Introduction

Burn injuries are a global public health problem, and their epidemiology presents regional variations.¹ Burns can result from several etiological agents, such as direct exposure to fire, chemicals, electrical current, radiation, or friction.² Additionally, these injuries are a significant cause of death worldwide.^1,2

The World Health Organization estimates that burns account for approximately 180,000 deaths annually.³ Infant mortality from burns is higher in lowand middle-income countries, being about seven times higher than that observed in high-income countries.³ These injuries are a severe issue, as the possible sequelae include scarring, physical limitations, or permanent damage,⁴ including deficits in fine motor skills or contractures when occurring near joints.⁵

Although burns are a significant global problem, the literature about their occurrence in Peru is scarce.⁶ This is worrisome, as understanding the epidemiology of burns is fundamental for public health planning and the development of potential interventions for Peru and Latin America. Therefore, we aimed to locate studies, describe bibliometric data, and define the epidemiological profile of burn patients in Peru through a systematic review.

Objectives

The primary objective of this study was to investigate the epidemiology of burns in Peru. Specific objectives included describing bibliometric data, identifying the locations of each study, synthesizing clinical and demographic data, and determining areas for further research.

Materials and Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).⁷ The Population, Exposure, Comparator, and Outcomes (PECO)⁸ strategy employed ‘Peru’ as the population and ‘burns’ as exposure, not including comparators or outcomes.

Inclusion criteria were the following: studies describing burn patients in Peru at any time period; in any language; cross-sectional, cohort, and case-control studies; conference proceedings, articles, and theses.

Exclusion criteria included incomplete studies or those lacking data from Peru. Also, randomized clinical trials, case series, case reports, and reviews were not considered.

We assessed the following four databases on March 9, 2023: Web of Science (WOS), Medline (PubMed), Embase, and the Virtual Health Library (VHL). There was no restriction on the year of publication, and we considered all articles published up to the time of access for screening. After article selection and reading, we reviewed their references to include new papers, the process was not repeated. We included “gray literature” studies⁹ because we noted that a part of the relevant literature was in conference proceedings and theses. We used the following terms for each database: “Burns”, “Electrical Burns”, “Ocular Burns”, “Inhalation Burns”, “Burn Treatment Units”, and “Peru”.

We saved the data in the Rayyan (Qatar Foundation) platform,¹⁰ a reference manager designed for review studies.¹¹ Two independent researchers reviewed the studies, through a blinded process. The consensus of all researchers solved any disagreements. We searched for complete texts of studies without abstracts or incomplete articles. Studies that did not meet the inclusion criteria were removed from the analysis.

Two independent researchers extracted and evaluated the data, solving any differences by consensus. Other researchers performed a second review of the extraction, which used the online spreadsheet software Google Sheets (Google LLC.).¹²

We obtained the coordinates of each hospital using the Google Maps (Google LLC.)¹³ and generated a map employing the QGIS (QGIS Development Team) software,¹⁴ which processed the locations of each study’s occurrence. These maps were based on OpenStreetMap (OSM Foundation)¹⁵ and the dataset from the United Nations Office for the Coordination of Humanitarian Affairs on the Humanitarian Data Exchange Database.¹⁶

To assess the quality of the articles, two independent reviewers used the 9-item Joanna Briggs Institute (JBI) checklist.¹⁷ We did not apply the checklist to conference proceedings. We collected the infection criteria defined in each study. “Unclear” refers to information not provided or of a dubious nature.¹⁷ We created a flowchart following the PRISMA guidelines (►Fig. 1)⁷ and ►Fig. 2 summarizes the quality of the studies according to the JBI criteria.¹⁷

Fig. 2 - The Joanna Briggs Institute (JBI) 9-item checklist for assessment of study quality.

Fig. 2 - The Joanna Briggs Institute (JBI) 9-item checklist for assessment of study quality.

Fig. 3 - Map of hospital locations in Peru.

Fig. 3 - Map of hospital locations in Peru.

We performed the statistical analysis using the RStudio (Posit PBC) software’s¹⁸ meta¹⁹ and metafor²⁰ packages. After extraction, we transposed data on the number of infections, deaths, and samples into the software to conduct the meta-analysis.²¹ We employed the inverse variance logit transformation (PLOGIT) method and a random-effects model. Using the Hartung-Knapp method, we adjusted the confidence intervals (CIs) of the combined effect to provide more conservative estimates.²¹

The I² statistics assessed the heterogeneity between studies, and values above 50% indicated heterogeneity. We also calculated and evaluated the Tau-squared (τ²) values.²¹ We conducted subgroup analyses using Chi-squared (χ²) tests to assess whether there was a significant difference between subgroups (studies that did or did not perform laboratory testing for infections).²¹

Results

During the evaluation period, which extended up to March 9, 2023, we retrieved 971 records, including 810 from databases and 161 from citation searches. From the records retrieved from the databases, we excluded 189 duplicates. We screened the remaining 621 records, of which 614 were excluded for not meeting the inclusion criteria, resulting in 6 articles, of which 1 was not retrieved, and 5 underwent eligibility assessment. Among the records retrieved from citations, we included 7 in the analysis (►Fig. 1). The final sample consisted of 12 studies, including 1 conference abstract,6 peer-reviewed articles, and 5 theses (►Tables 1-3).^1,6,22-31

Table 1 - Publication characteristics and study location

Table 2 - Study sampling characteristics

As shown in ►Fig. 2, while the descriptions of subjects and settings were strong (items 4 and 5), two studies lacked clarity regarding their sampling methods, as they relied on convenience samples or failed to calculate sizes correctly (items 1-3). Clinical assessments may differ among physicians, and the studies did not use clear guidelines for evaluation (items 6 and 7). Furthermore, some studies failed to specify the criteria used for evaluating burns and infections. Although most of them focused on data description, bivariate and multivariate analyses were adequate, and reported proportions. Only two studies investigating infections did not use multivariate adjustment (item 8). The response rate was not described or was inadequate in five studies, with a response rate below 80% (item 9). ►Table 4 shows the evaluation of each study.^1,6,22-31

Table 3 - Clinical and epidemiological characteristics

Table 4 - Assessment of each study

The present review included a total of 3,446 patients from 12 studies, as detailed in ►Table 2. Sample sizes varied considerably across studies, ranging from 36 to 848 participants. The most prevalent age group differed among studies, as each adopted specific inclusion criteria. However, children were consistently identified as the group most affected by burns.

Regarding sex, men were the most affected, and the most common location for burn occurrence was in a domestic setting. As for etiology, the most common cause was contact with hot fluids, followed by direct fire. The limbs were the most frequently affected body parts. Only two studies^29,31 reported the average percentage of body surface area burned: the lowest average was 10.23%²⁹ and the highest, 17.7%.³¹

Most studies occurred in hospitals in Lima, and seven studies described burn units, with the primary hospital being Hospital Nacional Arzobispo Loayza. The mapin ►Fig. 3 shows the locations of the hospitals.

Fig. 1 - Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flowchart: study selection and inclusion. The flowchart shows the steps for selecting and excluding articles.

Fig. 1 - Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flowchart: study selection and inclusion. The flowchart shows the steps for selecting and excluding articles.

The pooled rate of infections was 31.7% (95%CI: 21.46- 44.08%; I² = 0%) among studies that did not use laboratory tests,^1,24,25 without significant heterogeneity. Among studies that used tests,^26,28 the pooled rate of infections was lower (7.74%), with a significant heterogeneity (I² = 94.2%). Studies using laboratory tests or not differed significantly (χ² = 4.80, p < 0.03).

Only five studies described the number of deaths among patients. Two studies did not report deaths during the evaluation period.^24,25 Cutipa et al.²⁶ included 350 patients and recorded five deaths. Sam et al.²⁹ analyzed 278 cases, with four deaths. In contrast, Chávez et al.²⁸ despite presenting the smallest sample (131 patients), reported 8 deaths. The analysis of deaths revealed a pooled rate of 1.88% (95%CI: 0.54-6.36%; I² = 67.1%), with significant heterogeneity.

We did not consider COVID-19 infections.²⁹ Also, despite describing 26 deaths in 1999, Delgado et al.²³ did not report the population at risk in the same year. As such, we did not include their study in this meta-analysis.

The publication of all studies, original articles, and theses occurred from 2002 to 2023 (►Table 2). Non-Peruvianjournals published half of the articles, with two in “Injury Prevention” and one in “Cirugía Plástica Ibero-Latinoamericana”. The theses were not subsequently published as articles.

Discussion

This study synthesized bibliometric and epidemiological characteristics and presented results consistent with those from other epidemiological studies regarding the etiology, location, and demographics of burns.^32,33 Although age is an essential demographic data, its comparative analysis had limitations due to heterogeneous inclusion criteria, lack of uniform etiological stratification, and use of different age ranges among the included studies, suggesting a point of improvement.

A study involving 8 centers in Africa, Asia, and Latin America (n = 15,344) revealed that the most common etiology of burns was hot fluids and flames, most frequently occurring at home, and in males.³² A scoping review of 40 studies on burns in Africa showed that the key etiology was scalding, and most patients were males.³³ However, these data diverge from the Global Burn Registry (n = 9,277), which showed that burns were more frequent in the domestic environment, with a greater number of female victims and flames as etiology.³⁴

Some studies addressing infections did not report on the method used for their identification. The meta-analysis showed a higher infection rate in studies that did not use laboratory tests. Among the studies that used laboratory tests,infection criteria can explain thedifferences;one included only wound infections,²⁶ while the other included any type.²⁸ Culture methods are more reliable, should be used whenever possible. Different inclusion criteria and study designs can explain the heterogeneity in mortality rates.

The number of published studies is scarce, and most are in Peruvian journals. Moreover, there was no publication of theses as articles upon completion. This fact may suggest a low level of engagement among researchers in investigating the topic and a problem within publications, as global journals may not be interested in publishing data from the country. A literature review of burn studies on PubMed showed that Peru had no first or senior authors in global publications. Meanwhile, Brazil was the leading country in Latin America, and the United States was the leading country in the Americas.³⁵ In this context, calls for papers targeting Latin American countries are critical for the development of studies in Peru and the greater visibility of regional data.

Overall, the analysis showed the scarcity of Peruvian literature. In terms of geographic distribution, studies generally focus on a single region, with only one survey study. Therefore, the accurate understanding of burn occurrence in Peru is complex, particularly because other areas differ in their economic development and population size. Additionally, these less developed regions have the highest risk of burns.³⁵ Survey studies, involving questionnaires and multicenter designs, can offer a better understanding of regions lacking hospital data.

The studies included in this analysis are valuable sources for understanding the epidemiology of burns in the Peruvian context, as they explore and report relevant aspects of the local reality and integrate a body of literature that deserves global attention. However, since they are mainly observational studies, it is worth noting some methodological limitations: (1) most occurred in a single hospital center, with varying inclusion criteria and reporting methods, making direct comparisons difficult; (2) regarding data quality, collection based on clinical records is subject to information bias; (3) clinical assessment can vary among professionals, resulting in an heterogenous patient classification and management.

Therefore, future multicenter studies should prioritize the standardization of clinical and methodological criteria in their protocols to improve the comparability and robustness of the data generated. The use of previous studies as a starting point is essential.

Conclusion

Although heterogeneous, the studies reveal fundamental data consistent with the epidemiology described in the literature. Based on these data, subsequent stages of studies in Peru should prioritize cooperation between geographically close centers, to optimize logistics and facilitate multicenter research. Moreover, in peripheric areas, the application of survey research (questionnaires) can significantly contribute to a deeper understanding of local epidemiology.