The authors used a binary logistic regression model in the research because the purpose of the research was to predict the presence of cancer using NTA and urine samples. Notably, the test of the presence of cancer could either be positive or negative. Therefore, the results of the test were binary (with two possible mutually exclusive outcomes). Besides, the predictive nature of the logistic regression was the reason for the authors’ choice of the model (Kilic, 2015) . With the binary logistic regression model, the researchers could determine the association between one dependent binary variable (the presence or absence of cancer) and one or more independent variables that could be measured on nominal, interval, ratio or ordinal level of measurement.
Using the binary logistic regression model was the most appropriate approach for the researchers since the model could be used to predict the probability of a person having cancer. In other words, the model could be used to predict the probability of getting prostate cancer under certain conditions or the factors expected to cause cancer. The analysis of the significant difference between NTA and TA groups was done using a two-tailed t-test. This test was appropriate because the analysis involved a comparison of two groups. Besides, the sample sizes of the two groups were different; the sample size of TA groups was n=26, and that of NTA was n=12 (Jarrard et al., 2019) . Therefore, a paired sample t-test or one way ANOVA could not be appropriate, making the independent sample t-test the most appropriate test. Notably, the sample sizes for the groups were both less than 30. Hence the student’s t distribution, which corresponds to the use of t-test, ought to be appropriately used.
Delegate your assignment to our experts and they will do the rest.
The authors display the results in both tables and figures. For instance, they use Table 1 to display the urine sample clinical and pathologic characteristics, and Table 2 to display the p values of prostate cancer with clinical factors along with ore with PLA2G16. Besides, the authors use Figure 1B, for example, to illustrate the percentages of methylation in the NTA and TA groups for different samples, which were found to be significantly different since the t-test p-value was p<0.001.
The results table stands alone since it is possible to interpret the study from it. Notably, the result table on clinical and pathologic characteristics indicates interpretable results through the highlighted p values of the patients’ characteristics. From the table, the significant characteristics of the patients are PSA, prostate size, and PSA density. Their corresponding p values of 0.03, 0.01, and 0.01 respectively are less than 0.05. On the other hand, the p-value of the age of the patients is 0.27, which is larger than 0.05, implying that this demographic characteristic is not statistically significant in assessing the presence of cancer in a patient.
Similarly, the results table showing the odds of prostate cancer with PLAG2G16 or clinical factors has the p values for each pair that shows the significance of the predictive accuracy of the model. For example, the predictive accuracy of PLA2G16 was the highest at 0.8 for CG2. The statistical significance of these results can be assessed using the corresponding p-value, which is p<0.001. Using a similar approach, the significance of the predictive accuracy of the other models can be assessed from the table by checking the corresponding p values and comparing them against 0.05 level of significance. The p-values that are less than 0.05 imply that the odds ratio has statistically significant predictive accuracy. On the other hand, the odds ratios with the p-value larger than 0.05 mean that they lack statistical significance, as far as their predictive accuracy are concerned.
References
Jarrard, W., Schultz, A., Etheridge, T., Damodaran, S., Allen, G., Jarrard, D., & Yang, B. (2019). Screening of urine identifies PLA2G16 as a field defect methylation biomarker for prostate cancer detection. PLOS ONE , 14 (6), e0218950. doi: 10.1371/journal.pone.0218950
Kilic, S. (2015). Binary logistic regression analysis. Journal of Mood Disorders , 5 (4), 191. doi: 10.5455/jmood.20151202122141