A cannabinoid trial can answer a carefully defined question without representing everyone who might be interested in its findings. Before interpreting an outcome, it helps to ask who could join, who actually joined, and whose circumstances remain outside the evidence. These are related questions, but each reveals a different limit.
Eligibility describes the intended study population
Inclusion criteria specify characteristics needed to enter a trial. Exclusion criteria identify characteristics that prevent entry. Depending on the research question, these might concern a diagnosis, age range, previous treatment, other medicines, or additional health conditions.
The US National Cancer Institute’s explanation of cancer trials describes eligibility criteria involving factors such as age, medical history, and previous treatment. It explains that criteria help identify appropriate participants and protect their safety. This is a cancer-research example, not a universal description of every cannabinoid trial.
Criteria also belong to a study’s planned methods. A registry entry can therefore help establish what population researchers originally intended to study. Our article on clinical trial registration and cannabinoid study plans explains how those plans fit into interpreting research.
Enrolment shows who actually participated
Eligibility is a boundary, not a participant list. People who meet the criteria may never hear about a study, may be unable to attend visits, or may choose not to participate. Consequently, a trial open to a broad age range can still enrol mostly people from one part of that range.
Compare the eligibility criteria with the baseline characteristics table and participant-flow information in the published report. Look for the numbers screened, enrolled, assigned to groups, and included in analyses, where reported. These stages help distinguish a broadly worded invitation from the population that ultimately contributed data. If subgroup details are missing, readers should retain that uncertainty rather than assume that the enrolled participants mirrored everyone who met the entry criteria.
The FDA’s December 2025 guidance on enhancing trial participation addresses eligibility, enrolment practices, and trial design to encourage more representative participation. It considers demographic characteristics and clinical characteristics such as comorbidities, organ dysfunction, and disability. This is US research guidance, not EU law or a finding about cannabinoid effects.
Random assignment has a different job
The word “randomized” can create a misleading impression of broad representation. As the NCATS definition of randomization explains, randomization assigns participants to study groups by chance. It concerns allocation within a recruited group. It does not mean researchers randomly sampled the wider population.
A trial can therefore use sound random assignment while enrolling a narrow population. Randomization helps support comparisons between the study groups; it cannot, by itself, establish that the results apply equally to people who were absent.
A hypothetical example
Imagine a fictional cannabinoid trial enrolling adults aged 40–65 with a specified condition. It excludes people taking a particular medicine, and most enrolled participants come from one specialist clinic. No outcome is assumed here.
Even if allocation is randomized, the study would leave questions about younger adults, older adults, people taking the excluded medicine, and patients in other settings. Exclusion alone would demonstrate neither benefit nor harm for those groups. It would identify an evidence gap. Equally, meeting the eligibility criteria would not guarantee that every eligible subgroup was adequately represented.
A practical reading checklist
- What diagnosis, age range, treatment history, and other conditions defined eligibility?
- Which eligible people actually enrolled, and how were they recruited?
- Which groups were excluded or scarcely represented?
- Does the report provide enough detail to judge those differences?
- Could differences in population, setting, intervention, or outcomes matter for the question being asked?
The Cochrane Handbook’s discussion of interpreting evidence treats applicability as distinct from risk of bias. Differences require judgment; every difference does not automatically invalidate a finding. Comparing populations across studies also helps explain why systematic reviews assess certainty in cannabinoid research rather than simply count positive and negative studies.
The useful conclusion is a bounded one: describe the population the evidence covers, identify important gaps, and keep uncertainty visible.