Prediction of Mortality After Major Burn: Physiological Versus Biochemical Measures

Objective. To compare the predictive power of serum lactic acid (LA) and physiological score (Tobiasen’s Abbreviated Burn Severity Index [ABSI]) after a major burn. This prospective, analytical study was conducted at the Liaquat University of Medical and Health Sciences (LUMHS) Burn Unit (Jamshoro, Pakistan) from June 2007 to May 2008. Methods. Eighty adult patients presenting with burn injuries of varying etiology with body surface area (BSA) of ≥ 20% were enrolled in the study. The outcome measures included acute phase death (≤ 3 days), and death within first week, second week, and fourth week of admission. The ABSI score and lactate values were stratified into 4 groups to facilitate comparison. The correlation of outcome variables with independent variables was analyzed to measure linear association with Pearson’s correlation coefficient. The proportionality of differences in hazard ratios was tested by the Cox proportional hazard method. The receiver-operator characteristics (ROC) curve analysis using the area under the curve (AUC) was determined using ABSI ≥ 12 and LA ≥ 4.0 mmol/L as a cutoff for a “positive test.” Results. The mean age of patients (47 men and 33 women) was 31 years. The mean body surface area affected was 42%. Mean BSA affected for patients who died was 64.9%. A statistically significant negative correlation of death during acute phase death and within the first week of admission was observed for LA compared to ABSI. When the Cox proportional hazard model was constructed, LA was found to be statistically significant (P = 0.001). The difference between areas under receiver-operator characteristics (AUROC) was insignificant. Considering ABSI and LA 4 mmol/L or greater as a “positive test,” LA was found to have a sensitivity of 89.7% at 100% specificity, while ABSI has a 79.3% sensitivity at 100% specificity. Conclusion. This study has once again shown the clinical usefulness of ABSI. Measurement of serum LA has emerged as a new promising approach and a predictive tool for early death after major burn.

Although the chances of survival after burn injury have steadily increased over the last 3 decades, the prediction of mortality from burn injury is still a subject of interest for burn surgeons. This is due to the fact that an accurate method that quantitatively summarizes severity of burn injuries has various practical applications. It provides clinicians not only a base for clinical decisions, but also assists in understanding the relative contributions of different prognostic criteria. These estimates would also be useful to patients’ families and medical professionals making medical and financial decisions regarding their care.¹ The simplest formula for the prediction of mortality from burns should be based on a minimal set of easily obtainable variables. A classic example that calculates the percent likelihood of mortality is the patient’s age in years plus the percentage of the body-surface area (BSA) that was burned.² This formula, while easy to remember, is only useful for early outcomes assessment and has various limitations regarding long-term survival or mortality prediction. The Abbreviated Burn Severity Risk Index (ABSI)³ has been a more reliable and more frequently used formula in clinical practice for more than 15 years.

Mortality prediction in individual patients by any scoring system is limited and in general, no better than good clinical judgment. Therefore, decisions for individual patients should never be based solely on a statistically derived injury severity score. Conversely, it can be assumed that biochemical analysis would have much better practical implication when compared to the statistically derived injury severity score since it eliminates the human element.

The role of LA as a frequent cause of acute metabolic acidosis is well established. It has recently been shown that elevated serum LA is a predictor of mortality in burn injuries.^4,5 A raised serum LA level is associated with an increased risk of death in patients with infection6 and sepsis^7–11—a common complication in burn victims who sustain acute phase burn shock. The Surviving Sepsis Campaign^12,13 suggests that serum LA measurement can help identify patients at high risk of death so that they can be managed more aggressively.¹⁴ Therefore, measurement of serum LA can be helpful not only to predict early mortality, but also to identify deaths as result of infection and sepsis.

This prospective case series was conducted at the Burn Emergency Unit at Liaquat University of Medical and Health Sciences (Jamshoro, Pakistan) from June 2007–May 2008. During the study period, adult patients presenting with major burn injury (N = 80) from different etiology having body surface area (BSA) affected 20% or greater were enrolled after obtaining written consent.

The BSA was calculated with Browder Charts. Resuscitation was done with the Parkland formula; the adequacy of resuscitation was monitored with urine output and mean arterial pressure. The ABSI recorded and the sample for LA was taken soon after admission. The sample for LA measurement collected in a blood sample bottle containing a small amount of iodoacetate (0.5 g/L). The iodoacetate stabilizes LA by inhabiting the glycolysis and lactic dehydrogenase activity without inducing pyruvate dehydrogenase. It also stabilizes the sample at room temperature for 160 minutes.¹⁵ The test performed on a HITACHI automated analyzer using a sera kit from ROCHE.

The outcome measures considered included the acute phase death (≤ 3 days) and death within the first, second, and fourth week of admission.

Statistics
The ABSI score and lactate values were stratified into 4 groups to facilitate comparison. ABSI score 2–7, 8–9, 10–11, ≥ 12 and lactate as low 0.0–2.0, intermediate 2.1–3.0, high 3.0–4.0, and very high > 4 mmol/L. The correlation of outcome variable with independent variables was analyzed to measure linear association with Pearson’s correlation coefficient. The Cox proportional hazard method was used to test proportionality of differences in the hazard ratio. ABSI ≥ 12 and LA ≥ 4.0 mmol/L were used as the cutoffs for a “positive test” to determine the sensitivity/specificity and ROC.

Table 1

The mean age of patients in this series was 31 years with a minimum of 18 to a maximum of 71 years. The male (47) to female (33) ratio was 1.43:1. The mean body surface area affected was 42% (range 21%–93%). In 30 cases, BSA affected was more than 40%. The most common cause of burns in this series was flame (58/80, 72.5%) followed by scald (18/80; 22.5%), chemical (2/80; 2.5%), and electrical (2/80; 2.5%). The total BSA affected with respect to age, serum LA, and ABSI are shown in Tables 2, 3, and 4, respectively.

The mean time elapsed between incidence and admission was 127.8 minutes (minimum 30, maximum 240) while mean time to sample collection for analysis after admission was 24 minutes (minimum 10, maximum 45).

During the study, 80 patients were admitted for acute burn injuries resulting from various causes, of whom 29 patients expired within 30 days of their admission. The acute phase deaths (within 72 hours of incidence) accounted for 31.1% of all deaths; deaths within first week (31.1%), second week (20.7%), at 30 days (17.3%). Mean BSA for patients expired during the period of study was 64.9%. The overall mortality was 36.3% (Table 5). Death within the first 7 days of admission accounted for 62% of all deaths.

The data were analyzed with Pearson’s correlation coefficient to measure the linear association between deaths with LA and ABSI. The outcome measures showed a negative linear relation with the independent variable.

A strong negative correlation of death during acute phase and first week (-0.679, -0.762) was found for LA when compared to ABSI (-0.270, -0.671; Table 6). However, for deaths between day 15 and day 30 from admission, the difference between correlation for both LA and ABSI was statistically insignificant.

When the Cox proportional hazard model was constructed, LA was found to be statistically significant (P = 0.001) with respect to the hazard ratio when compared to the ABSI (P = 0.982). The Exp (B) in Table 7 shows that the predicted change in the hazard is likely to increase 8-fold with a unit increase in the serum LA as compared to ABSI where it is less then the double. The chance of mortality increases 8-fold when serum LA increases from 3 mmol/L to 4 mmol/L. However, with changes in ABSI score the results are not expected to be as dramatic.

Serum LA and ABSI are considered diagnostic tests. The performance of these diagnostic variables was quantified by calculating the area under the ROC curve (AUROC). The ideal test would have an AUROC of 1, whereas, a random guess would have an AUROC of 0.5. Considering both as diagnostic tests, they were found to be almost equally effective with respect to 30-day mortality (Table 5, Figure 1).

ABSI ≥ 12 and LA ≥ 4.0 mmol/L were used as the cutoffs for a “positive test” to determine the sensitivity/specificity and ROC. Lactic acid found 89.7% sensitive while ABSI was 79.3% sensitive at 100% specificity (Table 8).

At 100% specificity, ABSI is 80% sensitive to predict death.

At 100% specificity, serum LA is 90% sensitive to predict death.

Improvements in intensive care over the last 20 years have resulted in significant reduction in mortality from a major burn. However, an extensive burn still carries high mortality. The overall mortality in this series was high in contrast to other relevant studies.^16–18 A simple explanation could be that this study was designed exclusively to assess the probability of death, and therefore, patients who had major burns were recruited. The mean BSA affected in this series of patients was 42% (50 cases with BSA between 21%–40% and 30 cases between 41%–93% BSA). This is in contrast to other studies^16–18 and that of Ryan et al¹⁹ where 77.5% of patients had an affected BSA of less than 20%. Inclusion of such low-risk cases yields no useful information regarding probability of death.

The results of the present study have shown that the increasing body surface was associated with increased risk of death (> 0.05). The results of this study show that LA, when compared to ABSI, is most sensitive to predict acute phase death and deaths within the first week of hospitalization (62% of all deaths), a finding that is consistent with the results of Choi et al²⁰ and Jeng et al.²¹ When considering deaths between the second and fourth weeks (38%), the difference between predictive powers of serum LA and ABSI were statistically insignificant.

The finding that LA is a sensitive predictor of early death will help redirect future management protocol in individual cases at the very early stage with an overall improvement in burn mortality. Of the total patients who died in the course of the present study, 18 (62%) died within the early period. In the future, a cohort of patients with a higher level of serum LA will be given more attention. Lactic acid was found to be more sensitive when compared to ABSI as a predictor of early burn mortality; however, as predictors of 30-day mortality, both LA and ABSI had almost identical sensitivity (100% specificity).

The scoring systems to predict morbidity/mortality of patients with burns have received increasing acceptance in recent years. These scoring systems help in understanding the relative influence of different prognostic variables on disease processes, for stratification of patient groups for various treatment modalities, and classification of severity of injury with reduced reliance on clinical suspicion. These play a vital role in the evaluation of economic burden of treatment and facilitate multicenter studies. It is important to recognize that for individual patients, scoring systems can never be a replacement for clinical decision-making. Any predictive scales for survival should assist in redirecting appropriate levels of care to maximize survival where feasible.

While the ABSI has been used in clinical burn practice for only the last 15 years, Karl Wilhelm Scheele first identified LA in sour milk in 1780 and German physician-chemist Johann Joseph Scherer demonstrated LA in human blood under pathological conditions in 1843 and 1851.²² Further investigations labeled LA simply as a metabolic dead-end waste product of glycolysis due to hypoxia. Resurgence of interest in LA arose when various studies demonstrated and established that elevated LA is one of the most common metabolic abnormalities in critically ill patients. Soon LA was being used as a diagnostic, therapeutic, and prognostic marker of tissue hypoxia in circulatory shock.^23–25 In shock, anaerobic production and impaired LA utilization (or both) can be operative.²⁶ Serum LA has recently been associated with an increased risk of death in patients with infection⁶ and sepsis.^7–11 The raised LA in body fluid is also helpful in diagnosing infection. Brook²⁷ showed that raised LA in peritoneal fluid is the hallmark of peritonitis. Soon Reynaert et al²⁸ described that peritoneal fluid to blood lactate gradient levels has a better correlation for detection of peritoneal infections. The marked rise in LA in cerebrospinal fluid is indicative of meningitis, but also can help differentiate bacterial from viral meningitis.^29,30

Early recognition of elevated blood lactate levels appears to be essential, as early interventions targeted on hemodynamic endpoints can decrease mortality by stratifying patients with major, extensive burns. It is not yet known whether interventions targeted specifically to normalize blood lactate concentrations can play any role in improving outcomes.

This study confirms the clinical usefulness of ABSI. Measurement of serum LA has emerged as a promising new predictive tool for early death after a major burn. We believe that lactate should always be measured routinely after a major burn, as it may help to detect critically ill patients either for adequacy of treatment or selection of other therapeutic options.

Address correspondence to: Dr. SM Tahir, FCPS, Department of Plastic & Burn Surgery, Liaquat University of Medical & Health Sciences, Jamshoro 35 Gulshan-e-Yasin, Unit No. 09 Latifabad, Hyderabad, Sindh Pakistan; Phone: 0300-3018532; E-mail: Syedsahib1@yahoo.com