Comparison of Revised Baux, Belgian Outcome of Burn Injury, and Abbreviated Burn Severity Index Scores as Model Predictors of Mortality in Burn Patients: A Retrospective Study
Article information
Abstract
Background
Burns continue to be a major cause of morbidity and mortality worldwide. Predictor models in burn patients have been developed to help clinicians determine mortality risk. These include the revised Baux (rBaux), Belgian Outcome of Burn Injury (BOBI), and Abbreviated Burn Severity Index (ABSI) scores. This study aims to compare rBaux, BOBI, and ABSI scores as model predictors of mortality in burn patients.
Methods
This is a retrospective study with an observational analytic design using secondary data obtained from research variables in the medical records of burn injury inpatients at Dr. Soetomo General Academic Hospital Surabaya, a tertiary center in Surabaya, from January 2020 to December 2022. Research subjects were selected using a total sampling method with inclusion and exclusion criteria. Diagnostic tests were carried out to obtain the results of sensitivity, specificity, negative predictive value, and positive predictive value.
Results
The study sample size was 197 patients. rBaux, BOBI, and ABSI scores were shown to have the same significance in predicting mortality in burn patients. Sensitivity and specificity and receiver operating characteristic analysis for each score were similar, making them all reliable in predicting mortality for burn patients.
Conclusion
Mortality scores in burn patients such as the rBaux, BOBI, and ABSI can be used effectively to determine outcomes. The simplicity of the rBaux score enables it to be used earlier to help determine patient mortality and the need for more intensive care.
Introduction
Burns are one of the major causes of morbidity and mortality worldwide. The World Health Organization estimates that 180,000 deaths annually are caused by burns [1]. A national study stated the incidence of burns in Indonesia at 2.2% with a mortality rate of 30% [1]. The survival rate has recently increased due to advances in burn care, but the mortality rate from burn injuries still remains high [2]. The survival of burn patients is affected by many factors including age, sex, presence of medical comorbidities, total body surface area (TBSA) involved in the burn, presence of inhalation injury, and burn depth [2]. Several predicting models have been developed to help physicians determine the mortality risk for burn injuries. Some of the most frequently used predicting models in Indonesia include the revised Baux (rBaux), Belgian Outcome of Burn Injury (BOBI), and Abbreviated Burn Severity Index (ABSI) scores [3].
The rBaux and BOBI score consist of similar variables, namely age, TBSA, and presence of inhalation trauma, but differ in score points and calculation methods [3]. The rBaux score, commonly used in clinical practice, is often regarded as a quick and simple tool to estimate mortality. However, the complexity of burn injuries frequently involves multiple organ systems and varying degrees of burn severity accompanied with various medical comorbidities, making other scoring methods also favorable [3]. The need for an accurate and reliable burn injury prediction model is critical due to the varying clinical outcomes and differences in burn care management across regions. With clearer prediction of burn patients’ prognosis, more advanced therapy can quickly be initiated as necessary to help decrease mortality. This study aims to help objectively determine the best scoring system for predicting mortality in burn injuries.
Methods
Study design and participants
This study was conducted at Dr. Soetomo General Academic Hospital Surabaya, a public tertiary level teaching hospital in Surabaya, Indonesia. With a capacity of more than 2,200 beds, it has the most well-equipped burn center in the region. With a retrospective observational analytic design, the study was conducted using secondary data from medical records of all inpatients admitted with burn injuries from January 2020 to December 2022. Research subjects were selected using the total sampling method. Patients with incomplete medical records were excluded. This study was approved by the Ethics Committee of Dr. Soetomo General Hospital (approval No. 0789/LOE/301.4.2/II/2022), and informed consent was obtained from all the study subjects.
Variables and measurements
Data obtained from medical records consisted of age, sex, TBSA, presence of inhalation trauma, burn injury depth, and occurrence of death. The acquired data was then calculated into rBaux, BOBI, and ABSI scores. The formula for rBaux score was (TBSA [in %] + age [in years] + [17×R]), with R=1 for the presence of inhalation injury. BOBI score was calculated from age (0–3 points: 0, <50 years; 1, 50–64 years; 2, 65–79 years; 3, ≥80 years), TBSA (0–4 points: 0, <20%; 1, 20%–39%; 2, 40%–59%; 3, 60%–79%; 4, ≥80%), and presence of inhalation injury (3 points). Total BOBI score can vary from 0 to 10 points, with each point representing mortality prediction from 0.1% to 99%. ABSI score was calculated from sex (0, male; 1, female), age (1–5 points: 1, 0–20 years; 2, 21–40 years; 3, 41–60 years; 4, 61–80 years; 5, 81–100 years), TBSA (1–10 points: 1, 0%–10%; 2, 11%–20%; 3, 21%–30%; 4, 31%–40%; 5, 41%– 50%; 6, 51%–60%; 7, 61%–70%; 8, 71%–80%; 9, 81%–90%; 10, 91%–100%), presence of inhalation injury (1 point), and presence of full-thickness burn (1 point). A total ABSI score of 2–3 points translated to >99% survival rate (very low threat), 4–5 points meant 98% survival (moderate threat), 6–7 points meant 80%–90% survival (moderately severe threat), 8–9 points meant 50%–70% survival rate (serious threat), 10–11 points had 20%–40% survival (severe threat), and 12–13 points had a survival rate of <10% (maximum threat).
Statistical analysis
Univariate analysis was done to report the demographic characteristics of the study subjects. The normality test was done using the Kolmogorov Smirnov test. All predicting models underwent receiver operating characteristic (ROC) analysis to determine the optimal cutoff for predicting mortality, followed by sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) analysis for mortality. Predicting models were compared using the McNemar test. A value of P<0.05 was considered statistically significant. All statistical analyses were conducted using the IBM SPSS version 26.0.
Results
A total of 197 burn patients were included in this study. Among these, 62 patients (31.5%) died and 135 patients (68.5%) survived. Table 1 shows the clinical demographic data of all subjects. A larger (71.0%) portion of the deceased patients were male. The highest number of deaths occurred in the age groups 46–55 and 56–65 years, each with respective percentages of 27.4% (17 deaths each). The largest portion of deaths occurred in patients with burns consisting of 61%–70% TBSA (21.0%). Mortality causes were frequently due to inhalation injury (69.4%) and sepsis (50.0%).
Clinical characteristics of subjects based on mortality
From the obtained data, ABSI, rBaux, and BOBI scores were calculated and compared between mortality groups. ABSI, rBaux, and BOBI all showed statistically significant higher scores in patients who died (10 [5–15] vs. 5 [2–11]; 113 [44–156.5] vs. 41 [7–113.5]; and 5 [1–8] vs. 1 [0–5], respectively, P<0.001) (Table 2).
Comparison of median score between ABSI, rBaux, and BOBI
The ROC curve is used to describe diagnostic accuracy and determine the optimal cutoff value of all predicting models. The optimal cutoff point for ABSI was ≥10.5 with area under the curve (AUC) value of 0.934 (very good) (Fig. 1A), the cutoff point for rBaux was ≥111.25 with AUC value of 0.920 (very good) (Fig. 1B), and the cutoff point for BOBI was ≥5.5 with AUC value of 0.910 (very good) (Fig. 1C). After that, the analysis was continued with diagnostic tests for sensitivity, specificity, PPV, and NPV (Table 3). ABSI scores with a cutoff point ≥10.5 had 45.16% sensitivity, 99.25% specificity, 96.55% PPV, and 79.76% NPV for predicting mortality. rBaux scores with a cutoff point ≥111.25 had 53.22% sensitivity, 98.52% specificity, 94.28% PPV, and 82.09% NPV for predicting mortality. BOBI scores with a cutoff point ≥5.5 had 48.3% sensitivity, 100% specificity, 100% PPV, and 80.3% NPV for predicting mortality. The corresponding odds ratios for ABSI, rBaux, and BOBI were 198.32, 85.91, and 30.99, respectively, with accuracy rates of 87.82% for ABSI, 85.28% for BOBI, and 87.82% for rBaux. The results of the chi-square test comparing the diagnostic values of sensitivity, specificity, PPV, NPV, accuracy, and odds ratio for the ABSI, BOBI, and rBaux scores revealed P-values of 0.564, 1.0, 1.0, 1.0, 1.0, and 1.0, respectively (P>0.05). These findings indicate no statistically significant differences in the diagnostic values among the ABSI, BOBI, and rBaux scores.
Receiver operating characteristics (ROC) curve analysis. ROC curves of ABSI score (A), rBaux score (B), and BOBI score (C). ABSI, Abbreviated Burn Severity Index; rBaux, revised Baux; BOBI, Belgian Outcome of Burn Injury.
Mortality of burn patients according to cutoff score of ABSI, rBaux, and BOBI
All three scoring systems were analyzed using the McNemar test to observe the relationship between scores (Table 4). A statistically significant difference was found between ABSI and rBaux scores as predictors of mortality in burn patients, whereas no significant difference was found between ABSI and BOBI, and between rBaux and BOBI in predicting burn patient mortality.
Comparison of mortality between ABSI, rBaux, and BOBI scores
The DeLong test was conducted to evaluate statistical differences in the ROC curves among the variables. The comparison between ABSI and BOBI resulted in a difference between AUC (DBA) of 0.024, with a standard error (SE) of 0.015 and a 95% confidence interval (CI) ranging from –0.005 to 0.053. The associated P-value of 0.101 indicates no statistically significant difference between these two variables. Similarly, the comparison between ABSI and rBaux yielded a DBA of 0.006, with an SE of 0.011 and a 95% CI of –0.015 to 0.028, accompanied by a P-value of 0.548, showing no significant difference. The comparison between BOBI and rBaux demonstrated a DBA of 0.017, an SE of 0.014, and a 95% CI of –0.010 to 0.045. The P-value of 0.218 indicates a statistically insignificant difference between these two variables. These results highlight varying levels of predictive performance among the variables (Table 5, Fig. 2).
Analysis comparison of ROC curves
Comparison graph of receiver operating characteristics curve. ABSI, Abbreviated Burn Severity Index; BOBI, Belgian Outcome of Burn Injury; rBaux, revised Baux.
Discussion
Mortality risk in burn patients is influenced by several factors including patient factors and trauma factors. Patient factors include age, sex, nutritional status, and preexisting comorbidities. Trauma factors include burn depth, TBSA affected by burn, burn location, presence of inhalation trauma, concomitant trauma, and individual responses [4].
The burn patients included in this study were mostly in the age range of 26–35 years with the highest mortality in the age range of 46–55 and 56–65 years. The prognosis of burns is generally worse in the very young and elderly population. This is because regulatory and immune systems in very young patients are not yet fully formed, whereas in elderly patients the immune systems have deteriorated significantly [5]. Sex may also affect mortality of burn patients. Research conducted by Lam et al. [6] showed that the mortality rate of women burn patients was lower than men. Women have greater resistance to inflammatory responses due to estrogen which decreases systemic inflammation by inhibiting the response of interleukin-1 in the hypothalamus [7].
Absorption of combustion products through inhalation can cause short-term airway obstruction, leading to local or systemic toxic effects [8]. This is in accordance with the fact that the mortality of burn patients with inhalation trauma has been stated to be up to 30% higher than those without inhalation trauma [4]. Burns affecting a larger TBSA percentage can cause increased evaporation of body fluids leading to local tissue damage, hypovolemia and coagulopathy, and therefore are associated with higher rates of infection, sepsis, organ failure, and death [9]. This also applies to burn injury depth. Full-thickness burn injuries have a poorer prognosis due to factors such as a prolonged healing process, which increases the risk of exposure to nosocomial infections. This in turn can lead to sepsis and even death [10].
ABSI, first published in 1982 by Tobiasen et al. [11], was the first scoring system used to predict mortality of burn patients. The next scoring system to be developed was BOBI which was developed in 2009 [12]. The original Baux score was modified into the rBaux score by Osler et al. in 2010 [13]. The rBaux score adds assessment of inhalation trauma status to the formula calculating the risk of death in burn patients.
Choosing the appropriate scoring system—rBaux, ABSI, or BOBI—depends on the characteristics of the patient population and the clinical context. While all have similar parameters, rBaux excels in detailed assessments involving inhalation injuries that can be used easily in the emergency department, whereas ABSI has a broader application that makes it versatile across more widespread age groups and clinical settings [14]. Our respective cutoff points for ABSI, BOBI, and rBaux at 10.5, 5.5, and 111.25 demonstrated specificity above 98% for all scores compared to other similar studies where the specificity was below 83% [14-16]. Inhalation injury was more prevalent in our study’s mortality group compared to those in the studies of Wardhana et al. and Iustitiati and Nata’atmadjaa (69.4% vs. 20.1% vs. 50.0%), which may have contributed to the higher score cutoff points, and, as a result, higher specificity compared to Wardhana et al. and Iustitiati and Nata’atmadjaa [14,15]. In addition, the proportion of full-thickness burns was also higher in our study than that of Wardhana et al. (88.70% vs. 60.45%). Wardhana et al. previously reported that inhalation injury, along with full-thickness burn injury and TBSA >35.75% were significant predictors for mortality in burn patients [16]. Our study also emphasized the importance of identifying inhalation injury and full-thickness burn injury as a mortality predictor; however further analysis is still necessary to justify expansive application of these findings. The characteristics of our sample group may have also acted as a factor in our higher mortality rate. Our hospital is the only tertiary center that has a specialized burn care unit in the region, therefore patients may present to our hospital with particularly severe burn injuries. A similar phenomenon was also observed in a previous study at Arizona Burn Center, which also reported lower sensitivity due to a larger proportion of high mortality burn cases [17].
The comparison of the three different scoring systems on burn patients in this study demonstrates that they all can predict mortality effectively and that various factors including age, sex, burn severity, presence of inhalation injury, and extent of burn area, are important to consider. Its simplicity notwithstanding, the rBaux score is as effective as other scoring systems when predicting mortality, which takes into account three variables, TBSA, age and the presence of inhalation injury. For this reason, it can easily be used in any medical facility and is ideal to use in emergency room settings to quickly identify when the risk of mortality is high [14].
However, this study also has several limitations, such as the single-center design, small sample size, and unbalanced high mortality rates. Further research, particularly multicenter studies with a more diverse patient population and proportional group distributions, is essential for refining these predictive models and ensuring their applicability across various clinical settings. Improved scoring systems could enhance early identification of high-risk patients, ultimately guiding more effective clinical interventions and improving patient outcomes in burn care.
Notes
No potential conflict of interest relevant to this article was reported.