The silent exclusion: rethinking eligibility and generalizability in clinical trials

Randomized controlled trials (RCTs) were introduced in the late 1940s, with one of the first landmark studies conducted by the British Medical Research Council in 1948, evaluating streptomycin for the treatment of pulmonary tuberculosis,¹ and are widely regarded as the gold standard in evidence-based medicine due to their ability to isolate the effect of an intervention by controlling for confounding factors through randomization, blinding, and standardized procedures. This methodological rigor ensures high internal validity, which has played a central role in advancing treatment strategies and improved patient outcomes. Yet, this often comes at a cost: limited generalizability to broader patient populations. One major but often overlooked limitation of contemporary RCTs lies in the systematic exclusion of a large proportion of the patient population. This exclusion commonly results from restrictive eligibility criteria aimed at reducing variability and insuring patient safety.² However, this exclusion can also occur in a more subtle way during the informal prescreening processes that go undocumented and unreported in the scientific literature. Trial designs frequently favor homogeneity to optimize internal validity, relying on narrowly defined inclusion and exclusion criteria. As a result, individuals with complex health profiles, such as older adults, patients with multiple comorbidities or on multiple medications, those with cognitive impairments, or a history of adverse drug reactions, are routinely excluded.³ While these exclusions may be justified on methodological grounds, they often result in study populations that fail to reflect those who would benefit most from the intervention in real-world practice.⁴

Before patients formally enter the recruitment process, many are informally assessed and excluded by investigators or clinical staff based on anticipated barriers to participation, such as likely non-adherence, cognitive challenge, language barriers, unstable living conditions, or perceived lack of motivation. These early decisions are typically undocumented, meaning that a large portion of the recruitment process remains invisible. This hidden layer introduces a bias that further narrows trial populations and remains underappreciated in academic discourse.⁵

In the BOREAS trial,⁶ for example, no patients from our COPD outpatient clinic qualified for inclusion despite screening a substantial cohort, highlighting how restrictive criteria can disconnect trials from routine clinical populations.

The implications are far-reaching. First, such exclusions undermine the external validity of trials, the degree to which their findings apply to the broader patient population. For example, an intervention that proves effective in relatively healthy, middle-aged trial participants may not translate well to frail, elderly patients with complex needs. Likewise, therapies tested in specialized academic centers may not perform as well in routine clinical settings with fewer resources.⁷

Second, these exclusion practices contribute to a cycle of inequity in clinical evidence. Populations that are often underrepresented in trials, such as the elderly, ethnic minorities, patients from lower socioeconomic backgrounds, or those with complex health conditions, are also the ones most likely to experience worse health outcomes. Excluding them from research not only limits the relevance of clinical guidelines but also perpetuates disparities in care. This results in a body of knowledge that is both scientifically narrow and ethically questionable.⁸

Addressing these shortcomings would require a paradigm shift in how trials are designed and reported. Sponsors and investigators should design studies with broader eligibility criteria that reflect the complexity of real-world patient populations, rather than avoiding it. Where exclusions are necessary for safety reasons, these should be transparently reported and justified. Prescreening procedures, which often operate as a “black box” in trial recruitment, should be documented and standardized across sites, providing clarity on who is excluded and why.⁹

One of the key challenges in clinical trials is ensuring the recruitment of a broad and representative sample of participants. The informed consent process, while essential for ethical reason, can itself become a barrier, particularly for individuals with limited capacity, language difficulties, or social vulnerabilities. In certain context, and with proper ethical oversight, waivers or modifications of informed consent procedures may be necessary to ensure inclusive enrollment. These approaches must be carefully considered to balance ethical principles with the goal of equitable representation.¹⁰

Moreover, regulatory agencies and academic journals also have a role to play in enhancing transparency by requiring detailed reporting of all screening phases, including prescreening. Tools such as CONSORT flow diagrams should be extended to include data on initial patient pools and reasons for exclusion at every stage.¹¹

Although RCTs remain the cornerstone for evaluating treatment efficacy, the design of many traditional, explanatory RCTs limits their applicability to real-world settings. Stringent inclusion criteria, tightly controlled conditions, and modest sample sizes often constrain their ability to capture outcomes that matter most to patients and healthcare systems, such as hospital admissions, long-term morbidity, and mortality. Complementing traditional RCTs with pragmatic trials and real-world evidence frameworks can help bridge the gap between efficacy and effectiveness.¹² However, a potential limitation of pragmatic, broadly inclusive trials is their increased susceptibility to challenges such as higher dropout rates, lower adherence, and treatment cross-over. These issues can attenuate observed effect sizes and increase the risk of failing to detect a true treatment benefit.

Ultimately, clinical research should aim to generate knowledge that is applicable to the complexities of real-life healthcare, not just to idealized trial settings. To achieve this, we must confront the structural and methodological blind spot that systematically exclude large portions of the population. Only by embracing greater inclusivity and transparency can we create evidence that is scientifically robust, ethically sound, and genuinely useful in improving patient care.

Contributors

PS conceptualised and wrote the initial draft. JUS, NDJ, AM, and TBS contributed supervision, writing - original draft (revision), writing - review & editing.

Concept of Paper – PS.

Critical Review & Writing of Paper - all authors contributed equally.

Declaration of interests

TBS has received research grants from Bayer, Novartis, Pfizer, Sanofi Pasteur, GSK, Novo Nordisk, AstraZeneca, Boston Scientific and GE Healthcare, consulting fees from Novo Nordisk, IQVIA, Parexel, Amgen, CSL Seqirus, GSK and Sanofi Pasteur, and lecture fees from AstraZeneca, Bayer, Novartis, Sanofi Pasteur, GE healthcare and GSK.

Alexander is the (unpaid) Chair of the European Respiratory Society’s Airway Pharmacology and Treatment Group. He has received non-financial support from Verona Pharma and holds stocks in Healthy Networks. He has also received honoraria from GlaxoSmithKline and consulting fees from Sanofi.

The other authors declare no competing interests.