Abstract
Introduction:
This global burden of burn injury is suffered disproportionately by people in low- and middle-income countries, where 70% of all burns occur. Models based in high-income countries to prognosticate burn mortality treat age as a linearly increasing risk factor. It is unclear if this relationship is similar in resource-limited settings.
Methods:
We analyzed patients from the Kamuzu Central Hospital Burn Registry in Lilongwe, Malawi from 2011-2019. We examined the relationship between burn-associated mortality and age using adjusted survival analysis over sixty days, categorized into group groups: 1) younger children <5 years 2) older children 5 to 17 years 3) adults 18-40 years 4) older adults >40 years.
Results:
2,499 patients were included. Most patients were <5 years old (n=1,444) with only 133 patients >40 years. Older adults had the highest crude mortality at 34.6% and older children with the lowest at 13.%. Compared to younger children, the hazard ratio adjusted for sex, %TBSA, and operative intervention was 0.59 (95% CI 0.44, 0.79) for older children and 0.55 (95% CI 0.40, 0.76) for adults. Older adults were statistically similar to younger children.
Conclusions:
We show in this cohort study of burn-injured patients in a resource-limited environment that the relationship between mortality and age is not linear and that the use of age-categorized mortality prediction models is more accurate in delineating mortality characteristics. Categorizing age based on local burn epidemiology will help describe burn mortality characteristics more accurately, leading to better-informed management strategies aimed at attenuating burn mortality for different populations.
Keywords: burn, sub-Saharan Africa, age, injury
Introduction
There are an estimated 11 million new burn injuries globally each year. This burden of burn injury is suffered disproportionately by people residing in low- and middle-income countries (LMICs), where 70% of all burns occur.1 The three WHO regions with the greatest prevalence of injury are the Eastern Mediterranean Region, the South-East Asian Region, and the African Region, with the African Region bearing nearly two-thirds of the total burden.2 The epidemiology of burn injuries in LMICs countries is unique compared to those in high-income countries (HICs) with studies from sub-Saharan African (SSA) countries demonstrating that up to 80-90% of burns occur in the home.3–6 Furthermore, data has shown that children, especially those under 5 years old and adults over 60 years, constitute the highest risk group of burn victims.7–9 Given the increased incidence of burn injury in SSA and the limited health care expenditure, burn-related mortality remains high.
Models to prognosticate burn mortality are ubiquitous in the burn literature. One of the most commonly utilized models is the Baux Score.10, 11 Almost all models include its three variables in their analysis of burn outcomes: age, percent total body surface area (%TBSA) burn, and inhalation injury. Typically, the entire spectrum of age and %TBSA are included in the model. However, because so many of these models were developed in high-income countries, they may not be generalizable in a resource-limited setting. This is especially true in regions like SSA, which lack reliable resources for burn care, putting certain groups like young children at risk of worse clinical outcomes than expected in high-resource settings.12, 13
Malawi, is a small, impoverished country in southeast Africa with a per capita gross national income (GNI) of just US$360 in 2018, the second lowest in the world.14 Malawi also has a high burn prevalence, particularly in young children, with a high burn mortality.15 The health care system is similar to the surrounding countries, allowing Malawi to serve as a proxy for burn care and clinical outcomes in the region. We hypothesized that age variably impacts outcomes in burns and age-specific models based on categories will more accurately predict mortality compared to models which use age as a continuum or single model for all ages. We, therefore, sought to characterize the effect of age when categorized into four cohorts: younger children, older children, adults, and older adults.
Materials and Methods
This study is a retrospective analysis of a prospectively collected burn registry data at Kamuzu Central Hospital (KCH), a public, 900-bed tertiary care hospital in the capital city of Lilongwe, which serves as a referral center for approximately 7.5 million people in the central region of Malawi. The KCH Burn Unit was established in 2011 and averages 25-40 admissions per month.16 Pediatric and adult patients are admitted to the same unit but separate areas. The KCH Burn Unit is a 31-bed unit with five full-time nurses, and two trained clinical officers, and a consultant plastic surgeon. The burn registry was established in June 2011 to record patient information and clinical outcomes, including demographics, burn injury characteristics, type of operative intervention, and clinical outcomes.
We included all patients admitted to the burn unit at KCH between June 2011 and December 2019 and had their age recorded in the registry. The primary aim of this study was to examine the relationship between age at the time of burn injury and burn-associated mortality. The outcome, crude mortality, was calculated from all in-hospital mortality among burn patients admitted to our burn unit. Only in-hospital mortality was included due to a lack of available data on out of hospital deaths. Age was defined as the exposure and was analyzed both as a continuous variable in years and as a categorical variable. The categorical age variable was defined as four groups: 1) younger children < 5 years 2) older children 5 to 17 years 3) adults 18-40 years 4) older adults > 40 years. The age forty was used to delineate older adults based on the life expectancy in Malawi of approximately 60 years.17 Children were separated at age 5 because most burn injuries at our center occur in those less than five and this group has special physiological considerations with resuscitation due to their young age.
We examined the characteristics of the study population by assessing the distribution of variables within the four study cohorts. We examined categorical and continuous variables by calculating the frequencies of categorical variables and the distribution of continuous variables. Bivariate analysis was used to compare these variables across the study cohorts and identify potential confounders of the relationship between age and burn-associated mortality. We used Pearson’s correlation for the categorical variables and 2-sample t-tests or one-way analysis of variance for continuous variables. Medians of non-normally distributed continuous variables were tested using a Kruskal-Wallis test. Means were reported with standard deviations (SD) and medians with interquartile ranges (IQR).
We used survival analysis to estimate the hazard ratio for in-hospital mortality at sixty days for the older children (5 to 17 years), adults (18-40 years), and older adults (> 40 years) using the younger children group (< 5 years) as the referent. We initially plotted an unadjusted Kaplan-Meier curve comparing the four patient groups and then utilized a Cox regression model to estimate the mortality hazard ratio between the four cohorts, adjusting for confounders. Sixty days was used as the time frame due to the extended period of hospitalization burns patients have at our center. Among potential confounders, we accessed and corrected for violation of the proportional hazard assumption. A potential confounder was included in the model if it substantially affected the adjusted hazard ratio. An adjusted hazard ratio and 95% confidence interval compared to the referent are reported. In addition, an adjusted Kaplan-Meier curve using the same potential confounders is reported.
For comparison, we also analyzed the effect of utilizing age as a continuous variable. We used a logistic regression model to plot the predicted probability of death versus increasing age. We adjusted the model based on potential confounders identified in our survival analysis. We initially examined this relationship assuming a linear relationship between age and mortality. However, we also tested whether there was a better model fit after the addition of a polynomial term, testing both quadratic and cubic terms.
All statistical analysis was performed using Stata/SE 16.0 (Stata- Corp LP, College Station, TX). Ethics approval and a waiver for informed consent was obtained from the University of North Carolina Institutional Review Board and the Malawi National Health Sciences Research Committee.
Results
During the study period, 2,527 patients were admitted to the KCH Burn Unit, with 28 patients (1.1%) missing a recorded age, leaving 2,499 total patients included in the analysis. Median age was 3 years (IQR 2-11) and 58.0% were males (n=1,447).
The age groups were as follows: young children (<5 years): 1,444 patients, 57.8%; older children (5-17 years): 541 patients, 21.7%; adults (18-40 years): 381 patients, 15.3%; older adults (>40 years): 133 patients, 5.3%. The baseline characteristics of the four groups are shown in Table 1. There were minor differences in sex, with a higher proportion of males in the older age groups. The time to presentation was very different between the four groups. Both children cohorts presented earlier than adults, with 72.0% of younger children presenting within 24 hours and 60.6% of older children, compared to only 55.5% of adults and 47.4% of older adults (p<0.001). Over a third of adults (34.7%) and almost half of older adults (45.1%) presented after 48 hours.
Table 1.
Characteristics of patients who suffered burn injury by age.
| Younger Children < 5 years (n=1,444) |
Older Children 5-17 years (n=541) |
Adults 18-40 years (n=381) |
Older Adults > 40 years (n=133) |
p value | |
|---|---|---|---|---|---|
| Sex: N (%) | |||||
| Female | 605 (41.9) | 259 (47.9) | 138 (36.2) | 45 (33.8) | 0.006 |
| Male | 835 (57.8) | 281 (51.9) | 243 (63.8) | 88 (66.2) | |
| Missing | 4 (0.3) | 1 (0.2) | 0 (0.0) | 0 (0.0) | |
| Use of Traditional Medicine | |||||
| Yes: N (%) | 155 (10.7) | 50 (9.2) | 25 (6.7) | 11 (8.3) | 0.08 |
| Time to Presentation | |||||
| 0-24 hours | 1,039 (72.0) | 328 (60.6) | 211 (55.4) | 63 (47.4) | <0.001 |
| 24-48 hours | 43 (3.0) | 31 (5.7) | 29 (7.6) | 9 (6.8) | |
| > 48 hours | 333 (23.1) | 170 (31.4) | 132 (34.7) | 60 (45.1) | |
| Missing | 29 (2.0) | 12 (2.2) | 9 (2.4) | 1 (0.7) | |
| Cause of Burn Injury: N (%) | |||||
| Cooking Related | 709 (49.1) | 165 (30.5) | 79 (20.8) | 22 (16.5) | <0.001 |
| Clothes Caught Fire | 147 (10.2) | 103 (19.0) | 40 (10.5) | 26 (19.6) | |
| Fell into Fire | 99 (6.9) | 128 (23.7) | 125 (32.8) | 58 (43.6) | |
| House Fire | 14 (1.0) | 6 (1.1) | 27 (7.1) | 4 (3.0) | |
| Explosion | 3 (0.2) | 7 (1.3) | 25 (6.6) | 3 (2.3) | |
| Mob Justice | 0 (0.0) | 1 (0.2) | 14 (3.7) | 0 (0.0) | |
| Other1 | 393 (27.2) | 117 (21.6) | 68 (17.9) | 17 (12.8) | |
| Missing | 79 (5.5) | 14 (2.6) | 3 (0.8) | 3 (2.3) | |
| Type of Burn: N (%) | |||||
| Scald Burn | 1,128 (78.1) | 248 (45.8) | 101 (26.5) | 24 (18.1) | <0.001 |
| Flame Burn | 285 (19.7) | 275 (50.8) | 253 (66.4) | 106 (79.7) | |
| Other | 23 (1.6) | 14 (2.6) | 25 (6.6) | 3 (2.3) | |
| Missing | 8 (0.6) | 4 (0.7) | 2 (0.5) | 0 (0.0) | |
| % Total Burn Surface Area (TBSA) Burn | |||||
| Median (IQR) | 13 (8-20) | 14 (7.5-22) | 15 (7.5-26) | 18 (9-30) | 0.02 |
| Inhalation Injury | |||||
| Yes: n (%) | 0 (0.0) | 1 (0.2) | 0 (0.0) | 0 (0.0) | 0.7 |
| Underwent Excision and/or Grafting | |||||
| Yes: N (%) | 200 (13.9) | 157 (29.0) | 130 (34.1) | 45 (33.8) | <0.001 |
| Hospital Length of Stay (Days) | |||||
| Median (IQR) | 10 (6-21) | 15 (6-39) | 21 (8-56) | 28 (5-67) | <0.001 |
Other causes of burns include: falls into hot liquid, domestic abuse, contact with hot objects, and unknown causes.
Among younger children, 49.1% had a cooking-related burn injury compared to just 30.5% in older children, 19.7% in adults, and 16.5% in older adults (p<0.001). Falling into a fire, house fires, explosions, and mob justice were all rare in the children cohorts but were a relatively common cause of burn in both adult cohorts. Scald burns were the majority of burn types for younger children at 78.1% compared to just 45.8% in older children, 26.5% in adults, and 18.1% in older adults (p<0.001) with flame burns much more common in the latter cohorts. 66.4% of adults and 79.7% of older adults had a flame burn. Median %TBSA ranged from 13 to 18, increasing with each age group (p=0.02). Far fewer younger children underwent excision and/or grafting at only 13.9% compared to 29.0% in older children, 34.1% in adults, and 33.8% in older adults (p<0.001). Hospital length of stay was also shorter for younger children with a median of 10 days (IQR 6-21) compared to 15 days (IQR 6-39) in older children, 21 days (IQR 8-56) in adults, and 28 (IQR 5-67) days in older adults (p<0.001).
In-hospital crude mortality was significantly different between the four age groups. Crude mortality was 17.0% (n=246) in the younger children cohort, 13.5% (n=73) in the older children cohort, 18.6% (n=71) in adults, and 34.6% among older adults (p=0.006). Initially, we used unadjusted Kaplan Meier survival analysis curve to compare 60-day in-hospital mortality among the four groups. The unadjusted curve demonstrated a survival advantage over time for the older children compared to adults and younger children, with older adults having the worst mortality over time. (Figure 1). The unadjusted Cox proportional-hazards model revealed a HR of 0.71 (95% CI 0.54, 0.92) for older children compared to younger children. For adults, the unadjusted HR was not significant at 0.96 (95% CI 0.73, 1.27) compared to younger children, while older adults had a higher risk of mortality with a HR of 2.05 (95% CI 1.46, 2.85). We repeated this analysis, adjusting for %TBSA, the presence of flame burns, and whether the patient underwent excision and/or grafting. Flame burns were used as a surrogate for burn depth because burn depth data was not available. Delays in presentation were not included in the model because it was not significant and did not change the model outcome. The adjusted Kaplan Meier survival analysis curve shows improvement in the adjusted survival curve for adults compared to younger children with a similar curve for older children and older adults. Overall, older adults and younger children are shown to have the worst adjusted survival. (Figure 2). The adjusted Cox proportional-hazards model revealed a HR of 0.59 (95% CI 0.44, 0.79) for older children compared to younger children. For adults, the adjusted HR was 0.55 (95% CI 0.40, 0.76) compared to younger children, showing a similar relative risk as older children. Older adults were not statistically different than younger children with an adjusted HR 1.27 (95% CI 0.86, 1.80).
Figure 1.
Unadjusted Kaplan-Meier Survival Estimates for 60-Day In-Hospital Mortality stratified by age groups: younger children (< 5 years), older children (5-17 years), adults (18-40 years), and older adults (> 40 years).
Figure 2.
Adjusted Kaplan-Meier Survival Estimates for 60-Day In-Hospital Mortality stratified by age groups: younger children (< 5 years), older children (5-17 years), adults (18-40 years), and older adults (> 40 years). Model adjusted for %TBSA, the presence of flame burns, and whether the patient had surgical excision and/or skin grafting.
We also explored the relationship between burn-associated mortality and age, treating age as a continuous variable and adjusting for the same variables used in our survival analysis, %TBSA, the presence of flame burns, and whether the patient underwent excision and/or grafting. We used logistic regression to plot the relationship between age and the adjusted predicted probability of death, as shown in Figure 3. This demonstrates a linear increase in the adjusted probability of death as age increases. The likelihood ratio test for this model resulted in a p-value of 0.11. In this model, the adjusted predicted probability of in-hospital death for a four-year-old is 0.14 (95% CI 0.13, 0.16), for a 17-year old 0.15, (95% CI 0.13, 0.18), a 30-year old 0.16 (95% CI 0.13, 0.20), and a 60-year old 0.19 (95% CI 0.12, 0.28). However, when a quadratic polynomial term is added to the logistic regression model, the predicted probability curve more closely aligns with the results shown in our survival analysis, demonstrating a nonlinear relationship between age and the adjusted probability of death. The curve more closely resembles a U-shape, with an increased risk of mortality at a younger age and then climbing again in older ages. (Figure 4). The likelihood ratio test for this model resulted in a p-value of <0.001. In this model, the adjusted predicted probability of death for a four-year-old is 0.14 (95% CI 0.12, 0.16), for a 17-year old 0.08, (95% CI 0.06, 0.11), a 30-year old 0.07 (95% CI 0.05, 0.11), and a 60-year old 0.29 (95% CI 0.18, 0.43).
Figure 3.
Adjusted logistics regression model plotting the adjusted predicted probability of death by area.
Figure 4.
Adjusted logistics regression model plotting the adjusted predicted probability of death by area with a cubic polynomial term.
Discussion
In this study, we show that the effect of age on burn mortality is better delineated when we utilize age-specific categories in regression models. Children less than five years and older adults over the age of 40, had a higher risk of death, while older children and adults had the lowest risk of mortality, independent of burn size. Traditional mortality models have historically been useful when used to predict outcomes for a larger population. Based on conventional modeling, had we considered age as a continuous variable, we would erroneously assume that mortality increases with increasing age without discrimination for extremes of age. The utilization of age as a continuum makes it less accurate in predicting outcomes in very young children and older adults.
In the United States, Taylor et al. analyzed data from the American Burn Association National Burn Repository to assess the independent influence of age as a predictor of mortality after controlling for burn size and inhalation injury.18 They built their regression models using age, either a continuous or categorical variable, with the three age categories being <18, 18-65, and >65 years, respectively. They found that in the pediatric subgroup analysis, predicted mortality decreased with age among children, in contrast to the continuous age model. In the older adult age group, they found that predicted mortality increased with age. Both of these findings are consistent with our results in a resource-limited setting, where younger children and older adults had the worst outcomes.
In Malawi and surrounding countries, it is well known that young children represent the largest age group admitted to the burn unit. 19 We show that infants and toddlers have a higher mortality rate than older children after controlling for burn size. Our findings are consistent with other reports from SSA, which have shown mortality rates as high as 40% among children, with %TBSA as the primary driver of mortality. 20 Reasons for the higher mortality rate in younger children is multifactorial. One reason may be the immunological immaturity seen in children in the post-burn period. IL-17 and GM-CSF, which are both immunostimulant, have levels were significantly lower in the pediatric burn patients when compared to adults for the first-week post-burn.21 Secondly, children have different pathophysiologic responses to burn injury. An understanding of the need for increased fluid resuscitation requirements and nuanced monitoring is essential to optimize their outcomes.22, 23 Lastly, the importance of nutrition and nutritional supplementation cannot be overstated in a burn-injured cohort in a resource-limited setting. In Malawi, the prevalence of underweight children ≥ 5 years living in Malawi increased from 12.1 to 16.7% between 2009 and 2014.24 In 2010, 48% of Malawian children ≥ 5 years were stunted. We have previously evaluated the role of nutritional status on outcome in our burn cohort, and the presence of pre-burn injury malnutrition predicted an increase in burn mortality among children.25 The widespread incidence of malnutrition in our study population likely contributes to worse outcomes for very young children.
Increased mortality in older adults is also well established, as traditional models of age in a linear regression analysis predicts increasing mortality. These models still show that %TBSA and the presence of inhalation injury are strong drivers of mortality. However, the presence of preexisting medical conditions and changing physiological reserves in older adults may portend poorer survival.26 Several studies show the effect of comorbidities, such as renal insufficiency, respiratory diseases, and diabetes mellitus on the increased risk of hospital-acquired infections and mortality.27–29 The relative protection of older children and young adults is likely explained by an absence of the factors previously discussed for young children while also avoiding the comorbidities of older age. Older children and young adults have the advantage of having more developed immune systems, predictable responses to resuscitation, and robust compensatory mechanisms during stress.
We acknowledge some study limitations due to the retrospective methodology. Furthermore, unlike other regions, the number of patients with inhalation injury presenting to our center was negligible, preventing us from accounting for this in our model. Other factors, such as comorbidities in older adults, are also not included in the model given the lack of available medical history in our population. These comorbid conditions likely explain some of the differences shown in our study between adults and young children.
Conclusion
We show in this cohort study of burn-injured patients in a resource-limited environment that the relationship between age and mortality is not linear over the full age spectrum and use of age-categorized mortality prediction models is more accurate in delineating mortality characteristics. Categorizing age based on local burn epidemiology will help describe burn mortality characteristics more accurately, leading to better-informed management strategies aimed at attenuating burn mortality for different populations.
Table 2.
Burn-associated mortality by age group.
| Younger Children < 5 years (n=1,444) |
Older Children 5-17 years (n=541) |
Adults 18-40 years (n=381) |
Older Adults > 40 years (n=133) |
p value | |
|---|---|---|---|---|---|
| Crude In-hospital Mortality | |||||
| N (%) | 246 (17.0) | 73 (13.5) | 71 (18.6) | 46 (34.6) | 0.006 |
| Unadjusted Hazard Ratio of Death | |||||
| Hazard Ratio (95% CI) | -- | 0.71 (0.54, 0.92) | 0.96 (0.73, 1.27) | 2.05 (1.46, 2.85) | <0.001 |
| Adjusted Hazard Ratio of Death | |||||
| Hazard Ratio (95% CI) | -- | 0.59 (0.44, 0.79) | 0.55 (0.40, 0.76) | 1.27 (0.86, 1.80) | <0.001 |
Acknowledgements
Study data were collected and managed using REDCap electronic data capture tools hosted at UNC. REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies, providing: (1) an intuitive interface for validated data entry; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for importing data from external sources.
Funding support: Provided by the Department of Surgery at the University of North Carolina
Financial Support
Financial support was provided by the Department of Surgery at the University of North Carolina for all aspects of the study including: design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article
References
- 1.Mock CPM, Peden M, Krug E, eds. A WHO plan for burn prevention and care. World Health Organization; 2008. [Google Scholar]
- 2.World Health Organization. WHO Health Estimates 2014 Summary Tables: Deaths and Global Burden of Disease 2014 [Available from: http://www.who.int/healthinfo/global_burden_disease/en/]. Accessed: 15 April 2020
- 3.Stokes M, Johnson W. Burns in the Third World: an unmet need. Ann Burns Fire Disasters. 2017;30(4):243. [PMC free article] [PubMed] [Google Scholar]
- 4.Rybarczyk MM, Schafer JM, Elm CM, et al. A systematic review of burn injuries in low- and middle-income countries: epidemiology in the WHO-defined African Region. Afr J Emerg Med. 2017;7(1):30–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nthumba PM. Burns in sub-Saharan Africa: a review. Burns. 2016;42(2):258–66. [DOI] [PubMed] [Google Scholar]
- 6.Ahuja RB, Bhattacharya S. Burns in the developing world and burn disasters. Bmj. 2004;329(7463):447–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tyson AF, Boschini LP, Kiser MM, et al. Survival after burn in a sub-Saharan burn unit: challenges and opportunities. Burns. 2013;39(8):1619–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Diop-Ndoye M, Bodjona J, Diouf E, et al. Management of thermal severe burns in children in Le Dantec University Teaching Hospital. Dakar Med. 2005;50(3):194–7. [PubMed] [Google Scholar]
- 9.Adamo C, Esposito G, Lissia M, et al. Epidemiological data on burn injuries in Angola: a retrospective study of 7230 patients. Burns. 1995;21(7):536–8. [DOI] [PubMed] [Google Scholar]
- 10.Osler T, Glance LG, Hosmer DW. Simplified estimates of the probability of death after burn injuries: extending and updating the baux score. J Trauma Acute Care Surg. 2010;68(3):690–7. [DOI] [PubMed] [Google Scholar]
- 11.Roberts G, Lloyd M, Parker M, et al. The Baux score is dead. Long live the Baux score: a 27-year retrospective cohort study of mortality at a regional burns service. J Trauma Acute Care Surg. 2012;72(1):251–6. [DOI] [PubMed] [Google Scholar]
- 12.Botman M, Beijneveld J, Negenborn V, et al. Surgical Burn Care in sub-Saharan Africa: A Systematic Review. Burns Open. 2019. [Google Scholar]
- 13.Joseph K, Trehan A, Cherian M, et al. Assessment of acute burn management in 32 low- and middle-income countries. World J Surg. 2016;40(4):791–800. [DOI] [PubMed] [Google Scholar]
- 14.The World Bank. Malawi: 2019. The World Bank Group; [Available from: https://data.worldbank.org/country/malawi.] Accessed: 15 April 2020. [Google Scholar]
- 15.Komolafe O, James J, Makoka M, et al. Epidemiology and mortality of burns at the Queen Elizabeth central hospital Blantyre, Malawi. CAJM. 2003;49(11). [PubMed] [Google Scholar]
- 16.Kiser M, Beijer G, Mjuweni S, et al. Photographic assessment of burn wounds: a simple strategy in a resource-poor setting. Burns. 2013;39(1):155–61. [DOI] [PubMed] [Google Scholar]
- 17.The World Bank. “Life expectancy at birth, total (years) - Malawi.” 2019. The World Bank Group; [Available from: https://data.worldbank.org/indicator/SP.DYN.LE00.IN?locations=MW]. Accessed: 15 April 2020 [Google Scholar]
- 18.Taylor SL, Lawless M, Curri T, et al. Predicting mortality from burns: the need for age-group specific models. Burns. 2014;40(6):1106–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Erickson EJ, Merrell SW, Saffle JR, et al. Differences in mortality from thermal injury between pediatric and adult patients. J Pediatr Surg. 1991;26(7):821–5. [DOI] [PubMed] [Google Scholar]
- 20.Olawoye O, Iyun A, Ademola SA, et al. Demographic characteristics and prognostic indicators of childhood burn in a developing country. Burns. 2014;40(8):1794–8. [DOI] [PubMed] [Google Scholar]
- 21.Devine RA, Diltz Z, Hall MW, et al. The systemic immune response to pediatric thermal injury. Int J Burns Trauma. 2018;8(1):6. [PMC free article] [PubMed] [Google Scholar]
- 22.Arbuthnot MK, Garcia AV, editors. Early resuscitation and management of severe pediatric burns Seminars in pediatric surgery; Elsevier; 2019:73–78. [DOI] [PubMed] [Google Scholar]
- 23.Allorto N, Oosthuizen G, Clarke D, et al. The spectrum and outcome of burns at a regional hospital in South Africa. Burns. 2009;35(7):1004–8. [DOI] [PubMed] [Google Scholar]
- 24.Malawi Unicef. “Nutrition Statistics in Malawi: 2018”. UNICEF; 2018. [Available from: https://www.unicef.org/malawi/sites/unicef.org.malawi/files/2018-09/UNICEF_Nutrition_Factsheet_2018.pdf]. Accessed: 15 Aprill 2020. [Google Scholar]
- 25.Grudziak J, Snock C, Zalinga T, et al. Pre-burn malnutrition increases operative mortality in burn patients who undergo early excision and grafting in a sub-Saharan African burn unit. Burns. 2018;44(3):692–9. [DOI] [PubMed] [Google Scholar]
- 26.Knowlin L, Stanford L, Moore D, et al. The measured effect magnitude of co-morbidities on burn injury mortality. Burns. 2016;42(7):1433–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Knowlin LT, Purcell L, Cairns BA, et al. Burn injury mortality in patients with preexisting and new onset renal disease. Am J Surg. 2018;215(6):1011–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Knowlin LT, Stanford LB, Cairns BA, et al. The effect of preexisting respiratory comorbidities on burn outcomes. Burns. 2017;43(2):366–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Knowlin L, Strassle PD, Williams FN, et al. Burn injury outcomes in patients with preexisting diabetic mellitus: risk of hospital-acquired infections and inpatient mortality. Burns. 2018;44(2):272–9. [DOI] [PMC free article] [PubMed] [Google Scholar]