Most prediction models for hospital mortality due to TBI are conducted in high-resource settings in North America and Europe. Nevertheless, sociocultural characteristics, prehospital care, and clinical practices may differ in Low- and middle-income countries (LMICs), leading to discrepancies in mortality rates and predictive factors. In our study, the in-hospital mortality rate was 55%, significantly higher than the hospital mortality rates of 19.1–33.2% reported in other studies conducted in Brazil [5, 9–12], and even higher when compared to mortality rates in North America (13.3%), Europe (25.5%), and China 20.0% [23–25]. Interestingly, the mortality rate 1 year after discharge (15.1%) was lower than the mortality rate reported in other long-term studies in China (30%) [26], Europe (30%) [27] and India (44%) [6]. Therefore, the mortality rate is substantially high in the setting of this study, possibly explained by the severity of the injuries associated with limited diagnosis capabilities of intracranial injuries, delayed or inadequate initial care, lack of monitoring, and specialized treatment. The reduced levels of mortality within 1 year may be partially explained by the excess in-hospital mortality [5].
The majority of the sample consisted of male individuals in the productive age range of 34 to 65. As expected, individuals from the survival group were younger. Additionally, the proportion of individuals over the age of 65 was significantly higher in the group of patients who died during hospitalization. Age greater than 65 years was associated with a higher risk of hospital mortality in the final logistic regression model for hospital mortality. Interestingly, our findings suggest that individuals aged between 34 and 65 were more susceptible to die within the first year. However, this may be influenced by the larger number of individuals in this age group in the sample and the smaller number of individuals in the 'death' group compared to the 'survivor' group. Despite that, older age was a stronger predictor of in-mortality in previous studies conducted in Brazil [5], Latin America [28], and high-income countries [8]. Only one study did not find an effect of age on mortality after severe TBI [29], which might be due to the small number of elderly patients included. The proportion of male individuals was higher in the survival group, with a potential to predict mortality in the individual analysis. However, in the final logistic regression model, sex was not associated with the outcome. This trend has been previously observed in other studies (5, 8, 30–31]. It appears that sex may be associated with the incidence and mortality of TBI, as men are more exposed to risk factors for TBI [28]. Nevertheless, when included in a multivariate model, other variables such as injury severity scores seem to be more effective in predicting the outcome.
In our study the ISS ≥ 25 was associated with hospital mortality. This finding was also previously reported by the literature in both LMICs and HICs [25, 32–33]. Furthermore, GCS at admission, another widely used predictor of in-hospital mortality, [12, 31, 34], surprisingly, was not associated with outcome in this study. Most patients' scores in the GCS at hospital admission ranged between 3–5, and even though some individuals had higher GCS scores at emergency, as part of a severe TBI sample, at some point they evolved to loss of consciousness and lower scores. In addition, changes in the pattern of pupillary responses and performing decompressive craniectomy were also associated with outcome, as previously reported by the literature in both LMICs and HICs [8, 11, 35].
Our findings show that the main cause of injury in the sample was fall and motor vehicle accidents in both groups. Interestingly, fall is the main etiology of TBI in HICs, mainly in European countries [36]. On the other hand, previous studies carried out in LMICs, evidenced that the main cause of severe TBI was road traffic accidents [25, 37] and constantly related to cultural and socioeconomic characteristics. Individuals who died had a major proportion of cause of injury by Fall and Gunshot. However, in the final model of logistic regression none of these etiologies were associated with hospital mortality.
Many pathophysiological markers could be associated with increased mortality rate in TBI patients. Although, in our study parameters such as systolic blood pressure (SPB), diastolic blood pressure (DBP) and heart rate (HR) did not statistically significantly differ between groups. However, respiratory rate (RR) was positively associated with higher risk of hospital mortality, while body temperature at hospital admission was negatively associated with hospital mortality. Hyperglycemia was considerably higher in the individuals who died during hospitalization; however, it was not associated with hospital mortality in the final model. Previous studies regarding the alterations in the cardiovascular system in TBI patients, also suggests that SBP, DBP and HR were indeed not associated with higher mortality rates [5, 29]. Conversely, body temperature, RR and high glycemic levels, as previously reported by studies in both LMICs and HICs, were found to be associated with death during hospitalization due to severe TBI [38–39]. At the one-year analysis, no variables other than age were associated with mortality. This suggests that long-term mortality might be influenced much more by sociodemographic factors such as post-discharge technology care, access to rehabilitation centers, family care, and others, rather than early clinical factors.
LIMITATIONS
One of the primary limitations of this study is its single-center setting. As the data were collected from only one hospital, the generalizability of the results to other healthcare facilities and patient populations might be limited. Variability in patient demographics, medical practices, and available resources between institutions could impact the transferability of the findings. Secondarily, the lack of data on computerized tomography (CT) scans is a significant limitation of this study. CT scan results are recognized as strong predictors of outcome after TBI; however, in the present study, they could not be considered due to the absence of a researcher trained in interpreting these exams. Despite the availability of CT scan reports in the medical records, their reliability was compromised as they were written by the on-duty medical staff or by the examination center.
Notwithstanding these acknowledged limitations, the present study possesses notable strengths that contribute to its robustness. Firstly, the data collection was conducted prospectively, ensuring the accuracy and reliability of patient information. Secondly, the study benefited from a large sample size derived from a referral trauma hospital, which enhances the representativeness of the findings. Notably, patients with severe TBI in the geographical areas covered by the study are predominantly treated at this participating reference center, suggesting the inclusion of a significant proportion of severe TBI cases within the region during the study period.