Stats Made Simple: Interpreting tables, graphs, and figures

What quickly sets experienced readers apart is how efficiently they can extract key information from figures and tables in a scientific article. Each type of scientific publication follows a relatively standardized structure and presents data in predictable ways, making it easier to identify the information needed. The EQUATOR Network provides reporting guidelines for each study design, and their checklists can be used to focus on the essential elements required for critical appraisal of a publication. 

In this text, we provide practical clues to understand tables, diagrams, and figures commonly encountered in clinical research (randomized trials, observational studies, and meta-analyses), and briefly describe the structure of fundamental/basic research articles (1–3).

A. Study design diagram

Flow diagrams are usually the easiest place to start, as they offer a rapid overview of the study design. These diagrams highlight the journey of participants or studies from recruitment to completion. Pay particular attention to potential sources of bias, such as absolute numbers at each step, reasons for exclusion, and adherence to follow-up (4).

B. Table 1

This naturally leads to the descriptive statistics section, which typically presents data in large tables summarizing patient characteristics, including age, sex, disease duration, disease activity measures, comorbidities, and other relevant variables. The purpose of these tables is to assess representativeness of study sample. Check carefully sample characteristics to capture common biases, such as more severe disease, fewer comorbidities, or other selection effects (4). Meta-analyses also present similar information, usually summarizing study characteristics according to the PICO structure (Population, Intervention, Comparison, Outcome).

C. Other tables in clinical research papers 

Once the study sample is well defined, clinical research articles focus on assessing associations between variables. This process generally occurs in two steps: first determining whether an association exists and then estimating the strength of that association. Statistical significance is commonly reported using p-values (typically considered significant if <0.05) or 95% confidence intervals (considered significant if they do not cross 1 in the context of ratios). Researchers then quantify the magnitude of the association. Depending on the study design, results may be reported using measures such as relative risk (RR), odds ratio (OR), hazard ratio (HR), Pearson or Spearman correlation coefficients, beta coefficients, chi-squared tests, Fisher’s exact test, absolute risk reduction (ARR), or number needed to treat or harm (NNT/H). While a detailed explanation of each statistical measure is beyond the scope of this article, the table below summarizes the key differences between the three most encountered ratios in rheumatology studies: OR, RR, and HR (5,6).

	Relative Risk (RR)	Hazard Ratio (HR)	Odds Ratio (OR)
Interpretation in clinic	Gives you a multiplication factor for the control group incidence	Gives you a multiplication factor for the control group incidence	Used as an estimation of RR/HR but overestimate the level of association if the incidence is high (common threshold at 10%)
Study design	Trials and Cohort studies	Trials and Cohort studies	Every (the only one usable for Case-Control studies)
Timeline	Static, based on cumulative incidences at the end of the study.	Dynamic, based on rates.	Static. Does not consider rates. Summarise an overall study.
Limitations	Requires a highly representative population.	Requires a highly representative population. Rate of change between groups needs to be constant	Not always useful, may exaggerate risk.

Modified from George et al. Cureus 2020

In addition to relative measures, it is important to examine absolute risks with and without the intervention. Absolute risk provides valuable insight into the real-world impact of an intervention on a population and often makes conclusions more understandable for non-specialists and patients. Measures derived from absolute risk, such as absolute risk reduction (ARR) and number needed to treat or harm (NNT/H), are also widely used in health economics (5).

D. Figures in clinical research papers

Fortunately, scientific publications do not rely solely on tables; figures are frequently used to visually illustrate key findings. The main types of graphs encountered in clinical research are illustrated below. 

A. Kivitz et al. Ann Rheum Dis 2025
S.V.J. Snoeck Henkemans et al. Ann Rheum Dis 2025
Pluma et al. Ann Rheum Dis 2025

Many pharmaceutical randomized controlled trials use bar plots to highlight their primary endpoint. As shown in example A, this phase II RCT demonstrates a difference in ACR20 response rates between placebo and two treatment conditions at a specific time point. While this figure is important from a research perspective to demonstrate drug efficacy, rheumatologists may be more interested in outcomes that are directly relevant to clinical practice, such as remission rates at 3- or 6-months using guidelines criteria (4). In cohort studies or trials reporting hazard ratios, Kaplan–Meier curves are commonly presented. These graphs allow rapid visualization of differences in event rates over time between groups and represent the graphical expression of the HR. The final example (C) illustrates a classical forest plot, typically used in meta-analyses. Individual studies are displayed on the left with their corresponding effect estimates ((incidence rate ratio is synonymous for RR), while the diamond at the bottom represents the pooled result. When combining heterogeneous studies or case-control designs, it is often necessary to convert association measures to OR before pooling (4).

E.  Basic research articles (Tables and Figures)

Outside of clinical research, fundamental or basic research articles follow a different logic. Tables are less frequently used in the main manuscript, and figures play a central role. Figures are typically organized to support new biological or mechanistic findings and may combine experimental workflows, visual representations of results (such as flow cytometry gating strategies, immunohistochemistry images, UMAP plots from single-cell RNA sequencing, or western blot images), and statistical analyses. The nature of these figures depends heavily on the experimental techniques used, and familiarity with these methods is often essential to critically assess the validity and relevance of the results.

In conclusion, the ability to critically interpret tables, graphs, and figures is essential to efficiently evaluate scientific literature in rheumatology. Understanding the structure of different study designs, recognizing common statistical measures, and identifying the strengths and limitations of visual data presentation, is key for evidence-based medicine.

References