Psychometric Properties of Food Frequency Questionnaires Used in Dentistry: A Systematic Review Free

High and frequent consumption of sugary foods or drinks is considered a risk factor for the development of chronic noncommunicable diseases, such as diabetes, hypertension, cardiovascular diseases, and dental caries, in addition to causing other damages to general health [Amezdroz et al., 2019]. The American Dental Association states that food consumption is a determining factor for good oral health and directly influences the development of dental caries [American Dental Association, 2022].

Dental caries is the most prevalent multifactorial disease [Moreira, 2016]. Socioeconomic and cultural factors, exposure to fluoride, and excessive and frequent consumption of foods with high free sugar content are the main factors associated with the development of the disease [Brasil, 2008; Rosa, 2015; Amezdroz et al., 2019]. The cariogenicity of foods or beverages is determined by the presence of fermentable carbohydrates with sticky textures and highly processed starches. These carbohydrates are used as a substrate by microorganisms in the oral cavity and consequently contribute to the formation of caries lesions [Rosa, 2015].

Determining the cariogenicity of the diet is essential for classifying the patients’ risk of illness. In addition to knowledge, training, and time for quality assessment, dental professionals require an accurate assessment of dietary control with the choice of a reliable dietary assessment instrument [Fernandes, 2015; Patenaude et al., 2020]. In clinical practice, the most common instruments to assess food intake are the 24-h recall (R24h), 3-day food diary (DA), and food frequency questionnaire (FFQ). FFQs are most frequently used in studies investigating the association between dietary consumption and clinical outcomes; however, the quality of the instruments available for use in dentistry has not been reported in the literature [Fernandes, 2015; Pedraza and Menezes, 2015; Rosa, 2015].

When choosing a dietary assessment instrument, especially for application at the population level, possible errors in measurement due to sources of systematic bias should be eliminated by the validation process [Fisberg et al., 2009]. This validation process is of paramount importance for dental professionals in order to have an effective, reliable, and recognized method for diagnosing eating behaviors at risk for dental caries. For the validation of the instruments, their quality, applicability, and psychometric properties (internal consistency, test-retest reliability, content validity, face validity, and construct validity), with emphasis on the levels of reliability, validity, responsiveness, and interpretability, need to be considered [Campos et al., 2019]. Thus, the data can be generalized in an appropriate way, that is, knowing that the information collected measures and evaluates food consumption reliably, measuring exactly what it intends to measure [Fernandes, 2015]. Therefore, the objective of this systematic review was to analyze the quality of the food frequency instruments/scales used in dentistry while considering their psychometric properties.

This review protocol was submitted for registration in the PROSPERO database (International Prospective Register of Systematic Reviews), under protocol number CRD42021223949, and performed according to the PRISMA 2020 guidelines (Transparent Reporting of Systematic Reviews and Meta-Analyses) [Page et al., 2021].

The construction of the study question followed the acronym CITYPOMP proposed [Munn et al., 2018] for the validation studies. The study question was as follows: What is the quality (reliability, validity, and responsiveness) of FFQs used in dentistry to assess the consumption of foods with high cariogenic potential? The acronym guided the selection of keywords.

• Construct of interest (CI): to evaluate the consumption of foods with high cariogenic potential

• Type of measurement instrument (TY): FFQ

• Population (PO): instruments aimed at any age group

• Measurement properties (MP): reliability, validity, and responsiveness

The databases used for the bibliographic survey were PubMed, Scopus, Web of Science, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), Dentistry and Oral Sciences Source (via EBSCO), the Latin American and Caribbean Health Sciences Literature Database (LILACS), and Brazilian Bibliography of Dentistry (BBO). Controlled vocabulary and free terms related by Boolean operators “OR” and “AND” were used according to the guiding question of the study. Each search strategy was adapted according to the specifics of each database. There were no restrictions on language or publication period. The search strategies for all databases are included in online supplementary material 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000529886). All results were imported into the endnote reference manager to remove duplicates and then into the Rayyan QCRI manager to perform the selection process.

Gray literature was also accessed from Proquest, Banco de Teses Capes and Biblioteca Digital de Teses e Dissertações (BDTD), Google Scholar, and the proceedings of the International Association for Dental Research (IADR) to find unpublished works in the main databases. The reference lists of the primary articles included were also checked to identify any titles not retrieved in the main search. Searches were carried out in all databases and gray literature from July 8, 2020, to October 10, 2020, with an update on December 12, 2022.

Studies on the construction and/or validation of instruments/scales to assess food consumption for use in dentistry (in clinical settings and/or epidemiological studies), with or without evaluation of their psychometric properties, were included. Review studies, letters to the editors, and research protocols were excluded.

The included articles were initially selected by reading the titles and abstracts, and duplicate studies from both the databases were eliminated. A complete reading was performed if there was any doubt regarding the eligibility of the study due to lack of information in the abstract. All included studies were read in full to confirm the eligibility criteria. The entire process was performed independently by two examiners (C.S. and S.V.R.), who strictly complied with the inclusion and exclusion criteria. Disagreements were resolved through discussion with the help of a third reviewer (J.S.R.). The agreement among reviewers was evaluated using the Kappa coefficient; it was acceptable when the values were above 0.7–0.8 (substantial agreement) and 0.8–1.0 (excellent agreement) [Landis and Koch, 1977]. The value found after reading the titles and abstracts was 0.77, which was considered a substantial agreement. This value increased to 0.81 after reading the full text, which was considered an excellent agreement [Landis and Koch, 1977].

Two reviewers independently performed data extraction (C.S. and S.V.R.) using an extraction form designed for this review, according to the characteristics identified in the studies. Before extraction, the form was pretested, and the researchers were trained. Disagreements among researchers were resolved through discussions with the help of a third reviewer (J.S.R.).

The methodological quality of the included studies was assessed using the Consensus-Based Standards for the Selection of Health Measurement Instruments (COSMIN) [Mokkink et al., 2018], which is a specific instrument for validation studies. This instrument allowed the evaluation of the quality of studies by considering the following psychometric properties: internal consistency, test-retest reliability, content validity, face validity, and construct validity [Campos et al., 2019]. It had 10 assessment categories that were used to determine whether the study met the standards of good quality, response theory, and generalizability of results [Mokkink et al., 2018].

Categories 1 and 2 referred to the validity of the content (requirements, general scope of the project, and validation by specialists); categories 3–5 referred to structural validity, internal consistency, and cross-cultural validity, respectively; together, they formed the internal structure. Categories 6–10 formed the properties of the remaining measurements: reliability, measurement error, hypothesis testing for construct validity, and responsiveness [Melo et al., 2014]. These categories consisted of items that assessed the measurement properties. Each item can be classified as (1) very good, (2) suitable, (3) doubtful, (4) inappropriate, or (5) not applicable (NA). The final score for each category was determined using the lowest ranked option; for example, if for a reliability study one item in a box is rated as “inadequate,” the overall methodological quality of that reliability study is rated as “inadequate” [Mokkink et al., 2018].

The risk of bias assessment was performed independently by two examiners (C.S. and S.V.R.). Differences were discussed until a consensus was reached, with the participation of a third reviewer (J.S.R.). All researchers were trained to use COSMIN. Inter-rater agreement was calculated using the intraclass correlation coefficient [Campos et al., 2019]. For each property, the percentage of items classified as “excellent,” “very good,” “adequate,” “doubtful,” or “inadequate” was calculated. For this, Wilcoxon’s nonparametric test for ordinal variables was performed to assess whether there was any systematic error; the significance level adopted was p ≥ 0.05, with no significant difference between the mean scores of the classification of each examiner. The intraclass correlation coefficient was 0.92, indicating an excellent inter-rater agreement. Wilcoxon’s nonparametric test resulted in a value of p = 0.166; therefore, there was no systematic error among the evaluators.

Data analysis was performed considering the following items: author, country, instrument language, sample, assessment instrument, instrument structure, type of food, comparison with clinical condition, adaptation of an existing instrument, purpose of the study, and psychometric properties. It was not possible to conduct a meta-analysis of the psychometric properties as only the studies by Psoter et al. [2008] and Devenish et al. [2017] expressed the Pearson correlation data. The structures of these instruments were very dissimilar; one was a FFQ with evaluation of foods and beverages with total and free sugars [Devenish et al., 2017], and the other was a FFQ regarding the consumption of sugary drinks and candies and the assessment of the amount of added sugars in food preparation [Psoter et al., 2008].

The article selection process is illustrated in Figure 1. After searching the databases, gray literature, and removing duplicate studies, a total of 6,421 articles was found. After reading the titles and abstracts, 8 articles were selected; however, at the stage of reading the texts in full, one study was excluded because it was not a specific instrument for assessing food consumption and the risk of caries. Finally, 7 articles were included in this literature review.

Table 1 summarizes the information found in the included articles [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Devenish et al., 2017; Amezdroz et al., 2019; Ngah et al., 2019; Devenish et al., 2020; Patenaude et al., 2020]. Most of the studies were conducted in the American continent, in the USA [Custodio-Lumsden et al., 2016], Canada [Patenaude et al., 2020], and Haiti [Psoter et al., 2008]. Two studies were carried out in Asia, in Japan [Shinga-Ishihara et al., 2014] and Malaysia [Ngah et al., 2019], and two in Oceania, both in Australia [Devenish et al., 2017; Amezdroz et al., 2019]. The instruments available in the literature are in English [Custodio-Lumsden et al., 2016; Devenish et al., 2017; Amezdroz et al., 2019; Patenaude et al., 2020], Japanese [Shinga-Ishihara et al., 2014], Creole [Psoter et al., 2008], and Malay [Ngah et al., 2019]. Only one study performed translation and cross-cultural validation of an instrument [Shinga-Ishihara et al., 2014], whereas the others were construction studies [Shinga-Ishihara et al., 2014], built and adapted the instrument from an English version [Evens, 1997] to Japanese. The English version of the study [Evens, 1997] was not found for complete reading, preventing the quality and validation of the original instrument from being evaluated, considering its psychometric properties.

Regarding the target population of the instruments, two studies [Custodio-Lumsden et al., 2016; Amezdroz et al., 2019] were intended for children but answered by parents, as they evaluated the diet of children up to preschool age. One instrument [Psoter et al., 2008] performed a joint dietary assessment of parents and their children. Two studies [Devenish et al., 2017; Ngah et al., 2019] obtained information on the frequency of food consumption from parents of children to assess the family’s habitual consumption. Four studies validated their instruments in an adult population and one [Shinga-Ishihara et al., 2014] applied the Japanese version of the FFQ to pregnant women. Only one [Patenaude et al., 2020] validated their instrument in the general population.

Regarding the objectives of the instruments, some evaluated the frequency of general food consumption while others focused on foods with high sugar content or associated with other measures. Those with more than one measure were evaluated in addition to food frequency, family behavior, or other risk factors related to the development of dental caries. One study [Devenish et al., 2017] evaluated general food consumption, considering not only cariogenic foods but also foods that protected general health. Three studies [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Devenish et al., 2017] evaluated food frequency related only to sugary food. Two studies developed an FFQ using other methods [Custodio-Lumsden et al., 2016; Patenaude et al., 2020]. The instrument proposed by [Patenaude et al., 2020] evaluated the frequency of food consumption with eating behavior. [Custodio-Lumsden et al., 2016] developed a tool called MySmileBuddy, which evaluated dietary practices, caregiver attitudes and beliefs, fluoride use, and family history, in addition to food consumption. Finally, the instrument by [Amezdroz et al., 2019] was a cariogenic scale that measured the cariogenicity of foods and beverages. Only three studies [Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Amezdroz et al., 2019] compared the results of the proposed instrument with the clinical condition of the patient. These studies compared the frequency of food consumption with the following clinical parameters: presence of caries using oral clinical examinations (ICDAS II) [Amezdroz et al., 2019] and assessment of the level of salivary mutans streptococci [Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016].

Table 2 presents the classification of the quality of the studies for each psychometric property considered in the COSMIN checklist [Mokkink et al., 2018]. Regarding the general evaluation of the included studies, all were classified as “inadequate.” Online supplementary material 2 presents the analysis of the quality of the studies for each psychometric property considered in the COSMIN checklist, so it is possible to follow in full all the questions evaluated in each domain of the instrument, as well as the detailed classification of each study. Complete validation of the instrument, considering the categories evaluated by COSMIN, was performed only by Shinga-Ishihara et al. [2014]. Five studies were identified [Psoter et al., 2008; Custodio-Lumsden et al., 2016; Amezdroz et al., 2019; Ngah et al., 2019; Patenaude et al., 2020] that analyzed at least one psychometric property (linguistic validation, test-retest reliability and reproducibility, expert panel, pilot test, criterion validity, and internal consistency). Only one study did not assess any psychometric property [Devenish et al., 2017].

Overall, among all the categories evaluated by COSMIN, the category “3” of internal structure and structural validity (3) was the most frequently evaluated as adequate. The construct validity category (9) was the worst evaluated, as it was not considered by most studies and only were classified as “adequate” [Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016]. The study by Shinga-Ishihara et al. [2014] received the best evaluation since it was the only study to evaluate all the psychometric properties used in COSMIN; however, it also had a general classification of “inadequate”, as not all criteria for each category were met.

All studies evaluated the reliability categories (test-retest); however, only one study [Shinga-Ishihara et al., 2014] obtained an “adequate” quality rating in this category. With respect to responsiveness, three studies [Shinga-Ishiharai et al., 2014; Devenish et al., 2017; Custodio-Lumsden et al., 2016] were classified as “doubtful.” The others presented “inadequate” quality for this category [Psoter et al., 2008; Amezdroz et al., 2019; Ngah et al., 2019; Patenaude et al., 2020], as no study had fully validated this psychometric property.

The assessment of food consumption is a crucial part of the approach to dental treatment [Fernandes, 2015], as intake of sugary foods and drinks is directly related to the development of dental caries [Tini and Long, 2015; Mokkink et al., 2018; Campos et al., 2019]. The correct evaluation of this consumption represents a great challenge for these professionals; therefore, the use of reliable instruments with validated psychometric properties [Campos et al., 2019] and the ability to effectively identify foods with potential risks for oral diseases [Fernandes, 2015; Rosa, 2015] is of paramount importance.

Most of the instruments found in this review are FFQs related to the consumption of foods and beverages with a high content of total and free sugars. They are aimed at the age group of children up to preschool age [Psoter et al., 2008; Custodio-Lumsden et al., 2016; Devenish et al., 2017; Amezdroz et al., 2019; Ngah et al., 2019] and are in English [Psoter et al., 2008; Custodio-Lumsden et al., 2016; Devenish et al., 2017; Amezdroz et al., 2019; Patenaude et al., 2020]. Regarding the quality of these instruments, all studies were classified as “inadequate” in the general assessment, with reliability (test-retest) being the most validated psychometric property [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Devenish et al., 2017; Amezdroz et al., 2019; Ngah et al., 2019; Patenaude et al., 2020]; only one study [Shinga-Ishihara et al., 2014] validated all the psychometric properties used in COSMIN, even if incompletely.

An important finding of this study was the predominance of diet instruments in English, with only three in other languages: Japanese, Creole, and Malay [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Ngah et al., 2019]. However, the diet has been evaluated in other languages and different countries in dentistry [Arheiam et al., 2016; Amezdroz et al., 2019; Ngah et al., 2019; Hancock et al., 2020; Mahboobi et al., 2021], as it is one of the factors related to dental caries, even without an instrument validation. Regarding the diet assessment methods most frequently used in published studies, three systematic reviews found that most studies used the DA method, followed by the FFQ and R24h, to assess the relationship between dental caries and food [Moynihan and Kelly, 2014; Hancock et al., 2020; Mahboobi et al., 2021]. These instruments are the reference methods to be used in epidemiological studies and/or in clinical practice to assess food consumption [Fernandes et al., 2013; Fernandes, 2015; Rosa, 2015]. To choose the best method, objective of the analysis (individual or epidemiological), psychometric properties, and possibility of adapting the measure for the study population should be considered, considering the regional diversity of eating habits [Evens, 1997; Shinga-Ishihara et al., 2014].

Both DA and/or R24h can be applied to any population; however, for epidemiological studies, these instruments do not encompass variability, habits, and seasonality of the diet and do not represent habitual consumption [Fisberg et al., 2009]. In addition, the compilation and analysis of their results are more complex because of the qualitative nature of the data, which are difficult to apply in large epidemiological studies [Fernandes 2015; Pedraza and Menezes, 2015]. In contrast, the FFQ estimates dietary consumption over time based on the possibility of measuring the intensity of exposure and classifying individuals according to the frequency of consumption. This makes it possible to estimate the association of consumption categories with the development of chronic noncommunicable diseases [Fernandes, 2015; Rosa, 2015]. This variability of methods found in the literature can make comparisons between studies difficult as some of them, such as DA and R24h, do not allow long-term evaluation. Considering the complexity of obtaining valid measurements of food intake in epidemiological studies, there is a need to adapt and validate existing FFQ [Kolodziejczyk et al., 2012; Fernandes, 2015] for different languages and cultures. In addition, the formulation of instruments with the help of nutrition professionals plays an important role in the reliable construction and evaluation of food consumption with cariogenic potential. The main focus is on the use by dentists for dental interventions to diagnose the consumption of potentially cariogenic foods and help in possible clinical outcomes. Therefore, the support of the nutrition professional, with his extensive knowledge in the area, adds to the improvement of the development of the evaluation instrument.

In the literature, there are few instruments validated for adults and the elderly [Moynihan and Kelly, 2014], which was confirmed by the findings of this review since most of the FFQs found were directed toward children. Once the disease burden reaches adulthood, it is important that the relationship between diet and dental caries be investigated in these groups [Moynihan and Kelly, 2014; Hancock et al., 2020; Mahboobi et al., 2021], and considering the specificities of this age group, it might be interesting to build and/or adapt an instrument. Even in the age group of children up to preschool age, food choices and frequency are totally dependent on parents and caregivers [Scaglioni et al., 2018], and children do not have the cognitive ability to fully self-report their food consumption; therefore, parents and caregivers must answer these questions [Devenish et al., 2017; Amezdroz et al., 2019; Ngah et al., 2019]. There are limitations to obtaining information about a child’s nutrition from parents and caregivers since most of them cannot stay with their children for most of the day and information can be hidden [Kolodziejczyk et al., 2012; Morikava et al., 2018]. However, these instruments are still relevant, given the importance of identifying risk factors for caries, including diet [Fernandes, 2015; Rosa, 2015]. This reinforces the importance of evaluating the construct validity of these instruments, especially convergent validity, by comparing their results with patients’ clinical conditions.

Regarding translations and cross-cultural evaluation, only one study [Shinga-Ishihara et al., 2014] adapted an existing instrument in English into Japanese, designed to be applied to children [Evens, 1997]. The original version had 35 food items with assessment of food consumption related to caries during the previous month [Evens, 1997]; however, this instrument was not found in its complete version, so its psychometric properties could not be evaluated. The Japanese version [Shinga-Ishihara et al., 2014] added foods that were regularly consumed in the country to increase the validity of the content. In this version, the questionnaire was validated exclusively for the population of pregnant women, as dietary preferences often change during this period and can lead to frequent ingestion of cariogenic foods [Shinga-Ishihara et al., 2014]. Studies [Psoter et al., 2008; Ngah et al., 2019] only translated the instrument, being translated/back-translated from a source language to another language using native speakers; it was not reported whether cross-cultural adaptation was carried out in the test-retest version. One study [Psoter et al., 2008] only validated reproducibility (test-retest).

For cross-cultural validation, adaptations must be made carefully since the instrument must be internally reliable; for international comparisons, ensuring the quality of the data and the information produced is important. The instrument must be translated/back-translated from a source language to another language using native speakers. Mastery of the other language, carried out by peers in a hidden way, carrying out the comparability of the translation results, adaptation of the culture, habits, and seasonality of the diet, and carrying out a pretest for necessary modifications demonstrates reliability and validity [Shinga-Ishihara et al., 2014; Ngah et al., 2019]. In the study by Shinga-Ishihara et al. [2014], all stages of translation and cross-cultural adaptation were carried out, ensuring reliability and comparability with the original study; however, there were flaws in the validation process as the criteria for each category were not fully complied with, and therefore, the overall rating of the study was “inadequate.”

Three studies [Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Amezdroz et al., 2019] performed construct validation and compared the results of the proposed instrument with the clinical condition of the patients. In two studies [Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016], salivary mutans streptococcal levels showed a statistically significant relationship with dietary intake, indicating that their higher levels were associated with higher scores for cariogenic food intake. One study [Amezdroz et al., 2019], with the use of clinical oral examinations (ICDAS II), showed a significant increase in caries lesions as advanced disease at the age of 5 years, when the child’s intake of discretionary foods that do not provide nutrients, including a cariogenic diet, increases. These relationships between the instrument and the clinical condition are essential for findings to be effectively generalized and for the information to reflect the individual’s real condition and be able to measure exactly what is proposed [Fernandes, 2015].

The most frequently evaluated property in the studies was reliability (test-retest) [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Amezdroz et al., 2019; Ngah et al., 2019; Patenaude et al., 2020]. Only one study obtained an “adequate” quality rating according to COSMIN [Shinga-Ishihara et al., 2014]. The studies [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Amezdroz et al., 2019; Ngah et al., 2019; Patenaude et al., 2020] validated this property incompletely, with the absence of an interval between two applications, lack of description of the training, and whether the interviewers were able to carry out the research, making the quality of the measurement instrument questionable [Cunha et al., 2016; Alves, 2017]. In addition, despite being a necessary quality, it must be evaluated together with other components within the psychometric properties to guarantee its accuracy.

Finally, the general results of the methodological quality of studies found showed that all studies included in this review were classified as “inadequate” since the general classification of the quality of each study was always determined by the worst condition found in the criteria evaluated in each category. The study by Shinga-Ishihara et al. [2014] received the best assessment according to the COSMIN scale, as it was the only one to assess all predicted psychometric properties. However, it was still classified as inadequate because the hypothesis testing and response capacity categories were incomplete [Shinga-Ishihara et al., 2014]. This showed that most studies have important limitations that influence the evidence on the instruments if they measure what they purport to measure [Vermelho, 2014]. The rigidity of the COSMIN general classification was to ensure an instrument’s validity, reliability, and reproducibility; the psychometrics of each measure must be carefully considered [Cunha et al., 2016; Mokkink et al., 2018; Campos et al., 2019].

It is possible to highlight other limitations that affected the methodological quality of the studies, such as the limitation of the sample population and failure to detail all psychometric properties performed [Psoter et al., 2008; Shinga-Ishihara et al., 2014; Custodio-Lumsden et al., 2016; Ngah et al., 2019; Patenaude et al., 2020]. Among the main flaws, one study [Devenish et al., 2017] did not have any psychometric properties for validation. When a psychometric instrument is not tested, its measurement is imprecise, which limits the validity of the inferences obtained from this instrument [Alves, 2017]. It is necessary to compare these instruments with the clinical conditions of oral health to ensure that they are measuring exactly what is expected [Fernandes, 2015; Cunha et al., 2016].

Regarding the limitations of this systematic review, it is important to note that it was not possible to have full access to one study [Evens, 1997]. The study by Shinga-Ishihara et al. [2014] has adapted an existing instrument in English into Japanese, designed to be applied to children [Evens, 1997]. However, all possible attempts were made to access complete articles.

Most of the instruments found in this literature review consisted of FFQs related to the consumption of foods and beverages with a high total and free sugar content, aimed at the age group of children of preschool age. All studies were classified as “inadequate” in the general assessment. Reliability (test-retest) was the most validated psychometric property, and only one study validated all psychometric properties used by COSMIN. Advances are needed in the validation process of these instruments, with appropriate and complete methods for validating their psychometric properties, to minimize measurement errors and avoid incorrect estimations.

There are no conflicts of interest to declare. There was no need to go through the Research Ethics Committee to deal with a systematic literature review. An ethics statement is not applicable because this study is based exclusively on published literature.

There are no conflicts of interest to declare.

This study was funded in part by the Integrated Research Center (CIP) of the Pontifical Catholic University of Paraná, Brazil.

All authors have reviewed and approved the complete manuscript. Caroline Souza dos Santos, Juliana Schaia Rocha, and Samuel Jorge Moysés conceived the ideas of the systematic review and the methodological approach. Caroline Souza dos Santos, Saulo Vinicius da Rosa, and Juliana Schaia Rocha performed the screening and data extraction of this literature review. Caroline Souza dos Santos, Saulo Vinicius da Rosa, Juliana Schaia Rocha, Rodrigo Nunes Rached, Marcia Helena Baldani, and Renata Iani Werneck performed the final selection. Gil Guilherme Gasparello helped with the English language and led the final writing of the manuscript.

All data generated or analyzed during this study are included in this article and its supplementary material files. Further inquiries can be directed to the corresponding authors.