Introduction
Periodontitis is a common dental disease affecting about 90% of the planet’s adult population. Just like caries, it leads to tooth loss. However, if almost everyone knows about the dangers, causes, and prevention of caries, then periodontitis is still a little-known disease for many (Preshaw & Bissett, 2019). The so-called periodontium surrounds human teeth. It consists of gums, blood vessels, and connective and bone tissue. The periodontium provides nutrition to the hard tissues of the tooth and the alveolar process – the part of the jaw in which the tooth sockets are located, and it also tightly holds the tooth in its place.
The paper provides a critical analysis of the study of this disease. In this situation, the article’s authors resorted to sex assessment using PCA analysis methods, which contribute to data dimensionality reduction with minor information loss and linear discriminant analysis with nonparametric statistic tools (Zhao et al., 2021). This choice is due to a small sample of 17 people, which is a problem in selecting suitable patients with all the appropriate input conditions for the experiment. The prevalence, complexity of the disease and potential harm to other systems of the human body dictate the need for its study.
A Review of the Data Analysis
Research Problem
Periodontitis is an infectious inflammation of the periodontium. It is often preceded by gingivitis, in which the gum’s surface becomes inflamed: there is bleeding and discomfort. With periodontitis, the inflammation penetrates more profoundly; the symptoms may intensify. If the disease is not treated, it turns into periodontal disease, in which bone tissue is destroyed, and teeth begin to fall out. Factors that create a favorable environment in the mouth for them to feed and reproduce cause this disease (Preshaw & Bissett, 2019). These include the following possible causes, which are common in most patients.
Firstly, poor oral hygiene: after eating, food particles remain in the gaps between the teeth, in the gum pockets – the depression between the gum and the tooth tissue: it is an excellent breeding ground for bacteria. The less or carelessly people brush their teeth, the more likely they are to develop inflammation. Secondly, the cause may be tartar: every day, dental plaque, a biofilm of bacteria, forms on a person’s teeth.
Even in developed countries, the prevalence of the disease and the lack of adequate health education have become fundamental prerequisites for posing the research problem. An integrated approach used in the in-depth diagnosis of the disease at the microbiota level requires constant study of various aspects of periodontitis. In addition, side effects can harm the patient beyond the oral cavity, all the way to the immune system (Preshaw & Bissett, 2019). In order to study this issue in more detail, it often resorts to the demographic and other characteristics of patients to closely monitor the risk group. The wide range of microbiome variations in these traits points to the complex nature of the problem, requiring detailed research on each possible aspect.
Data Collection and Source
As mentioned above, the sample for this experiment is relatively tiny because rather severe conditions were set for selecting the necessary patients. As a result, 17 people were selected, eleven of whom were men. Inclusion criteria were between 20 and 44 years of age, resulting in a median of 35 and 34 years for males and females, respectively (Zhao et al., 2021). In addition, the presence of at least fifteen natural teeth was highlighted, while there was no fact of the introduction of antibiotics, treatment of periodontitis, bridges, implants, and prostheses, or systemic diseases, including infectious ones in the oral cavity. Finally, the patients had to be non-smokers to exclude bad habits’ harmful and robust influence on the experiment results.
Participants were validated by professionals in the field who are practicing dentists. A grade of periodontitis was included from mild to the most severe grade 4 (Zhao et al., 2021). The study was officially registered, and each participant went through the appropriate consent procedures before any necessary verification and control procedures for future data.
The experiment included procedures specific to this area for oral interviews, inspections, and taking samples with their special storage. At least nine health indicators were assessed for the appropriate classification of patients according to morbidity and other characteristics. The obtained samples were subjected to DNA extraction, sequencing, and bioinformatics analysis using appropriate software, and finally, the obtained results were evaluated using several particular professional databases and analyzed the principal components of PCA and linear discriminant LEfSe (Zhao et al., 2021). Therefore, it can be concluded that the participation of patients was limited to oral interviews, examinations, and taking appropriate samples for further study using applied tools.
Variables
The fundamental division of the research experiment was the differentiation of groups by sex; however, a deep further analysis assumed the specific integration of multiple indicators. The demographic characteristics of patients are essential in the context of the study of this disease. Age did not show statistical significance in this experiment at p > 0.05 (Zhao et al., 2021). Gender in this context is the independent variable, while the dependent variable is the composition of the microbiome, in particular the number of specific species of representatives for pure marks, OTU, the Chao1 index, Simpson and Shannon, and the PD tree (Zhao et al., 2021). As a result, data differentiated by types and abundance of representatives of the microbiome were obtained and presented in the form of heat maps, cladograms, effect size diagrams, and tables.
The use of the linear discriminant analysis method is due to the continuity of the dependent variables, which in this case, do not meet the requirements of a normal distribution. In turn, the independent sex variable is discrete, simplifying the search for linear combinations of variables for dependence. Due to the broad differentiation of independent variables, different statistical analysis methods were used for each index.
Sample Size Estimation
When conducting studies determining the prevalence of a characteristic in a population, the calculation of the sample size is necessary for the resulting estimates to have the desired degree of accuracy. In effect-finding studies, sample size estimation is essential to ensure that it is likely to be found if a clinically or biologically significant effect exists, i.e., the analysis will yield statistically significant results (Wang & Ji, 2020). If the sample size is small, even if there are significant differences between groups, it will not be possible to prove that they result from something other than sample variability.
In this case, the second aspect is used when the effect of the sex-specific influence of various specific indices and microbiota composition is assessed. Accordingly, such checks were carried out primarily for dependent variables. The significance level is the threshold value for the p-score below which the null hypothesis must be rejected, and it is concluded that there is evidence of an effect – it was taken as 0.05 (Zhao et al., 2021). The null hypothesis was the lack of a relationship between sex and microbiota structure in young people with severe periodontitis (Zhao et al., 2021). At the same time, the purpose of the study was also exploratory and prognostic in the specifics of the influence of specific microbiota representatives on the disease’s development and its course.
In this regard, such a small sample is justified in the framework of this experiment. There is no need to project the results to a significantly larger population due to the specifics of representatives of the same race and age and the emphasis solely on the gender of the participants. The clinically significant effect size is the slightest difference between group means or percentages of events within the odds ratios closest to unity risk that can still be considered biologically or clinically significant (Wang & Ji, 2020). The sample size should be such that if such differences exist, then the study would produce statistically significant results. Within the framework of this experiment, the results turned out to be statistically significant, which led to the cancellation of the null hypothesis. Formally, the information was confirmed by multiple checks as part of the evaluation of dependent variables, which will be discussed below.
Appropriateness of Statistic
Initially, the analysis was carried out to understand the statistical significance of age as a separate independent variable. The assessment showed that age differences are not statistically significant, which allows us to ignore this variable as influencing the course and results of the experiment under these conditions (Zhao et al., 2021). The Illumina MiSeq and QIIME software tools and the clustering of individual microbiota sequences showed thinning curves with a trend toward more excellent saturation, confirming that the sample was sufficient to capture statistically significant information from the results.
Indexes such as Simpson, Shannon, Chao1, and tree PD measurements were tested using the Wilcoxon rank sum nonparametric statistical test with a significance level of less than 1%. The results showed that the Simpson and Shannon indices did not differ in patients with different stages of the disease close to advanced at a confidence level already greater than 5% (Zhao et al., 2021). The advantage of this test is that it gives significant results on pretty small samples. Using its assessment of differences between two samples by the level of any trait, measured quantitatively, allows for building a heat map with a broad differentiation of dependent variables that do not have a normal distribution.
Further results have already been evaluated using other statistical analyses. Differences directly in the microbiota composition in quantitative characteristics of abundance with weighted voting and analysis of PCA and PCoA showed a difference in sex groups with a statistical significance of less than 0.1% (Zhao et al., 2021). At the same time, the authors of the study resorted to non-metric multidimensional NMDS scaling based on unweighted UniFrac distances, which partially offset the small sample sizes.
The evaluation of the effect of linear discriminant analysis LEfSe has already been used as a link between the predictive function of the experiment to identify specific representatives of the microbiota that create a statistically significant difference between groups. As a result, taxa characteristics of both males and females were identified, which makes an essential contribution to the study of this disease, albeit on such a local scale.
The further development of diseases associated with immunity was dictated by such a difference in the differentiation of the microbiota composition. Identifying specific representatives using the KEGG pathway mechanism made it possible to implement the predictive function of potential hazards for patients with severe forms of periodontitis. Although the Wilcoxon tests were found to be different in level for each particle identified, statistically significant gender differences were found (Zhao et al., 2021). The immune system of women appeared to be more prepared and enriched, while that of men was at greater risk. At the same time, the potential for the development of cardiovascular diseases caused by these representatives was revealed, with a noticeable difference between the sexes confirmed by the tests.
Given the specifics of data processing, the use of nonparametric methods that do not exclude statistical inference and the possibility of constructing a distribution in some instances is justified due to the broad differentiation of dependent variables. Parametric tests are more powerful than nonparametric ones in the case of normal population distribution. Nonparametric tests are less sensitive than their parametric counterparts, and if it is essential to detect even slight deviations, special care should be taken in choosing the test statistic (Davis et al., 2019). However, some problems may arise with the interval scale if the data are not presented in standardized estimates. The human oral microbiota is still under active study; therefore, no general standards for its assessment have yet been identified. In addition, checking the distribution for normality requires rather complex calculations, the result of which is not known in advance. More often, the distribution of signs differs from the normal one; then, one has to turn to nonparametric criteria.
Nonparametric tests are devoid of the above limitations. However, they do not allow a direct assessment of the level of such important parameters as the mean or variance; with their help, it is impossible to assess the interactions of two or more conditions or factors that affect the change in signs. In addition, nonparametric methods are most justified with small samples, which is entirely consistent with this case.
Data Display
The reflection of the data obtained is an essential illustration due to the broad differentiation of the representatives of the microbiota and the multi-stage evaluation of the study. Although ultimately, a statistically significant difference between the two groups of patients by gender is considered, the specifics of processing the data obtained require the use of a large number of different tools with non-parametric methods of evaluation. Each step of the study is illustrated either by a table or a suitable diagram, both in heat maps and three-dimensional boxplots (Zhao et al., 2021). Considering that the experiment stages have different functions, from finding dependence to predicting the development of the disease, methods specific to the medical field are used to demonstrate data.
In fact, the prediction stage uses bioinformatic resources to assess the impact of the identified representatives of the microbiota on the health of the patient as a whole. Metabolic pathways for specific input data may reflect the potential development of the course of the disease in severe forms, at least within this small sample. Extrapolation to larger populations is limited by many factors, including demographics, which, according to the article’s literature review, also have a statistically meaningful effect on disease progression.
Nevertheless, such a comprehensive analysis makes it possible to visualize the data obtained in several diverse figures in the article. The illustration of the results is crucial for the convenience of their potential implementation and use as a foundation for further research. When integrating related knowledge, the reflection of the output data can at least help in the fine work with the disease of the representatives of this sample, which can potentially lead to better results in the treatment, diagnosis, and control of the disease. Although the professional interpretation of the study’s results is currently imprecise, it provides food for more targeted practical experiments in this field of medicine.
Data Analysis Evaluation
The data analysis and literature review section of many studies are critical for several reasons. First, these chapters form the basis of the experiment, postulating a problem mentioned in other works, causing the motivation and necessity of such an activity. Secondly, they provide an understanding of potential blind spots in this area, which, when analyzed in detail, determines the primary tasks in the study of a particular disease. Finally, any inconsistency or misunderstanding in these matters on the part of the authors may lead to a distortion of the results, which, as a rule, are used in practical medicine. In this regard, a preliminary critical assessment of this section is always necessary.
Nurses often skip this section due to its volume with limited time resources, entirely relying on the conscientiousness of the authors of the articles. At the same time, in this situation of increased responsibility, especially with patients with severe forms of diseases, only professional knowledge is not enough. Statistical analysis provides essential indicators of the degree of confidence in the results obtained and understanding the motivation for sample size; choice of confidence interval should also be subjected to preliminary critical evaluation.
Projecting the preceding onto this study, it should be said that the identified representatives of the microbiota by sex groups can similarly affect patients only in the specified age range, the selected race, and the set of conditions given in the article as input data. First, the apparent influences of smoking, dietary habits, the presence of systematic treatment, or other oral cavity diseases can significantly affect each stage of the experiment, completely changing, for example, the predictive conclusions of the study. Secondly, in addition to these aspects, there are many more hidden ones that should also be focused on: for example, the cultural characteristics of the behavior of patients of a particular race, chronic diseases of the immune system, taking certain drugs, up to a genetic predisposition. Although patients’ DNA is analyzed in this situation, many features of this structure are still under active study and can give completely polar results without a preliminary critical assessment of these sections.
Conclusion
In this paper, the article What is the Sex Variations in the Oral Microbiomes of Youths with Severe Periodontitis was critically evaluated for the adequacy and practical significance of the statistical methods used. The analysis showed that, despite a small sample, a step-by-step examination of specific aspects of the disease with a differentiated assessment of statistical significance using non-parametric methods provides an understanding of the difference in the course of the disease by sex and is also predictive of the context of the demographic characteristics of the sample. Finally, the importance of the literature review sections, data analysis, and understanding of statistical tools in the professional and applied nursing practice was assessed.
References
Davis, S. E., Greevy Jr, R. A., Fonnesbeck, C., Lasko, T. A., Walsh, C. G., & Matheny, M. E. (2019). A nonparametric updating method to correct clinical prediction model drift. Journal of the American Medical Informatics Association, 26(12), 1448-1457.
Preshaw, P. M., & Bissett, S. M. (2019). Periodontitis and diabetes. British Dental Journal, 227(7), 577-584.
Wang, X., & Ji, X. (2020). Sample size estimation in clinical research: from randomized controlled trials to observational studies. Chest, 158(1), S12-S20.
Zhao, Y. Q., Zhou, Y. H., Zhao, J., Feng, Y., Gao, Z. R., Ye, Q. & Guo, Y. (2021). Sex Variations in the Oral Microbiomes of Youths with Severe Periodontitis. Journal of Immunology Research, 2021.