This is one of the first observational studies in Brazil that exclusively assessed pediatric oncology inpatients and compared two nutritional screening tools as well as the performance and efficacy of both tools on these inpatients. Similarly, Bicakli and Kantar (2021) studied 170 pediatric inpatients in an oncology hospital in Turkey, and their findings revealed a prevalence of oncological disease close to what we found, with 68.2% of patients presenting with solid tumors and 21.8% presenting with hematological cancer [41].
Malnutrition is defined by Mehta et al. as “an imbalance between nutrient requirements and intake that results in cumulative deficits of energy, protein, or micronutrients that may negatively affect growth, development, and other relevant outcomes” [42]. We know that the malnutrition rates are higher in hospitalized children and that they can also be related to the primary disease. This can cause numerous damages to the child’s life and survival as well as have social and economic impacts [43–45].
In regard to the prevalence of malnutrition in the sample, 21.6% of the inpatients were determined to be malnourished. These results are similar to those found in a study conducted by Carniel et al., who assessed 242 patients and classified approximately 20% of these patients as being malnourished or at nutritional risk using SGNA. This same study also showed a high sensitivity of SGNA to the nutritional risk diagnosis, a finding that was also observed in our study, where we found that SGNA had a sensitivity of > 90%. Moreover, these same authors reported that SGNA can be applied to patients not only at the time of admission in order to identify the nutritional risk early, but it can also be used effectively as a nutritional assessment method [27].
Other studies also found similar results, such as the one presented in 2015 by Costa and Pastore, who carried out a longitudinal study in Brazil and identified a prevalence of malnutrition through S/A and W/A parameters equivalent to 22.1% and 20.8%, respectively [46]. Likewise, in a meta-analysis by Iniesta et al., a prevalence of malnutrition of 20% for all types of pediatric cancers was observed [47].
Nutritional screenings are crucial to prevent malnutrition or to identify patients at risk of malnutrition. Among all pediatric nutritional screening methods available, there is only one developed for pediatric oncology patients; however, to the best of our knowledge, no tool is considered as the gold standard for the screening of these patients or even has been validated in Brazil [48].
The objective assessment was used as a standard to compare the nutritional screening tools and was carried out through anthropometric measurements. This method can be considered as a gold standard, since it uses parameters accepted and validated worldwide to achieve the final result in the objective assessment [14, 16, 49].
A cross-sectional study published in 2015 conducted in a pediatric hospital in southern Brazil aimed to compare SGNA and StrongKids with the objective assessment at hospital admission and to associate them with the period of hospitalization. They found significant values but minimum concordance (κ = 0.148) between StrongKids and the objective assessment. When analyzing the concordance between SGNA and the objective assessment, the results were even lower (κ = 0.043; p = 0.063) [50]. In our study, this information agrees with the values found in the comparison between StrongKids and the objective assessment, which also obtained a low concordance (κ = 0.192); however, it disagrees with respect to the values found in the comparison between SGNA and the objective assessment, where significant results were obtained (p = 0.021) with a reasonable concordance level (κ = 0.235).
In the Master's thesis conducted by Maciel in Brasília, Brazil (2018), which had the objectives of analyzing the accuracy of the screening tool StrongKids and estimating the prevalence of malnutrition and nutritional risk in 271 children from 10 public hospitals in the Federal District, the author found malnutrition and nutritional risk percentages of 12.1% and 33.9%, respectively. With respect to the accuracy analysis, StrongKids identified that 78.8% of the malnourished inpatients were at nutritional risk. Among the well-nourished inpatients, the tool showed that 87.9% had no nutritional risk. The author discusses the grouping carried out in the StrongKids final classification, which classified ‘at risk’ inpatients as ‘at high risk’ by the tool; this may have had an impact on the loss of sensitivity (12.1%) and the increase of specificity (97.9%) [30].
With nutritional screening in pediatrics, the objective is to identify individuals who present a nutritional risk so that any intervention can be carried out. Therefore, higher values for sensitivity and PPV demonstrate a higher probability that the patient classified as being at nutritional risk indeed carries this risk. Lower specificity values, as seen in the comparison of both screening tools in the present study, bring a higher chance of false-positive results, which means incorrectly classifying inpatients at no nutritional risk as being at risk. Nevertheless, it is expected that a screening tool shows high sensitivity values so that there are few false-negative results and nutritional interventions can be implemented early. The results obtained in the current study may be linked to the structure of each nutritional screening method, with SGNA being a more complex tool that requires higher technical skills to assess the nutritional status, thus resulting in a higher sensitivity, versus StrongKids, which is a practical and easily applicable tool.
Some authors have evaluated StrongKids as a simple, fast, and easily applicable tool [41]. SGNA, on the other hand, is considered as a more complex screening method that requires training and experience by the evaluator and consequently demands more time to be carried out [51, 52]. Secker and Jeejeebhoy, the SGNA authors, do not report on how much time is needed to complete it or the level of training required of the evaluators, definitely relevant pieces of data, which are limitations of this tool [12]. In the current study, we found that the StrongKids screening tool took around 5 min to complete, while the SGNA screening tool demanded approximately 15–20 min to complete.
Of note, there is no specific question for inpatients who are exclusively on enteral or parenteral nutrition in either questionnaire; therefore, the assessment is incomplete, since all questions made in relation to feeding apply only to oral feeding. No discussion on this topic is available in the literature.
There are some additional limitations of this study that must be addressed. First, the sample size was limited, which may have had an impact on the significance found in some analyses. In addition, many patients may have postponed doctor’s appointments due to the COVID-19 pandemic and consequently delayed the start of oncological treatment. The statistical difficulty presented by the results related to the score/classification in StrongKids was another limitation, making it necessary to make adjustments to the final classification of inpatients so that we could carry out more conclusive analyses.
A strength of this study was that all analyses were conducted by a single evaluator who always used the same equipment; therefore, the variability bias among evaluators and/or differences in results related to the equipment used was eliminated.