{"entity":{"id":"trial-population-transportability","kind":"term","name":"Transporting trial evidence to another population","aka":["clinical trial transportability","trial generalizability","eligibility and target population"],"tldr":"A trial result describes the people and setting studied. Applying it elsewhere requires checking the differences that could change benefit or harm.","summary":"A primary methodological framework used trial-participation propensity scores to assess similarity between trial participants and a target population. Its worked example was outside oncology; it provides a way to examine generalizability, not a validated cancer-treatment matching rule. Weighting measured characteristics depends on assumptions and sufficient overlap. Disease, treatment setting, prior therapy and other possible effect modifiers still matter, and an average trial effect is not an individual forecast.","asOf":"2026-11-10","links":[{"label":"Primary framework for trial generalizability (2011)","url":"https://doi.org/10.1111/j.1467-985X.2010.00673.x"},{"label":"Study Designs for Extending Causal Inferences From a Randomized Trial to a Target Population.","url":"https://pmc.ncbi.nlm.nih.gov/articles/PMC8536837/"},{"label":"Dahabreh and colleagues: transportability methods, May 2019 v1","url":"https://arxiv.org/pdf/1905.07764"},{"label":"NCI: finding a trial and checking eligibility","url":"https://www.cancer.gov/research/participate/clinical-trials-search/steps"}],"tags":[],"related":["external-validation","subgroup-analysis","confounding-by-indication"],"cancers":[],"sections":[],"technologies":[],"targets":[],"drugs":[],"companies":[],"institutions":[],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":[],"keyPapers":[],"journals":[],"dependsOn":[],"notes":["Eligibility asks whether someone meets protocol entry rules. Transportability asks whether a treatment-effect estimate can extend to a specified target population or setting. Randomisation assigns treatment within an enrolled sample; it does not randomly select that sample from everyone who might receive treatment. Eligible, invited, enrolled and followed people can differ. First specify the target population and align disease, treatment history, intervention and background care, comparator, endpoint and clock. A change of sequence or comparator may change the question itself. A prognostic factor predicts outcomes, while an effect modifier changes a treatment contrast on a specified scale. Even a constant relative effect can yield different absolute benefits when baseline risks differ. A marker or cancer label shared between heavily pretreated metastatic and early postoperative settings does not bridge differences in biology, competing risks, tolerance and endpoints.","Standardisation can combine subgroup effects in the proportions found in a target population; weighting can make trial observations represent that mixture. Both require appropriate trial and target data, comparable well-defined treatments, measurement of relevant differences and informative overlap. Weighting cannot invent a response for an absent target profile or remove an unmeasured effect modifier. In a fictional randomised trial measuring recurrence by three years from assignment, subgroup risk differences are -5 and -15 percentage points. A 50/50 mixture averages -10 points. An 80/20 target mixture would give 0.8 × (-5) + 0.2 × (-15) = -7 points only if those effects transport and relevant differences have been addressed. A missing third group leaves its contribution unresolved. The arithmetic illustrates an assumption; it establishes neither transportability, eligibility nor an individual's benefit.","Protocol eligibility can depend on disease, prior treatment, organ function, safety history and assay definitions, and needs study-team confirmation. Meeting criteria guarantees neither benefit, enrolment nor an available slot. If a fictional trial allows older adults but enrols almost nobody with a particular organ limitation, matching its age rule does not establish its average effect for that group. Actual representation, safety data, treatment context, overlap and the assumptions bridging populations remain necessary, with follow-up, harms and uncertainty considered. No universal similarity score certifies that bridge. An adjusted estimate is not automatically established, and even a well-supported population effect is not a person's guaranteed outcome. The examples here are fictional and provide no actual eligibility or access determination."],"category":"Trials"},"route":"/terms/trial-population-transportability/","neighbours":{"term":[{"id":"confounding-by-indication","kind":"term","name":"Confounding by indication","route":"/terms/confounding-by-indication/"},{"id":"measurement-to-benefit-evidence","kind":"term","name":"Evidence from measurement to patient benefit","route":"/terms/measurement-to-benefit-evidence/"},{"id":"external-validation","kind":"term","name":"External validation","route":"/terms/external-validation/"},{"id":"subgroup-analysis","kind":"term","name":"Subgroup analysis (forest plots)","route":"/terms/subgroup-analysis/"}]}}