Determining true difference between treatment groups

Abstract

In this article, the author reviews the P value and how it is used to determine true difference of outcome in treatment groups. P value, standard deviation, standard error of the mean, bias, and confidence interval are discussed in common language, with a minimum of jargon and with clinical examples.

A New Measure of Overlap: An Alternative to the p--value

“…This is a standard set-up and the theory using p-values well documented. See, for example, Cardinal (2016) and Pocock (2006). Now H 0 will come under suspicion of being wrong if X, the sufficient statistic for θ, which is itself the sample mean of the observations from the individuals allocated to treatment T 1 , is too large.…”

Understanding data and its analysis for interpretation – Correct interpretation and safety matters!

“…Clinical studies are conducted to test the superiority usually (inferiority and equivalence trials are also conducted) of an intervention over a control (cervical epidural along with general anesthesia vs. general anesthesia alone in this study). [ 4 5 6 7 ] The outcomes of such studies are evaluated based on two aspects– statistical (result-based merit) and clinical relevance of the outcome (process-based merit or non-statistics merits). [5-8] Both need to be considered in a study for making a meaningful conclusion.…”

“…The two statistical tools—P value and Confidence Interval (CI)—are usually reported. [ 6 7 ] They are related tools but provide different information of the data collected in the study. The P value does not prove the superiority of an intervention over the other but is just denotes whether the null hypothesis rejected based on observed data analysis.…”

“…However, this interpretation solely based on P value may be misleading as it does not provide any estimate of the true difference among the morphine consumption (or other parameters) in the two groups. [ 7 8 9 ]…”

“…CI provides magnitude of effect, variability and describes the uncertainty in the estimated difference (e.g., in the morphine consumption). [ 4 7 8 9 ] In other words, it provides a range of true effect of difference in the outcome parameters of the study. If there is no difference in means of the observed value among groups, then the difference shall be zero.…”