median GCS (3.5 (3, 13)). This subgroup had the highest mortality (44.5%, 268).
Injury characteristics and outcomes across all three cohorts stratified by injury force type (any penetrating injury vs blunt-only) are presented in online supplemental table 2. Within the TBI cohort, patients with penetrating injuries had more than double the crude mortality of those with blunt injury (13.4% vs 5.6%), a shorter median time to death (18.4 hours (2.9–43.5) vs 62.3 hours (22.6–145.7)) and a higher rate of death during EMS transport (2.3% vs 0.1%). Within the haemorrhage cohort, crude mortality was similar between mechanisms (5.2% in penetrating vs 6.1% in blunt), though penetrating injuries received more critical interventions (surgery: 18.1% vs 6.8%; tranexamic acid (TXA): 27.0% vs 10.8%; blood products: 30.5% vs 16.9%) and, when death occurred, did so more rapidly (median 5.9 hours (1.7–72.8) vs 45.2 hours (8.7–189.1)).
After adjusting for severity, as defined using the mild, moderate and severe definitions of injury-pattern-specific severity from tables 1 and 2, we observed a 1.87 (95% CI 1.60 to 2.19) times higher risk of mortality among haemorrhage+TBI patients compared with isolated TBI and a 3.38 (95% CI 2.96 to 3.85) times higher risk of mortality among haemorrhage+TBI patients compared with isolated haemorrhage (table 6). These associations persisted after we additionally stratified by injury force and mode of prehospital transport (haemorrhage+TBI vs TBI: RR=1.84, 95% CI 1.57 to 2.16; haemorrhage+TBI vs haemorrhage: RR=2.98, 95% CI 1.5, 3.48; table 6). When we adjusted for severity using NISS categories rather than injury-pattern-specific definitions, the RRs remained large in magnitude and significant (table 6). The crude and stratum-specific RRs underlying the Mantel-Haenszel adjusted RRs in table 6 are presented in online supplemental table 3. In all analyses, there was strong evidence of heterogeneity across strata with the largest RRs observed among lower acuity patients.
Table 6. Adjusted risk ratios comparing mortality among haemorrhage+TBI to isolated TBI or isolated haemorrhage.
| Haemorrhage+TBI vs isolated TBI (reference group) | Haemorrhage+TBI vs isolated haemorrhage (reference group) | |||
|---|---|---|---|---|
| N | MH aRR (95% CI) | N | MH aRR (95% CI) | |
| Adjusted using injury-pattern-severity strata | 2468 | 1.87 (1.60 to 2.19) | 6058 | 3.38 (2.96 to 3.85) |
| Adjusted using injury-pattern-severity strata, injury force type, and mode of prehospital transportation | 1971* | 1.84 (1.57 to 2.16) | 5364† | 2.98 (1.5 to 3.48) |
| Adjusted using NISS-based severity strata | 2465‡ | 2.33 (1.92 to 2.83) | 6058 | 1.86 (1.61 to 2.14) |
497 patients were excluded from the MH adjusted analysis due to positivity violations caused by zero observed deaths in at least one exposure category.
694 patients were excluded from the MH adjusted analysis due to positivity violations caused by zero observed deaths in at least one exposure category.
3 patients missing NISS scores and therefore excluded from this analysis.
CI, confidence interval; MH aRR, Mantel-Haenszel adjusted risk ratio; NISS, New Injury Severity Score; TBI, traumatic brain injury.
Discussion
In this South African trauma population, we describe three distinct trauma phenotypes relevant to acute care—haemorrhage, TBI and combined haemorrhage+TBI. The haemorrhage+TBI cohort experienced higher morbidity and mortality than the other cohorts. Within each cohort, moderate-to-severe injuries consistently—and expectedly—had worse physiologic perturbations, greater need for critical interventions and worse outcomes. Notably, almost half (44.5%) of the moderate-to-severe haemorrhage+TBI subgroup died.
Our findings align with literature demonstrating higher mortality among multisystem injured patients in similar settings; a cross-sectional study at a tertiary hospital in Ghana found 33% mortality among polytrauma patients,26 and a prospective cohort study of adult polytrauma patients at a tertiary referral hospital in Tanzania found a 7-day mortality of 20%.27 Crude mortality in our haemorrhage+TBI cohort was 31.3%, approximately fourfold higher than in isolated TBI (7.7%) and nearly sixfold higher than in isolated haemorrhage (5.4%). The mortality differences were mirrored in other measured outcomes: patients with haemorrhage+TBI exhibited the most severe physiologic derangements and required greater intensity of non-surgical emergency interventions, including airway management and transfusions. ICU admission was more common among haemorrhage+TBI patients (35.9%) than among those with isolated TBI (13.1%) or isolated haemorrhage (9.8%). These ICU admission proportions are comparable to or exceed those reported in similar resource-limited trauma settings, such as 14.2% in a Brazilian trauma cohort28 and 16.8% in a Tunisian trauma cohort,29 highlighting the morbidity underlying the elevated mortality risk observed specifically within the haemorrhage+TBI population. The relatively short ICU length of stay observed even among the most severely injured patients (median 2.6 days among haemorrhage+TBI) warrants cautious interpretation: given that time from injury to death was a median of 20 hours in this cohort, abbreviated ICU stays may partly reflect early in-ICU mortality among the most critically injured rather than rapid clinical stabilisation; distinguishing between these possibilities was beyond the scope of this descriptive study.
In all stratified, adjusted models, the risk of mortality among haemorrhage+TBI patients consistently remained approximately 2–3 times higher than among patients with isolated haemorrhage or TBI. Interestingly, the relative increase in mortality associated with multisystem trauma was even greater among lower acuity patients. This finding likely reflects the relatively low absolute risk of mortality in this group, but also suggests that even minor or moderate multisystem trauma can meaningfully increase mortality.
Collectively, these findings suggest that injury factors alone do not fully account for the elevated mortality observed in the combined haemorrhage+TBI cohort. While this study did not directly measure health system-level factors, other contributors to mortality in complex or multisystem trauma—including delayed recognition and triage of multisystem traumas, inefficient transfer processes to higher levels of care, and overburdened surgical resources—have been described in resource-limited trauma settings, including within this study’s own health system.18 19 30 31 Our findings characterise haemorrhage+TBI as a distinct, high-risk trauma phenotype warranting efficient identification and intentional resource allocation, particularly in high-volume, resource-constrained trauma settings.
Patients with TBI predominantly sustained blunt trauma from strikes/hits and vehicular injuries, consistent with injury patterns in LMICs.32 Although airway interventions were common, few other interventions were performed—even among moderate-to-severe TBI, less than 1 in 7 patients received non-airway interventions and only 1 in 14 underwent surgery. Despite receiving fewer procedures than other cohorts, the moderate-to-severe TBI subgroup had the highest proportion of transfers to higher levels of care (62.9%), second highest proportion of ICU admissions (22.1%), and longest median time to death (34.4 hours). TBI patients require vital health system resources like EMS transportation and critical care services that are often strained in LMICs,33 34 highlighting the need for system-level planning and triage protocols specifically for these patients.
Patients with haemorrhage experienced mostly penetrating injuries from stabbings followed by firearms—demonstrative of the global epidemic of interpersonal violence that is particularly damaging in LMICs.35 Those with moderate-to-severe haemorrhage had the most pronounced haemodynamic instability reflected in vital sign and laboratory abnormalities. Despite these derangements, this cohort had the lowest mortality. Haemorrhage deaths occurred earlier in the postinjury course, a finding well described in high-income and low-income settings alike.3 This low crude mortality may reflect survivor bias as patients with catastrophic bleeding may die rapidly prior to study enrollment. However, it may also reflect the life-saving impact of haemostatic resuscitative efforts. While interventions like direct pressure and tourniquets are unreported in this study, other haemorrhage interventions were frequently used in this cohort—among patients with moderate-to-severe haemorrhage, one in three received blood products, one in four received TXA and one in five underwent surgery.
This study aggregates and reports several important patient-level outcomes and health system resource consumption across three key trauma cohorts in a South African setting. While mortality data and post-mortem reviews draw attention to high-risk disease processes and preventable deaths, positing areas for facility-based or community-based interventions in clinical training and violence prevention,18 19 understanding clinical care patterns and specific resources required to meet the demands of different injuries can inform system-level planning and resource allocation, such as staffing models, blood bank capacity and critical care bed availability. A review of patient-level data from a Cameroonian trauma registry identified deficits in primary survey interventions, underscoring the need for provider training and protocols to identify and treat life threats.36 Similarly, our data on quantifiable resources consumed across each cohort could be combined with facility-level trauma burdens to inform facility resource allocation, trauma staffing models and health system budgeting within tiered health systems.
This secondary analysis has several limitations. Data are collected for the EpiC study, limited to variables available in clinical records. Residual confounding from unmeasured or incompletely measured factors cannot be excluded. Cohort definitions were informed by global trauma literature but adapted for local use given epidemiological differences and health system constraints. Thus, findings may not generalise to settings outside South Africa. Further analyses will compare our findings to global trauma literature to determine external validity and explore underlying drivers of differences in outcomes. Finally, unmeasured survivor bias may be present, particularly in this patient population that often arrives by private transport (55% in this cohort) and that experiences long intervals from injury to contact with the health system. Among patients transported by ambulance in a related study in this region, the median time from injury to ambulance arrival was 0.7 hours (0.3–1.5), consistent with findings from this secondary analysis (table 4), while patients who self-presented to a health facility did so with a median time from injury to facility arrival of 1.9 hours (1.9–2.6).37 Patients sustaining the most severe injuries are more likely to die early—potentially on scene prior to contact with an ambulance or before self-presentation at a participating health facility—and therefore are not captured within the EpiC study. Results may not generalise to settings with different ambulance utilisation or response patterns, and we cannot account for time to initial care and its influence on the outcomes observed here. While survivor bias likely affects all injury types, differences in patient characteristics and causes of death cannot be determined within the scope of this analysis.
Conclusions
Combined haemorrhage+TBI represents a high-risk trauma cohort, exhibiting the highest morbidity burden and highest crude and severity-adjusted mortality compared with haemorrhage and TBI cohorts within a South African population. Within each injury type, higher injury severity resulted in increased crude mortality, care intensity and resource use. These findings establish cohort-specific morbidity and mortality for an LMIC and may inform approaches to early risk stratification, escalation decisions and resource prioritisation in this and other similar trauma settings.
Supplementary material
Acknowledgements
We are thankful to Western Cape Government Health and Wellness for their support of the EpiC study and resulting substudies. We would like to thank and acknowledge the EpiC study research staff who abstracted data from medical records: Mandy Abels, Razia Amien, Lauren Barron, Raeesah Bray, Anthony Carr, Melany Codling, Zuleiga Davids, Ameerah Davids, Aneeka Domingo, Magzane Ghall, Chloe Ile, Daisy Inglis, Alana Jooste, Arfia Kamroodien-Cader, Gadija Khan, Farren Kleinsmith, Vusumzi Kose, Mudiwa Llobell, Jani Marais, Lindi Martin, Kelsey McIntyre, Minnie Minnaar, Fabio Moriera, Tanisha Morris, Tevarus Naicker, Anna Notten, Azia Parker, Justin Petersen, Tahrir Rasool, Astrid Samuels, Kirsten Scoble, Regan Smith, Anronel Stofberg, Andre Swanepoel, Dylan van Staden, Alexa van der Walt, Catherine Wandie, Shariefa Williams.
Footnotes
Funding: Epidemiology and Outcomes of Prolonged Trauma Care (EpiC) is funded by the US Department of Defence (award numbers W81XWH-19-20055 (to N-KM-M, JMD, SdV and HJL) and W81XWH-20-20042 (to N-KM-M, JMD, SdV and HJL).
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2026-120519).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Data availability free text: The datasets generated and/or analysed for this study are not publicly available due to contractual limitations from the grant sponsor and a data share contract. However, data can be made available on reasonable request to the principal investigator.
Collaborators: The “EpiC Study Site Collaborators” author group is comprised of: Mohammed Mayet, MBChB; Lesley Hodsdon, MBChB; L’Oreal Snyders, MBChB; Leigh Wagner, MBChB; Karlien Doubell, MBChB; Denise Lourens, MBChB. All are employees of Western Cape Government Health and Wellness, Cape Town, South Africa.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Contributor Information
EpiC Study Site Collaborators:
Mohammed Mayet, Lesley Hodsdon, L’Oreal Snyders, Leigh Wagner, Karlien Doubell, and Denise Lourens
Data availability statement
Data are available on reasonable request.
References
- 1.World Health Organization Injuries and violence. 2024
- 2.Callcut RA, Kornblith LZ, Conroy AS, et al. The why and how our trauma patients die: A prospective Multicentre Western Trauma Association study. J Trauma Acute Care Surg. 2019;86:864–70. doi: 10.1097/TA.0000000000002205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bernhardt IM, Moore EE, Sauaia A, et al. Timing of trauma deaths due to uncontrolled bleeding have not changed in three decades: A multicenter study of patients in hemorrhagic shock. Am J Surg. 2025;250:116510. doi: 10.1016/j.amjsurg.2025.116510. [DOI] [PubMed] [Google Scholar]
- 4.Naik A, Bederson MM, Detchou D, et al. Traumatic Brain Injury Mortality and Correlates in Low- and Middle-Income Countries: A Meta-Epidemiological Study. Neurosurgery. 2023;93:736–44. doi: 10.1227/neu.0000000000002479. [DOI] [PubMed] [Google Scholar]
- 5.Indorewala Y, Nasef Y, Jayagopi K, et al. Permissive hypotension in adult trauma: A systematic review of outcomes across clinical settings, injury type, and resuscitation strategies. Am J Emerg Med. 2026;105:150–7. doi: 10.1016/j.ajem.2026.04.021. [DOI] [PubMed] [Google Scholar]
- 6.McDonald SJ, Sun M, Agoston DV, et al. The effect of concomitant peripheral injury on traumatic brain injury pathobiology and outcome. J Neuroinflammation. 2016;13:90. doi: 10.1186/s12974-016-0555-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Leung LY, Wei G, Shear DA, et al. The Acute Effects of Hemorrhagic Shock on Cerebral Blood Flow, Brain Tissue Oxygen Tension, and Spreading Depolarization following Penetrating Ballistic-Like Brain Injury. J Neurotrauma. 2013;30:1288–98. doi: 10.1089/neu.2012.2715. [DOI] [PubMed] [Google Scholar]
- 8.McHugh GS, Engel DC, Butcher I, et al. Prognostic Value of Secondary Insults in Traumatic Brain Injury: Results from The IMPACT Study. J Neurotrauma. 2007;24:287–93. doi: 10.1089/neu.2006.0031. [DOI] [PubMed] [Google Scholar]
- 9.Wiles MD. Blood pressure in trauma resuscitation: “pop the clot” vs. “drain the brain”? Anaesthesia. 2017;72:1448–55. doi: 10.1111/anae.14042. [DOI] [PubMed] [Google Scholar]
- 10.Tobin JM, Dutton RP, Pittet JF, et al. Hypotensive resuscitation in a head-injured multi-trauma patient. J Crit Care. 2014;29:313. doi: 10.1016/j.jcrc.2013.11.017. [DOI] [PubMed] [Google Scholar]
- 11.Senthilkumaran S, David SS, Manikam R, et al. Permissive hypotension in a head-injured multi-trauma patient: Controversies and contradictions. J Anaesthesiol Clin Pharmacol. 2015;31:428–9. doi: 10.4103/0970-9185.161745. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rosenkrantz L, Schuurman N, Arenas C, et al. Maximizing the potential of trauma registries in low-income and middle-income countries. Trauma Surg Acute Care Open . 2020;5:e000469. doi: 10.1136/tsaco-2020-000469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Feldhaus I, Carvalho M, Waiz G, et al. Thefeasibility, appropriateness, and applicability of trauma scoring systems in low and middle-income countries: a systematic review. Trauma Surg Acute Care Open . 2020;5:e000424. doi: 10.1136/tsaco-2019-000424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Mock C, Joshipura M, Arreola-Risa C, et al. An estimate of the number of lives that could be saved through improvements in trauma care globally. World J Surg. 2012;36:959–63. doi: 10.1007/s00268-012-1459-6. [DOI] [PubMed] [Google Scholar]
- 15.Calvello EJB, Broccoli M, Risko N, et al. Emergency care and health systems: consensus-based recommendations and future research priorities. Acad Emerg Med. 2013;20:1278–88. doi: 10.1111/acem.12266. [DOI] [PubMed] [Google Scholar]
- 16.Matzopoulos R, Prinsloo M, Pillay-van Wyk V, et al. Injury-related mortality in South Africa: a retrospective descriptive study of postmortem investigations. Bull World Health Organ. 2015;93:303–13. doi: 10.2471/BLT.14.145771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Suresh K, Dixon JM, Patel C, et al. The epidemiology and outcomes of prolonged trauma care (EpiC) study: methodology of a prospective multicenter observational study in the Western Cape of South Africa. Scand J Trauma Resusc Emerg Med. 2022;30:55. doi: 10.1186/s13049-022-01041-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Dixon J, Verster J, Geduld H, et al. Preventable trauma deaths and health system strengthening recommendations: findings from a multi-disciplinary expert panel review of trauma deaths in the Western Cape, South Africa. BMC Public Health. 2025;25:3835. doi: 10.1186/s12889-025-25139-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Dixon J, de Vries S, Fleischer C, et al. Preventable trauma deaths in the Western Cape of South Africa: A consensus-based panel review. PLOS Glob Public Health . 2024;4:e0003122. doi: 10.1371/journal.pgph.0003122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Petrucelli E, States JD, Hames LN. The abbreviated injury scale: Evolution, usage and future adaptability. Accident Analysis & Prevention. 1981;13:29–35. doi: 10.1016/0001-4575(81)90040-3. [DOI] [Google Scholar]
- 21.Cheema B, Twomey M. In: Health Do, editor. The South Africa Triage Scale (SATS), Training Manual 2012: Emergency Medicine Society of South Africa. Cape Town: Western Cape Government Department of Health. 2012. https://emssa.org.za/wp-content/uploads/2011/04/SATS-Manual-A5-LR-spreads.pdf Available.
- 22.Wong HS, Curry NS, Davenport RA, et al. A Delphi study to establish consensus on a definition of major bleeding in adult trauma. Transfusion. 2020;60:3028–38. doi: 10.1111/trf.16055. [DOI] [PubMed] [Google Scholar]
- 23.Arboleda NS, Diaz AR, Zuñiga AJ, et al. Shock Index, Lactate, and Base Deficit as Bleeding Predictors in Trauma Patients from a Complex Emergency Department: A Prospective Cohort Study. Panam J Trauma Crit Care Emerg Surg. 2023;12:42–6. doi: 10.5005/jp-journals-10030-1416. [DOI] [Google Scholar]
- 24.Marino AN, Mui M, Dobryakova E, et al. Classification Systems for Moderate-Severe Traumatic Brain Injury: Review and Recommendations. Curr Phys Med Rehabil Rep. 2025;13:32. doi: 10.1007/s40141-025-00504-7. [DOI] [Google Scholar]
- 25.Osler T, Baker SP, Long W. A modification of the injury severity score that both improves accuracy and simplifies scoring. J Trauma. 1997;43:922–5. doi: 10.1097/00005373-199712000-00009. [DOI] [PubMed] [Google Scholar]
- 26.Seidu AS, Alhassan AR, Buunaaim ADB. Epidemiology of Polytrauma at a Teaching Hospital in Northern Ghana: A Cross‐Sectional Study. Int J Clin Pract. 2024;2024:4131822. doi: 10.1155/2024/4131822. [DOI] [Google Scholar]
- 27.Premji EN, Kilindimo SS, Sawe HR, et al. Patterns and Predictors of Timely Presentation and Outcomes of Polytrauma Patients Referred to the Emergency Department of a Tertiary Hospital in Tanzania. Emerg Med Int. 2022;2022:9611602. doi: 10.1155/2022/9611602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rio T, Nogueira LS, Lima FR, et al. Performance of severity indices for admission and mortality of trauma patients in the intensive care unit: a retrospective cohort study. Eur J Med Res. 2023;28:559. doi: 10.1186/s40001-023-01532-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kahloul M, Bouida W, Boubaker H, et al. Value of anatomic and physiologic scoring systems in outcome prediction of trauma patients. Eur J Emerg Med. 2014;21:125–9. doi: 10.1097/MEJ.0b013e32836188ce. [DOI] [PubMed] [Google Scholar]
- 30.Osebo CD, Munthali VJ, Rwanyuma LJ, et al. Time-critical care gaps and systemic delays linked to higher mortality in severe trauma patients in Tanzania. BMC Emerg Med. 2026;26:73. doi: 10.1186/s12873-026-01498-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Reynolds TA, Stewart B, Drewett I, et al. The impact of trauma care systems in low- and middle-income countries. Annu Rev Public Health. 2017;38:507–32. doi: 10.1146/annurev-publhealth-032315-021412. [DOI] [PubMed] [Google Scholar]
- 32.Allen BC, Cummer E, Sarma AK. Traumatic Brain Injury in Select Low- and Middle-Income Countries: A Narrative Review of the Literature. J Neurotrauma. 2023;40:602–19. doi: 10.1089/neu.2022.0068. [DOI] [PubMed] [Google Scholar]
- 33.Pant A, Suksaroj TT, Rattanapan C, et al. Emergency medical services in low- and middle-income countries: a systematic review of systemic barriers and integration challenges. Prev Med Rep. 2026;67:103522. doi: 10.1016/j.pmedr.2026.103522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.African Critical Illness Outcomes Study (ACIOS) Investigators The African Critical Illness Outcomes Study (ACIOS): a point prevalence study of critical illness in 22 nations in Africa. Lancet. 2025;405:715–24. doi: 10.1016/S0140-6736(24)02846-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Matzopoulos R, Bowman B, Butchart A, et al. The impact of violence on health in low- to middle-income countries. Int J Inj Contr Saf Promot. 2008;15:177–87. doi: 10.1080/17457300802396487. [DOI] [PubMed] [Google Scholar]
- 36.Christie SA, Zheng D, Dissak-Delon F, et al. How trauma patients die in low resource settings: Identifying early targets for trauma quality improvement. J Trauma Acute Care Surg. 2023;94:288–94. doi: 10.1097/TA.0000000000003768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Mould-Millman N-K, Baidwan NK, Beaty B, et al. Prolonged casualty care: Extrapolating civilian data to the military context. J Trauma Acute Care Surg. 2022;93:S78–85. doi: 10.1097/TA.0000000000003675. [DOI] [PMC free article] [PubMed] [Google Scholar]