Exclusion Criteria

Exclusion criteria are the characteristics that disqualify a study or participant from being included in a research study or systematic review. These criteria are established before data collection begins and serve to create a well defined study population. They work in tandem with inclusion criteria to determine which participants, studies, or data points are eligible for analysis.

Clearly defined exclusion criteria are essential for the scientific rigor and reproducibility of research. In clinical trials, they protect participant safety by excluding individuals for whom an intervention might be harmful (such as pregnant women or patients with severe kidney disease). In systematic reviews, they help limit the evidence synthesis to studies of appropriate design, population and outcomes. Well crafted exclusion criteria create a more homogeneous sample, which reduces variability (and, in observational studies, can reduce confounding through restriction), making it easier to detect treatment effects. However, overly restrictive exclusion criteria can limit the external validity of the findings, creating a tension between internal validity and generalizability that researchers must carefully balance.

Exclusion criteria are defined during the study planning phase and documented in the protocol, grant application, or systematic review registration (such as a PROSPERO registration). For clinical trials, exclusion criteria typically fall into several categories: safety related (excluding individuals at high risk of harm from the intervention), methodological (excluding individuals whose conditions might confound the results), and practical (excluding individuals unlikely to complete the study).

In a systematic review, exclusion criteria apply at the study level. Reviewers specify which study designs, populations, interventions, comparators, outcomes, and settings fall outside the scope of the review. During the screening process, reviewers apply exclusion criteria first at the title and abstract stage, then at the full text stage, documenting reasons for exclusion at the latter stage.

The decision about what to exclude requires careful justification. Each exclusion criterion should have a clear scientific rationale; exclusions that appear arbitrary or that disproportionately exclude certain demographic groups (such as older adults, women, or racial minorities) are increasingly criticized for reducing the applicability of research findings. In the US, the FDA's 2020 guidance Enhancing the Diversity of Clinical Trial Populations: Eligibility Criteria, Enrollment Practices, and Trial Designs recommends that sponsors examine each exclusion criterion to see whether it is necessary, avoid broad, overgeneralized exclusion categories, and consider broadening eligibility as safety data accumulate, for example between Phase II and Phase III. Reporting guidelines also ask for transparency: CONSORT 2025 asks trial reports to state the eligibility criteria for participants and to show participant flow in a diagram, including losses and exclusions after randomization with reasons, and PRISMA 2020 asks systematic reviews to state their eligibility criteria and to explain why studies that might appear to meet the inclusion criteria were excluded.

Exclusion criteria should be considered, alongside inclusion criteria, for any research study or systematic review that selects participants, records or studies. They are most critical in clinical trials, where they directly affect participant safety and the interpretability of results. In systematic reviews, they define the boundaries of the evidence synthesis. In survey and observational research, they clarify which respondents or records are outside the scope of the investigation. The criteria should be specified a priori (before data collection or screening begins) to prevent data driven decisions that could introduce bias.

A common error is failing to justify exclusion criteria, listing them without explaining the scientific or safety rationale behind each one. Overly restrictive criteria that exclude large proportions of the target population can produce findings that are internally valid but clinically irrelevant because they do not apply to typical patients. In systematic reviews, a frequent mistake is applying exclusion criteria inconsistently, with different reviewers making different decisions about borderline cases. This can be addressed by having two independent reviewers screen studies and using a third reviewer to resolve disagreements. Another mistake is modifying exclusion criteria after data collection has begun without documenting the change and its rationale, which introduces the risk of outcome driven decisions.

A clinical trial of a new diabetes medication specifies the following exclusion criteria: age under 18, pregnancy or planned pregnancy, eGFR below 30 mL/min/1.73 m² (indicating severe kidney impairment), current use of insulin, and history of diabetic ketoacidosis. During recruitment of 2,000 screened patients, 680 (34%) are excluded: 180 for kidney impairment, 220 for insulin use, 150 for age, 90 for pregnancy related criteria, and 40 for ketoacidosis history. The CONSORT flow diagram documents each exclusion reason and the numbers at each stage.