Significance of Early Acidosis in Predicting Early Mortality Among Flame Burned Patients in a Kenyan National Hospital


 Background: Burn pathophysiology, fluid therapy and mortality have been assessed by various laboratory parameters including lactate and base deficit serum markers. This study targeted flame injured cohort with an objective to determine the significance of early acidosis, through measurement of three acid base indices at admission in predicting 5-day (early) mortality among the flame burned patients.Methodology: A cohort study. Eighty flame injured patients presenting within 24 hours of incident were assessed for arterial blood pH, lactate and base deficit (BD). Mortality was recorded until day 28. Receiver operating characteristics curves were drawn; area under curve, cut offs, sensitivities and specificities for BD, lactate and pH were determined for 5-day mortality. The cut offs were used to derive contingency tables for calculation of predictive values. Odds ratios were calculated at 95% confidence interval. 28 day survival curve was generated. Level of significance was <0.05.Results: 28-day mortality was 39%. Five-day mortality was 24%. The latter was predicted by a lactate level of 2.36 mmol/L, BD of -10.05 mmol/L and pH of 7.344 with 75%,74% and 95% sensitivities respectively. Odds of patients dying at these levels or worse were 6.3, 11.4 and 36.9 respectively all with significant p-values.Conclusion: Arterial pH, base deficit and lactate are good predictors of 5-day mortality among fire victims in the Kenyan context.

48% of re victims presenting to a Kenyan National Hospital present with signi cant metabolic acidosis.
24% of the re casualties died within ve days and 39% of them died within 28 days of hospitalization.
Arterial lactate, base excess and pH showed a good prediction of early mortality among ame burned patients.

Background
Burns is a type of trauma that has a icted man since he rst invented the use of re 1.9 Million years ago (1). Major burns are a signi cant cause of morbidity and mortality in many hospitals. In one of the national referral hospitals in Kenya, close to 600 major burns patients were admitted annually. Nearly 40% of these were as a result of open ames out of which 30% succumbed to their injuries. These re related deaths accounted for up to 76% of overall burn mortality (2). Standard initial care included uid and electrolyte therapy, intubation and oxygen supplementation in suspected inhalational injury and wound care. Alkali therapy was not part of standard of care in acidotic ame burned patients.
Well known predictors of mortality among the many determinants of burn related deaths are total burn surface area, depth of the burn and presence of inhalational injury as described by J. Tobiasen et al in 1982 (3). These are more severe in ame burns compared to other burn mechanisms and cause greater derangement of physiology which begin soon after the burn incident (2). Key among these are perturbations in body uids metabolism, peripheral perfusion, gaseous exchange and acute in ammatory response (4,5). The resulting pathophysiology tends to cause acidosis and can be detected by various clinical and laboratory parameters including arterial blood pH, base de cit and lactate levels. These three parameters measured at admission formed the basis of this study. Their signi cance in predicting 5-day mortality was assessed premised on the fact that such early deaths would more likely be due to primary physiological effects of the burn as opposed to secondary complications such as wound sepsis (6).

Methods
This was a prospective observational cohort study which was carried out between April 2017 and February 2018. Data collection spanned six months. It involved ame burned patients aged 13 years or older presenting within 24 hours of re incident to the accident and emergency department of the national referral hospital. All patients received standard of care according to the hospital's protocol. The sample size was determined by Fisher's formula with nite population correction (7). The target population was re casualties who presented to authors' hospital in half a year. Mortality of these patients from previous studies done at the same centre was reported to be 30% (2). Therefore, the calculated sample size was 80 participants.
Requisite approvals, including publication of ndings, were obtained from the institutional research and ethics committee.
Eligible participants were recruited consecutively as soon as they arrived at the accident and emergency department. Written consent was obtained either from the patients or their next of kin as appropriate. Immediately thereafter, total burn surface area (%TBSA) was estimated using Lund-Browder chart. Two mililiters of arterial blood was drawn aseptically into heparinized syringe from the radial or femoral artery and submitted to the laboratory within 20 minutes. Among other parameters, three acid base indices of interest (the pH, base de cit and lactate) were measured using Siemens RapidPoint R 500 blood gas analysis machine. Mortality within the rst 28 days of patients' admission was recorded with special interest of early deaths that occurred within the rst 5 days.
Statistical package for social sciences (SPSS) version 23 was used for analysis. Receiver operating characteristic (ROC) curves were generated to determine cut off values and associated sensitivity and speci city for base excess, lactate and pH. The cut offs were used to derive contingency tables with 5-day mortality as the primary outcome. Positive predictive value (PPV), negative predictive value (NPV) and odds ratios (OR) at 95% con dence interval were calculated. Kaplan Meier survival curve was generated for 28 days' survival. All statistical tests were considered signi cant at a p-value of <0.05.

Arterial blood acid base parameters
Three arterial blood gas parameters were targeted: the pH, base de cit and lactate. These had a mean of 7.30, -7.87mmol/L and 2.91mmol/L respectively [Table3]. The three blood gas parameters were correlated with %TBSA. All were found to worsen with increasing %TBSA on Pearson's moment product correlation [ Table 4]. Prediction of 5-day mortality based on cut off values for the three acid base indices The cut off values for each of the three acid base indices were used to generate contingency tables.
Positive predictive values, negative predictive values and odds ratios were calculated at 95% con dence interval with 5-day mortality as the primary outcome [ Table 5].

Discussion
Burn injury induces complex physiological stress to the victim and untold anguish. Many determinants of burn mortality have been studied and include extremes of age, presence of comorbid conditions and inadequate or delayed resuscitation (8, 9). However, strong determinants of burn mortality are %TBSA, depth of the burn and presence of inhalational injury (3). The resulting burn pathophysiology lead to derangements in uid metabolism, circulatory and respiratory perturbations, systemic in ammatory and metabolic responses. These alterations are worse in re injuries due to greater thermal energy transfer and the potential for inhalational injury. These account for many deaths in re victims which tend to occur at the scene of incident or in early course of hospitalization compared to deaths from other burn mechanisms (6, 10).
The burned skin loses its uid conservation quality resulting in transdermal exudation of uid and electrolytes (11). Moreover, blood vessels in the burned tissue suffer endothelial necrosis and breakdown causing further uid and even blood loss (12). Thirdly, the burn injury initiates acute in ammatory process locally which is associated with vasodilation, increased vascular permeability and thereby third spacing of body uids. The locally released cytokines may lead to systemic in ammatory response syndrome (SIRS). Furthermore, catecholamine surge due to the injury causes vasoconstriction that increases total peripheral resistance (13). These factors lead to contracted vascular volume and reduced perfusion resulting into hypovolemic and redistributive shock of various degrees. A reduction of oxygen tension in tissues and accumulation of metabolic waste in the interstitial uid follows. If present, physicochemical inhalation injury to the respiratory epithelium leads to ventilation-perfusion mismatch which worsens hypoxemia. Besides, should there be asphyxiants such as carbon monoxide in the inhaled smoke, further hypoxemia results. These perfusion, in ammatory and respiratory disturbances lead to metabolic derangements including acidosis which was the target for this study.
The basic burns pro le and mortality ndings in this study were similar to prior studies carried out at the same institution involving the ame injured cohort. The mean %TBSA was 30.9%. Overall 28-day mortality was 38.8%. Sixty-one percent of the deaths occurred within the rst ve days and additional 7% occurring by day 7 of admission.
All the three acid base indices taken at admission were noted to worsen with increasing %TBSA indicating worsening pathophysiology. At derived cut offs, odds of early (5-day) mortality were signi cantly high.
Base de cit was found to be a good predictor of 5-day mortality at cut off value of -10.05 mmol/L (p<0.001) with odds ratio of 11.4 (95% CI, 3.4-37.9). Choi et al reported that abnormal increase in BD (<-6 mmol/L) after a burn injury represents mal-perfusion state which may not be recognized by traditional parameters such as urine output (5). Mutschler et al lent credence to this observation asserting that BD increases with deepening degrees of hypovolemic shock in trauma patients. He noted that a BD of ≤ -10mmol/L represented a class IV hypovolemic shock (14). Choi et al reported that patients with a BD of < -6 mmol/L recorded within 24 hours of a burn injury had a more orid SIRS, were more prone to acute respiratory distress syndrome and experienced more severe multiple organ dysfunction syndrome hence a higher mortality outcome compared to those with BD of >-6 mmol/L (5). Our study echoed Choi's and Mutschler's nding such that a BD of -10.05mmol/L predicted mortality with a sensitivity of 73.7% and speci city of 80.3%.
Cochran et al reported that an early rise in serum lactate was an independent risk factor for death after a burn injury (15). This was demonstrable in our study where at 2.36 mmol/L or more, lactate showed an acceptable discrimination for 5-day mortality with 6.3 times more likelihood of death compared to patients with <2.36 mmol/L (95% CI, 1.1-36.9). Therefore, just as BD, lactate elevations at day 0 is a useful parameter to separate survivors from non survivors of burns a position also observed by Andel et al (16).
The BD and lactate lower the blood pH. A pH of 7.344 or less predicted early mortality with a sensitivity of 94.7% and speci city of 67.2% (p<0.001). The odds of burn patients dying at these levels was 36.9 times more than those with a higher pH (95% CI,4.6-296.4). A higher pH predicted 5-day survival of up to 97.6%.
From the foregoing, the three acid base parameters discussed can be utilized to identify burned patients at higher risk of death. Additionally, literature highlights importance of lactate and BD in particular, in estimating uid requirement and adequacy of resuscitation in major burns. However, except for the work by Peaston, in 1968 there is paucity of data on the utility of alkali therapy in treatment of acidosis in burns as applied to other causes of severe acidosis (14,16,17,18). Therefore, whether or not alkali therapy can help better outcomes in severe acidosis associated with ame burns is a potential area of further study.

Conclusion
Flame burned patients undergo complex pathophysiological processes that explain the tendency for an elevated arterial blood base de cit (BD), a raised lactate and a lower pH. Measured at admission, these parameters can predict ve-day mortality with good sensitivity and high odds of death at cut offs of -10.05 mmol/L, 2.36 mmol/L and 7.344 respectively. Therefore, these tests can help identify early, burned patients at higher risk of death and help inform and guide a more vigorous early resuscitation and enhanced treatment. Moreover, it is plausible to deduce that alkali treatment may be a useful addition to the resuscitation therapy administered to selected burns patients that present with severe acidosis.

Study Limitation
Only 39 of the 80 recruited participants had their samples tested for lactate due to frequently unavailable reagent for lactate measurement at the study hospital. None the less, the results and analysis obtained were signi cant and therefore included in this article.

Recommendations
Findings of this study may be used to identify physiologically high risk ame injured patients for a more aggressive uid therapy and enhanced treatment. A question worth further research is whether alkali therapy may have a useful role in burn related severe acidosis. Declarations Acknowledgement Wycliffe Ayieko of School of Public Health, University of Nairobi is acknowledged for data analysis and Ms Esther Kimaru, an accident and emergency care nurse at Kenyatta National Hospital, is acknowledged for assistance in data collection.

Funding
This research did not receive any speci c nancial or non-nancial grant from funding agencies in the public, commercial or not for pro t sectors.
Authors' contribution ENM was the principal investigator, drafted the manuscript and is the Corresponding author.
SOK did the concept paper and edited the draft manuscript.
JMN reviewed literature and did the second manuscript.
COA helped interpret and contextualise the laboratory data and typeset the nal draft of the manuscript.

Availability of data and materials
The data set for this study is available on reasonable request to the corresponding author.

Competing interests
Authors declare that they have no competing interests.
Ethics and consent to participate The joint Kenyatta National Hospital and University of Nairobi Ethics and Research Committee's approval was obtained to carry out the study and to publish the results under reference number P744/10/2016. ROC curve for arterial blood pH vs 5-day mortality