Joanna Grudziak

The effect of anatomic location of injury on mortality risk in a resource-poor setting

INTRODUCTION Injury is a significant cause of death, with approximately 4.7 million people mortalities each year. By 2030, injury is predicted to be among the top 20 causes of death worldwide. We sought to characterize and compare the mortality probability in trauma patients in a resource-poor setting based on anatomic location of injury. METHODS We performed a retrospective analysis of prospectively collected data using the trauma database at Kamuzu Central Hospital (KCH) in Lilongwe, Malawi. We included all adult trauma patients (≥16years) admitted between 2011 and 2015. We stratified patients according to anatomic location of injury, and used descriptive statistics to compare characteristics and management of each group. Bivariate analysis by mortality was done to determine covariates for our adjusted model. A Cox proportional hazard model was performed, using upper extremity injury as the baseline comparator. Descriptive statistics were used to describe the trend in incidence and mortality of head and spine injuries over five years. RESULTS Of the 76,984 trauma patients who presented to KCH from 2011 to 2015, 49,126 (63.8%) were adults, and 8569 (17.4%) were admitted. The most common injury was to the head or spine, seen in 3712 patients (43.6%). The highest unadjusted hazard ratio for mortality was in head and spine injury patients, at 3.685 (95% CI=2.50-5.44), which increased to 4.501 (95% CI=2.78-7.30) when adjusted for age, sex, injury severity, transfer status, injury mechanism, and surgical intervention. Abdominal trauma had the second highest adjusted hazard of mortality, at 3.62 (95% CI=1.92-6.84) followed by thoracic trauma (HR=1.3621, 95% CI=0.49-3.56). CONCLUSION In our setting, head or spine injury significantly increases the hazard of mortality significantly compared to all other anatomic injury locations. The prioritization of timely operative and non-operative head injury management is imperative. The development of head injury units may help attenuate trauma- related mortality in resource poor settings.

Epidemiology, Management, and Functional Outcomes of Traumatic Brain Injury in Sub-Saharan Africa

BACKGROUND Trauma accounts for 4.7 million deaths each year, with an estimated 90% of these occurring in low- and middle-income countries (LMICs). Approximately half of trauma-related deaths are caused by central nervous system injury. Because a thorough understanding of traumatic brain injury (TBI) in LMICs is essential to mitigate TBI-related mortality, we established a clinical and radiographic database to characterize TBI in our low-income setting. METHODS This is a review of prospectively collected data from Kamuzu Central Hospital, a tertiary care center in the capital of Malawi. All patients admitted from October 2016 through May 2017 with a history of head trauma, altered consciousness, and/or radiographic evidence TBI were included. We performed descriptive statistics, a Cox regression analysis, and a survival analysis. RESULTS There were 280 patients who met inclusion criteria; of these, 80.5% were men. The mean age was 28.8 ± 16.3 years. Median Glasgow Coma Scale (GCS) score was 12 (interquartile range, 8-15). Road traffic crashes constituted the most common injury mechanism (60.7%). There were 148 (52.3%) patients who received a computed tomography scan, with the most common findings being contusions (26.1%). Of the patients, 88 (33.0%) had severe TBI, defined as a GCS score ≤8, of whom 27.6% were intubated and 10.3% received tracheostomies. Overall mortality was 30.9%. Of patients who survived, 80.1% made a good recovery. Female sex was protective, and the only significant predictor of poor functional outcome was presence of severe TBI (hazard ratio, 2.98; 95% confidence interval, 1.79-4.95). CONCLUSIONS TBI represents a significant part of the global neurosurgical burden of disease. Implementation of proven in-hospital interventions for these patients is critical to attenuate TBI-related morbidity and mortality.

The effect of pre-existing malnutrition on pediatric burn mortality in a sub-Saharan African burn unit

INTRODUCTION Nutritional status predicts burn outcomes in the developed world, but its effect on burn mortality in the developing world has not been widely studied. In sub Saharan Africa, burn is primarily a disease of children, and the majority of children in sub-Saharan Africa are malnourished. We therefore sought to determine the prevalence and effect of malnutrition on burn mortality at our institution. METHODS This is a retrospective review of children aged 0-5, with anthropomorphic measurements available, who were admitted to our burn unit from July 2011 to May 2016. Age-adjusted Z scores were calculated for height, weight, weight for height, and mid-upper arm circumference (MUAC). Following bivariate analysis, we used logistic regression to construct a fully adjusted model of predictors of mortality. RESULTS Of the 1357 admitted patients, 839 (61.2%) were aged 0-5. Of those, 512 (62.9%) had one or more anthropomorphic measurements available, and were included in the analysis. 54% were male, and the median age was 28 months. The median TBSA was 15%, with a majority of burns caused by scalds (77%). Mortality was 16%. Average Z-score for any of the indicators of malnutrition was -1.45±1.66. TBSA (OR: 1.08, 95% CI: 1.06, 1.11), decreasing Z-score (OR: 1.19, 95% CI: 1.00, 1.41), and flame burn (OR: 2.51, 95% CI: 1.40, 4.49) were associated with an increase in mortality. CONCLUSION Preexisting malnutrition in burn patients in sub-Saharan Africa increases odds of mortality after controlling for significant covariates. Survival of burn patients in this region will not reach that of the developed world until a strategy of aggressive nutritional support is implemented in this population.

Anatomic Location and Mechanism of Injury Correlating with Prehospital Deaths in Sub-Saharan Africa

IntroductionTrauma is a large contributor to morbidity and mortality in developing countries. We sought to determine which anatomic injury locations and mechanisms of injury predispose to prehospital mortality in Malawi to help target preventive and therapeutic interventions. We hypothesized that head injury would result in the highest prehospital mortality.MethodsThis was a retrospective analysis of all trauma patients presenting to Kamuzu Central Hospital in Lilongwe, Malawi, from 2008 to 2015. Independent variables included baseline characteristics, anatomic location of primary injury, mechanism of injury, and severity of secondary injuries. Multivariable logistic regression was used to assess the effect of primary injury location and injury mechanism on prehospital death, after adjusting for confounders. Effect measure modification of the primary injury site/prehospital death relationship by injury mechanism (stratified into intentional and unintentional injury) was assessed.ResultsOf 85,806 patients, 701 died in transit (0.8%). Five hundred and five (72%) of these patients sustained a primary head injury. After adjustment, head injury was the anatomic location most associated with prehospital death (OR 11.81 (95% CI 6.96–20.06, p < 0.0001). The mechanisms of injury most associated with prehospital death were gunshot wounds (OR 38.23, 95% CI 17.66–87.78, p < 0.0001) and pedestrian hit by vehicle (OR 2.62, 95% CI 1.92–3.55, p < 0.0001). Among head injury patients, the odds of prehospital mortality were higher with unintentional injuries.ConclusionsHead injuries are the most common causes of prehospital death in Malawi, while pedestrians hit by vehicles are the most common mechanisms. In a resource-poor setting, preventive measures are critical in averting mortality.

Paediatric road traffic injuries in Lilongwe, Malawi: an analysis of 4776 consecutive cases

This was a retrospective review of all children aged ≤16 who were treated in the casualty department at the central hospital in Lilongwe, Malawi, between 1 January 2009 and 31 December 2015. A total of 4776 children were treated for road traffic injuries (RTIs) in the study period. There was an increase in incidence from 428 RTIs in 2009 to a maximum of 834 in 2014. Child pedestrians represented 53.8% of the injuries, but 78% of deaths and 71% of those with moderate to severe head injuries. Pedestrians were mostly injured by cars (36%) and by large trucks, buses and lorries (36%). Eighty-four (1.8%) children were brought in dead, while 40 (0.8%) children died in the casualty department or during their hospital stay. There has been a drastic increase of RTIs in children in Lilongwe, Malawi. Child pedestrians were most affected, both in terms of incidence and severity.

Pre-burn malnutrition increases operative mortality in burn patients who undergo early excision and grafting in a sub-Saharan African burn unit: Methodological issues

Drs. Ashrafi-Asgarabad and Safiri raise several questions about the methodology we employed in our article [1]. Thank you for this opportunity to respond to their letter. The first question has to do with selection of variables for our multivariable logistic model. As stated in the article, we used variables that were either (a) significantly associated with mortality in our univariable analysis, but only if the confidence intervals were narrow and we had sufficient data about a particular variable; or (b) have been reported to be associated with burn mortality in LMICs in other studies. The second question is regarding potential collinearity of our variables. This question is not relevant to our final model as collinearity does not affect or negate the overall predictive model of a multivariable logistic regression model. Individual covariates can be affected by this concept, but our article emphasizes the interaction between nutritional stress, operative timing, and %TBSA burned. We did not set out to negate or minimize the contribution of any of these variables to our operative mortality. Our article aims to develop a multicausation predictive model of burn mortality in our burn patients, which is represented by the full multivariable logistic regression model. Regarding the large odds ratio and wide confidence interval of the association of amputation with mortality: it is widely accepted that wide confidence intervals indicate sparse data. However, amputation was not used in our multivariable logistic regression model, and we reported this finding merely in the interest of completeness of data. Finally, we make no claims regarding the validation of our z-score model in the current article. We are in the process of testing this model to validate it. R E F E R E N C E