To the Editor: I agree with Dr. Glick and colleagues (1) that there is a need to reconsider the current model of inpatient psychiatric care, which consists of ultrashort stays and a sole focus on safety and crisis stabilization. However, I disagree with the proposal for a decision model, which begs the question as to what the interventions are “that cannot be safely undertaken in an outpatient setting.” “Outpatient setting” is too vague a term in this context, and this proposal completely ignores current international research on community-based alternatives to hospitalization.

Similar question begging and unwarranted generalizations are found in the following statements. “The patient population • includes those who most clinicians would agree require a 24-hour inpatient stay.” It is well known that clinicians' threshold for both admitting patients and keeping them in hospital varies significantly both within countries and between countries. It partly depends on the availability of community-based alternatives and also on clinicians' willingness to use those alternatives. Dr. Glick and colleagues also state that “for most patients in an acute psychiatric crisis, hospital stays are the only option.” The reference they cite to support this claim has nothing to do with the statement (2). It refers to a simple patient satisfaction study that compared patients in inpatient wards with those in residential alternatives and found that the latter were more satisfied. Ironically, the supplement of the British Journal of Psychiatry that featured this article consists of seven reports all describing research on residential alternatives to inpatient care, the general tenor of which is cautiously affirmative that residential alternatives to acute admission should be part of the spectrum of acute services. For example, a comparison of hospitalized patients with those in six residential alternatives found no significant differences in these patients' risk of intentional or unintentional self-harm, social functioning, social problems, and recent self-harm. One residence even treated involuntarily detained patients (3).

I agree that inpatient treatment is an essential component of any mental health system. The only way one can assess effectiveness of inpatient care is—as the authors state—with controlled studies. But what should be the comparator? Ethically, one cannot use a no treatment condition. One must use one of the three hospital alternatives: intensive home treatment, day hospitals, or residential alternatives. International research has shown that these models—particularly intensive home treatment—sometimes in combination with crisis residential alternatives, can allow clinicians to dispense with inpatient care for about 40% of patients, even psychotic and suicidal ones, plus enable early discharge for about 40% of those who are admitted to a hospital (4). The implementation phase of Dr. Glick and colleagues' model can often be accomplished equally as well—sometimes better—outside of the hospital (5).

Debating whether hospitalizations should be short, medium, or long is unlikely to influence governments or managed care companies. It also misses the most important issue: How can we develop a mental health system in which patients receive treatment that is not too much and not too little, with an intensity that varies in response to their fluctuating needs, and that is the least disruptive to their lives.