Psychiatric advance directives aim to help people retain control over their treatment when incapacitated. Although strategies exist to help consumers develop directives without provider involvement, many consumers engage a provider in creation of directives to facilitate decisions about treatment, confirm details of previous treatment, ensure timely access to the document, or provide clarification.

Joint crisis plans, which promote these goals through meetings between consumers and providers facilitated by an independent mental health professional, have been associated with reduced involuntary hospitalizations in the United Kingdom. In 2005–2006, we pilot tested this intervention in the United States to determine the feasibility of its completion and use.

The study was conducted at two publicly operated clinics and one affiliated nearby hospital in New York City serving a predominantly low-income Latino population. The inpatient unit and crisis response team directors agreed to store copies of plans if participants wished, but this was not possible at the psychiatric emergency room.

We included consumers admitted to a psychiatric inpatient service at least once in the past two years with a diagnosis of psychotic or bipolar illness. Inpatients and those in mandated community treatment were excluded.

After recruitment, an investigator not affiliated with clinic services met with the participant and case manager (or psychiatrist, if no case manager was involved) to discuss and record the participant's preferences for the plan's content. To finalize it, the investigator held a meeting with the participant, case manager, psychiatrist, and other relevant professionals. The investigator encouraged the participant to bring a caregiver, friend, or advocate. The aim was to discuss stakeholders' views on crisis management and to negotiate mutually acceptable solutions. The selection of information and the exact wording were the participant's choice alone. Legal documentation was not used because New York State advance directive legislation does not cover advance instructions.

Besides a description of routine care and regular medication, content included relapse indicators, what had or had not helped in previous relapses, steps to avert hospitalization, and procedures if hospitalization occurs. For example, a participant said that he disliked the nickname used by some clinic attendants and that it made him more irritable during relapse. Many participants identified medications that had caused side effects or asked that their usual medication simply be restarted; similarly, all stated whom they wanted with them in the emergency room and who could and could not visit them if hospitalized. One participant asked whether he could be placed in a single room with encouragement to share after a couple of days.

After confirming the draft with the participant and, if the participant requested significant changes, with the providers, the investigator distributed the final copy to the participant and others whom the participant designated. Another interviewer administered a questionnaire to participants for their views of the intervention shortly thereafter and again 12 months later.

The intervention was completed for all 20 participants. Of 17 participants interviewed, all but one felt it probably or definitely reflected their wishes. At 12 months, all 16 participants interviewed agreed that they would recommend the plan to others, and most agreed that they had become more involved with their care (N=9) and had a better understanding of their illness (N=10).

Case managers, who also received questionnaires, thought participants learned about their illness and felt less anxious about what might happen in a crisis. One stated that the process allowed participants to talk about problems that providers had felt were too sensitive; another thought it had helped a participant avoid hospitalization. Two cited its usefulness for family members. None raised liability concerns.

Of eight participants hospitalized, one consulted the plan during this crisis and during another crisis that did not lead to hospitalization. One participant's relative brought the plan to the emergency room for physicians' review while the participant was there. The director of the inpatient unit, where copies were kept for all hospitalized participants, stated that only one participant's care was incongruent with the plan's content, when the patient had made a serious suicide attempt; the seriousness of the attempt, rather than the written preferences, determined care. Two other participants reported difficulties in implementing the plan: one whose preferences were not carried out in the emergency room and another who struggled to remember the content during a crisis. In summary, the intervention was welcomed by consumers and providers, and the results provide evidence that it promoted shared decision making. Further work is needed if such plans are to be used by emergency service providers.

This study was supported by the MacArthur Research Network on Mental Health and the Law and the Center for Homelessness Prevention Studies at Columbia University Mailman School of Public Health.