Nonmaleficence

Summary

Nonmaleficence is the principle that physicians should not cause harm to patients. It is often confused with the Latin maxim primum non nocere — “above all, do no harm” — but that phrase does not actually appear in the Hippocratic writings. As a formal principle in bioethics, nonmaleficence requires more than good intentions: it demands that clinicians exercise due care, avoid negligence, and justify any action that sets back a patient’s interests. In practice, the principle raises difficult questions about when to stop treatment, whether withdrawing care differs morally from never starting it, and how to make decisions for patients who cannot decide for themselves. These are not abstract puzzles — they shape what happens in intensive care units, neonatal wards, and hospice programs every day.

The Principle

What Nonmaleficence Means

The principle of nonmaleficence obligates physicians to abstain from causing harm to others. It sounds simple, but the simplicity is deceptive. The celebrated maxim primum non nocere — “above all, do no harm” — is often proclaimed as the foundational commitment of the Hippocratic tradition. Yet the phrase does not appear in the Hippocratic writings. A passage sometimes translated this way is a strained rendering of a single Hippocratic line. What the Hippocratic Oath actually says incorporates both nonmaleficence and beneficence together: “I will use treatment to help the sick according to my ability and judgment, but I will never use it to injure or wrong them.” (Tom L. Beauchamp, James F. Childress, 2013)

The myth matters because it distorts the principle’s weight. If primum non nocere were truly the first principle of medicine, then avoiding harm would always take priority over providing benefit — a physician would be obligated to do nothing rather than risk harm. Beauchamp and Childress argued that this gets the relationship between nonmaleficence and beneficence wrong. The two principles must be kept distinct, but combining them into a single principle — or ranking one automatically above the other — obscures critical moral distinctions. Obligations not to harm others are sometimes more stringent than obligations to help them, but the reverse is also true. A physician who causes a minor injury (swelling from a needlestick, say) while saving a patient’s life has rightly given beneficence priority over nonmaleficence in that instance. No rule in ethics gives nonmaleficence automatic precedence over beneficence in every circumstance. (Tom L. Beauchamp, James F. Childress, 2013)

The Concept of Harm

What counts as harm? Beauchamp and Childress defined harm as a thwarting, defeating, or setting back of some party’s interests. This definition is deliberately broad, but it comes with an important qualification: a harmful action is not always wrong. Justified amputation of a consenting patient’s leg, justified punishment of a physician for incompetence, justified demotion of an employee for poor performance — all of these set back someone’s interests, yet none are wrongful. The principle of nonmaleficence does not prohibit all harm; it functions as a prima facie obligation, meaning it requires justification for every harmful action. (Tom L. Beauchamp, James F. Childress, 2013)

This distinction between harm and wrong is what makes the principle workable in clinical medicine, where nearly every intervention carries some risk of harm. Surgery causes tissue damage; chemotherapy destroys healthy cells alongside cancerous ones; even diagnostic procedures can injure. What nonmaleficence demands is not the avoidance of all harm but the justification of every harm — a demonstration that the setback to the patient’s interests is warranted by the expected benefit.

Harm and Due Care

The Standard of Due Care

Nonmaleficence in practice is enforced through the standard of due care. Due care means taking appropriate care to avoid causing harm, as the circumstances demand of a reasonable and prudent person. The standard requires that the goals pursued justify the risks imposed to achieve them. (Tom L. Beauchamp, James F. Childress, 2013)

The concept is contextual. What counts as due care for a surgeon performing a delicate operation differs from what counts as due care for a general practitioner prescribing a common medication. The standard is set by what a competent professional in the same field, with the same training, would do under similar circumstances.

Negligence

Negligence is the failure to meet the standard of due care. Beauchamp and Childress identified two forms. The first is advertent negligence, or recklessness — intentionally imposing unreasonable risks of harm. A physician who performs a procedure knowing they lack the training to do it safely is reckless. The second is inadvertent negligence, or carelessness — unintentionally but carelessly imposing risks of harm. A physician who fails to check a patient’s allergy history before prescribing a medication is careless. (Tom L. Beauchamp, James F. Childress, 2013)

Both forms are failures of nonmaleficence. The distinction between them matters legally (recklessness often carries harsher penalties) but the moral point is the same: the physician owed the patient a level of care and fell short.

Withholding and Withdrawing Treatment

The Moral Irrelevance of the Distinction

One of the most consequential arguments in the nonmaleficence literature concerns the distinction between withholding treatment (never starting it) and withdrawing treatment (stopping it after it has begun). Many clinicians and families experience these as morally different — it feels worse to turn off a ventilator than never to have turned it on. But Beauchamp and Childress argued that the distinction is morally irrelevant and potentially dangerous. Withdrawing can happen through omission (not recharging batteries that power a respirator, not refilling a feeding tube), and in multi-staged treatments, a decision not to begin the next stage can be tantamount to stopping treatment even if earlier phases continue. Both withholding and withdrawing can be instances of allowing to die; both can be instances of killing. The moral question is whether a duty to treat exists, not whether the treatment has already started. (Tom L. Beauchamp, James F. Childress, 2013)

Overtreatment and Undertreatment

The danger of privileging withholding over withdrawing runs in two directions. The more obvious risk is overtreatment: if stopping a treatment once started is felt to be morally worse than never starting it, clinicians may continue burdensome interventions long after they have ceased to benefit the patient, trapping patients and families in biomedical technology with no clear exit. The less obvious but equally real risk is undertreatment: patients and families who fear being trapped by technology they cannot later refuse may decline interventions that could genuinely help them. A patient who refuses a trial of mechanical ventilation because they believe it can never be stopped is being harmed by the very distinction meant to protect them. (Tom L. Beauchamp, James F. Childress, 2013)

Artificial Nutrition and Hydration

A related distinction — between medical treatments and artificial nutrition and hydration (AN&H) — carries similar risks. The emotional weight of “feeding” is different from the emotional weight of “treatment,” and some religious and legal traditions have treated the withdrawal of nutrition and hydration as categorically different from the withdrawal of other technologies. Beauchamp and Childress rejected this distinction. Artificial nutrition and hydration is a medical treatment, delivered through medical technology (nasogastric tubes, intravenous lines, percutaneous endoscopic gastrostomy), and subject to the same ethical analysis as any other life-sustaining intervention. No morally relevant difference exists between them. (Tom L. Beauchamp, James F. Childress, 2013)

The Rule of Double Effect

The Four Conditions

The rule of double effect (RDE) is a moral principle developed within Roman Catholic moral theology to address situations in which a single action produces both a good and a bad effect — paradigmatically, administering pain medication that relieves suffering but hastens death. Classical formulations of the RDE identify four conditions that must be satisfied for such an act to be justified: (1) the act itself must be good or at least morally neutral, independent of its consequences; (2) the agent must intend only the good effect, not the bad one; (3) the bad effect must not be a means to the good effect; and (4) the good effect must be proportionate to the bad one — the benefit must outweigh the harm. (Tom L. Beauchamp, James F. Childress, 2013)

The Intention-Foresight Problem

The second condition — that the agent intend only the good effect — is the load-bearing element of the RDE, and it is the one Beauchamp and Childress found most difficult to sustain. The RDE depends on a sharp distinction between what an agent intends and what an agent merely foresees. A physician who administers morphine to relieve pain, foreseeing but not intending that it will hasten death, acts permissibly under the RDE. A physician who administers the same dose with the intention of hastening death does not.

But Beauchamp and Childress proposed a model of intention based on what is willed rather than what is wanted. On this model, intentional actions and intentional effects include any effect specifically willed in accordance with a plan — including effects that are tolerated rather than desired. A physician who administers morphine knowing it will hasten death, and who incorporates that foreseeable consequence into a deliberate plan of action, has willed the patient’s earlier death even if they do not want it. Under this conception, the distinction between what agents intend and what they merely foresee in a planned action is not viable. (Tom L. Beauchamp, James F. Childress, 2013)

This critique does not mean that the RDE captures nothing real. The moral difference between a physician who genuinely aims to relieve suffering and a physician who aims to kill is significant. But the RDE’s attempt to locate that difference in the formal structure of intention, rather than in the totality of the agent’s reasoning and the proportionality of the outcome, is — on Beauchamp and Childress’s analysis — conceptually untenable. (Tom L. Beauchamp, James F. Childress, 2013)

Treatment Decisions for Incompetent Patients

Obligatory, Optional, and Wrong to Treat

Beauchamp and Childress proposed replacing the traditional categories of ordinary and extraordinary means — categories that had long governed end-of-life decision-making in Roman Catholic moral theology and in law — with three framework categories: treatment that is obligatory (wrong not to treat), treatment that is obligatory not to provide (wrong to treat), and treatment that is optional (neither required nor prohibited). The framework relies heavily on quality-of-life considerations and on the balance of probable benefits against probable harms and burdens. (Tom L. Beauchamp, James F. Childress, 2013)

This tripartite scheme has the advantage of making the moral reasoning explicit. Rather than asking whether a ventilator is “ordinary” or “extraordinary” — a question that conflates technological complexity with moral obligation — the framework asks whether the intervention, for this particular patient in these particular circumstances, produces a net benefit that justifies its burdens.

Futility

The concept of medical futility has been invoked to justify decisions not to treat, but Beauchamp and Childress argued that the term is overused and should be handled with care. In its proper sense, futility applies only to treatments that have no chance of being successful given acceptable medical goals. A genuinely futile intervention — one with no possibility of achieving any acceptable outcome — is morally optional and in many cases should not be initiated or continued. But the term is frequently stretched to cover interventions that have some chance of success but a low one, or interventions whose outcomes the physician judges to be poor quality of life. These are legitimate considerations, but calling them “futile” smuggles evaluative judgments into a term that sounds factual. Wherever possible, the term should be replaced with more precise language about benefit-burden ratios. (Tom L. Beauchamp, James F. Childress, 2013)

Quality of Life and Best Interests

When patients cannot make decisions for themselves, treatment decisions must rest on some standard. Beauchamp and Childress argued that quality-of-life judgments are central to — not a departure from — responsible decision-making for incompetent patients. But such judgments must be carefully constrained. Intellectual disability is irrelevant to whether treatment is in the patient’s best interest. The value of the patient’s life to others — to family members, to society, to the health care system — should not be confused with the quality of life for the patient. Criteria focused on the incompetent patient’s best interests should be decisive, even when those interests conflict with familial or societal interests in avoiding burdens or costs. (Tom L. Beauchamp, James F. Childress, 2013)

For never-competent patients — critically ill newborns, young children, individuals who have never possessed decisional capacity — the best-interest standard is the appropriate framework. The standard asks what reasonable persons would consider the highest net benefit, weighing probable benefits of different treatments against their probable harms and burdens to the patient. Parents and other surrogates for never-competent patients can legitimately use predictions about survival and quality of life, evaluated according to the patient’s interests, to determine whether treatments are obligatory, optional, or wrong. (Tom L. Beauchamp, James F. Childress, 2013)

History of Nonmaleficence

The history of “do no harm” as a medical concept is considerably more specific — and more clinically grounded — than its current use as an abstract ethical slogan suggests.

The Hippocratic context: The maxim “to help and not to harm” appears in the Epidemics I not as a general principle but as a prognostic imperative. Jonsen traces the passage: the author is teaching physicians to recognize which patients can benefit from treatment and which cannot. “Helping and not harming” means that the physician should undertake care only when the disease “does not exceed the strength of men’s bodies” and the patient can cooperate in recovery.(Jonsen, 2000) The harm to be avoided is specifically the subjection of incurable patients to dangerous therapies — bleeding, cutting, burning, and strong drugs that were genuinely lethal in the wrong hands.(Jonsen, 2000) This is a clinician’s concept of harm, rooted in the knowledge that Hippocratic therapeutics were powerful enough to kill. The later abstract formulation primum non nocere smooths away this clinical specificity.

The double-effect tradition: The medieval scholastic development that most directly addressed the problem of harm-in-the-course-of-healing was the doctrine of double effect, first formulated by Thomas Aquinas for self-defense. Catholic moral theologians applied it to medical interventions — amputation, sterilization, and other procedures ordinarily forbidden — to justify actions that caused harm when their purpose was saving life. Jonsen notes that these theologians also developed the doctrine of “ordinary versus extraordinary means” to determine when such harm was actually required: no one was obliged to undergo extraordinary suffering or accept futile treatment.(Jonsen, 2000) These medieval theological distinctions traveled directly into twentieth-century bioethics, where the language of ordinary/extraordinary means was still in use in the 1952 papal address on ventilators.

The organ transplantation challenge: Joseph Murray’s first successful kidney transplant (1954) confronted the maxim “do no harm” with a structurally new problem: excision of a healthy organ from a healthy donor for another’s benefit. Under the traditional Hippocratic reading, any harm done must contribute to the ultimate benefit of the patient who bears the harm. A healthy donor’s excised kidney did nothing of the kind. Jonsen describes how the medical profession and law had to reconceptualize harm to accommodate what transplantation actually involved: either the donor’s benefit was redefined (as psychological benefit from saving a family member) or the concept of harm was recognized as always requiring qualification by consent and proportionality of benefit.(Jonsen, 2000) The kidney transplant case showed that a simple “do no harm” formula was not sufficient to handle modern medical innovation.

These three moments — the Hippocratic clinical calculus, the medieval theological refinements, and the twentieth-century challenge from transplantation — illustrate that nonmaleficence has always been a more historically complex and contextually sensitive concept than its aphoristic form suggests.

See Also

• Beneficence
• Autonomy
• Medical Ethics
• Edmund Pellegrino
• Hippocratic Oath
• Physician-Patient Relationship
• Virtue Ethics
• Paternalism
• Informed Consent

Sources

• Beauchamp, T. L. & Childress, J. F. (2013). Principles of Biomedical Ethics, 7th ed. Oxford University Press. [beauchamp-childress-principles-of-biomedical-2013] — Lead authority
• Jonsen, Albert R. (2000). A Short History of Medical Ethics. Oxford University Press. [jonsen-short-history-medical-2000]