Presenting Descriptive Statistics in a Paper

Everything in this module comes together in one place: Table 1, the 'baseline characteristics' table that opens almost every clinical paper. Its job is to describe who was actually studied, so readers can judge whether the results apply to their own patients. Learning to build and read Table 1 is a practical capstone for descriptive statistics. The structure is simple and conventional. Rows are the variables (age, sex, comorbidities, key labs); columns are the groups being compared, for example treatment versus control, often with a total column. Each cell summarizes one variable in one group using the rules you have now learned: continuous symmetric variables as mean (SD); continuous skewed variables as median (IQR); and categorical variables as n (%). Choosing the right summary for each variable type is exactly the skill this module has built, and getting it wrong, reporting a mean for skewed length-of-stay data, for instance, is a giveaway of weak statistical care. A well-made Table 1 lets a reader answer three questions at a glance. Who was studied? The summaries describe the sample, which determines external validity, whether you can generalize the findings to your patients. Were the groups comparable at baseline? In a randomized trial the two columns should look similar; large baseline differences hint at failed randomization or confounding. And is the sample size adequate and clearly reported? Every variable should show how many patients contributed, making missing data visible rather than hidden. One important modern caution concerns p-values in Table 1. In a randomized controlled trial, comparing baseline characteristics with p-values is discouraged by major journals and the CONSORT guideline, because any differences are by definition due to chance (randomization guarantees it), so a 'significant' baseline difference is meaningless. Instead, reviewers increasingly prefer standardized differences to flag imbalances that might matter. In observational studies, baseline tables legitimately highlight differences between groups that may need adjustment later. Practical tips make your Table 1 trustworthy: state the unit and summary type for every row; report denominators when data are missing; keep decimal places sensible (age to whole numbers, not two decimals); and never present a percentage without its count. Done well, Table 1 is not a formality, it is the foundation of the reader's trust: if the description of the sample is careless, why would anyone believe the analysis that follows? Mastering it means you can both build an honest table and spot a misleading one.