Almost every clinical paper opens its Results with Table 1: who the participants were before the exposure or treatment. Reviewers read it to judge whether the groups are comparable and whether the sample resembles the patients your findings will be applied to. The format is conventional, and a few choices (which summary statistic, P values or not) are where most revision requests come from.
What goes in Table 1
- Rows: the baseline variables a reader needs to judge comparability and generalizability: demographics, disease characteristics, comorbidities and the key confounders. Not outcomes.
- Columns: one per group (treatment arm or exposure group), plus an overall column, each headed with its number of participants.
- The right summary for each variable:
- continuous and roughly normal: mean (SD);
- continuous and skewed (lab values, length of stay): median (IQR);
- categorical: No. (%), with the percentages of each group's total.
- Missing data: the number missing per variable, in the row or a footnote.
The format
| Characteristic | All (N = 120) | New treatment (n = 60) | Standard (n = 60) | Standardized difference |
|---|---|---|---|---|
| Age, mean (SD), y | 62.7 (10.3) | 64.0 (9.4) | 61.4 (11.0) | 0.25 |
| Female, No. (%) | 54 (45.0) | 34 (56.7) | 20 (33.3) | 0.48 |
| BMI, mean (SD) | 27.2 (4.0) | 28.0 (4.4) | 26.5 (3.4) | 0.36 |
| C-reactive protein, median (IQR), mg/L | 5.9 (3.3-10.0) | 6.1 (3.3-9.0) | 5.8 (3.3-10.8) | |
| Diabetes, No. (%) | 33 (27.5) | 12 (20.0) | 21 (35.0) | 0.34 |
| Stage, No. (%) | ||||
| I | 35 (29.2) | 16 (26.7) | 19 (31.7) | |
| II | 60 (50.0) | 34 (56.7) | 26 (43.3) | |
| III | 25 (20.8) | 10 (16.7) | 15 (25.0) |
Footnote: "Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); IQR, interquartile range. Standardized differences above 0.10 indicate imbalance."
How to describe it in the text
Of the 120 patients, 54 (45.0%) were women, and the mean (SD) age was 62.7 (10.3) years (Table 1). Patients receiving the new treatment were more often women (56.7% vs 33.3%) and less often had diabetes (20.0% vs 35.0%) than those receiving standard care.
Then say how you handled the imbalance (for example, adjustment in the regression model).
P values or standardized differences?
- Randomized trials: do not test baseline differences. CONSORT advises against it: any difference is due to chance by design, so a P value answers no useful question.
- Observational studies: P values depend on sample size and are a poor guide to balance. Report the standardized difference instead (difference in means or proportions divided by the pooled SD); values above 0.10 are conventionally treated as meaningful imbalance. In the example, sex (0.48), BMI (0.36), diabetes (0.34) and age (0.25) are all imbalanced.
Mistakes reviewers catch
- Mean (SD) for skewed variables. A lab value with a long right tail needs median (IQR).
- "±" for the SD. Medical journals following AMA style write mean (SD), so "±" leaves readers unsure whether it means SD or SE.
- Percentages that do not say their denominator, especially when data are missing.
- Outcomes in Table 1. It is for baseline characteristics only.
- P values in an RCT's Table 1.
Let KyroStat do the write-up for you
KyroStat builds Table 1 from your data: it picks mean (SD) or median (IQR) by the shape of each variable, adds counts and percentages, computes standardized differences, and formats the table for your journal.
Frequently asked questions
Should the overall column come first or last? Either; follow the journal's examples. Many put it first.
How many decimals? One for means, SDs and percentages is usual. See how to report statistics in AMA style.
Which reporting guideline covers Table 1? STROBE item 14 for observational studies and CONSORT for trials. See the STROBE checklist guide and the CONSORT 2025 guide.
What if my journal uses APA style? The same structure works; use APA's number formatting. See how to report descriptive statistics in APA 7.