Attention to building design as an element in the treatment of patients with psychiatric and substance use disorders predates current codes, standards, and debates about the clinical ethics of facility design—such as whether facilities should be designed with private or semiprivate bedrooms. This essay explores the ethical basis of the design of facilities for psychiatric patients as it relates both to the architecture of past asylums and to contemporary thinking about psychiatric hospital design. The argument is made that the limitation of patients' autonomy and rights by the built environment is legitimate and ethical.

Early 19th century asylum design was based largely on a strong belief by alienists that architectural design was of "cardinal importance" and inseparable from any plan of treatment for the insane (1). The impulse to rationalize and reform the management of "the mad" expressed itself most visibly in the creation of the asylum (2). Early alienists believed that for psychiatric patients to be cured they must be separated from the home and community environment and that "a lunatic can be cured only in an institution" (3,4). In an 1846 statement typical of the period, John Galt, superintendent of the Eastern State Lunatic Asylum in Virginia, observed in The Treatment of Insanity that all persons who managed lunatics believed that no cure could take place unless patients were removed from their familiar home environment and installed in a suitable "therapeutic space" (4,5). That space was singularly and unambiguously believed to be the asylum (4).

The design of such a novel building type, which manifested a series of tensions between home and institution and benevolence and surveillance, was a challenge to architects (4). Many architects who set the earliest course for asylum architecture were guided by prison design (6). Although early asylums may appear to the untrained eye to be no different from prisons, there was an acknowledged attempt to include design elements that would assist in "the cure." G. T. Hine, an early asylum architect and arguably the most prolific and influential specialist asylum architect of the period, stated in a 1901 lecture to the Royal Institute of British Architects that "the architect must remember that he can materially assist the doctor in his cure, as well as his protection, of the patient by the careful consideration he gives to the many details of planning and construction" (7). In this way the ethics of asylum design, firmly rooted in beneficence, was born, and since then spatial relations have been central to the function and management of persons treated in institutional settings (8).

T. S. Kirkbride's designs for asylums, first fully described in his 1854 book The Construction, Organization and Arrangement of Hospitals for the Insane, rationalized an existing system of asylum administration that was well established and that catered to the needs of the "curable insane." Kirkbride, superintendent of Pennsylvania Hospital for the Insane, found English architect Isaac Holden's original design deficient and set about to make changes to support his notion of a "moral architecture" that would move beyond the simple tenets of separation and observation. Wards would be used to categorize patients by severity of disease as well as by "social traits," with the most severely ill patients confined to ground-floor wards furthest from the center building. Supporters of nonlinear plans, however, found Kirkbride's "shallow V" designs to be overly specialized. Galt observed that American asylum architecture had "produced dysfunctional buildings whose prison-like features belied their therapeutic function" (3). Similar criticism of the asylum was made by George Cook, who wrote "The asylum's forbidding mass and barely concealed means of confinement created a fearful penal milieu hardly conductive to cure" (3). Critics of the linear plan proposed the import of several European design innovations that relied on nonlinear or "cottage" architecture.

The focus on architecture in the care of patients with mental illness was evident in Kirkbride's 1854 book, but by 1900 creating a therapeutic architecture was not the key goal of physicians who worked in asylums. It had become evident that environmental determinism, as it had come to be understood at the time, had not produced many cures (3). Calls for reform in these early public institutions from the 1930s through the 1950s were not uncommon (9,10), and by the late 1960s patient advocates were calling for a new architecture. A new design ethic, in part made possible by new pharmaceutical treatments for mental illness, broke the large asylums into smaller units, dispersed those units throughout a community, and emphasized the right of a patient to be treated in the least restrictive environment as a new prima facie value.

The number of patients in public asylums, now more often called psychiatric hospitals, would fall from 550,000 in 1950 to fewer than 90,000 by the mid-1990s (11). The patient population remaining in institutions could be described as simultaneously acute and chronic and was increasingly viewed as containing "dangerous individuals" (12) to whom an acute care paradigm of informed consent, in which a patient's consent to treatment is assumed, could be ethically applied (2,13,14). The ethics of asylum architecture shifted again, and now security was emphasized (11). For the second time since the early days of Bethlem and the Great Confinement (15), many institutionalized patients were viewed not only as having an illness but also as posing a security risk. The popularity of Freudian psychoanalysis and other "talk therapies" in conjunction with proven psychoactive drugs made the need for an "architectural cure" in an institutional setting largely a part of a distant orthodoxy. The psychiatric hospital would now, more than ever before, have to prevent elopements and acts of self-harm by patients. The hospital became a setting in which patients regularly had their freedom of movement limited by architectural design mostly out of concern for security rather than cure.

The ethical principle placed most at risk by the "architecture of madness" is patient autonomy. Today, with few exceptions, inpatient treatment for individuals with mental and substance use disorders is provided in a locked facility. That is, patients' freedom of movement is restricted from the moment of admission until the day of discharge. Exit doors are locked and windows bolted shut; patient movement, activity, and privacy are closely monitored by staff in a modern purpose-built architectural surveillance system (2).

Patient autonomy, the right to refuse care, the right of association, and the right of freedom of movement are all set aside by a principle of beneficence when patients are thought to be a risk to themselves or others and suitable for admission to the locked unit of a psychiatric hospital. The wholesale suspension of these patient rights is believed to be permitted because the risk of suicide and elopement is presumed for all patients. No matter what the actual clinical presentation, all patients are treated in a like manner by the prevailing architectural ethic: they are all at risk and must be protected.

Current building code language, perhaps best codified by the American Institute of Architects, is likewise wholly presumptive of patients' intent to harm themselves (16). Today's asylum architect therefore proceeds in limiting patients' freedom within ethical boundaries established not by patient autonomy but by a principle of beneficence. Patient autonomy and personal freedom are routinely set aside "for the patient's own good" by the beneficence of current asylum architecture. Contemporary psychiatric architecture replete with its locked doors, lack of privacy, and structured social spaces is justifiable, therefore, by the application of the ethical principle of beneficence (17).

However, this straightforward notion of beneficence—of what is in the best interest of a person with mental illness—can be obfuscated by the consideration that freedom of movement is a prima facie liberty of utmost importance. This becomes an especially difficult issue for patients whose admission to a locked psychiatric unit is triggered exclusively by behavior and not by jurisprudence. Libertarians who make such an argument, which is similar to the arguments of the antipsychiatry proponents of the 1970s, assert that no restriction of freedom as a result of aberrant behavior is justified because "good mental hygiene" is a fabricated social construct. However, some restrictions of patients do withstand scrutiny through a libertarian lens, and even a strict libertarian would likely acknowledge that restriction of freedom of movement is permitted when there is a suspicion that a third party is at risk. An admission decision based on a need to protect third parties is supported by John Stuart Mill's (18) assertion that the "only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others."

If positive beneficence—an action undertaken to benefit others as a matter of obligation—is an insufficient justification to override autonomy, then the ethics of making choices to restrict a patient's freedom require that we enter the area of paternalism. Paternalism, the intentional overriding of a person's known preference that is justified by a goal of preventing harm to that person, is the ethic required to justify the admission to the locked unit if we assume that personal freedom is a prima facie right.

A question still remains, however, that requires a separate consideration of voluntary and involuntary psychiatric admissions. It can be argued that although the ethic of paternalism applies easily to the involuntary admission of a patient to a locked unit (a clear and unambiguous overriding of freedom of movement), the ethics of a voluntary admission are less clear. If an admission is voluntary—and a nonacute model of informed consent is employed—are locked doors and lack of privacy necessary or even permitted? How are locked doors and a lack of personal privacy justified within the same architectural envelope for these two very different types of patients: one who willingly surrenders rights and consents to treatment and the other from whom rights are taken and whose consent is implied or provided by surrogates? Some may argue that there should there be two different architectural standards and that ethical principles other than beneficence are necessarily dominant.

Unfortunately, the ethics of architecture of behavioral health facilities must remain blind to the issue of any ethical or design difference between voluntary and involuntary admissions to a locked unit. The architect affirms through the design of the environment of care that he or she sees no difference in the autonomy of these patients, and the unfettered application of the ethic of paternalism is permitted simply because the intentions of the patient cannot be reliably determined (19). Even antipaternalists do not object to paternalistic interventions that prevent anticipated acts of patient self-harm because they believe that substantially autonomous actions are not at stake (17). Reliable patient clinical assessments aside, all patients admitted to the locked unit are presumed by the architect to be at risk of self-harm and elopement. The law and ethical principles require that providing treatment in the least restrictive setting be an underlying value in psychiatric hospital design (19). However, the architecturally least restrictive environment is inevitably determined by the single patient who is most at risk on the unit.

In sum, the ethical justification for permissible architectural paternalism in the design of contemporary behavioral health facilities may best be measured against criteria proposed by Beauchamp and Childress (17) for clinical paternalism. Specifically, paternalism in architectural design is permitted when the harms prevented by the design outweigh the loss of independence; when the patient's condition seriously limits his or her ability for an autonomous choice; when the design would be universally justified under similar circumstances; and when the person who will be limited by the paternalistic design has consented, will consent, or would, if rational, consent to choices made on his or her behalf. Although the days of asylums and of alienists' near-obsession with architecture have dissipated, the moral issues of psychiatric hospital architecture and design remain caught between the biomedical ethics principle of patient autonomy and the good of paternalism. In the absence of a reliable determination of a patient's true intent for self-harm or elopement, the scale must remain justifiably tipped toward beneficence and the good of paternalism.

The author is grateful for the support and encouragement of the National Association of Psychiatric Health Systems and the members of the Health Guidelines Revisions Committee of the Facilities Guidelines Institute.

The author reports no competing interests.