Risk factors for cooking-related burn injuries in children, WHO Global Burn Registry

Abstract Objective To assess the characteristics of cooking-related burn injuries in children reported to the World Health Organization Global Burn Registry. Methods On 1 February 2021, we downloaded data from the Global Burn Registry on demographic and clinical characteristics of patients younger than 19 years. We performed multivariate regressions to identify risk factors predictive of mortality and total body surface area affected by burns. Findings Of the 2957 paediatric patients with burn injuries, 974 involved cooking (32.9%). More burns occurred in boys (532 patients; 54.6%) than in girls, and in children 2 years and younger (489 patients; 50.2%). Accidental contact and liquefied petroleum caused most burn injuries (729 patients; 74.8% and 293 patients; 30.1%, respectively). Burn contact by explosions (odds ratio, OR: 2.8; 95% confidence interval, CI: 1.4–5.7) or fires in the cooking area (OR: 3.0; 95% CI: 1.3–6.8), as well as the cooking fuels wood (OR: 2.2; 95 CI%: 1.3–3.4), kerosene (OR: 1.9; 95% CI: 1.0–3.6) or natural gas (OR: 1.5; 95% CI: 1.0–2.2) were associated with larger body surface area affected. Mortality was associated with explosions (OR: 7.5; 95% CI: 2.2–25.9) and fires in the cooking area (OR: 6.9; 95% CI: 1.9–25.7), charcoal (OR: 4.6; 95% CI: 2.0–10.5), kerosene (OR: 3.9; 95% CI: 1.4–10.8), natural gas (OR: 3.0; 95% CI: 1.5–6.1) or wood (OR: 2.8; 95% CI: 1.1–7.1). Conclusion Preventive interventions directed against explosions, fires in cooking areas and hazardous cooking fuels should be implemented to reduce morbidity and mortality from cooking-related burn injuries.


Introduction
Globally, burn injuries remain among the most prominent and preventable injuries in children. Childhood burns can have devastating physical, psychological and socioeconomic consequences. 1 Cooking-related burns have emerged as a particularly damaging subset of childhood burns, representing up to an estimated 85% of childhood scalds, and these types of burns carry a greater patient injury burden and long-term morbidity than burns of other etiologies. [2][3][4] Young children and girls are more likely to experience paediatric burns. 1,[5][6][7][8] Low socioeconomic status and its associated circumstances, such as lack of water supply and volatile cooking materials, are also reported to be key risk factors for childhood cooking burns. 9,10 Research on paediatric cooking burns has been largely restricted to local or national studies, limiting generalizability of conclusions, especially to global systemic factors underlying disparities in burn injuries. 4,5,11,12 Wide variability in research methods has made meaningful comparisons of clinical outcomes and risk factors between high-and lowincome countries difficult. While rapid declines in mortality from childhood burns have been reported in high-income countries over the past decades, 13 the same trend has not been documented in low-and middle-income countries. 9 We aim to address this gap in the literature by analysing data from the first large-scale, global data repository for burn injuries, the World Health Organization (WHO) Global Burn Registry. 14 The registry, launched in January 2018, is a global instrument to standardize the reporting of hospital admissions due to burn injury across the world and was developed for public health officials and researchers after a thorough evaluation of existing national and international burn registries (Box 1). The registry comprises data submitted voluntarily by hospital staff members. A study evaluating the instrument reported that mostly physicians entered data into the system, followed by nurses. 15 The registry records hospital admissions, injury characteristics and patient demographics. The aim of our research was to identify the key risk factors predictive of paediatric cooking burns and mortality following these burns by using the data submitted to the registry, and to inform strategies for prevention and management.

Data source
We searched the Global Burn Registry for all cooking-related burns injuries on 1 February 2021. From the registry, we retrieved details about patients' demographics, burn characteristics (e.g. total body surface area), mode of contact causing the burn injury, cooking fuel sources, hospital characteristics (e.g. length of stay) and discharge outcomes of the index admission. We de-identified and uploaded the data onto a preformatted Excel spreadsheet, Version 16.33 (Microsoft, Redmond, United States of America, USA). The data in the registry are anonymized and available to academic institutions, and therefore are not considered protected health information so institutional review board approval is not required. We followed the terms and conditions given by WHO for using the data and/or information in the registry. 16 Objective To assess the characteristics of cooking-related burn injuries in children reported to the World Health Organization Global Burn Registry. Methods On 1 February 2021, we downloaded data from the Global Burn Registry on demographic and clinical characteristics of patients younger than 19 years. We performed multivariate regressions to identify risk factors predictive of mortality and total body surface area affected by burns. Findings Of the 2957 paediatric patients with burn injuries, 974 involved cooking (32.9%). More burns occurred in boys (532 patients; 54.6%) than in girls, and in children 2 years and younger (489 patients; 50.2%). Accidental contact and liquefied petroleum caused most burn injuries (729 patients; 74.8% and 293 patients; 30.1%, respectively). Burn contact by explosions (odds ratio, OR: 2.8; 95% confidence interval, CI: 1.4-5.7) or fires in the cooking area (OR: 3.0; 95% CI: 1.3-6.8), as well as the cooking fuels wood (OR: 2.2; 95 CI%: 1.3-3.4), kerosene (OR: 1.9; 95% CI: 1.0-3.6) or natural gas (OR: 1.5; 95% CI: 1.0-2.2) were associated with larger body surface area affected. Mortality was associated with explosions (OR: 7.5; 95% CI: 2.2-25.9) and fires in the cooking area (OR: 6.9; 95% CI: 1.9-25.7), charcoal (OR: 4.6; 95% CI: 2.0-10.5), kerosene (OR: 3.9; 95% CI: 1.4-10.8), natural gas (OR: 3.0; 95% CI: 1.5-6.1) or wood (OR: 2.8; 95% CI: 1.1-7.1). Conclusion Preventive interventions directed against explosions, fires in cooking areas and hazardous cooking fuels should be implemented to reduce morbidity and mortality from cooking-related burn injuries.

Study cohort
The study cohort included all paediatric burn injuries recorded in the Global Burn Registry from 1 January 2018 to 1 February 2021. Following the National Institute of Child Health and Human Development, we considered patients younger than 19 years as paediatric patients. 17

Variables
The primary outcome was burn severity as determined by two proxy metrics: (i) total percentage of body surface area affected; and (ii) mortality. We as-sessed five primary risk factors: sex, age, World Bank income category of country of residence, mode of contact causing burn injury, and cooking fuel. We used the National Institute of Child Health and Human Development age group categories, 17 a validated categorization of the biopsychosocial changes that occur from birth to adulthood. 18 The age categories are: children 2 years and younger; 3-5 years; 6-11 years; and 12-18 years. We included sex and age as risk factors, since they are well-documented predictors in regional studies and therefore merit further investigation on an international scale. [5][6][7][8] The global nature of the registry also allows study of the relationship between childhood burn injuries and socioeconomic factors, for which World Bank income classification is often used as a proxy measure. 19 The World Bank classify countries into four groups based on their gross national income per capita: high income, uppermiddle income, lower-middle income and low income. 20 Information on the mode of contact causing the burn injury and cooking fuel used are entered for every patient with a cooking burn. These data provide valuable insight to inform preventive measures. One of the following modes of contact can be chosen when entering data: deliberate movement (e.g. deliberate touch to hot surface); accidental contact (e.g. fall/spill); exposure to open fire in the cooking area; explosion in the cooking area; or an unspecified other. Cooking fuel was categorized as: natural gas, liquefied petroleum, kerosene (paraffin), electricity, wood, charcoal and other. We categorized fuel sources that were infrequently reported in the registry as other (e.g. ethanol, traditional biomass, and unspecified sources).

Statistical analysis
For data analysis we used SPSS software, Version 26 (IBM Corp., Armonk, USA). Frequencies and percentages were calculated for categorical variables. Medians and interquartile ranges (IQR) were calculated for continuous data. We found that total percentage of body surface area affected was a skewed continuous variable (skewness: 2.1), therefore the Kruskal-Wallis test was used to find factors associated with increased total body surface area affected and mortality, and the Mann-Whitney U test was used to compare subgroups. We assessed the five selected risk factors as predictors for mortality and the severity of total body surface area affected, using logistic regression. The median total body surface area affected of the cohort (15%) was used as the threshold for logistic regression. All statistical tests were considered to be significant at two-sided P < 0.050.

Results
In total, 17 countries across all WHO regions reported 6965 burn injuries to the registry (Table 1). Of these, 2957 (42.4%) documented cases involved paediatric patients. Of the paediatric burn injuries, 974 (32.9%) were explic-

Box 1. Description of the Global Burn Registry
The Global Burn Registry is a global instrument developed by WHO to standardize the reporting and registry of burn admissions across the world. The registry comprises voluntary, anonymized burn data submitted by hospitals for use by public health officials and academic researchers. Data available include: country and city of patient, WHO region, World Bank Income category, respondent (e.g. patient, sibling), age, sex, pre-hospital duration hours, burn surface area, smoke injuries, associated anatomic injuries (e.g. head and neck, trunk), cause of burn, cooking-related factors (e.g. contact method, fuel, cooking height), contributing factors, number of people injured, surgery during admission, hospital stay, condition of discharge (e.g. dead, discharged without physical impairment), submission date and health facility information.
WHO: World Health Organization.

Sex and age
Burn injuries occurred more often in boys (532 patients; 54.6%) and chil-dren 2 years and younger (489 patients; 50.2%; Table 2). Girls suffered burns with a higher mortality rate (P value: 0.002) compared to boys. Although children 2 years and younger experienced the highest number of burns, they suffered less deadly burns than children in other age groups (vs 3-5 years P value: 0.003; vs 6-11 years P value: 0.114; vs 12-18 years P value: 0.002; Table 2).

Country income classification
Most paediatric cooking burns in the registry were reported in upper-middleincome countries (475 injuries; 48.8%).
There was no significant difference in total percentage of body surface area affected by burns (P value: 0.520) or in mortality (P value: 0.614) between the different income categories ( Table 2).

Mode of contact
Most children were exposed to burns by accidental contact (729 patients; 74.8%) followed by deliberate touch (108 patients; 11.1%). Compared to the least deadly burns (those caused by deliberate touch), burns caused by explosions, fires in the cooking area and unspecified other sources caused significantly higher mortality (all P values < 0.001). Explosions (P value < 0.001), fires in the cooking area (P value < 0.001), other sources (P value: 0.003) as well as accidental contact (P value: 0.014) caused significantly larger body surface burns than deliberate touch (Table 2).

Cooking fuel
The most commonly used cooking fuel in childhood burns was liquefied

Risk factors
Children injured by a fire in the cooking area or an explosion were more likely to have more than 15% of the total body surface area affected than children who were injured by deliberate touch (odds 2) were more likely to cause burn injuries affecting more than 15% of the total body surface area (Table 3). Children were more likely to have died if they had been injured by an explosion (OR: 7.5; 95% CI: 2.2-25.9), a fire in the cooking area (OR: 6.9; 95% CI: 1.9-25.7) or if an unspecified contact method was reported to the registry (OR: 6.6; 95% CI: 1.9-23.0) than children who were injured by deliberate touch. Compared to liquefied petroleum, children injured by the fuel sources charcoal (OR: 4.6; 95% CI: 2.0-10.5), kerosene (OR: 3.9; 95% CI: 1.4-10.8), natural gas (OR: 3.0; 95% CI: 1.5-6.1) or wood (OR: 2.8; 95% CI: 1.1-7.1) were more likely to die (Table 3).

Discussion
We used WHO Global Burn Registry to understand the cause of and risk factors for cooking burns in children. Unlike previous work that used singlecentre and national databases, this study compared burn causes and health outcomes on an international scale to better inform and advance global surgical initiatives.
Our findings corroborate previous research on the demographics of children injured by cooking-related burns. 5,7,9,21 Although more than half of these burns occurred in boys, girls were nearly twice as likely to die from these burns. The female predominance in burn severity may relate to the relegation of women to cooking roles and hence girls might spend more time in places with dangerous cooking fires and hot liquids than boys. [22][23][24] We found that half the burn injuries reported to the registry were in children 2 years and younger. Infants are known to be disproportionally affected by all types of burn injuries. 6,25 Younger children are more impulsive, curious and lack self-awareness, putting them at greater risk of accidental burns exposure. 9 To prevent burn injuries in young children, caretakers should be informed about preventive measures. For example, written and pictorial education material should be available at places frequented by such caretakers. 26 In our study sample, we found no correlation between age and burn size or severity.
Previous regional work has shown that paediatric burns are more common and more deadly in low-and middleincome countries. 9 Here we report that burn injuries did not differ significantly by burn size or mortality across country income group, and no income group was an independent risk factor for burn severity. This contradictory finding is likely due to the skewed distribution of countries in the Global Burn Registry. Of the 974 patient reports analysed, only 68 reports are from high-income countries, and of those 65 are from the Russian Federation. Despite the vast range of middle-income countries in the registry, comparing the burden of burn injuries in countries is difficult without more data from countries that Table 3.

Risk factors associated with total burn surface area and mortality of paediatric cooking burn patients, 2018-2021
Characteristic OR (95% CI) progressed the most in burn care, particularly the United States and western European countries.

Sex
The findings of the study suggest that stakeholders should focus their preventive work on the mode of contact and cooking fuel. These two factors were both significant risk factors for burn size and mortality. Burn injuries caused by explosions or fires in the cooking area were associated with the highest mortality compared to any risk factor in this study. Although the most common fuel source was liquefied petroleum, the findings suggest that this is a relatively safe cooking fuel, as is electricity. In contrast, we found kerosene to be a significant risk factor for burn mortality along with natural gas, charcoal and wood. A cooking fuel and its propensity to cause a fire or explode are likely linked to one another, and further analyses are needed to dissect this relationship. To reduce severity and mortality from burns, prevention efforts should focus on explosions and fires rather than the most common burns in scalds, spills and accidental touches. Additionally, our results indicate that preventive initiatives that shift cooking methods away from hazardous fuels, like wood and kerosene, towards safer cooking fuels, like electricity and liquefied petroleum, could result in large gains in severe burn prevention. Currently, many countries subsidize rather than disincentivize the use of fuels such as kerosene; the current analysis adds a medical perspective to the growing momentum of international and economic research advising against these subsidies. 27,28 This study has several limitations. First, entry of data in the Global Burn Registry is voluntary and subject to reporting bias and selective data input. Although interrater reliability is potentially problematic in multicentre databases, we mitigated this risk by studying objective and quantitative variables, such as mortality and total body surface area affected. Second, the registry is a relatively new instrument, with so far limited breadth and number of hospitals involved, as well as the period of data collection. At the time of study, only 17 countries had reported data for up to 3 years. Third, our cohort of cooking-related burns came from 14 of those countries, with an uneven case distribution, limiting generalizability to socioeconomic groups as a whole. Lastly, though this study reports on possible risk factors of paediatric cooking burns, these associations are observational and no causal links can be identified from this work.