Inclusion and exclusion criteria
Based on Wikipedia: Inclusion and exclusion criteria
In 1935, the federal government drew red lines around Black neighborhoods on city maps and declared them unfit for investment. The practice was called redlining, and its effects persist ninety years later.
A similar, though less visible, act of boundary-drawing happens inside sterile clinical trial rooms every single day, yet it determines who lives, who gets sick, and who simply disappears from the medical record. When a researcher designs a study to test a new drug for heart disease or hearing loss, they do not simply open the doors to everyone in need. Instead, they construct a fortress of rules known as inclusion and exclusion criteria. These are not arbitrary gates; they are the architectural blueprint of medical truth. They define exactly who gets to be counted when science asks, "Does this work?" and just as importantly, who is left out so that the answer remains clean.
To understand why a pregnant woman might be barred from a life-saving trial, or why a man with a specific type of diabetes could be denied entry to a cardiovascular study, one must look beyond the dry jargon of protocol manuals. This process is the invisible filter through which all modern medicine flows, and it carries a heavy, often unacknowledged weight.
The fundamental tension in any clinical trial lies between two competing goals: scientific purity and real-world application. Inclusion criteria are the positive list, the specific characteristics that prospective subjects must possess to enter the study. According to the International Council for Harmonisation (ICH) E3 guideline on reporting clinical studies, these criteria concern the properties of the target population. They define the very group to which the study's results should be generalizable. If a researcher is studying a new therapy for late-stage lung cancer, their inclusion criteria might mandate that participants have a specific tumor stage, a particular history of previous treatments, or fall within a narrow age range.
These requirements sound logical on paper. They are meant to ensure that the people in the trial actually have the condition being studied and that their biology is similar enough to allow for clear data collection. But here lies the first trap: when these criteria become too strict, they create a "perfect" study population that barely exists outside the laboratory.
Exclusion criteria, by contrast, are the negative list. They define the reasons why patients from the target population must be turned away. These rules concern properties of the study sample itself. While inclusion asks "Who belongs?", exclusion asks "Who threatens the data?" or "Who is at risk?"
The reasons for exclusion fall into three distinct categories, each with its own moral and scientific calculus.
First, there are ethical exclusions. These are the barriers erected to protect vulnerable populations from harm. Children, pregnant women, patients with severe psychological illnesses, or those unable to sign an informed consent form are routinely excluded. The rationale is clear: these groups require special protection because they cannot fully grasp the risks of a new intervention or because the intervention might cause irreversible damage to a developing fetus. The state acts as a guardian here, but in doing so, it often leaves these populations without data-driven treatment options until decades later.
Second, there are practical exclusions. These arise from the logistical nightmare of running a complex study. If a trial relies on a written questionnaire to assess outcomes, participants who cannot read are excluded not because their biology is different, but because the tool used to measure them fails them. This creates a blind spot in medical knowledge where entire demographics are rendered invisible simply because they cannot navigate the bureaucracy of the assessment.
Third, and perhaps most insidiously, there are scientific exclusions designed to eliminate confounding variables. If a patient has a comorbidity—a secondary disease like diabetes or heart failure—researchers often exclude them to ensure that any observed effect is due solely to the drug being tested, not the interaction of multiple conditions. This makes the data clean. It makes the p-value significant. But it also means the results may be useless for the elderly patient who walks into a clinic with three chronic illnesses and a new prescription.
The danger is that these exclusions can introduce bias so profound that the study's conclusions become lies of omission. A systematic review of hearing loss in adults recently highlighted this stark reality. The review found that while studies were representative of the US population in terms of sex, they failed catastrophically to represent racial or ethnic diversity. The exclusion criteria, whether explicit or implicit, created a clinical picture of hearing loss that looked nothing like the actual human experience in America.
When we talk about poorly justified reasons for exclusion, we are talking about criteria that have no direct bearing on the condition, intervention, or results. Excluding someone simply because they live too far from the clinic is a convenience bias, not a scientific one. It limits the study to those with transportation and time, skewing the socioeconomic makeup of the trial.
However, there are strongly justified reasons for exclusion that no reasonable person would argue against. If an intervention would be harmful to a specific group, they must be excluded. This includes situations where there is a lack of equipoise—the scientific uncertainty about which treatment is better—because if the medical community already believes one arm of the trial is dangerous, it is unethical to recruit patients into it. Similarly, if a patient cannot provide informed consent, they are rightly protected from being used as data points.
Then there are the "potentially justified" reasons, where the line blurs. Researchers may exclude individuals who they predict will not adhere to the protocol or who are unlikely to complete follow-up visits. While this sounds pragmatic, it often results in the exclusion of people with unstable lives—the very people who might benefit most from a new treatment but lack the stability required by a rigid study design. It creates a feedback loop where only the privileged, stable, and compliant get tested, leaving the rest as statistical ghosts.
Consider the specific case of coronary heart disease research. A rigorous trial in this field would set minimum outcomes at coronary deaths and non-fatal myocardial infarctions. To ensure the data is sound, researchers require appropriate measures of Framingham variables: age, sex, LDL cholesterol, HDL cholesterol, total cholesterol, diabetes status, smoking history, and hypertension levels. They might look for a cohort study, a nested case-control design, or a cardiovascular trial follow-up study that estimates the predictive value of a novel risk factor after adjusting for these established variables.
The exclusion list for such a study is equally specific. Patients are excluded if they have no data available, or if they belong to a sub-population with known coronary disease where the intervention would be too late to test prevention. They are excluded if the study design does not measure Framingham variables appropriately or if the article format is wrong. While these rules seem technical, they act as a sieve that filters out complexity.
The human cost of this filtering is often invisible in the final paper, which might conclude with a triumphant headline: "New Drug Reduces Heart Attack Risk by 40%." What remains unsaid is that this number applies only to white men, aged 50 to 65, without diabetes, who have reliable transportation and can read complex forms. The woman of color with the same heart condition but a history of diabetes might find the drug ineffective in her body because her specific physiology was never included in the test. She is not an outlier; she is the reality that the study was designed to ignore.
This issue extends beyond race and gender into the very structure of how we conduct science. The U.S. Food and Drug Administration (FDA) and the European Medicines Agency have issued guidelines, such as those found on the ICH website, urging for more representative trials. Yet, the inertia is immense. Clinical trial management companies and researchers often default to exclusion because it makes their jobs easier. It reduces the noise in the data. But science that does not reflect the population it serves is not just incomplete; it is dangerous.
The history of medical research is littered with examples where exclusion led to tragedy. For decades, women were excluded from cardiovascular trials because researchers believed their hormonal cycles made their data too variable. The result was that heart disease symptoms in women went undiagnosed and undertreated for generations. When a woman walked into an emergency room with chest pain, she was often dismissed as anxious or having indigestion, because the medical textbooks were written based on trials of men.
Similarly, pregnant women have been systematically excluded from almost all drug trials due to ethical concerns about fetal safety. The consequence is that when a pregnant woman gets sick, doctors are forced to prescribe medications with no data on their safety in pregnancy. They must guess. This "therapeutic orphan" status leaves mothers and fetuses vulnerable because the scientific community decided it was too risky to study them.
The solution is not to abandon exclusion criteria entirely; that would lead to chaos where data becomes uninterpretable. The solution lies in critical scrutiny of every single criterion. Researchers must ask, as a matter of routine: "Does this exclusion protect the patient, or does it just make our data cleaner?" If a criterion excludes people because they are poor, or disabled, or belong to a minority group, and that exclusion has no direct biological bearing on the drug's mechanism, then it is a failure of design.
We must also recognize that the definition of "diversity" in clinical trials often stops at the superficial. A trial might include women, but if they are all from one socioeconomic class and one geographic region, the results still lack generalizability. The systematic review of hearing loss mentioned earlier serves as a grim reminder: sex is not enough. Racial and ethnic diversity must be central to the inclusion criteria, not an afterthought.
The path forward requires a shift in mindset for everyone involved, from the FDA regulators to the clinical trial management companies listed on career sites. It demands that we value real-world applicability over laboratory perfection. If a drug works only in a highly controlled, homogenous group, it is not a cure for the population; it is a cure for a subset.
When you read about a breakthrough treatment in the news, look closely at the fine print. Who was allowed to be there? Who was kept out? The answers to those questions determine whether that headline applies to you, your mother, or your neighbor. Inclusion and exclusion criteria are not just administrative hurdles; they are the moral boundaries of medicine. They decide who is worthy of being counted in the pursuit of health.
The bias introduced by poorly justified exclusions is not a minor statistical error. It is a systemic erasure. When we exclude individuals because their lives are too complex, their backgrounds too varied, or their consent processes too difficult to manage, we are implicitly stating that some lives matter less in the equation of science. This is a profound ethical failure.
We need strong justification for every barrier. If an individual cannot adhere to the protocol, can we support them better rather than excluding them? If data is missing, can we invest in better collection methods rather than discarding the patient? If a population lacks reliable information, is it because they are inaccessible, or because the system has failed to reach them?
The goal of clinical research is to improve human health for everyone. But you cannot improve what you do not measure. And you cannot measure those you have refused to see. The rigid walls of exclusion criteria must be breached, not by lowering scientific standards, but by raising our commitment to truth. A study that does not reflect the diversity of the people it aims to help is a study that has already failed its primary mission.
As we move into an era where precision medicine promises tailored treatments for individuals, the irony deepens: we are failing to include the very individuals who need that precision most. The exclusion criteria of the past have created gaps in our knowledge that threaten to widen into chasms of inequality. It is time to rewrite the rules, not with a pencil, but with a conscience.
The next time you see a clinical trial recruitment ad, or read the results of a major study, remember the invisible lines drawn on the page. Remember the pregnant woman waiting for data that doesn't exist yet. Remember the elderly patient with three chronic conditions who was told "no" because their biology was too messy for the math. They are not just footnotes in a protocol; they are the people whose lives hang in the balance of our scientific choices.
Science is only as good as the population it serves. If we exclude the world to study a sample, we will never understand the whole. The challenge of inclusion and exclusion criteria is not merely statistical; it is deeply human. It asks us to decide who belongs in the story of medicine, and who gets left out of the history books.
We must choose wisely. Because in the end, the only data that truly matters is the kind that saves lives—every single life.