Using a Hypothetical Scenario to Inform Psychiatric Advance Directives

In December 1991, the Patient Self-Determination Act (Omnibus Budget Reconciliation Act of 1990) was implemented. The act requires federally funded hospitals and other health care facilities to inform patients that they have a right to complete an advance directive in accordance with state law and to provide educational material and the appropriate form ( 1 ).

Advance directives should ensure that patients' medical health care preferences guide their future clinical care and should avoid ethical and legal complications when patients may not be able to make or communicate their choices ( 2 ). Similarly, psychiatric advance directives allow a competent individual to document his or her psychiatric treatment preferences and identify a surrogate decision maker to serve as a proxy during periods of future compromised decision making ( 3 ). A psychiatric advance directive is described as empowering patients to assert control in treatment decisions; decreasing perceived coercion; increasing treatment collaboration, motivation, and adherence; enhancing communications about treatment preferences; facilitating appropriate and timely treatment; and avoiding conflicts over treatment and medication issues ( 4 , 5 , 6 , 7 ).

Despite these reported benefits, several surveys of mental health providers reveal a reluctance to embrace psychiatric advance directives because of perceptions that such directives may disallow appropriate treatment interventions or hospitalization ( 8 , 9 , 10 ). On the other hand, numerous studies document the interest in and willingness to complete psychiatric advance directives by persons with serious mental illness ( 3 , 11 , 12 , 13 ).

In previous reports, we used medical illness scenarios to describe the Health Care Preferences Questionnaire (HCPQ) ( 14 , 15 ) and its utility in eliciting end-of-life care preferences for persons with serious mental illness ( 16 ). The purpose of this survey-based study was to evaluate the utility of using the HCPQ psychiatric scenario to inform psychiatric advance directives for persons with serious mental illness. Specific questions include the following: Are hypothetical scenarios about health status a feasible approach for eliciting psychiatric treatment preferences? What are the views of persons with serious mental illness regarding involuntary psychiatric care? What characteristics are associated with consumer treatment preferences?

Methods

Participants in this study were clients of the Massachusetts Department of Mental Health receiving services from one of five contracted community mental health agencies. Participant recruitment began in November 2000 and ended in July 2001. The first 150 eligible volunteers were included in this study. All participants met state criteria for a serious and persistent mental illness. Participants were 18 years of age or older, able to communicate in English, and provided written informed consent. Two institutional review boards approved the study protocol and the informed consent document. Written informed consent was obtained from participants after they received a complete description of the study. A previous publication provides a detailed description of the sample and data collection methods ( 15 ).

The HCPQ is divided into five parts and evaluates participants' attitudes and preferences for treatment-specific choices and assesses basic health care values. HCPQ topic areas include health status, assessed with the 12-item Short Form Physical and Mental Health Summary Scales (SF-12) ( 17 , 18 ); advance care planning; treatment preferences concerning scenarios about medical and psychiatric state; beliefs, values, and concerns about the end of life; and interview feedback and follow-up, including questions regarding comfort and stress associated with the interview and talking about the end of life ( 15 , 16 ). Participants' health care values and beliefs were obtained by agency clinicians who were trained to administer the HCPQ.

The HCPQ was supplemented by an addendum where clinicians subjectively evaluated participants' understanding of the questions, obvious comfort or distress during the interview, apparent cognitive impairment or mental status abnormalities (each rated on a 5-point scale), and expressions of a negative reaction to any question.

The hypothetical psychiatric health state scenario describes the progressive deterioration of a psychiatrically ill person and asks about providing medications, emergency interventions (restraint or seclusion), and electroconvulsive therapy (ECT). Clinician interviewers first read the scenario with an imaginary patient as the subject and then asked the participant to select the most appropriate treatment choice for this person. Participants were then asked to choose among the same treatment options for himself or herself. Where applicable, scenarios included elements pertaining to judicial review, in accordance with Massachusetts legal requirements. Results from the hypothetical medical health state scenarios are described in previous work ( 16 ).

Psychiatric treatment preferences were examined through cross-tabulation analyses. Correlations were identified between treatment preferences and having a proxy decision maker; medication noncompliance; age; physical and mental well-being; psychiatric diagnosis; cognitive impairment; understanding of questions; and comfort, stress, and negative reactions to the interview.

Results

Description of participants

As previously reported ( 15 ), 67 participants (45%) were female, the mean±SD age of the sample was 42.0±10.4 years, and 137 participants (91%) were Caucasian. Participants were predominantly educated through high school or beyond (N=116, or 77%), and most had never married (N=129, or 86%). Seventeen participants (11%) had a guardianship of person.

Ninety-nine participants (66%) had a primary psychiatric diagnosis of schizophrenia or other psychosis, 42 (28%) had a diagnosis of a mood disorder, and nine (6%) had a diagnosis of personality disorder or posttraumatic stress disorder. Participants had demographic, clinical, and functional characteristics similar to a larger sample of Massachusetts Department of Mental Health consumers, as assessed by the Current Evaluation of Risk and Functioning ( 15 , 19 ). Participants' mental and physical health functioning was similar to that in samples from other studies of persons with serious mental illness, as measured by the SF-12 ( 15 , 18 ).

The psychiatric vignette

The psychiatric scenario has four stages where questions are asked of participants. The first stage evaluates preferences for providing medication against one's will. The vignette reads as follows: "Imagine that a person with a serious mental illness becomes frightened of others, upset, and confused. Over a few weeks things get worse and the person is placed in a hospital. He believes that other people want to hurt him and that the food is poisoned. He refuses to take any medications. At this point, he is so sick that he is unable to make good decisions. If medications will help this person to get better, is it OK to give them to him against his will? If this were you, would you want medication to be given to you against your will? If you had a person who was your health care proxy, would you want your proxy to approve medication for you, even if you were refusing it?"

As shown in Table 1 , a total of 112 participants (79%) believed that medication should be given to the imaginary other person; however, only 99 participants (70%) would want to receive the medication against their own will. There were significant correlations between approval for providing the medication and greater understanding of the interview (r=-.188, p<.05), adherence to medications (r=-.180, p<.05), and older age (r=-.200, p<.05). Finally, having a proxy was positively correlated with believing that the proxy should approve the medication (r=.228, p<.01).

Enlarge table

The second stage of the scenario describes increased aggression and the need for emergency interventions. The vignette continues as follows: "Let's say this person assaulted a hospital employee because he is afraid the employee wants to hurt him, even though that employee was trying to help. In this situation, should the physicians use an emergency intervention, such as putting the person in a locked quiet room or limiting his movements by tying him down to a bed with restraints?"

Participants had similar responses for self and other in the use of emergency interventions. Most believed that emergency interventions were appropriate measures in this situation for the imaginary other person (72%) or for themselves (73%). Among the 101 participants who said that emergency interventions should be administered, seclusion in a locked quiet room was preferred over physical restraints (N=81, or 80%, versus N=20, or 20%). Affirmative responses to the use of restraint or seclusion were correlated with greater comfort with the interview (r=.244, p<.01) and having picked a health care proxy (r=.191, p<.05).

The third stage of the scenario asks about involuntary medications that the patient has been refusing to take and that could improve his or her thinking. "Now he is so confused and fearful that he is attacking others. In this situation, should the physician order an injection of medication that would calm the person and that might help to treat his mental illness?"

Most participants would want the physician to provide an injection of medication under these circumstances. However, participants were more likely to want the physician to give the injection to the imaginary person than to themselves (88% versus 76%). Here, 94 of 145 participants (65%) had specific medications that they would never want to receive. The medications that were most commonly cited included haloperidol (indicated by 32 participants) and thorazine (mentioned by 24 participants).

Younger age (r=.181, p<.05) and nonadherence to medication routines (r=.191, p<.05) were associated with refusing specific medications. A total of 118 participants (81%) also had preferences regarding which medications were the most appropriate. Having lower mental well-being, according to the mental component of the SF-12 (r=.222, p<.01), or a mood disorder (r=-.197, p<.05) instead of a thought disorder was associated with preferences about the best medication to treat one's mental illness. Finally, adherence to medication was positively correlated with choosing that the health care proxy would inform the physician or judge about particular medication preferences (r=-.167, p<.05).

The final stage of the scenario addresses the provision of electroconvulsive therapy against the patient's wishes. "Now let's imagine that this person's mental illness symptoms might be quickly and effectively treated with ECT (electroconvulsive therapy, or 'electric shock therapy') treatments. In this situation, he is refusing ECT. If a judge says that the person is not able to decide for himself at this time because of his illness, is it OK to give these treatments?"

The scenario goes on to ask, "If this patient's situation worsened so that his life was in danger due to his mental illness (for example, he is in danger of dying from starvation or self-injury) and other treatments were not helping, do you think electric shock treatments should be ordered, even if he were refusing them at the time?"

Most participants were unwilling to receive ECT as a potential quick and effective treatment (other, 40%; self, 32%). They were more likely to agree to ECT when life was in danger because of psychiatric symptoms (other, 57%; self, 60%). Worse physical well-being was associated with believing that the judge should decide to provide the participant with ECT (r=.180, p<.05). In addition, more medical problems were associated with believing that ECT should be ordered in a life-threatening situation (r=-.202, p<.05).

Distress

As previously reported (16), 29 participants (22%) experienced some level of distress or negative reaction associated with the HCPQ interview (subjectively rated by the clinician interviewer). Among these participants, eight were distressed by the psychiatric vignettes in particular and two were distressed by the entire interview (7% total). A total of six participants terminated the interview prematurely because of negative reactions, and two of these terminated after they had completed the psychiatric vignette portion (4% total). Negative reactions to the HCPQ interview were associated with the belief that a physician should not order a medication injection for the participant (r=-.212, p<.05).

Understanding

Clinician interviewers indicated that most participants understood HCPQ questions (extremely well or very much, N=90, or 68%; somewhat, N=34, or 26%; not at all or not very much, N=9, or 7%). Greater understanding was associated with affirmative responses regarding the receipt of medications via injection (r=-.259, p<.01) and a wish to have the health care proxy tell a judge (r=-.214, p<.05) or physician about ECT preferences (r=-.236, p<.01). In contrast, the interviewer's subjective perception of the participant' impaired decision making (N=17, or 13%) was correlated with negative responses to providing medications against one's will (r=.200, p<.05) or having the proxy recommend medications against one's will (r=.202, p<.05), receiving medications by injection (r=.265, p<.01), and wanting the proxy to tell the physician about one's wishes regarding ECT (r=.191, p<.05).

Discussion

The findings from this study suggest that consumers are able to identify their preferences for psychiatric advance directives by responding to hypothetical psychiatric scenarios. Only a small number of participants experienced significant distress during the administration of the HCPQ, and most participants had a good understanding of the questions and concepts. Participants provided similar, though somewhat lower, approval of involuntary psychiatric treatments for themselves compared with an imaginary patient. Of note, involuntary treatments, such as the provision of medications, medication injection, and seclusion or restraint, were supported by over two-thirds of participants. In contrast, use of ECT as a "quick and effective treatment" was supported by less than one-third of participants.

These findings supplement our previous work, which found that the HCPQ provides a feasible, acceptable, and useful standardized approach to medical advance care planning for persons with serious mental illness ( 15 ). Similar to our previous analyses that conclude that persons with serious mental illness were able to designate treatment preferences in response to end-of-life health scenarios ( 16 ), this study shows that the same is true for psychiatric health scenarios.

Mental health providers, family members, and consumers support the use of psychiatric advance directives as an important means of strengthening communication, avoiding unwanted treatment, obtaining needed treatment, and empowering consumers ( 7 , 13 ). One-half to two-thirds of persons with serious mental illness consistently express interest in completing psychiatric advance directives ( 7 , 13 ), and one recent study found that 61% completed psychiatric advance directives when this process was facilitated through a structured session with a manual ( 20 ).

A significant barrier that limits the use of advance directives among individuals with serious mental illness is difficulty obtaining appropriate supports to complete a psychiatric advance directive ( 7 ). One recent study found that 44% of consumers needed nontechnical support to complete psychiatric advance directives (terminology paraphrasing, discussions of the purpose, when they can be activated, role of surrogate decision makers) and 35% wanted help from a case manager ( 4 ). In contrast to software-produced advance directives, the HCPQ is a semistructured interview that provides support to consumers while it is being completed and allows for alliance building between the clinician and the consumer.

Clinicians can use the HCPQ "imaginary person" methodology to support consumers in designating advance directives. The initial presentation of health state scenarios through the "imaginary person" point of view allows the consumer to think of the situation in the third person, as a story, and ask questions about it. This process facilitates active discussion of treatment preferences while ensuring that individuals have the opportunity to request clarification and enhance their understanding of treatment choices. It can reduce consumer anxiety, improve understanding, and set the stage for thoughtful responses. This process also can increase clinician sensitivity to proposed treatments and their impact on a consumer. Once completed, documented preferences can guide health care providers, proxies, and court-appointed guardians in choosing treatments when the consumer may be unable to either make or communicate his or her choices.

Our findings regarding treatment preferences are similar to recent reports suggesting that consumers prefer methods for deescalating crises over restraint and ECT ( 3 ). Most participants supported the use of involuntary treatments in an emergency situation (medications, 70%; medication injection, 76%; seclusion and restraint, 73%). Participants were least likely to support the use of ECT (32%), which aligns with findings by Srebnik and colleagues ( 3 ) and Backlar and colleagues ( 21 ), who found that 72% and 57% of consumers, respectively, refused ECT. Of note, participants in our study were more likely to support the use of ECT when their life was in danger (60%).

As noted in other studies, several characteristics were associated with psychiatric treatment preferences. The provision of medications was associated with greater understanding, compliance with medications, and age. Sixty-five percent of participants identified one or more medications that they did not want to receive. This is slightly lower than the 71% that has been reported elsewhere in the literature ( 20 ). Preferences about specific medications that one would never want to receive were associated with younger age and medication nonadherence.

There are several limitations that should be considered in the interpretation of this data. As mentioned in previous studies using this data set ( 15 ), the study sample was drawn from one urban and suburban area, was not culturally or racially diverse, and used a purposive sampling method of self-selected participants. In addition, although the study sample consisted of individuals receiving mental health treatment services, the extent of participants' experience or knowledge with each health issue or procedure is unknown. Health care preferences may be shaped by poorly formed understanding of treatment procedures; however, this concern is broadly applicable to general health care decision making, as well as for health care proxies who speak on behalf of others. Furthermore, stigma and misperception may influence decisions about some psychiatric treatment choices. For example, the low acceptance of ECT, which aligns with findings from other studies of persons with serious mental illness ( 3 , 21 ), may indicate that persons receiving mental health treatment experience the same distrust and misperception of ECT that is held by the general population ( 22 , 23 ).

Future studies may wish to examine the utility of the HCPQ among members of ethnic minority groups, other cultures, or individuals in rural settings. Future studies also may wish to examine the utility of the HCPQ in enhancing the therapeutic alliance between consumers and clinicians and in ensuring that consumers receive preferred interventions during periods of compromised decision making.

Conclusions

In summary, our study illustrates that persons with serious mental illness may benefit from the use of a semistructured interview technique for eliciting preferences to be included in a psychiatric advance directive. The use of a hypothetical psychiatric scenario was intended to prepare persons with serious mental illness to identify treatment preferences in a nonintimidating manner. This approach was associated with minimal distress, high rates of understanding, and the elicitation of treatment preferences that are similar to those reported by others. This study complements existing data ( 15 , 16 ) and supports the use of the HCPQ as a means for eliciting medical and psychiatric treatment preferences for advance directives.

Acknowledgments and disclosures

This study was funded by grant 35497 from the Robert Wood Johnson Foundation through the Promoting Excellence in End-of-Life Care Program. We are grateful to Steven Carreras, Ph.D., Anna Rubley, L.I.C.S.W., Meredith Hanrahan-Boshes, R.N., and Sally Neylan Okunfor, B.S.N., M.M.H.S., for their assistance as project directors. We are also grateful to Keith Miles, M.P.A., and Stephen Bartels, M.D., M.S., for assistance in developing the evaluation program. The HCPQ is further described at www.promotingexcellence.org/i4a/pages/Index.cfm?pageID=3825.

The authors report no competing interests.