Paediatric Partial-Thickness Burn Therapy: A Meta-Analysis and Systematic Review of Randomised Controlled Trials

Background: Paediatric second-degree burn injuries are a significant source of medical challenges to the population that may cause severe, lifelong complications. Currently, there are dozens of therapeutic modalities and we aimed to summarise their reported outcomes and determine their effectiveness, compared to the widely used silver sulphadiazine (SSD). Methods: We conducted the meta-analysis and systematic review of randomised controlled trials (RCTs), which investigated the performance of dressings in acute paediatric partial-thickness burns. The evaluated endpoints were time until wound closure, grafting and infection rate, number of dressing changes and length of hospitalisation. Results: Twenty-nine RCTs were included in the qualitative and 25 in the quantitative synthesis, but only three trials compared SSD directly to the same intervention (Biobrane). Data analysis showed a tendency for faster healing times and a reduced complication rate linked to biosynthetic, silver foam and amnion membrane dressings. A substantial difference was found between the number of dressing changes associated with less pain, narcosis and treatment duration. Conclusions: Considerable between-study heterogeneity was caused by the unequal depth subcategory ratio and surface area of the injuries; therefore, no significant difference was found in the main outcomes. Further research is necessary to establish the most effective treatment for these burns.


Introduction
Nearly one hundred thousand (viz.,~96,000) children suffer a fatal injury from preventable, mostly flame-related (~93%) burn traumas, each year, that is 263 cases per day, according to the WHO's latest global report. The likelihood of a non-fatal injury is assumed to be at least ten times higher (i.e., 1-7,000,000/year), and due to the absence of a successfully coordinated prevention, treatment or rehabilitation strategy, almost half of them (49%) suffer from some form of irreversible disability after the burn [1,2]. Complications such as extensive contractures and amputations constitute physical impairments, but even a relatively minor scar or the memory of the trauma can provoke lifelong psychological disorders [3][4][5]. Compared to adults, children, especially infants, have thinner skin and However, numerous studies revealed several disadvantages to the use of SSD, which led to the development of a wide range of alternative topical treatments. Nevertheless, their efficacy in the management of paediatric PT burns remains largely unclarified. A summary of each intervention that was analysed and compared to SSD in the present study can be found in Text S1.
We performed a literature search to systematically review the available treatment options for paediatric PT burns, then conducted a meta-analysis to obtain insights into the dressings' healing potential and complication rate. We aimed at collecting randomised controlled trials (RCTs) about PT burn treatments that reported the time to reepithelialisation (TTRE), grafting and infection rate, number of dressing changes and length of hospital stay (LOS) in patients younger than 18 years old at the time of the injury.

Search Methods for Identification of Records
On 29 October 2020, a systematic search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA) (Table S1) [28]. We searched for RCTs that compared at least two different interventions in patients under the age of eighteen with PT burns in the MEDLINE (via PubMed), Embase, Web of Science and CENTRAL databases ( Figure 1). the need for surgical intervention. Thus, we chose this therapy as a comparator because most of the articles reported their findings correlated to SSD due to its historical relevance. However, numerous studies revealed several disadvantages to the use of SSD, which led to the development of a wide range of alternative topical treatments. Nevertheless, their efficacy in the management of paediatric PT burns remains largely unclarified. A summary of each intervention that was analysed and compared to SSD in the present study can be found in Text S1. We performed a literature search to systematically review the available treatment options for paediatric PT burns, then conducted a meta-analysis to obtain insights into the dressings' healing potential and complication rate. We aimed at collecting randomised controlled trials (RCTs) about PT burn treatments that reported the time to reepithelialisation (TTRE), grafting and infection rate, number of dressing changes and length of hospital stay (LOS) in patients younger than 18 years old at the time of the injury.

Search Methods for Identification of Records
On 29 October 2020, a systematic search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA) (Table S1) [28]. We searched for RCTs that compared at least two different interventions in patients under the age of eighteen with PT burns in the MEDLINE (via PubMed), Embase, Web of Science and CENTRAL databases ( Figure 1).
Our search keys can be found in Text S2, and the records identified by them were exported without the use of filters.

Study Selection, Data Extraction and Management
The studies identified by the search were screened by two independent review authors (AL and MT) to assess their eligibility. The eligible articles were collected in End-Note X9 (Clarivate Analytics, Philadelphia, PA, USA), then the outcomes recorded in Microsoft Excel (Microsoft Corporation, Albuquerque, NM, USA) by two authors Our search keys can be found in Text S2, and the records identified by them were exported without the use of filters.

Study Selection, Data Extraction and Management
The studies identified by the search were screened by two independent review authors (AL and MT) to assess their eligibility. The eligible articles were collected in EndNote X9 (Clarivate Analytics, Philadelphia, PA, USA), then the outcomes recorded in Microsoft Excel (Microsoft Corporation, Albuquerque, NM, USA) by two authors independently (AL and MT). Discrepancies were resolved by consensus after re-checking the original article.

Data Synthesis
Statistical analysis was performed by an expert biostatistician (AV) using the methods recommended by the working group of the Cochrane Collaboration [30]. In the metaanalysis, the calculated effect sizes were visualised in forest plots using Comprehensive Meta Analysis (Version 3) statistical software (Biostat Inc., Englewood, NJ, USA). Heterogeneity was tested with Cochrane's Q (χ 2 ) test and the I 2 statistic. Q test was considered significant when p-values were less than 0.1. Based on the suggestion of the Cochrane Handbook, I 2 values from 30% to 60% represent moderate and between 50% and 90% substantial heterogeneity. Due to the groups' generally high heterogeneity, DerSimonian and Laird random-effects models were used in all analyses [31].
For continuous outcomes: means, and for dichotomous outcomes: event rates with 95% confidence intervals, were pooled in each subgroup to compare the differences between the intervention groups. In the case of some subgroups, there were studies with more than one intervention group; therefore, we combined these groups based on the suggestion of the 6.5.2.10 section of the Cochrane Handbook [30]. When the means and standard deviations (SD) of the effects were not reported, we derived these data from the graphical representation of the outcomes or by estimation based on the work of Wan et al. 2014 with the use of medians, minimum, maximum or quartiles [32]. In three trials, indicators of SD were not reported [21][22][23], thus, we obtained them from a previous meta-analysis [33], which included the required data.

Search Results
The search identified 1853 potentially relevant records after duplicate removal that were screened by title ( Figure 1). After exclusion, 474 abstracts were assessed. The full texts of a total of 196 articles were retrieved; then, 152 trials were excluded because they had an unmatching or unknown study population or design. Ten studies did not contain the specified outcomes and five full texts could not be obtained; thus, these were also excluded from the analysis. Finally, 24 RCTs containing 21 full-text articles and 3 conference abstracts [27,34,35] were included in this meta-analysis. Another full-text article [26] was identified when the reference lists of the eligible papers were checked; it was included in the qualitative synthesis. In the systematic review, an additional three full-text RCTs [36][37][38] and one conference abstract [39] were included.

Methodological Quality of the Included Studies
A summary of the risk of bias assessment is shown in Figure S1. Generally, the risk of bias was considered high, and the articles often lacked essential information. Randomisation protocols were generally not discussed, but studies reported the use of lottery [16,21], tables of random numbers combined with lottery [20] and a randomization schedule [34]. Computergenerated individual unit block randomization [19] and randomization tables [27,35,47,49]including one that was further stratified by age and area [17]-along with a statistician generated age-stratified permutated block method [45] were also used. One article divided the treatment groups by even and odd admission days [37], and one study contained seven patients with resident preference-based randomisation in addition to a computer-generated randomisation table [22].
While evaluating the funding sources, we found that eight articles received either financial or material donations from the manufacturer [17,20,21,25,45,46,53] and two were supported by solely independent grants [37,47], although most of these researchers stated that they had no conflict of interest, with two exceptions. One of the funders supervised the design of the study, the data analyses and the development of the manuscript [17], and another intervention was developed by the first author, who is also the director of the company that sells it [43].
Overall, two studies reported using the CONSORT criteria while conducting the research [45,46], which may be the reason behind the missing data, such as randomisation or concealment protocols.

Time to Reepithelialisation (TTRE)
Our primary outcome to determine the interventions' effectiveness was the mean TTRE or complete wound closure time. A total of 623 participants (ranging from 4 to 145 in the different studies with an average of 30) from 17 trials were included in this meta-analysis. Interventions with similar characteristics were pooled together to rank this outcome because direct comparisons were only published for SSD and Biobrane in a sufficient quantity. In total, 265 children received SSD with a mean TTRE of 17.89 days, which was the slowest among the analysed interventions, although the difference was not statistically significant (p = 0.70). Lower TTRE was seen in 224 children treated with NPWT This forest plot of studies pools mean TTRE inter days, lasting from the time of the paediatric PT burn injury until wound closure. Black squa dicate the TTRE in each study. The size of the black squares represents the individual study w and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue dia indicates the overall effect, and its outer edges characterise the Cis [16,18,[20][21][22][23][24][25][26][27]35,[42][43][44][45].
Further analysis was conducted to find out the reason behind the groups' cons ble heterogeneity (which was indicated by a high I 2 of 75.35-99.85). Not surpris when mean TTRE was stratified by depth, a significant difference (p = 0.0004) was between II/B (20.53 days), II/A (13.77 days) and combined PT (12.43 days) burns (F S2). This difference also clearly indicates that PT burn subcategories should be ana separately, even though most of the articles [18,19,21,23,24,26,37,40,[43][44][45]47,48,53] previous review [33] pooled them together. Due to the low number of eligible stud each subgroup, we were not able to conduct a meta-analysis on the individual inte tion's TTRE stratified by the depth of the burn. Nevertheless, we pooled and ranke treatment options according to their depth; in II (Table 2), II/A (Table S2), II/A + MD ( S3) and II/B (Table S4)   This forest plot of studies pools mean TTRE intervals in days, lasting from the time of the paediatric PT burn injury until wound closure. Black squares indicate the TTRE in each study. The size of the black squares represents the individual study weight, and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue diamond indicates the overall effect, and its outer edges characterise the Cis [16,18,[20][21][22][23][24][25][26][27]35,[42][43][44][45].
Further analysis was conducted to find out the reason behind the groups' considerable heterogeneity (which was indicated by a high I 2 of 75.35-99.85). Not surprisingly, when mean TTRE was stratified by depth, a significant difference (p = 0.0004) was found between II/B (20.53 days), II/A (13.77 days) and combined PT (12.43 days) burns ( Figure S2). This difference also clearly indicates that PT burn subcategories should be analysed separately, even though most of the articles [18,19,21,23,24,26,37,40,[43][44][45]47,48,53] and a previous review [33] pooled them together. Due to the low number of eligible studies in each subgroup, we were not able to conduct a meta-analysis on the individual intervention's TTRE stratified by the depth of the burn. Nevertheless, we pooled and ranked the treatment options according to their depth; in II (Table 2), II/A (Table S2), II/A + MD (Table S3) and II/B (Table S4) PT burns.  Numbers marked with a single star (*) were converted from median to mean. Two stars (**) signify that the number could not be converted from median due to missing IQR75 or reporting the range in 10 and 90 percentiles (II = partial-thickness burn injury (PT); AM = amnion membrane; NPWT = negative pressure wound therapy; NR = not reported; PMB = polymyxin b; SSD = silver sulphadiazine; SUM = the summarised values of the same interventions; TBSA% = burned area of the total body surface; TT = traditional treatment; TTRE = time to reepithelialisation; TTRE red% = the percentage of time reduction, with the addition of the intervention; TBSA%/TTRE = the area of regeneration per day).
Another strong correlation between the burn area and TTRE was observed when we developed a novel ratio of TBSA% to TTRE (T%/T), which indicates what percentage of the TBSA regenerates each day and can also be used to standardise the burn sizes (Table 2 and Tables S2, S4 and S5). A critical limitation of using T%/T is that in smaller (under 5 TBSA%) burns, the ratio will be low, even though reepithelialisation was rapid. The reason behind this, in our hypothesis, is that there seems to be a minimum physiological time for wound regeneration, which is unrelated to the burn size and takes approximately 5-7 days. In studies that did not report TBSA% [17,25,[34][35][36]39] or TTRE [19,41,47], the T%/T ratio could not be calculated (Table S3). Since TTRE alone seems insufficient to determine the additional interventions' effectiveness on wound closure (e.g., vitamins or heparin), we also calculated these therapies' TTRE reduction percentage (TTRE red%). The additional interventions were compared to their control treatment, where they received a placebo (or nothing) instead, on top of the traditional treatments ( Table 2, Tables S2 and S4).
For SSD only, TTRE was also compared in subgroups divided by depth (SSD II: 17.11; SSD II/A: 17.05 days; p = 0.99) and by area (<10 TBSA%: 17.03; 10-25 TBSA%: 20.59 days) ( Figures S4 and S5), but the differences were not significant between the subgroups (p = 0.59). This may indicate that the burns categorised as PT were mostly II/A injuries. Some articles reported the fraction of wound closure on the tenth day (day 10 RE%), which is summarised in Table S5.
Only three trials [18,21,22] reported the TTRE of the same comparator (SSD: 17.94 days) and intervention (Biobrane: 14.27 days), which were analysed separately. Despite the lack of a significant difference between the two treatments (p = 0.61), every article reported improved results with Biobrane compared to SSD ( Figure S6).

Grafting Rate and Non-Grafted Rate
If conservative treatment is unable to heal the injury, a permanent skin transplantation is needed to facilitate wound closure. First, the interventions' mean percentages of how many conservatively treated patients required grafting related to the whole study population were calculated. The ratio of grafted patients was 19.3% in SSD-, 20.5% in biosynthetic-, and 18.9% in silver foam-treated patients (p = 0.99) ( Figure S7). Because every treatment without any grafted patient (zero outcomes) had to be excluded from the previous meta-analysis, here we used the reverse approach; that is, the comparison of the percentage of the non-grafted patients among the treatments (Figure 3). how many conservatively treated patients required grafting related to the whole study population were calculated. The ratio of grafted patients was 19.3% in SSD-, 20.5% in biosynthetic-, and 18.9% in silver foam-treated patients (p = 0.99) ( Figure S7). Because every treatment without any grafted patient (zero outcomes) had to be excluded from the previous meta-analysis, here we used the reverse approach; that is, the comparison of the percentage of the non-grafted patients among the treatments (Figure 3). Mean non-grafted population ratios. They should be interpreted as grafting rates of paediatric PT burns when subtracted from 1 (100%). Black squares indicate the TTRE in each study. The size of the black squares represents the individual study weight, and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue diamond indicates the overall effect, and its outer edges characterise the CIs [18,19,21,22,24,26,43,44,46,48,49].
With this method, we found that by subtracting the non-grafted population percentages from 100%, the grafting rate was 13.2%, 13.4%, 13.1% and 9.8% in patients treated Figure 3. Mean non-grafted population ratios. They should be interpreted as grafting rates of paediatric PT burns when subtracted from 1 (100%). Black squares indicate the TTRE in each study. The size of the black squares represents the individual study weight, and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue diamond indicates the overall effect, and its outer edges characterise the CIs [18,19,21,22,24,26,43,44,46,48,49].
With this method, we found that by subtracting the non-grafted population percentages from 100%, the grafting rate was 13.2%, 13.4%, 13.1% and 9.8% in patients treated with SSD, silver foam, biosynthetic and Biobrane, respectively. These results indicate that among Biobrane-treated children, grafting was required 25.8% less often compared to SSD; however, the difference between the treatments did not reach the level of significance (p = 0.98).

Dressing Changes
There was not enough data to conduct a meta-analysis of the required dressing changes between the two interventions. Nevertheless, the mean frequency of dressing reapplications showed a great variance among interventions, and they positively correlated with pain and discomfort levels. Furthermore, dressing changes were proportional to the rate of anaesthesia induction as well as to the time required for the healthcare professionals and the operating theatre for the administration of the treatments. SSD seemed to be the least efficient option with an extremely high 65.5 mean dressing changes if the wounds were treated openly and 9.6 dressing changes with closed wound treatment. Interventions with three or fewer dressing reapplications were Acticoat + Mepitel and Tilapia (number of changes 3.0 for both), Acticoat (2.7), NPWT + Acticoat + Mepitel (2.  Table 4).  Numbers marked with a single star (*) were converted from median to mean. (II = Partial-thickness burn injury (PT); II/A = superficial PT; AM = amnion membrane; ex. Mostaque = this study was excluded from a part of the analysis due to its open treatment regime; MD = mid-dermal or mixed-depth burn injury; NPWT = negative pressure wound therapy; NR = not reported; PMB= polymyxin b; RCTs = randomised controlled trials; SD = standard deviation; SSD = silver sulphadiazine.)

Infection Rate and Non-Infected Rate
The patients' percentage that showed signs of infection during their treatment was calculated similarly to grafting needs. The infection rates in the cases of different interventions were as follows: biosynthetic dressings: 21.8%; silver foam dressings: 12.4%; Biobrane: 11.7%; SSD: 9.2% (p = 0.65) ( Figure S9). By subtracting the non-infected children's percentages from the whole population, the calculation revealed slightly different results. In this case, biosynthetic dressings still had the highest microbial contamination rate of 19.4%, among which the rate for Biobrane was 11.7%. SSD showed an even lower rate of 7.4% infections, while the percentage of infected patients was 7.0% in silver foam dressings and 3.5% in Acticoat treatment groups (p = 0.24) (Figure 4).
Individual intervention effect analysis indicated potential alternatives with more advantageous effects on infection rates, such as Aquacel Ag foam (2.4%) and PMB combination therapies such as collagenase or AM with Nystatin (2.0% and 1.9%, respectively) or Acticoat with Mepitel (0%) ( Table 5).
Additionally, the infected population rate in the case of SSD (13.9%) was similar to the rate of Biobrane (9.9%), without a significant difference (p = 0.91) between the treatments ( Figure S10).

Length of Stay (LOS)
The length of hospital stay-the time spent inside the hospital from admission until discharge-is associated with the total cost of care, and it enormously impacts the children's discomfort levels. Sufficient data for a meta-analysis was only available for SSD-and AMbased treatments, for which treatments the mean LOS were 12.5 and 8.3 days, respectively ( Figure 5).
While this indicates a 33.6% shorter LOS in the case of AM-based compared to SSDbased treatments, the difference was non-significant (p = 0.43). The analysis of specific interventions showed that without antibiotic coverage, LOS is similar in the case of amniotic membrane and SSD treatments (11.37 vs. 13.77 days), while the addition of nystatin and PMB can reduce LOS to 2 days (Table S6). It is important to note that in the cases of several treatments, such as EZDerm (LOS: 3.4 days), Mepilex (LOS: 3.1 days), Biobrane (LOS: 2.4 days) and AM + Nystatin + PMB (LOS: 2 days), the children could be discharged even before the complete reepithelialisation of their injuries, whereas patients treated with SSD and collagenase stayed in the hospital for the entire duration of dressing changes.  The results should be interpreted as infection rates when subtracted from 1 (100%). Black squares indicate the TTRE in each study. The size of the black squares represents the individual study weight, and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue diamond indicates the overall effect, and its outer edges characterise the CIs [18,19,[21][22][23][24]43,44,[46][47][48][49].
Individual intervention effect analysis indicated potential alternatives with more advantageous effects on infection rates, such as Aquacel Ag foam (2.4%) and PMB combination therapies such as collagenase or AM with Nystatin (2.0% and 1.9%, respectively) or Acticoat with Mepitel (0%) ( Table 5).
Additionally, the infected population rate in the case of SSD (13.9%) was similar to the rate of Biobrane (9.9%), without a significant difference (p = 0.91) between the treatments ( Figure S10).  The results should be interpreted as infection rates when subtracted from 1 (100%). Black squares indicate the TTRE in each study. The size of the black squares represents the individual study weight, and the horizontal lines show their corresponding 95% confidence intervals (CIs). A blue diamond indicates the overall effect, and its outer edges characterise the CIs [18,19,[21][22][23][24]43,44,[46][47][48][49].   While this indicates a 33.6% shorter LOS in the case of AM-based compared to SSDbased treatments, the difference was non-significant (p = 0.43). The analysis of specific interventions showed that without antibiotic coverage, LOS is similar in the case of amniotic membrane and SSD treatments (11.37 vs. 13.77 days), while the addition of nystatin and PMB can reduce LOS to 2 days (Table S6). It is important to note that in the cases of several treatments, such as EZDerm (LOS: 3.4 days), Mepilex (LOS: 3.1 days), Biobrane (LOS: 2.4 days) and AM + Nystatin + PMB (LOS: 2 days), the children could be discharged even before the complete reepithelialisation of their injuries, whereas patients treated with SSD and collagenase stayed in the hospital for the entire duration of dressing changes.

Discussion
Even though SSD is widely used as a treatment for burns, our study concluded that it has some disadvantages that can outweigh its beneficial effects, which are mainly its applicability, low cost and notable antibacterial efficacy (i.e., an infection rate of 9.22%) [16][17][18][19][20][21][22][23][24][25][26][27]. However, SSD was associated with slow wound closure (TTRE II/A: 11.0 days; II/B: 25.7 days; II: 18.3 days and 0.39 T%/T) and prolonged hospital stay (LOS II: 13.8 days) as well as with frequent, time-consuming dressing changes (on average 9.6 times; every 1-3 days in PT burns)-also causing pain and anxiety-and a substantial need for grafting (i.e., in 21.5% of the patients). Furthermore, its known side effects include allergic reactions, argyria and neutropenia [54], and it also causes the wound bed's discolouration, which can render wound evaluation and depth determination difficult [55].
Compared to SSD, collagenase combined with PMB showed no difference in TTRE in PT burns, whereas Silvasorb led to improved healing times in II/A and MD burns with 12.4 days (though TBSA% was not reported in Silvasorb-treated patients) [25]. Collagenase + PMB treatment markedly reduced the infection rates (to 2%) in the burned children, but it was associated with a high grafting rate (32%) and prolonged LOS (11.3 days). In the case of both Silvasorb and collagenase + PMB treatments, the dressing change rate was exceptionally high (13.5 and 11 times, respectively), which raises concerns about the recommendation of these treatments in paediatric PT burns.
Among the modern biosynthetic dressings, Biobrane and Transcyte had excellent efficacy with TTRE in PT burns of 10.63 days and T%/T of 0.63% for Biobrane, and TTRE 7.50 days and T%/T of 0.66% for Transcyte. In contrast, EzDerm was less efficient (TTRE in PT burns: 18.75 days; 0.23 T%/T) than SSD. The rates of infection and grafting were high in the case of EZDerm (37.0 and 23.9%, respectively) and Biobrane (12.9% for both),

Discussion
Even though SSD is widely used as a treatment for burns, our study concluded that it has some disadvantages that can outweigh its beneficial effects, which are mainly its applicability, low cost and notable antibacterial efficacy (i.e., an infection rate of 9.22%) [16][17][18][19][20][21][22][23][24][25][26][27]. However, SSD was associated with slow wound closure (TTRE II/A: 11.0 days; II/B: 25.7 days; II: 18.3 days and 0.39 T%/T) and prolonged hospital stay (LOS II: 13.8 days) as well as with frequent, time-consuming dressing changes (on average 9.6 times; every 1-3 days in PT burns)-also causing pain and anxiety-and a substantial need for grafting (i.e., in 21.5% of the patients). Furthermore, its known side effects include allergic reactions, argyria and neutropenia [54], and it also causes the wound bed's discolouration, which can render wound evaluation and depth determination difficult [55].
Compared to SSD, collagenase combined with PMB showed no difference in TTRE in PT burns, whereas Silvasorb led to improved healing times in II/A and MD burns with 12.4 days (though TBSA% was not reported in Silvasorb-treated patients) [25]. Collagenase + PMB treatment markedly reduced the infection rates (to 2%) in the burned children, but it was associated with a high grafting rate (32%) and prolonged LOS (11.3 days). In the case of both Silvasorb and collagenase + PMB treatments, the dressing change rate was exceptionally high (13.5 and 11 times, respectively), which raises concerns about the recommendation of these treatments in paediatric PT burns.
Among the modern biosynthetic dressings, Biobrane and Transcyte had excellent efficacy with TTRE in PT burns of 10.63 days and T%/T of 0.63% for Biobrane, and TTRE 7.50 days and T%/T of 0.66% for Transcyte. In contrast, EzDerm was less efficient (TTRE in PT burns: 18.75 days; 0.23 T%/T) than SSD. The rates of infection and grafting were high in the case of EZDerm (37.0 and 23.9%, respectively) and Biobrane (12.9% for both), whereas Transcyte had a low rate of 5.0% for both. The need for reapplication was considerably low in the case of all three biosynthetic dressings, as shown by the small number of dressing changes in the case of EZDerm (n = 5), Biobrane (n = 3.4) and Transcyte (n = 1.5). Based on these results, biosynthetic treatments in children with PT burns are promising interventions, but in order to reduce the susceptibility to infection, and potentially the need for grafting, it is suggested that they should be applied in a combination with antimicrobial agents.
Silver foam dressings were mostly studied in small burns (<5 TBSA%) [43][44][45][46][47][48][49][50][51], though the wound's area [17] and closure time [47] were not reported in the RCTs in the case of Aquacel products. By the 10th day of the treatment, the reepithelialisation was remarkably high in the case of Acticoat (93%) and Aquacel Ag (94%). Accordingly, the TTRE and T%/T in PT burns were reasonable in the case of Acticoat (14.2 days and 0.23%, respectively) and Mepilex (10.3 days and 0.28%, respectively). The number of dressing changes, infection rates and grafting needs were relatively low in the case of Acticoat (n = 2.7, 3.5% and 20.9%, respectively) and Mepilex Ag (n = 4.0, 16.4% and 3.3%, respectively). The LOS in the hospital was notably short (only 3.1 days) in children treated with Mepilex Ag. Aquacel Ag was also associated with a small need for dressing changes (n = 1.0) and low susceptibility to infections (2.3%). These results suggest that the silver foam dressings are efficient interventions in PT burns of children. However, before they can be firmly recommended for general practice, further studies are warranted to test their effect on more extensive burns as well.
Similarly to silver foam dressings, the combination therapies were mainly analysed on smaller burns, which could contribute to their favourable TTRE in PT burns, viz., 15.0 days for Biobrane + Recell, 10.6 days for Acticoat + Mepitel and 8.7 days for NPWT + Acticoat + Mepitel, as well as to the low T%/T values, which were 0.35% for Biobrane + Recell, 0.14% for Acticoat + Mepitel and 0.17% for NPWT + Acticoat + Mepitel. As an exception, the treatment with the combination of Biobrane and Acticoat resulted in a longer TTRE of 21 days and a higher T%/T of 0.87%. By the 10th day of the treatment, the reepithelialisation percentage was modest in the case of Acticoat + Mepitel (42.5%), and NPWT + Acticoat + Mepitel (68.6%), whereas it was remarkably high (95%) in children treated with Biobrane + Recell. Every intervention performed better than SSD in terms of the lower need for dressing changes, which were on average 6.9 for Biobrane + Acticoat, 4.8 for Biobrane + Recell, 3 for Acticoat + Mepitel and 2.4 for NPWT + Acticoat + Mepitel. It should be noted that the price of these treatments was higher than the cost of SSD, but the nursing and operating theatre time, along with the anaesthetic use and total cost, were high in the SSD group as well. The higher initial cost of the combination treatments is one of the main obstacles that prevent them from widespread use as burn therapies. Compared to SSD, the grafting rates were also reduced for Biobrane + Acticoat (13.9%), Acticoat + Mepitel (6.8%), NPWT + Acticoat + Mepitel (2.1%) and Biobrane + Recell (0%), but it should be mentioned that low grafting rates were also found with the sole treatment of Biobrane (6.2%), Transcyte (5%) and Mepilex Ag (3.3%). The infection rates were exceptionally high in the case of Biobrane + Acticoat (60%) and Biobrane + Recell (40%), which suggests that these combinations suppress the antimicrobial efficacy, while no infections were reported in children treated with Acticoat + Mepitel, which may indicate a powerful antimicrobial effect.
Radiation-sterilised AM allografts combined with antimicrobial agents [20,40,41] and tilapia xenografts [16] seem to be a surprisingly effective, low-cost solution, but their procurement and storage may be challenging. Their application seems comfortable and less painful during and in-between dressing changes, and they were also associated with the least number of average dressing changes: Tilapia (n = 3.0), AM (n = 1.3) and AM + Nystatin + PMB (n = 0.5). Moreover, the times needed for wound closure were among the lowest reported values as indicated by TTRE and T%/T in paediatric PT burns for Tilapia (10.1 days and 1.11%), AM (13.3 days and 0.56%) and AM + Nystatin + PMB (6.0 days and 2.00%). The infection rate in the case of AM + Nystatin + PMB was also very low (1.9%).
We collected several additional interventions that could reduce the time for reepithelialisation (as indicated by TTRE red%, see Methods for details) when they were supplemented to the treatment. The list of these interventions (with the corresponding burn severity and TTRE red%) included NPWT (II/A:12.6%; II/B: 14.3%), rhEGF (II/A: 20.2%; II/B: 20.7%), bFGF (II: 21.1%), vitamin E + C + Zinc (II: 23.7%), rhGM-CSF (II/B: 27.9%), wIRA (II/A: 30.8%) and heparin (II: 40.0%). While in several cases the cost of these interventions presents a considerable obstacle to their use, supplementation with vitamins, minerals and heparin can be promising and inexpensive adjuvants in burn therapies. It must be noted, however, that we could identify only a single report for each treatment, which warrants further research to establish the true efficacy of these additional interventions. For additional information on the reported advantages and drawbacks of each analysed treatment option see Text S1.
While a similar meta-analysis was previously conducted about the management of partial-thickness burn wounds in children in 2014 by Rashaan ZM et al., they could only compare the effects of non-silver treatment related to silver sulfadiazine due to the scarcity of articles [33]. In recent years, several new research studies were published on this topic; thus, we were able to analyse subgroups such as biosynthetic or silver foam dressings as well. Another systematic review was published by Vloemans et al. in 2014 called the Optimal treatment of partial-thickness burns in children [56]. They separated their findings based on evidence level (RCTs, cohort studies, case reports), but they were not able to conduct statistical analysis for similar reasons as Rashaan et al. Since then, more than twice as many RCTs have been published on this subject, new interventions have been testedsuch as the combination or additional therapies-and more articles have been issued for the already existing therapies, which have now been added to this updated summary.
Limitations of our study must also be discussed. We aimed at collecting and evaluating articles strictly with the highest evidence level, namely RCTs; therefore, we had to exclude a lot of potentially relevant observational and case studies. It was surprising that despite the thorough review of the databases, we could identify a relatively low number of articles that fulfilled our inclusion criteria, especially if we consider the vast number of available treatment options. These RCTs were often describing a small and significantly heterogeneous population-due to the burns' mixed sub-depth ratios and various average areas. The low number of studies limited our options for a more extensive meta-analysis on individual interventions and resulted predominantly in a qualitative synthesis of the available data. As another limitation of our study, it can also be mentioned that secondary outcomes were scarcely reported, and even then, they were assessed in diverse ways, mainly in the cases of cost, pain sensation and scar formation; hence, we were not able to compare these three endpoints. Moreover, some of the research was conducted over two decades ago when many of the more accurate diagnostic devices for burn depth classification and area determination (such as Laser Doppler Imaging) were not as widely available as now [57]. Thus, the preciseness of the older measurements might be questionable, and they were often unverifiable without photo documentation.
The assessed risk of bias was also high in general, largely resulting from the lack of reporting randomisation and blinding as well as the absence of (pre)trial protocols. Furthermore, most of the studies did not follow the CONSORT criteria-which may be one of the reasons behind the cause of missing data-or disclose funding sources and conflict of interests. In those cases when the founders were mentioned, they were usually the manufacturers of the evaluated intervention, which poses further risks for bias.

Conclusions
There are still many pieces missing from the grand picture of paediatric partialthickness burn therapies; this review's main goal was to summarise our current knowledge on the topic. Although the results presented in this article will most probably change over time, we aimed at highlighting currently unclear areas in our understanding and at facilitating further clinical studies in the field. A future network meta-analysis would provide sufficient information to differentiate between the efficacy of individual interventions, but a lot more RCTs are needed before we will be able to properly compare them.
Our primary recommendation for investigators is that superficial and deep seconddegree burns in children should be analysed separately due to their significantly different characteristics. Furthermore, researchers should follow the CONSORT criteria and report predetermined outcomes of general interest (e.g., TBSA%, TTRE, T%/T, infection and grafting rates, number of dressing changes and LOS) along with their unique observations. Establishing a single, internationally accepted standard for pain and scar evaluation in paediatric burns would greatly advance this process. Another interesting future aspect could be the analysis of optimal dressing change rates. While every intervention could facilitate the healing of second-degree paediatric burn wounds, individual data analysis showed remarkable differences in secondary outcomes that could not have been statistically proven because of the aforementioned limitations. When choosing the preferred intervention in paediatric PT burns, physicians should consider treatments with little need for dressing changes because these options require the lowest number of anaesthesias, as well as cause the least pain and discomfort for the children. Moreover, by reducing the reapplication rate, the operating theatres' availability can be increased, and time can be saved for the healthcare providers, the advantages of which may result in a decrease in overall costs.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/life12050619/s1, Figure S1: Risk of bias assessment. The RCTs compared paediatric patients with PT burns. Figure S2: Pooled TTRE of the interventions, adjusted by burn depth. The groups were II/A, II and II/B injuries in children, and the TTRE was measured in days. Figure S3: Mean TTRE of the interventions, modified by burn area. The categories were <5, 5-10, 10-25 TBSA% paediatric PT injuries, and the TTRE was calculated in days. Figure S4: SSD's mean TTRE stratified by depth. The classifications were paediatric II/A and II burns and the TTRE was determined in days. Figure S5: Average TTRE of SSD, modified by burn area. The area cohorts were <10 and 10-25 TBSA% PT injuries, and the TTRE was defined in days. Figure S6: The mean TTRE of SSD and Biobrane. The meta-analysis of TTRE in children with PT was measured in days. Figure S7: Average grafted percentages. Treatment groups were paediatric PT patients with SSD, silver foam and biosynthetic dressings. Figure S8: Mean non-grafted patient ratio with SSD and Biobrane. The percentage comparison of successful conservative treatment in PT injuries of children. Figure S9: Average infected population percentages. Treatment comparison of SSD, biosynthetic and silver foam dressings as well as Biobrane only in paediatric PT injuries. Figure S10: Mean non-infected patients' ratio of SSD compared to Biobrane. The meta-analysis compares the frequencies of complication-free paediatric PT burn treatments, in percentages. Table S1: RoB2 analysis. Table S2: Mean TTRE of paediatric II/A treatments. TTRE values were calculated in days. (II/A = Superficial partial-thickness injury; NA = not applicable; NPWT = negative pressure wound therapy, NR = not reported; SD = standard deviation; SSD = silver sulphadiazine; SUM = the summarized values of the same interventions; TBSA% = burned area of the total body surface; TT = traditional treatment; TTRE = time to reepithelialisation; TBSA%/TTRE= the area of regeneration per day; TTRE red% = the percentage of time reduction, with the addition of the intervention; VIS = visible spectrum light; wIRA = water-filtered infrared A.) Table S3: The average TTRE in paediatric II/A and MD burns. The TTRE was measured in days. No description was provided about the TBSA% in these instances, thus the T%/T ratios could not be calculated. (II/A = Superficial partial-thickness injury; MD = mid-dermal burn; NR = not reported; SD = standard deviation; SSD = silver sulphadiazine; SUM = the summarized values of the same interventions; TBSA% = burned area of the total body surface; TTRE = time to reepithelialisation.) Table S4: Mean TTRE summary of II/B injuries in children. The TTRE was calculated in days. (II/B = Deep partial-thickness burn injury; AM = amnion membrane; NA = not applicable; NPWT = negative pressure wound therapy; NR = not reported; SD = standard deviation; SSD = silver sulphadiazine; SUM = the summarised values of the same interventions; TBSA% = burned area of the total body surface; TTRE = time to reepithelialisation; TBSA%/TTRE = the area of regeneration per day.) Table S5: Average percentage of reepithelialisation on the tenth day (day 10 RE%) in paediatric PT injuries. (II = Partial-thickness burn injury (PT); II/A = superficial PT; AM = amnion membrane; day 10 RE = the fraction of reepithelialisation on the tenth day; NA = not applicable; NPWT = negative pressure wound therapy; NR = not reported; SSD = silver sulphadiazine; TBSA% = burned area of the total body surface; TTRE = time to reepithelialisation; TBSA%/TTRE = the area of regeneration per day.) Table S6: Mean length of stay. The time children spent (days) inside the hospital while being treated with different regimens for PT burns. (II = Partial-thickness burn injury (PT); II/A = superficial PT; AM = amnion membrane; LOS = length of hospital stay; NPWT = negative pressure wound therapy; NR = not reported; PMB = polymyxin b; SD = standard deviation; SSD = silver sulphadiazine.) Text S1: Reviews of the analysed interventions. Text S2: Search keys. References [16][17][18][19][20][21][22][23][24][25][26][27][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51]53, are mentioned in the Supplementary Materials.