The Problem with “Evidence-Based”

by Brenton L. Delp

Few phrases carry more authority in modern medicine, psychology, addiction treatment, education, and public policy than the phrase “evidence-based.” It sounds sober, scientific, objective, and responsible. It appears to separate disciplined knowledge from mere opinion, superstition, ideology, tradition, charisma, and personal anecdote. In principle, this is exactly what it should do. Human beings are easily deceived. Clinicians are easily seduced by theories that confirm their experience. Institutions often continue doing what they have always done simply because they have always done it. Treatment systems become loyal to their own language. Patients are harmed when authority is not checked by evidence. In this sense, the demand for evidence is not merely technical. It is ethical.

And yet the phrase “evidence-based” has itself become a problem. Not because evidence does not matter, but because the invocation of evidence can create the illusion that interpretation has been overcome. The phrase often functions as though the evidence simply speaks for itself. But evidence never speaks for itself. Evidence must be gathered, selected, organized, measured, interpreted, compared, translated, and applied. Between the data and the decision stands judgment. Between the study and the suffering person stands interpretation. Between the statistical average and the individual case stands the difficult question of meaning.

This is the central point: evidence-based practice does not mean practice without interpretation. It means interpretation disciplined by evidence. That distinction matters enormously. If evidence is ignored, treatment collapses into ideology, charisma, intuition, institutional habit, or wishful thinking. But if evidence is treated as self-interpreting, treatment collapses into another kind of ideology: the ideology of method without wisdom. In that case, “evidence-based” becomes a slogan rather than a discipline.

The original meaning of evidence-based medicine was far more modest and more intelligent than many of its later popular uses. David Sackett and his colleagues famously defined evidence-based medicine as the “conscientious, explicit, and judicious use of current best evidence” in making decisions about patient care, but they also insisted that this evidence had to be integrated with clinical expertise and the values of the patient.¹ That second part is often forgotten. Evidence-based medicine was never supposed to mean that research replaces judgment. It was supposed to mean that judgment must be answerable to research. It was a corrective against unexamined authority, not a machine for eliminating clinical wisdom.

The distinction between evidence-based research and evidence-based practice is therefore essential. Evidence-based research asks: What does the data show under particular conditions? Evidence-based practice asks: How should this evidence guide action with this person, this population, this institution, this history, and this form of suffering? These are not the same question. A study may show that a certain intervention improves outcomes on average. That does not automatically tell us why it worked, for whom it worked, under what circumstances it worked, whether it can be reproduced in ordinary clinical settings, or whether it should dominate an entire field of treatment.

Research produces findings. Practice requires judgment.

This is why the question “Who is qualified to interpret evidence?” cannot be answered simply by saying “scientists,” “clinicians,” “researchers,” “doctors,” “therapists,” or “experts.” Each of these words can conceal as much as it reveals. A person may have advanced statistical training and still misunderstand the lived reality of addiction, compulsion, trauma, despair, shame, religious language, social isolation, or institutional coercion. Another person may have decades of clinical experience and still misunderstand research design, selection bias, randomization, effect size, confounding variables, publication bias, replication, or the difference between correlation and causation. Interpretation requires more than one kind of competence.

The qualified interpreter of evidence must possess at least three forms of literacy. The first is methodological literacy: the ability to understand what kind of study one is reading, what it can prove, what it cannot prove, what its limitations are, and whether its conclusions actually follow from its data. The second is clinical or practical literacy: the ability to understand how findings enter the complexity of real human lives. The third is ethical and interpretive literacy: the ability to explain what is known, what is not known, what is uncertain, and what must not be overstated.

This last form of literacy is often the rarest. It is not enough to know the evidence. One must know how not to abuse the evidence.

Modern research operates through abstraction. It must do so. To study anything scientifically, one must isolate variables, define outcomes, standardize methods, and reduce the overwhelming complexity of life into measurable forms. This is not a defect in science; it is one of the conditions that makes science possible. But every act of measurement is also an act of limitation. What can be measured is not always what matters most. What matters most is not always easy to measure. And what can be demonstrated under controlled conditions may not survive unchanged when it enters the disorder of ordinary life.

This is especially true in the human sciences. A blood pressure medication, a surgical technique, and a psychotherapy model do not function in exactly the same way. The more an intervention depends on meaning, relationship, motivation, culture, narrative, identity, language, and expectation, the more difficult it becomes to isolate the active ingredient. Addiction treatment is not simply a chemical or behavioral problem. It is also a problem of despair, repetition, attachment, shame, symbolic failure, social collapse, and the search for relief from consciousness itself. To evaluate treatment for addiction as though it were only a technical procedure is already to misunderstand the phenomenon being studied.

This does not mean research is useless. It means research requires humility.

One of the great dangers in discussions of evidence-based treatment is the confusion between association and causation. A study may show that people who participate in a certain program have better outcomes than people who do not. But that does not necessarily mean the program caused the outcome. The people who remain in the program may already be more motivated. They may have better transportation, stronger family support, fewer psychiatric complications, greater social stability, more religious compatibility with the program, or less severe impairment. Those who leave may not have “failed”; they may have been precisely the people for whom the model was least suitable. Unless these possibilities are carefully examined, evidence can easily become a flattering mirror for the institution that already holds power.

This matters deeply in addiction treatment, especially when discussing Alcoholics Anonymous and twelve-step approaches. AA has helped many people. That should be acknowledged plainly and without resentment. For some, it provides fellowship, structure, confession, repetition, ritual, accountability, humility, sponsorship, and a language through which suffering can be shared rather than hidden. It offers something many clinical systems fail to offer: a place to go when the appointment ends. In an age of isolation, that is not a small thing.

But acknowledging that AA helps many people is not the same as proving that AA, as a doctrine, is the universal or necessary form of recovery. A person may benefit from AA because of fellowship rather than theology, routine rather than surrender language, social belonging rather than step work, narrative coherence rather than metaphysical belief. The mechanism matters. If the mechanism is fellowship, then the field should develop many forms of fellowship. If the mechanism is accountability, then accountability can be built in different ways. If the mechanism is hope, then hope need not always be mediated through the same religious or quasi-religious vocabulary. If the mechanism is community, then the dominance of one model may actually suppress the emergence of other communities.

This is where “evidence-based” can become politically and institutionally ambiguous. Once a model becomes dominant, the evidence surrounding it may partly reflect its dominance. If most treatment centers recommend or require twelve-step participation, if courts and institutions refer people to twelve-step meetings, if insurance structures and treatment cultures assume twelve-step language, then participation is not occurring in a neutral environment. The model is not merely being tested; it is already embedded in the system. That does not invalidate all positive findings, but it does require careful interpretation.

The same problem appears in many fields. Evidence is never gathered in a vacuum. It emerges within funding structures, institutional priorities, professional incentives, publication pressures, cultural assumptions, and available categories of measurement. John Ioannidis’s famous warning that many published research findings are false was not an attack on science, but a warning about how easily research can be distorted by small samples, weak effects, flexible methods, selective reporting, and the pressure to publish positive results.² The lesson is not cynicism. The lesson is vigilance.

Cochrane’s original vision of evidence-based medicine was animated by a similar concern. Medicine needed better evidence because professional confidence was not enough.³ But the development of evidence hierarchies, systematic reviews, and randomized controlled trials created a new temptation: to treat certain forms of evidence as automatically superior in every context. Randomized controlled trials are powerful tools, especially when the question is narrow, the intervention is discrete, and the outcome is clearly measurable. But not every human question can be adequately answered by the same method. As Nancy Cartwright and Jeremy Hardie argue in the context of evidence-based policy, the fact that something worked “there” does not automatically mean it will work “here.”⁴ One must ask what support factors were present, what mechanisms were operating, and whether those same conditions exist in the new setting.

In clinical language, this is the problem of external validity. A treatment may work in a study and fail in ordinary practice. Or it may work in ordinary practice for reasons the study did not adequately understand. Or it may work statistically while leaving untouched the deeper form of suffering that brought the person to treatment in the first place. A person may stop drinking and remain spiritually dead. Another may relapse and yet begin, for the first time, to tell the truth. Which outcome matters? The answer cannot be determined by statistics alone, because the meaning of recovery is not reducible to a single measurable endpoint.

This is not an argument against measurement. It is an argument against the idolatry of measurement.

There is a profound difference between saying “we need evidence” and saying “only what can be measured is real.” The first statement is responsible. The second is metaphysics disguised as science. Human beings are not only organisms with symptoms. They are meaning-making creatures who suffer through memory, fantasy, loss, desire, guilt, shame, longing, despair, and failed transcendence. A treatment system that measures abstinence but ignores meaning may produce compliance without transformation. A treatment system that measures symptom reduction but ignores the person’s relation to existence may mistake management for healing.

The best traditions of evidence-based practice understand this. The American Psychological Association’s formulation of evidence-based practice in psychology, for example, explicitly integrates best available research with clinical expertise and patient characteristics, culture, and preferences.⁵ This is crucial. Evidence does not enter an empty space. It enters a life. That life has a history, a language, a family, a culture, a body, a set of wounds, a set of defenses, and a certain relation to hope. To ignore those factors in the name of evidence is not scientific rigor. It is abstraction pretending to be care.

The interpreter of evidence must therefore be able to say several things at once. This study is strong, but narrow. This result is promising, but not definitive. This intervention works on average, but not for everyone. This model helps some people, but its dominance may exclude others. This correlation is real, but causation has not been established. This outcome is measurable, but not exhaustive. This method is useful, but not absolute. This evidence should guide us, but it should not relieve us of responsibility.

That is the kind of speech the public rarely hears. Instead, it hears slogans. “Science says.” “Studies prove.” “Evidence-based treatment.” “Gold standard.” “Best practice.” “Clinically proven.” These phrases are not always false, but they are often incomplete. They can function as rhetorical shields. They allow institutions to present themselves as objective while hiding the interpretive decisions behind their conclusions. Once the phrase “evidence-based” is attached to a practice, criticism can be treated as ignorance, resistance, or anti-science. But the most scientific question is often the critical one: What exactly is the evidence? What exactly does it prove? What does it not prove? Who interprets it? Who benefits from this interpretation? Who is excluded by it?

The public does not need to become expert in every research method. That is impossible. But the public does need honest translation. Translation is not simplification in the vulgar sense. It is not dumbing down. It is the ethical act of making complexity intelligible without making it false. A good interpreter does not merely announce conclusions. A good interpreter teaches the layperson how to hold uncertainty. The best explanation may sound like this: “Here is what we know with reasonable confidence. Here is what remains uncertain. Here is where the evidence is strong. Here is where it is weak. Here is where the evidence supports the intervention, but not the ideology that has grown around it.”

This distinction between intervention and ideology is especially important. A method may work while the theory explaining it is partly wrong. A ritual may help while the doctrine attached to it remains questionable. A community may heal while its official language alienates many who need community. AA may help many people while still being overextended by treatment systems that lack imagination. Cognitive-behavioral therapy may be useful while still failing to reach the symbolic depths of a person’s suffering. Medication-assisted treatment may save lives while still requiring a broader account of meaning, attachment, and social repair. No method should be exempt from interpretation merely because it has evidence. Evidence grants responsibility; it does not grant immunity.

The deeper issue is that modern society often wants certainty from evidence. It wants research to perform the role once played by religious or metaphysical authority. It wants the study to settle the question, the guideline to replace judgment, the protocol to remove anxiety, the expert to deliver certainty. But the honest use of evidence does not abolish uncertainty. It disciplines uncertainty. It teaches us how to act responsibly without pretending to possess absolute knowledge.

That is why evidence-based practice should produce humility rather than arrogance. The more one understands research, the less one should be tempted by slogans. Good evidence narrows the field of fantasy. It protects us from harmful nonsense. It challenges beloved assumptions. It forces clinicians, institutions, and traditions to answer to something beyond themselves. But evidence also has limits. It does not interpret itself. It does not decide values. It does not define the whole meaning of recovery. It does not tell us, by itself, what kind of human being we are trying to help someone become.

In addiction treatment, the ethical question is not simply “What works?” The question is: What works, for whom, under what conditions, at what cost, according to which measure, and toward what vision of recovery? If recovery means mere symptom reduction, one kind of evidence will suffice. If recovery means restored agency, another kind may be needed. If recovery means social belonging, the evidence must include community. If recovery means truthfulness before one’s own life, the evidence must leave room for forms of transformation that are difficult to quantify. If recovery means endurance without illusion, then no single model, religious or secular, clinical or communal, can claim final authority.

The phrase “evidence-based” should therefore be reclaimed, not rejected. To reject evidence would be irresponsible. Too many people have been harmed by charismatic certainty, moralism, superstition, punitive treatment, and institutional laziness. But to reclaim evidence means rescuing it from slogan and restoring it to judgment. Evidence-based practice should mean disciplined inquiry, methodological honesty, clinical wisdom, cultural humility, and ethical translation. It should mean that no model gets to hide behind tradition, popularity, institutional dominance, or spiritual authority. It should also mean that no model gets to hide behind statistics it does not understand.

The real enemy is not evidence. The real enemy is unearned certainty.

A mature evidence-based culture would teach clinicians how to read research carefully. It would teach researchers how to respect clinical complexity. It would teach institutions how to distinguish between what is proven, what is plausible, what is merely customary, and what is ideologically convenient. It would teach the public that evidence is not magic. It would make room for lived experience without allowing anecdote to overrule disciplined knowledge. It would make room for disciplined knowledge without allowing abstraction to silence lived experience.

Such a culture would be especially valuable in addiction treatment, where suffering is both bodily and symbolic, chemical and existential, social and spiritual. The addicted person does not need ideology disguised as science, nor science reduced to bureaucracy. He needs truth, care, structure, interpretation, community, and freedom from false promises. Evidence can help provide these things, but only if it is interpreted wisely.

The final formulation, then, might be this:

Evidence-based practice is not the worship of studies. It is the disciplined interpretation of evidence by people competent to understand both its power and its limits, followed by the ethical translation of that interpretation into forms of care that ordinary people can understand and use without being misled.

That is the standard. Anything less risks turning “evidence-based” into another language of authority, another institutional badge, another way of silencing doubt. But rightly understood, evidence does not silence thought. It demands better thought. It does not remove judgment. It educates judgment. It does not end interpretation. It makes interpretation accountable.

Notes

1. David L. Sackett, William M. C. Rosenberg, J. A. Muir Gray, R. Brian Haynes, and W. Scott Richardson, “Evidence Based Medicine: What It Is and What It Isn’t,” British Medical Journal 312, no. 7023 (1996): 71–72.
2. John P. A. Ioannidis, “Why Most Published Research Findings Are False,” PLoS Medicine 2, no. 8 (2005): e124.
3. Archie Cochrane, Effectiveness and Efficiency: Random Reflections on Health Services (London: Nuffield Provincial Hospitals Trust, 1972).
4. Nancy Cartwright and Jeremy Hardie, Evidence-Based Policy: A Practical Guide to Doing It Better (Oxford: Oxford University Press, 2012).
5. American Psychological Association Presidential Task Force on Evidence-Based Practice, “Evidence-Based Practice in Psychology,” American Psychologist 61, no. 4 (2006): 271–285.
6. Trisha Greenhalgh, Jeremy Howick, and Neal Maskrey, “Evidence Based Medicine: A Movement in Crisis?” British Medical Journal 348 (2014): g3725.
7. Peter M. Rothwell, “External Validity of Randomised Controlled Trials: ‘To Whom Do the Results of This Trial Apply?’” The Lancet 365, no. 9453 (2005): 82–93.
8. John F. Kelly, Keith Humphreys, and Marica Ferri, “Alcoholics Anonymous and Other 12-Step Programs for Alcohol Use Disorder,” Cochrane Database of Systematic Reviews 3 (2020): CD012880.
9. Hans-Georg Gadamer, Truth and Method, trans. Joel Weinsheimer and Donald G. Marshall, 2nd rev. ed. (New York: Continuum, 1989). Gadamer is useful here not as a clinical authority, but as a philosophical reminder that understanding is always interpretive.
10. Aristotle, Nicomachean Ethics, especially Book VI on phronesis, or practical wisdom. The point is not that clinical practice is ancient ethics, but that action in concrete situations requires a form of judgment that cannot be reduced to technical rule-following.