How to make sense of clinical evidence: Why statistical thinking matters

By Dr Kristof Theys, expert-trainer of the Essentials of Statistical Thinking for Medical Affairs course.

Interpreting clinical evidence requires more than identifying statistically significant results. A meaningful understanding of study outcomes hinges on the ability to critically appraise the robustness, precision, and applicability of the findings. This requires statistical literacy: a structured approach to understanding uncertainty, validity, and relevance in the context of clinical research.

At the heart of this is a critical question: to what extent do the observed results reflect a true treatment effect, and how confident can we be in their applicability to clinical practice?
This can be systematically addressed by examining three core domains: internal validity, precision, and external validity (generalisability).
 

• ✅ Are the results trustworthy?

  The first thing to assess is validity. Internal validity refers to whether the study results accurately reflect what happened within the study population. This depends on the design and execution of the study. Common threats include selection bias, measurement bias, and confounding, all of which can distort results and lead to incorrect conclusions.

  External validity, or generalisability, concerns whether the results apply beyond the study sample. For instance, findings from a highly controlled hospital setting might not extend to broader primary care populations or to patients with different comorbidities.
   

• 📏 How reliable are the findings?

  Once internal and external validity have been considered, the next step is to assess precision: how much uncertainty surrounds the effect estimate.

  Larger sample sizes tend to yield more precise results by reducing random error. Confidence intervals are key here: narrower intervals indicate greater precision. However, they reflect only statistical uncertainty, and interpretation assumes that the study design and model assumptions are sound.
   

• ⚖️ Do the results matter in practice?

  Even precise and valid findings must be evaluated for clinical relevance. A statistically significant result (e.g., p < 0.05) does not always translate into a meaningful improvement for patients. It’s important to weigh the magnitude of effect, the confidence in the estimate, and the balance of benefits versus harms.
   

What role do study design and statistical methods play?

Different types of studies offer varying levels of evidence. Randomised controlled trials (RCTs) are typically the gold standard for estimating causal effects, but even well-designed RCTs can suffer from issues like loss to follow-up or inadequate blinding. Observational studies, while more reflective of real-world settings, require careful handling of confounding and bias.

Equally critical is how data are analysed. Statistical methods must align with the study design and appropriately address factors such as data type, missing values, confounders, and multiplicity (i.e., multiple hypothesis testing). Misapplication of statistical techniques can lead to false inferences, regardless of how well the study was designed.
 

Statistical thinking as a core competency

At its core, statistical thinking helps us deal with uncertainty. It allows us to turn raw data into interpretable insights: the foundation for sound clinical judgment, evidence-based decisions, and improved patient outcomes.

In a landscape where evidence drives regulatory, clinical, and commercial outcomes, the ability to think statistically is not optional, it is essential.