The Hidden Overlap Between Patient Group Means in Bariatric Randomized Controlled Trials

Abstract

Introduction

Reliance on summary data such as averages may not fully represent the breadth of individual patient responses that occur within a randomized controlled trial. As a result, a large portion of reported patient outcomes may be reasonably expected regardless of the trial arm to which a study subject is assigned. This study aims to investigate the extent of results overlap that exists between interventions in bariatric randomized controlled trials, despite significant P values by analyzing differences in trial means and standard deviations (SDs).

Methods

A comprehensive literature review was conducted on bariatric RCTs from 2010 to 2023, sourced from PubMed, MEDLINE, Cochrane Library, and EMBASE. Bariatric surgery trials examining percent weight loss were selected due to the continuous nature of the data. The inclusion criteria for the data were outcomes reported as mean ± SD, and normally distributed. The data distributions for each study were visualized using histograms to assess overlaps in mean weight loss across different interventions. Using provided sample means and SDs from each selected randomized controlled trial, percentage of overlap between patient group distributions of each study was determined.

Results

Out of 27 initially identified RCTs, six were included. These showed significant overlap between means, based on P values, for different bariatric interventions. The mean percent overlap of patients across all interventions of the 6 studies was 84.58%, with a minimum of 68.42% and maximum of 98%. This indicates that across all studies, only an average of 15.42% of patients fell outside the overlapping distribution and could be considered to have a weight loss solely as a response to the specific treatment.

Conclusions

While means are essential for statistical analyses, it is crucial to examine deeper nuances in data to understand prior to assigning causation for an individual patient response. Such insights are pivotal in the era of evidence-based and precision medicine, ensuring that treatment decisions are tailored not just based on group averages but also considering the potential range of individual outcomes.