Blunt chest trauma is among the most consequential yet frequently under-triaged components of major trauma, contributing directly or indirectly to a substantial share of trauma-related mortality worldwide. Injury to the thoracic cage and its contents the lungs, heart, great vessels, tracheobronchial tree, and mediastinal structures can simultaneously compromise ventilation, oxygenation, and circulation, and patients who develop cardiorespiratory failure after such injuries frequently require prolonged organ support in an intensive care unit (ICU). Road traffic collisions remain the principal mechanism globally: the World Health Organization estimates that approximately 1.16 million people die each year as a result of road traffic crashes, that these crashes are the leading cause of death among people aged 5 to 29 years, and that 92% of these fatalities occur in low- and middle-income countries, even though such countries own barely 60% of the world’s registered vehicles1. Thoracic injury is a major driver of this burden: in a 20-year registry of 176,346 road-traffic casualties from the Rhône region of France, thoracic trauma was recorded in only 3.6% of casualties yet accounted for 16.2% mortality within that subgroup and was judged the primary cause of death in 42.7% of fatalities, with traumatic brain injury (odds ratio 27.9), severe thoracic injury (Abbreviated Injury Scale ≥ 3; odds ratio 12.4), and advancing age independently associated with death2.

The epidemiological picture in South Asia mirrors, and in some respects exceeds, this global burden. In a prospective series of 1,258 chest trauma patients managed at a level 1 trauma centre in New Delhi, 83.5% of injuries were blunt, road traffic collisions accounted for nearly 60% of admissions, rib fractures were the single most common radiological finding, and overall mortality was 11%, rising to 11.65% among patients with blunt-mechanism injuries3. Comparable patterns predominantly young, male patients injured in road traffic incidents, with rib fractures, pulmonary contusion, and haemopneumothorax as the dominant pathology have been documented at other high-volume South Asian trauma centres4. Although the great majority of blunt chest trauma can be managed non-operatively with tube thoracostomy and supportive care3, a clinically important subset develops respiratory failure, haemodynamic instability, or multi-organ dysfunction severe enough to warrant ICU admission, and it is this subgroup that carries a disproportionate share of the morbidity, resource utilisation, and mortality associated with chest injury.

A substantial body of work has sought to identify the factors that determine survival in this critically injured population. Systematic reviews and meta-analyses have consistently implicated advancing age, injury severity score, number of rib fractures and flail segment, associated head or abdominal injury, pre-existing cardiopulmonary comorbidity, and delayed presentation as predictors of mortality after blunt chest wall trauma5,6, while ICU-specific cohorts have further highlighted admission Glasgow Coma Scale, physiological derangement at presentation, and the need for mechanical ventilation as powerful discriminators between survivors and non-survivors7. A 2024 systematic review and meta-analysis of chest injury outcomes across sub-Saharan Africa similarly identified advanced age, delayed hospital arrival, associated head, spinal, or cardiac injury, comorbidity, and higher injury severity scores as predictors of death, with pooled mortality reaching 9% overall and 11.65% specifically among blunt-mechanism injuries  figures that varied considerably with case-mix, trauma-system maturity, and time to definitive care8. A parallel review of clinical prediction models for blunt chest trauma has noted that existing scoring systems were derived and validated predominantly in North American and European trauma populations, with limited external validation elsewhere and inconsistent performance when applied outside their original setting9.

This body of evidence points to two related gaps. First, much of the literature on ICU outcomes after blunt chest trauma originates from health systems whose trauma networks, staffing ratios, and pre-hospital transport times differ substantially from those typical of South Asian tertiary care hospitals, which limits the direct applicability of existing risk-prediction models to this setting. Second, published South Asian data on chest trauma, while informative on injury pattern and overall outcome, have generally described general trauma or emergency department cohorts rather than the specific subgroup requiring intensive care, and few have used multivariable analysis to isolate independent determinants of mortality within this higher-risk population3,4. Given the differences in case-mix, referral patterns, and resource availability between these settings, locally derived evidence is needed to identify which clinical, demographic, and injury-related factors most reliably predict poor outcomes in South Asian ICU patients with blunt chest trauma, rather than extrapolating from prediction models developed elsewhere.

This retrospective study was therefore designed to determine the determinants of mortality and other clinical outcomes — including intensive care and hospital length of stay, duration of mechanical ventilation, and major in-hospital complications — among patients admitted to the intensive care unit with blunt chest trauma at [insert study site / hospital name]. By characterising the demographic, injury-related, and physiological factors independently associated with survival in this population, the study aims to generate context-specific evidence to inform risk stratification, resource allocation, and early clinical decision-making for this patient group. This retrospective study was conducted at Ayub Teaching Hospital, Abbottabad a medical college affiliated community hospital equipped with dedicated trauma and acute care services, including a mixed medical-surgical intensive care unit (ICU). Medical records of all patients admitted to the ICU with blunt chest trauma (BCT) between January 2023 and December 2025 were reviewed. Information regarding the patients and their injuries was primarily obtained from the institutional trauma registry and patients files. Where relevant data were incomplete, additional information was obtained through a detailed review of the patients’ electronic medical records. Data were subsequently entered into a standardized data collection form developed for the study. The variables assessed included demographic characteristics, initial vital signs, Glasgow Coma Scale (GCS) score, mechanism of trauma, anatomical distribution and severity of injuries, associated injuries, treatment received, duration of hospital and ICU stay, and ICU mortality. Overall injury severity was assessed using the Injury Severity Score (ISS), New Injury Severity Score (NISS), and Acute Physiology and Chronic Health Evaluation II (APACHE II) score. ISS and NISS were calculated in accordance with the Abbreviated Injury Scale 2008 guidelines.10 For outcome assessment, patients were categorized into two groups according to their ICU outcome: survivors and non-survivors. The clinical characteristics, injury patterns, severity scores, management, and outcomes of the two groups were subsequently compared. The study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations. General consent for medical care, including the use of clinical information for research and scientific purposes, had been obtained from patients or their legally authorized representatives. Ethical approval for the study was granted by the Institutional Ethical Review Committee of Ayub Medical College, Abbottabad Data were initially entered into Microsoft Excel (Excel for Mac, Version 16.81; Microsoft Corporation, Washington, USA) and subsequently analyzed using IBM SPSS Statistics version 28 (IBM Corp., Chicago, IL, USA). Continuous variables were summarized as mean ± standard deviation (SD), while categorical variables were reported as frequencies and percentages. For comparisons between survivors and non-survivors, the Mann–Whitney U test was applied to continuous or ordinal variables when appropriate. Categorical variables were compared using Fisher’s exact test or the chi-square test, depending on the distribution and applicability of the respective test. A two-sided p value of ≤0.05 was considered statistically significant. The relationship between individual clinical and injury-related variables and ICU mortality was initially assessed using univariable logistic regression.11 Given the small number of mortality events (6 deaths among 56 patients), multivariable analysis was performed cautiously to minimize the risk of model overfitting. Variables demonstrating statistically significant associations with mortality on univariable analysis were considered for inclusion in the multivariable model. These variables were entered into a forward stepwise logistic regression model using the likelihood-ratio method to identify factors independently associated with ICU mortality. The final model was interpreted as exploratory, given the limited number of outcome events and the potential for model instability. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. All statistical tests were two-sided, and a p value ≤0.05 was considered statistically significant.

During the study period, 887 patients presented with blunt chest trauma, of whom 56 (6.3%) required admission to the intensive care unit (Figure 1). The ICU cohort was young and predominantly male, with a mean age of 30.1 ± 15.5 years and 49 patients (87.5%) being men. On admission, patients were tachycardic and tachypnoeic, with a mean systolic blood pressure of 127 ± 28 mmHg, pulse rate of 100 ± 28 beats/min, respiratory rate of 23 ± 7 breaths/min, and oxygen saturation of 96 ± 7%; the mean Glasgow Coma Scale (GCS) score was 12 ± 4, reflecting a cohort with a substantial burden of altered consciousness. Injury severity was correspondingly high, with a mean Injury Severity Score (ISS) of 22 ± 10, New Injury Severity Score (NISS) of 25 ± 11, chest Abbreviated Injury Scale (AIS) score of 3 ± 1, and head AIS score of 3.2 ± 1.3; the mean APACHE II score at ICU admission was 10 ± 8 (Table 1).

Table 1. Demographic and physiological characteristics on ICU admission (n = 56).

Values are mean ± SD unless otherwise stated. AIS, Abbreviated Injury Scale; APACHE II, Acute Physiology and Chronic Health Evaluation II.

Lung contusion was the most frequent thoracic injury, present in 40 patients (71.4%), followed by pneumothorax in 23 (41.1%) and multiple rib fractures in 22 (39.3%) (Figure 2). Hemothorax was documented in 11 patients (19.6%), and fractures of the first and/or second rib and surgical emphysema were each present in 7 patients (12.5%). Four patients (7.1%) had an isolated single rib fracture, while sternal fracture, cardiac injury, and great vessel injury were each identified in 2 patients (3.6%). Diaphragmatic injury and flail chest were each recorded in a single patient (1.8%). Because many patients sustained more than one thoracic injury, the cumulative percentages exceed 100% (Table 2).

Table 2. Pattern of thoracic injuries among ICU-admitted patients (n = 56).

Percentages sum to more than 100% because many patients sustained more than one thoracic injury.

Figure 1. Patient selection and outcome flow diagram for the study cohort.

Figure 2. Frequency of thoracic injury patterns among ICU-admitted patients (n = 56). Percentages sum to more than 100% because many patients sustained more than one thoracic injury.

Motor vehicle collisions were the predominant mechanism of injury, accounting for 42 patients (75.0%). Car occupants constituted 30 patients (71.1%), including 23 drivers, while 12 patients (28.6%) were pedestrians. Nine patients (16.1%) sustained injuries following a fall from a height greater than 1 metre, and 5 patients (8.9%) sustained occupational injuries.(Table 3).

Extrathoracic injury was common: the head and spine were each affected in 24 patients (42.9%). Over half of the cohort required blood transfusion (30 patients, 53.6%) or operative intervention (29 patients, 51.8%), and chest tube insertion was performed in 10 patients (17.9%). Twenty-seven patients (48.2%) required intubation and mechanical ventilation at the time of ICU admission; ventilatory support was most often driven by severe head injury, extensive lung contusion, cardiothoracic injury, or complications arising during the clinical course, including septicaemia, thromboembolism, and multiorgan dysfunction. The mean ICU and total hospital lengths of stay were 11.2 ± 14.0 and 18.7 ± 16.9 days, respectively (Table 3).

Table 3. Mechanism of injury, associated extrathoracic injury, and ICU interventions/course (n = 56).

Values are n (%) unless otherwise stated. Motor vehicle collision subcategories and occupational injury are not mutually exclusive with the mechanism-of-injury category above them, so component percentages need not sum to the group total.

Six of the 56 patients died during their ICU admission, an overall ICU mortality of 10.7% (Figure 1). Neither chest AIS (p = 0.62) nor head AIS (p = 0.07) differed significantly between survivors and non-survivors, although head AIS showed a trend toward higher values among patients who died.

On univariable analysis, higher GCS at ICU admission was associated with lower odds of mortality, with each one-point increase corresponding to a 23% reduction in the odds of death (OR 0.77, p = 0.002). Conversely, each one-point increase in APACHE II score was associated with higher odds of mortality (OR 1.17, p < 0.001). Intubation and mechanical ventilation were also associated with mortality; univariable logistic regression yielded an OR of 12.09 (p = 0.021), while Fisher’s exact test demonstrated a statistically significant association (p = 0.006). Age, systolic blood pressure, pulse rate, respiratory rate, oxygen saturation, ISS, and NISS were not statistically significantly associated with mortality on univariable logistic regression. (Table 4).

Table 4. Univariate analysis of continuous and physiological variables associated with ICU mortality.

OR, odds ratio (unadjusted, univariate logistic regression). * p < 0.05.

Categorical comparison between survivors and non-survivors using Fisher’s exact test confirmed intubation as the only variable significantly associated with mortality (p = 0.006); the proportion of deaths was markedly higher among patients who required intubation than among those who did not. Sex, occupational injury, blood transfusion, lung contusion, hemothorax, pneumothorax, and rib fracture pattern (first/second rib or multiple rib fractures) were not significantly associated with mortality (Table 5).

Table 5. Univariate analysis of categorical variables associated with ICU mortality.

Dash (—) indicates the variable was not entered into the univariate logistic regression model. * p < 0.05.

In forward stepwise exploratory multivariable logistic regression incorporating age, sex, occupational injury, systolic blood pressure, pulse rate, respiratory rate, oxygen saturation, GCS, blood transfusion, ISS, NISS, intubation, lung contusion, hemothorax, and APACHE II score, only the APACHE II score remained an independent predictor of ICU mortality: each one-point increase was associated with a 15% increase in the odds of death (OR 1.15; 95% CI 1.04–1.28; p = 0.008) (Table 6).Given the small number of mortality events, these findings should be interpreted as exploratory and require confirmation in larger cohorts.

Table 6. Multivariable logistic regression model for predictors of ICU mortality (forward stepwise, likelihood ratio).

Candidate variables entered into the model: age, sex, occupational injury, systolic blood pressure, pulse rate, respiratory rate, oxygen saturation, Glasgow Coma Scale score, blood transfusion, Injury Severity Score, New Injury Severity Score, intubation, lung contusion, hemothorax, and APACHE II score. Only the APACHE II score was retained in the final model. CI, confidence interval; OR, odds ratio.

Receiver operating characteristic analysis confirmed good to excellent discriminatory performance of the APACHE II score for ICU mortality, with an area under the curve of 0.884 (90% CI 0.679–1.000). An APACHE II score of approximately 15, corresponding to a predicted probability of death of 0.131, offered the optimal balance between sensitivity (90.0%) and specificity (81.7%) for identifying patients at risk of death (Figure 3). Taken together, these findings indicate that physiological derangement at ICU admission, as captured by the APACHE II score, was a stronger determinant of mortality than any individual pattern of thoracic injury.

Figure 3. Receiver operating characteristic (ROC) curve for the APACHE II score in predicting ICU mortality.

In this retrospective analysis of 887 patients presenting with blunt chest trauma, only 6.3% required intensive care, yet this subgroup accounted for an ICU mortality of 10.7% a figure that sits squarely within the range reported for blunt-mechanism thoracic injury elsewhere in the literature, from 9% pooled across sub-Saharan African cohorts to 11.65% among blunt-injury patients at a comparable South Asian level 1 trauma centre8,3. That a critically injured, ICU-selected population achieved a mortality broadly comparable to rather than substantially higher than general ward-and-ICU trauma cohorts suggests that, once patients reach a facility capable of intensive organ support, aggressive resuscitation and critical care can meaningfully offset the physiological burden of severe thoracic injury. The demographic and mechanistic profile of the cohort young, overwhelmingly male, and dominated by road traffic collisions, particularly among car occupants and pedestrians mirrors both the global epidemiology of blunt chest trauma and previously reported South Asian trauma registries1–4. This concentration of severe injury among young adults of working age carries substantial socioeconomic weight beyond in-hospital mortality, and reinforces road traffic injury prevention vehicle safety standards, pedestrian infrastructure, and enforcement of speed and restraint legislation as a public health priority in this setting rather than a purely clinical one. Lung contusion, rather than flail chest, was the dominant thoracic injury in this cohort (71.4% vs 1.8%), a pattern that echoes a broader shift in the blunt chest trauma literature away from rib fracture pattern as the principal determinant of respiratory compromise and toward the extent of underlying parenchymal injury5–7.Flail segments are a visually dramatic and historically emphasised marker of severe chest wall injury, but the low prevalence observed here suggests that, at least in this population, diffuse pulmonary contusion and its downstream physiological consequences hypoxaemia, reduced compliance, and secondary complications such as pneumonia or acute respiratory distress syndrome represent a more common final pathway to critical illness than mechanical chest wall instability per se. The principal finding of this study was that APACHE II score at ICU admission showed the strongest association with ICU mortality in this cohort. In the exploratory multivariable analysis, APACHE II was the only variable retained in the final model, with each one-point increase associated with higher odds of mortality. Its discriminatory performance was also relatively strong on ROC analysis. In a cohort of 1,019 critically injured trauma patients, APACHE II achieved an area under the curve of 0.77, more than 20 points higher than the Injury Severity Score (0.54), with each one-point increase in APACHE II conferring an odds ratio of death of 1.18 strikingly close to the adjusted odds ratio of 1.15 observed in the present analysis14. A recent Indian single-centre study similarly found that APACHE II outperformed the Injury Severity Score, Revised Trauma Score, and Trauma and Injury Severity Score in predicting trauma mortality, particularly among patients requiring ICU admission15. Because APACHE II aggregates acute derangement in oxygenation, haemodynamics, renal and metabolic function, and level of consciousness into a single composite score, it plausibly captures the cumulative physiological insult of multi-system injury including, but not limited to, its thoracic component more completely than any single anatomic injury score13.However, these findings should be interpreted in the context of the small number of mortality events, which limits the precision and stability of multivariable estimates. The results therefore suggest a potentially useful association rather than establishing APACHE II as a definitive mortality prediction tool for patients with blunt chest trauma. Neither ISS nor NISS reached statistical significance in this cohort, a finding that should be interpreted cautiously rather than as evidence against their clinical relevance. All patients included had already been triaged to intensive care, which restricts the range of injury severity under study and can obscure associations that are more readily detected across a broader, less selected trauma population, as in the meta-analyses underpinning current risk models for blunt chest wall trauma5,6. In addition, with only 6 deaths among 56 patients, the events-per-variable ratio for the 15-variable candidate model fell well below the threshold of approximately 10 events per predictor generally recommended for stable logistic regression estimates, raising the possibility that the stepwise procedure retained the variable with the strongest signal (APACHE II) while lacking power to detect smaller, real effects of anatomic severity scores16. The wide 90% confidence interval around the reported area under the curve (0.679–1.000) reflects this same limitation and should temper overinterpretation of the point estimate. Admission GCS and the need for intubation were both strongly associated with mortality on univariate analysis but did not survive multivariable adjustment, most plausibly because both are physiologically and, in the case of GCS, mathematically embedded within the APACHE II score itself. This pattern parallels a UK ICU cohort in which GCS below 15 was among the strongest independent predictors of blunt trauma mortality when modelled without a composite physiological score, and in which a customised physiological severity score achieved a similarly high area under the curve of 0.907. In practice, this suggests that GCS and the requirement for intubation remain useful rapid bedside indicators of risk at first contact before a full APACHE II score can be calculated even though their independent statistical contribution is subsumed once the composite score is available. From a practical standpoint, an APACHE II score of approximately 15 at ICU admission identified patients at high risk of death with a sensitivity of 90.0% and specificity of 81.7% in this cohort. In resource-constrained South Asian trauma centres, where ICU beds, invasive monitoring, and senior critical care input are often limited, a simple, already widely available bedside score of this kind could support early risk stratification flagging patients who may benefit most from prioritised ICU admission, closer monitoring, or early escalation discussions with patients' families without requiring additional investigation beyond what is already collected as part of routine ICU admission assessment. This study adds to a relatively sparse literature specifically addressing ICU, rather than general trauma-ward or emergency department, outcomes after blunt chest trauma in South Asia3,4, and applies multivariable analysis to isolate independent determinants of mortality within this higher-risk subgroup. Several limitations nonetheless warrant emphasis. The retrospective, single-centre design is subject to the inherent biases of chart-based data collection and limits generalisability to trauma systems with different referral patterns, staffing, or resource availability. The overall sample size, and in particular the small absolute number of deaths, constrains statistical power, widens confidence intervals, and increases the risk of both type II error for weaker predictors and instability in the stepwise variable-selection process16. Important potential confounders pre-hospital transport time, comorbidity burden, transfusion volume, and details of operative and ventilatory management were not modelled, and post-discharge functional outcomes were not captured. Prospective, multi-centre studies with larger event numbers are needed to externally validate the APACHE II cutoff identified here before it is adopted into routine triage protocols.

Blunt chest trauma requiring intensive care carried a substantial mortality of 10.7% in this South Asian cohort, occurring predominantly among young men injured in road traffic collisions and characterised more often by diffuse lung contusion than by flail chest. Physiological derangement at ICU admission, as captured by the APACHE II score, was the only independent predictor of death identified on multivariable analysis, with strong discriminatory performance (AUC 0.884) that exceeded that of anatomic injury-severity measures such as the ISS and NISS. An APACHE II score of approximately 15 offered a practical threshold for identifying patients at high risk of death. These findings support incorporating simple, physiology-based severity scoring into early risk stratification for blunt chest trauma patients requiring intensive care in similar resource settings, while underscoring the need for larger, prospective, multi-centre studies to validate this threshold and to more precisely define the independent contribution of anatomic injury pattern once adequately powered.