Statistical Analysis in Medical Research: A Case Study

Research purpose and hypothesis

Q1: What was the research purpose of this study? What was the research hypothesis? (Please present both null and alternate hypotheses in your answer. (4 points) 

Q2: What was the primary outcome? How was the primary outcome measured? Was this primary outcome a continuous variable, an ordinal variable, or a binary variable? (Please choose the most appropriate answer.) (4 points)

Q3: This was a hypothesis testing study, and the test power and sample size were specified even before the trial started. In the first paragraph in "Statistical Analysis" section, they estimated a sample size of 44 (the total of both groups) would achieve 94% power to detect a minimum of 1.5 points mean difference in pain scores between treatment and placebo. Unlike Cohen's method used in Literature Review 1, the sample size determination here used results from previous studies, which is the most common way for sample size calculation.

 

Q4: Table 1 summarizes the demographic and clinical characteristics by randomization group at the baseline. P values from demographic comparisons between these two groups (the saline group and the 4% Lidocaine group) were not presented in Table 1. If we want to compare with the following demographic and clinical characteristics between these two groups at the baseline, which statistical test would you recommend to use? You should identified the exact test (t, U, Chi square tests), two-sided or one-sided, paired or unpaired. (6 points)

 

Q5. Figure 2 showed the median pain scores during or after intercourse as recorded in weekly diaries using the Numeric Rating Scale. There were three p values in the figure. 

 

Q5a) What does p =0.007 indicate? You should first identify which two samples were compared (the same group of patients at different time points or different groups of patients at the same time?) and which statistical test was used and make a conclusion about the significance of the difference between two samples. (3 points)

 

Q5b) What does p < 0.001 indicate? You should first identify which two samples were compared (the same group of patients at different time points or different groups of patients at the same time?) and which statistical test was used and make a conclusion about the significance of the difference between two samples. (3 points)

 

Q5c) What does p =0.004 indicate? You should first identify which two samples were compared (the same group of patients at different time points or different groups of patients at the same time?) and which statistical test was used and make a conclusion about the significance of the difference between two samples. (3 points)

 

Q5d) If we want to compare two samples at "the end of the blinded 4 weeks" (i.e. the same time point), which statistical test would you recommend to use based on the "Statistical analysis" section they provided? (1 point)

 

Q6: Based on all information you get from the article, answer the following questions in regard to the nature of this study?

 

Q6a) Was this study a retrospective study or prospective study? (1 points)

 

Q6b) What was the design of this study, RCT Parallel, Case-Control, RCT Crossover, and cohort study? (1 point)

 

Q6c) If we want to define this study as one of the following trials: Superiority, non-inferiority, and equivalence, which one would you recommend? (1 point)

 

Q7). In this study, they used a t test-based procedure (mean difference) to get the sample size (for the power of 94%), but for analyzing the data they got, they used Mann-Whitney test and Wilcoxon Signed Rank test. Whats your own opinion about this issue, i.e. different tests were used for sample size calculation and comparing the real data, respectively? (3 points)