When persons with mental disabilities are believed to be. Use of Leverage Over Patients Money to Promote Adherence to Psychiatric Treatment

ORIGINAL ARTICLES Use of Leverage Over Patients Money to Promote Adherence to Psychiatric Treatment Paul S. Appelbaum, MD,* and Allison Redlich, PhD Abstract: Persons who handle money for patients with mental disabilities often use access to funds as leverage to promote adherence to treatment. Drawing on data from a five-site study involving interviews of 1011 patients at community mental health centers, this paper explores who provides money assistance, the predictors of leverage use, and its impact on patients. A substantial minority of patients reported having experienced leverage. Substance abuse and frequent hospitalization were the strongest clinical predictors, suggesting that functional impairment is a key determinant. Patients experiencing money leverage were also likely to experience other forms of leverage, and although they were more likely to feel coerced, they were also more likely to believe that money leverage was effective. amily members seemed to provoke fewer negative feelings as money managers. The frequency with which money leverage is used suggests attention is needed to both the fairness of leverage and the means of minimizing negative effects. Key Words: Coercion, leverage, compliance, representative payee. (J Nerv Ment Dis 2006;194: 294 302) When persons with mental disabilities are believed to be unable to manage their funds, both the Social Security Administration and the Veterans Administration appoint representative payees to receive disability and pension payments on their behalf. Of the 2.8 million people under age 65 who receive SSI payments for mental disabilities, more than 1.6 million have representative payees appointed (Social Security Administration, 2003). An additional, unknown number of persons with mental disabilities have someone who receives and/or manages their money for them on an informal basis. In general, representative payees are charged with helping to insure that recipients basic needs for food, clothing, housing, and medical care are met, and conversely that the money they *Department of Psychiatry, University of Massachusetts Medical School, Worcester, Massachusetts; and Policy Research Associates, Delmar, New York. Supported by the Research Network on Mandatory Outpatient Treatment of the John D. and Catherine T. MacArthur oundation. Send reprint requests to Paul S. Appelbaum, MD, Department of Psychiatry, University of Massachusetts Medical School, Worcester, MA 01655. Copyright 2006 by Lippincott Williams & Wilkins ISSN: 0022-3018/06/19404-0294 DOI: 10.1097/01.nmd.0000207368.14133.0c receive is not dissipated in frivolous or harmful ways (Social Security Administration, 2001). A growing literature has documented various aspects of the representative payee system, especially for persons with mental illness. Need for representative payees has been estimated in different populations (Rosen et al., 2002a, 2002b), agency-based models of providing services have been described (Congress and Chernesky, 1993; Conrad et al., 1999; Hanrahan et al., 2002; Luchins et al., 1998), and recipients satisfaction with these services has been explored (Cogswell, 1996; Dixon et al., 1999; Elbogen et al., 2003a, 2005; Luchins et al., 2003; Rosen et al., 2003b; Rosen et al., 2003c). Persons who have representative payees have been characterized as having a high incidence of substance abuse and nonaffective psychosis (Conrad et al., 1998; Elbogen et al., 2003c; Neale and Rosenheck, 2000; Ries and Comtois, 1997), greater severity of illness (Neale and Rosenheck, 2000; Ries and Comtois, 1997), and higher lifetime rates of hospitalization and homelessness (Conrad et al., 1998; Neale and Rosenheck, 2000; Ries and Comtois, 1997). Although controlled trials of the use of representative payees are lacking, uncontrolled studies have reported positive effects on housing stability (Luchins et al., 2003), reduced hospitalization (Luchins et al., 1998, 2003), and involvement in treatment (Elbogen et al., 2003a; Luchins et al., 2003). How representative payees and recipients interact regarding the disbursement of funds has been subject to surprisingly little systematic evaluation. The potential for conflict is evident, as recipients desire access to their money and payees seek to insure its appropriate expenditure (Gallmeier and Levy, 1998; Greener, 1975). Moreover, recipients have few due process protections in the appointment of a representative payee, and the possibility of abuse is real. Anecdotal reports suggest that styles of interaction vary substantially, from merely encouraging recipients to use their money wisely to exercising much tighter control over funds (Cogswell, 1996; Congress and Chernesky, 1993; Rosen et al., 2002a). In addition, it seems clear that at least some representative payees use control of recipients funds to leverage their compliance with recommended treatment, including attendance at treatment sessions, adherence to medication, and abstinence from alcohol and other substances (Brotman and Muller, 1990; Hanrahan et al., 2002; Luchins et al., 2003; Lucksted and Coursey, 1995; Ries and Dyck, 1997; Rosen et al., 2003a). This function appears to go beyond the role envisioned by the Social Security Administration for its representative payee program. The only previous study to explore the frequency of 294 The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006

The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 Leverage Over Patients Money such leverage reported that roughly one third of subjects said they anticipated that their discretionary funds would be withheld if they failed to comply with some aspect of treatment (Elbogen et al., 2003b). In an earlier paper from the present study, we reported that 15% to 32% of subjects recruited from community mental health centers in five sites across the United States who had someone who assisted them in managing their money said that this person used access to their funds as leverage to promote treatment compliance (Monahan et al., 2005). Use of money as leverage was associated with subjects who were significantly younger, had less insight about their illness, and were more likely to suffer from a psychotic disorder, to abuse substances, and to have higher rates of hospitalization (Monahan et al., 2005). Here we expand on that report to describe additional aspects of the population subject to such leverage, including modeling the predictors of leverage, and examining the relationships among money leverage, compliance with treatment, and attitudes toward treatment and use of leverage. The goal is better to inform discussions of the legitimate extent of leverage on persons with mental disabilities and the procedures that should regulate these practices. METHODS Procedures The methodology of this study is described in detail elsewhere (Monahan et al., 2005). In brief, approximately 200 outpatients from publicly funded programs were sampled from each of five sites: Chicago, Illinois; Durham, North Carolina; San rancisco, California; Tampa, lorida; and Worcester, Massachusetts. Although we specified that participants had to be treated for mental disorders (rather than for a substance abuse disorder without co-occurring mental illness), we did not restrict recruitment by diagnosis or level of acuity. The specific inclusion criteria for the samples were as follows: (1) age: 18 to 65 years; (2) language: English-speaking or Spanish-speaking; (3) service use: currently in outpatient treatment of a mental disorder with a publicly supported mental health service provider (operationally defined as at least one appointment/visit in the prior 6 months); and (4) duration of service use: first service contact as an adult occurred at least 6 months ago. At the Worcester, Tampa, and San rancisco sites, potential subjects were recruited sequentially in the waiting rooms of outpatient clinics of community mental health centers. In Durham, a list of potentially eligible subjects was created from management information system data, and these patients were randomly selected to be approached regarding the study. Both of these recruitment strategies were used at the Chicago site; half of their sample was obtained using the waiting room approach and the other half using the eligibility list approach. Refusal rates varied from 2% to 13% across sites. After complete description of the study to the subjects, written informed consent was obtained, per the procedures approved by IRBs at each site. Across the five sites, the mean age of the subjects ranged from 41 to 47 years, the percent male from 32% to 65%, the percent white from 35% to 69%, and there was moderate variation in the percent diagnosed with schizophrenia (42% to 50%), bipolar disorder (14% to 18%), and major depression (28% to 31%). More detail about study subjects can be found in Monahan et al. (2005). Measures Leverage We assessed subjects lifetime experience of four specific leverages derived from the social welfare system (money and housing as leverage) and the judicial system (the criminal sanction and outpatient commitment as leverage) (Monahan et al., 2005). Only those subjects who had a formal representative payee and/or an informal money manager were questioned in greater detail about their experience with money leverage. Subjects who reported that they had ever had a representative payee were asked whether that person had ever made getting money dependent on adherence to treatment. Similarly, subjects acknowledging ever having had someone who handled their money for them (a role not necessarily played by representative payees, but sometimes assumed informally by family members or others) were asked whether that person ever had conditioned disbursement on adherence. The identity of the representative payee or money handler was ascertained, and subjects were asked about the extent to which they believed that use of money leverage helped people with mental disorders to stay well. inally, an overall leverage score was created. Subjects who endorsed at least one of the four leverages were considered as having experienced any leverage, as compared with subjects who had experienced no leverage. Subject Characteristics In addition to obtaining objective demographic and diagnostic information by chart review, the anchored version of the Brief Psychiatric Rating Scale (BPRS; Woerner et al., 1988) was used to assess current psychiatric symptoms. Additionally, Global Assessment of unctioning (GA; American Psychiatric Association, 1994) scores were created to assess current functioning levels. Insight into mental illness was assessed with the Insight and Treatment Attitudes Questionnaire (McEvoy et al., 1989). Self-reported alcohol and drug use for the past 30 days was obtained. If participants had drunk any alcohol or taken street or nonprescribed drugs, follow-up questions adapted from the Michigan Alcoholism Screening Test (Selzer, 1971) and the Drug Abuse Screening Test (Skinner, 1982) were asked. We combined alcohol and drug abuse and dichotomized into one or more substance abuse symptoms versus no substance abuse symptoms. To assess treatment compliance, participants were asked to rate their level of compliance in the past 6 months on a 5-point Likert scale (1 never missed an appointment, 5 avoided keeping appointments altogether). Medication compliance was assessed in a similar manner on a 6-point scale (1 never missed taking medication, 6 never took medication). To measure treatment satisfaction, a series of 14 questions adapted from Mental Health Statistics Improvement Program Consumer Survey (Ganju, 2000) was asked. Items included, I felt free to complain, and, As a direct 2006 Lippincott Williams & Wilkins 295

Appelbaum and Redlich The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 result of services I received, I am better able to control my life. A mean composite score (Cronbach.89) was created to have one global measure of treatment satisfaction (higher scores indicate less satisfaction). Perceived coercion was measured using the MacArthur Perceived Coercion Scale (Gardner et al., 1993). Questions, which were adapted to reflect perceptions about outpatient treatment, included, I felt free to do what I wanted about going to the mental health center. As in other studies (Swartz et al., 2003), for most analyses, a dichotomous perceived coercion score was created and used by labeling those with a score below the median as low perceived coercion and labeling those with a score above the median as high perceived coercion. inally, a set of six Likert-scaled items was used to assess participants experience of autonomy in everyday life. Examples include, How much say did you have about what you would do during the day? and, How much say did you have about how much of your money you could spend? Higher scores indicate perceptions of more autonomy. The scale items were all significantly intercorrelated and demonstrated good internal reliability (Cronbach.83). Statistical Analysis We used logistic regression to examine the joint associations between participants demographic, clinical, and treatment characteristics and their money leverage experiences. or the purpose of multivariable modeling, pooling the data across sites offered the advantage of greater statistical power, but also posed two problems that required adjustment in the analyses. irst, we had to account for site effects and site-bycovariate interactions associated with leverage. To examine and control for these site effects, we used the Zelen test of the homogeneity of odds ratios (StatXact, 2003, pp. 511 517; Zelen, 1971). The Zelen statistic allowed us to test the null hypothesis that the relative risk for leverage did not differ across the five sites, but represented a sampling distribution from a common population. If the Zelen test showed that the sites odds ratios for a given variable were homogeneous, we then pooled the data for that variable and calculated a common odds ratio across sites. The second problem was that pooling the data could have distorted statistical inferences, insofar as the observations within each site were not independent. Without an adjustment for the clustered nature of the data, the SEs around the pooled estimates would have been understated, leading to overly liberal tests of statistical significance. Thus, we used the same specialized statistical software (StatXact, 2003) to adjust significance tests and confidence intervals around the common (pooled) odds ratios. or multivariable analysis, we used a companion statistical package designed to conduct multivariable logistic regression with stratified data (PROC LOGXACT; LogXact, 2002, pp. 83 103). These techniques provided the appropriate correction of variance estimates taking into account within-site correlation of observations. Specifically, the software uses the Cochran-Armitage method, as adapted by Rao and Scott (1992), to adjust the effective sample size for design effects that occur with a clustered sample (LogXact, 2002, pp. 755, 774). RESULTS Representative Payees In the present sample, across the five sites, the percent of patients who ever had a representative payee ranged from 31% to 53% (Table 1). Of those who ever had a representative payee, the majority (ranging from 69% to 92% across sites) either currently had a representative payee or had one in the past 6 months. Between 4% and 15% of the entire sample, TABLE 1. requency and Identify of Representative Payees, and Use of Leverage Chicago (N 205) Durham (N 204) San rancisco (N 200) Tampa (N 202) Worcester (N 200) Overall Range (N 1011) Representative payee ever 108 (52.7) 76 (37.2) 62 (31.0) 84 (41.6) 79 (39.5) 31% to 53% Representative payee current or in last 6 mo a 86 (40.5) 70 (34.3) 46 (23.0) 58 (28.7) 61 (30.5) 23% to 41% (78.9) (92.1) (74.2) (69.0) (77.2) (69% to 92%) Representative payee leverage ever a 31 (15.1) 15 (7.4) 8 (4.0) 20 (9.9) 20 (10.0) 4% to 15% (28.7) (19.7) (12.9) (23.8) (25.3) (13% to 29%) Representative payee leverage in last 6 mo b 12 (5.9) 7 (3.4) 3 (1.5) 11 (5.4) 10 (5.0) 2% to 6% (14.0) (10.0) (6.5) (19.0) (16.4) (7% to 19%) Representative payee identity c (N 107) (N 75) (N 62) (N 84) (N 78) (N 406) Parent 8 (7.5) 20 (26.7) 4 (6.5) 29 (34.5) 12 (15.4) 7% to 35% Spouse/partner 1 (1.0) 6 (8.0) 2 (3.2) 4 (4.8) 0 (0) 0% to 8% Other relative 6 (5.6) 11 (14.7) 6 (9.7) 21 (25.0) 16 (20.5) 6% to 25% Nonrelative 4 (3.7) 4 (5.3) 13 (21.0) 8 (9.5) 2 (2.6) 3% to 21% MH/social worker 88 (82.2) 22 (29.3) 31 (50.0) 7 (8.3) 44 (56.4) 8% to 82% Housing-related 0 (0) 12 (16.0) 4 (6.5) 11 (13.1) 4 (5.1) 0% to 16% Other 0 (0) 0 (0) 2 (3.2) 4 (4.8) 0 (0) 0% to 5% a Percentages in parentheses in this row are of those subjects who ever had a representative payee. b Percentages in parentheses in this row are those of subjects who had a representative payee in the last 6 mo. c N values may vary because of missing data. 296 2006 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 Leverage Over Patients Money and between 13% and 29% of the subsample who ever had a representative payee, reported that their representative payee had made receipt of money contingent upon their treatment adherence. Lower rates of leverage experience (7% to 19%) in the last 6 months were reported in the subset that had a representative payee at the time of interview or within the past 6 months. The identity of persons acting as representative payees varied by site (Table 1). In Chicago, where representative payees tend to be commonplace, 82% were mental health professionals or social workers. In San rancisco and Worcester, mental health professionals were the representative payees approximately half of the time. In contrast, in Tampa, mental health professionals accounted for only 8% of representative payees; instead, a parent was most often cited. In Durham, parents and mental health professionals were equally likely to be the representative payees. Regardless of whether patients had a representative payee, they were asked if they had someone who helped manage or handle their money. Overall, 13.4% of the sample had only a representative payee, 11.1% had only a money handler, and 27.4% had both a money handler and a representative payee. Money handler rates were very similar to those of representative payees (Table 2). Across the sites, between 27% and 46% of subjects had a money handler, and of those, between 14% and 27% had experienced leverage. Also, the identity of the money handler was very similar to that of the representative payees. Indeed, of those who had both a money handler and a representative payee, the person was the same 88% of the time. Characteristics Associated With Money Leverage Because representative payees and money handlers were the same persons the majority of the time, we created one money leverage variable that compared people who had ever experienced money leverage with those who had not. To examine the characteristics associated with having experienced money leverage, we then grouped independent variables into four domains: (1) demographics and social environment, (2) clinical characteristics, (3) money management relationship, and (4) treatment and systems utilization. Numerous significant bivariate relationships emerged between individual predictors and money leverage (see the first column in Table 3). (The only variable not to pass the Zelen test for homogeneity of odds ratios across study sites was living in a staffed residence, and therefore, no significance test is reported for that variable.) The second column in Table 3 displays adjusted common odds ratios for the variables within each domain that had a p value of 0.20 or less in the multivariable logistic regression analysis controlling for site after testing for homogeneity with the Zelen test. Analyses regarding domain 1 (demographics and social environment) revealed that money leverage was significantly more common among males, persons with a low income (below $700 a month), persons who were not living independently, and those who had other people providing direct assistance with their treatment (e.g., reminding them to take medication and helping them get to their appointments). In domain 2 (clinical), money leverage was more common among patients with a diagnosis of schizophrenia, substance abuse problems, lower GA scores, and less insight into their mental problems. Those who reported having been violent in the recent past tended to be more likely to have experienced money leverage, but this relation only approached statistical significance. In domain 3 (relationship to representative payee/money handler), we found that, in comparison to those for whom a mental health professional fulfilled the representative payee/money handler role, those who had a relative other than a parent or spouse/partner performing the function were significantly less likely to have experienced money leverage. Analyses from domain 4 (treatment and systems utilization) showed that money management leverage was more common among those with a history of multiple hospitalizations. When all significant variables in the domain models were included in the final logistic regression model that controlled for site differences (column 3 of Table 3), a more TABLE 2. requency and Identity of Money Handler, and Use of Leverage Chicago (N 205) Durham (N 204) San rancisco (N 200) Tampa (N 202) Worcester (N 200) Overall range (N 1011) Money handler 94 (45.9) 93 (45.6) 54 (27.0) 91 (45.0) 58 (29.0) 27% to 46% Money handler leverage a 25 (12.2) 13 (6.4) 8 (4.0) 22 (10.9) 15 (7.5) 4% to 12% (26.6) (14.0) (14.8) (24.2) (25.9) (14% to 27%) Money handler identity b (N 93) (N 93) (N 54) (N 91) (N 58) (N 389) Parent 3 (3.2) 20 (21.5) 11 (20.4) 27 (29.7) 8 (13.8) 3% to 30% Spouse/partner 4 (4.3) 12 (12.9) 2 (3.7) 9 (9.9) 4 (6.9) 4% to 13% Other relative 9 (9.7) 15 (16.1) 5 (9.3) 23 (25.3) 6 (10.3) 9% to 25% Nonrelative 4 (4.3) 10 (10.8) 12 (22.2) 7 (7.7) 1 (1.7) 2% to 22% MH/social worker 72 (77.4) 21 (22.6) 21 (38.9) 9 (9.9) 36 (62.1) 10% to 77% Housing-related 0 (0) 15 (16.1) 2 (3.7) 14 (15.4) 3 (5.2) 0% to 16% Other 1 (1.1) 0 (0) 1 (1.9) 2 (2.2) 0 (0) 0% to 2% a Percentages in parentheses are of those subjects with money handlers. b N values may vary because of missing data. 2006 Lippincott Williams & Wilkins 297

Appelbaum and Redlich The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 TABLE 3. Cross-Site Multivariable Models of Characteristics Associated With Money Leverage Bivariate Domain models inal model Common OR a 95% CI b p Adjusted OR 95% CI b p Adjusted OR 95% CI b p Model 1: demographics and social environment Age above median 0.72 (0.49 1.06) ( 44 y) emale 0.55 (0.37 0.82) ** 0.56 (0.35 0.89) * 0.81 (0.52 1.26) Nonwhite 1.34 (0.90 1.98) Cohabitation 0.61 (0.34 1.08) High school or above 0.91 (0.61 1.35) Income above median 0.46 (0.29 0.75) ** 0.56 (0.34 0.92) * ( $700) Social support for 2.07 (1.37 3.13) *** 2.23 (1.39 3.56) *** treatment (past 6 mo) Social support 0.76 (0.49 1.19) 0.68 (0.41 1.13) Independent living 0.42 (0.28 0.64) **** 0.58 (0.36 0.92) * Homeless 0.85 (0.25 2.90) Living with family 1.90 (1.08 3.34) * Staffed residence None (0.47 3.01) c Model significance Likelihood ratio 38.85 df 5 **** Model 2: clinical variable model Schizophrenia 2.23 (1.50 3.30) *** 2.13 (1.42 3.21) *** 1.07 (0.68 1.67) Nonpsychotic with 0.48 (0.26 0.88) * voluntarily commitment Substance abuse 2.03 (1.32 3.13) ** 1.95 (1.24 3.07) ** 1.99 (1.22 3.26) ** BPRS score above 1.35 (0.91 1.99) median ( 30) GA score above median 0.46 (0.29 0.71) *** 0.54 (0.4 0.84) ** ( 47) Insight score above 0.64 (0.43 0.93) * 0.63 (0.42 0.93) * 0.66 (0.42 1.02) median ( 18) Violence 2.11 (1.05 4.25) * 2.02 (0.96 4.24) Model significance Likelihood ratio 43.2 df 5 **** Model 3: money model Rep-payee/money handler relation Parent 2.11 (1.27 3.53) ** 1.45 (0.80 2.60) Spouse/partner 0.55 (0.21 1.46) 0.48 (0.17 1.34) Other relative 0.41 (0.20 0.84) * 0.40 (0.19 0.86) * 0.45 (0.22 0.93) * Nonrelative 0.69 (0.29 1.61) 0.61 (0.25 1.48) MH/social worker 1.35 (0.84 2.18) (reference category) Housing related 0.61 (0.23 1.63) 0.52 (0.18 1.49) Other 1.40 (0.26 7.45) 1.17 (0.21 6.45) 2.09 (0.37 11.7) Model significance Likelihood ratio 15.93 df 6 * Model 4: treatment and systems utilization Past hospitalizations 2.25 (1.48 3.42) *** 2.25 (1.48 3.42) *** 1.73 (1.08 2.78) * above median ( 3) Time in treatment above 1.06 (0.71 1.58) median ( 20 y) Picked up by the police in past 6 mo (treatment) 2.37 (1.26 4.47) ** (Continued) 298 2006 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 Leverage Over Patients Money TABLE 3. (Continued) Bivariate Domain models inal model Common OR a 95% CI b p Adjusted OR 95% CI b p Adjusted OR 95% CI b p Picked up by the police in 1.28 (0.52 3.12) past 6 mo (arrest) Outpatient visits above 1.52 (0.97 2.39) median ( 2 monthly) Involuntary Commitment 1.30 (0.84 2.00) Drug Attitudes Inventory 0.88 (0.60 1.30) above median ( 8) Medication compliance 1.16 (0.78 1.71) above median ( 3) Treatment compliance above median ( 3) None (0.57 2.90) Model significance Likelihood ratio 15.36 Likelihood ratio 24.75 df 1*** Df 7*** *p 0.05; p 0.10 (trend); **p 0.01; ***p 0.001; ****p 0.0001. a A common odds ratio with cluster-corrected CI is given only if all five sites odds ratios were determined by the Zelen test (p 0.05) to represent a sampling distribution from a common population. Results were produced by logic regression. b Where there is no common odds ratio, the interval gives the range of odds ratios across sites. c Significant values (OR 1.0) at three sites (#1, 2, 3). limited number of associations emerged. In the final model, having experienced money leverage was significantly more likely among substance abusers and those with a history of more than three prior hospitalizations, and significantly less likely among those whose representative payee and/or money handler was a relative other than a parent or spouse/partner. Other Leverages and Outcome Variables Bivariate relationships between money leverage and other forms of leverage are presented in Table 4. There was a strong likelihood for outpatients in this study to have experienced more than one type of leverage. Persons who had experienced money leverage were nearly four and a half times more likely to have experienced at least one of the other forms of leverage: they were twice as likely to have experienced leverage via the criminal or civil (i.e., outpatient commitment and related judicial orders) justice systems, and almost four times as likely to have been directed to participate in treatment as a condition of their housing arrangement. There was no significant relationship between money leverage and treatment satisfaction. Outpatients who had experienced money leverage were more likely to feel coerced, and to feel pressure to attend treatment and to take medications. They were less than half as likely to report a perceived sense of autonomy in everyday affairs in comparison to those who had not experienced any money-related leverage. inally, our respondents were asked to rate the effectiveness of money leverage. We collapsed these ratings into a scale with two categories (agree versus neutral/disagree). Outpatients who experienced money leverage were more than twice as likely as those who had not to believe that holding back money to gain treatment adherence was effective in helping people to stay well (Table 4). amily Versus Nonfamily Representative Payees/Money Handlers To compare perceptions of autonomy, coercion, and other variables according to whether the participant had money management by a relative versus a nonrelative, we performed a series of 2 (money relation: family versus nonfamily) 5 (data collection site) ANOVAs, with perceptions of autonomy, treatment satisfaction, coercion, medication and appointment pressure, and money leverage effectiveness perceptions as dependent variables. As shown in Tables 1 and 2, TABLE 4. Bivariate Relationships Between Experiencing Money Leverage and Other Leverages and Outcomes (Controlling for Site) Common OR a 95% CI p Criminal justice leverage 2.36 (1.57 3.56) 0.0001 Housing leverage 3.70 (2.49 5.51) 0.001 Outpatient commitment 2.10 (1.32 3.35) 0.01 leverage At least 1 of the 3 4.42 (2.82 6.92) 0.0001 nonmoney leverages Treatment satisfaction 1.24 (0.84 1.83) NS Pressure to take 4.64 (3.00 7.18) 0.0001 medications Pressure to attend 3.34 (2.24 4.98) 0.0001 treatment Autonomy 0.36 (0.23 0.56) 0.0001 Perceived coercion 1.75 (1.19 2.58) 0.01 Perception of the effectiveness of the leverage 2.58 (1.75 3.81) 0.0001 a A common odds ratio with cluster-corrected CI is given only if all five sites odds ratios were determined by the Zelen test (p 0.05) to represent a sampling distribution from a common population. Results were produced by logistic regression. 2006 Lippincott Williams & Wilkins 299

Appelbaum and Redlich The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 the majority of nonrelatives in this role were mental health professionals or social workers. Means and SDs are presented in Table 5. Respondents with a family member as the representative payee/money handler were more satisfied with their treatment compared with respondents with a nonfamily member in that role ( 1, 507 4.83, p 0.05). The main effect of site and the interaction effect between relation and site were not significant. As seen in Table 5, respondents with a family member who helped with their money tended to feel less pressure to take their medications, although the relationship did not reach statistical significance ( values 1/4, 490 2.11, p values 0.09; the interaction between the two variables was not significant). Consistent with the trend found with pressure to take medications, respondents with a family member who helped with their money reported feeling significantly less pressure to keep their mental health appointments than those with a nonfamily member in that role ( 1, 506 4.64, p 0.05, without a significant interaction effect). With regard to feelings of autonomy, a significant interaction between site and money relation emerged ( 4, 507 2.37, p 0.05). In the Tampa and Durham sites only (see Table 5), patients with family member money helpers reported feeling more autonomous than patients with nonfamily money helpers. Although several sites reported significant differences on perceived coercion depending on the identity of the money manager, there were no consistent findings across the sites. In Durham, patients with family member money helpers felt significantly more coercion, whereas in Tampa, these patients felt significantly less coercion in comparison to patients with nonfamily money helpers. inally, the degree to which patients agreed that it was helpful to treatment to make receipt of money contingent upon treatment adherence did not differ as a function of data collection site or the family/nonfamily money management role. DISCUSSION This study provides the most extensive look to date at the frequency and characteristics of the use of money to leverage adherence to psychiatric treatment. Our initial data suggested that use of money leverage tended to focus on sicker patients who use higher levels of services, and that although the incidence of money leverage was substantial, the majority of persons with a representative payee or informal money handler do not experience leverage as part of that relationship. It is clear from the data presented here that there is wide variation in who serves as representative payee across sites, perhaps as a function of whether formal programs exist assigning that role to persons in the local mental health agency. However, with the exception of relatives other than parents or spouses/partners, the identity of the payee had no significant impact on the likelihood of whether leverage was applied. On the other hand, subjects whose relatives helped with money management perceived them as exerting less pressure to keep appointments compared with nonrelatives, and reported themselves more satisfied with treatment, suggesting some advantage in having relatives serve in that role. TABLE 5. Mean (SD) Differences by amily () Versus Nonfamily (N) Representative-Payee/Money Handler a Chicago (N 205) Durham (N 204) San rancisco (N 200) Tampa (N 202) Worcester (N 200) Overall (N 1011) N (N 304) (N 213) N (N 56) (N 37) N (N 36) (N 78) N (N 65) (N 21) N (N 47) (N 53) N (N 100) (N 24) 1.86 (0.50) 2.04 (0.54) 2.13 (0.65) 2.05 (0.48) 2.02 (0.66) 2.30 (0.63) 1.97 (0.66) 2.07 (0.54) 2.05 (0.54) 2.19 (0.45) 2.01 (0.62) 2.13 (0.54) Treatment satisfaction 1.32 (1.39) 1.83 (1.87) 1.36 (1.61) 1.04 (1.58) 0.70 (0.98) 1.08 (1.53) 1.40 (1.52) 1.61 (1.73) 0.91 (1.29) 1.48 (1.66) 1.23 (1.46) 1.46 (1.72) Pressure to take meds 0.83 (1.34) 1.25 (1.73) 1.00 (1.49) 1.15 (1.48) 0.76 (1.18) 0.62 (0.91) 1.01 (1.26) 1.64 (1.76) 0.62 (1.09) 1.09 (1.58) 0.90 (1.30) 1.12 (1.54) Pressure to keep appointments Autonomy 3.45 (1.02) 3.67 (0.92) 3.86 (0.99) 3.32 (1.17) 3.56 (1.30) 3.78 (1.01) 3.63 (1.03) 3.24 (1.03) 4.05 (0.92) 3.90 (0.98) 3.73 (1.04) 3.63 (1.02) Perceived 5.04 (3.38) 5.96 (3.65) 7.55 (4.46) 5.98 (3.32) 4.95 (3.96) 5.55 (3.45) 5.38 (3.56) 7.39 (4.26) 6.08 (3.59) 6.96 (3.41) 5.96 (3.92) 6.23 (3.62) coercion Money leverage 2.83 (1.31) 3.16 (1.28) 3.25 (1.16) 3.43 (1.26) 3.19 (1.25) 3.32 (1.32) 3.37 (1.29) 3.37 (1.24) 3.35 (1.11) 3.13 (1.15) 3.26 (1.23) 3.26 (1.26) effectiveness a Significant main effects of money management relation are in bold. 300 2006 Lippincott Williams & Wilkins

The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 Leverage Over Patients Money Predictors of the use of money leverage include many of the indicia of greater severity of illness and functional impairment, from being diagnosed with schizophrenia to being unable to live independently. Combining all predictors in a multivariate model that accounted for possible site differences, only substance abuse and a higher rate of hospitalization predicted leverage, whereas getting money assistance from a relative other than a parent or spouse/partner predicted its absence. unctional impairment and the extent to which the person controlling access to the patient s money is motivated to encourage adherence (more distant relatives presumably being less motivated) appear to be key to whether leverage is employed. That patients experiencing money leverage also tended to experience other forms of leverage suggests that the interventions all respond to a common dilemma: the failure of patients to adhere to treatment. Perhaps more unexpected is the absence of a relationship between money leverage and treatment satisfaction, since we might expect leverage to be applied to those who are less satisfied and hence less adherent. Although patients subject to leverage reported feeling more coerced and less autonomous, they also were more likely to see money leverage as an effective tool for promoting compliance. Hence, while patients may not have liked being dealt with in this way, they appear to recognize money leverage as an effective intervention. Among the limitations of this study is the unknown representativeness of the sites involved and of the samples recruited; however, the geographic diversity of the sites and the substantial size of the sample at each site provide some assurance in this regard. Insofar as many of the data, including those regarding the use of leverage, derive from patients self-report, the validity of the findings depends on the accuracy of their responses, which were not independently verified. Moreover, our focus on use of leverage to promote treatment adherence leaves us unable to comment on other uses of leverage, including pressure to discourage substance use and to encourage allocation of funds to more responsible purposes (e.g., food, clothing, savings). inally, note must be taken that many of the measures were obtained at the time of the interview, whereas use of leverage was reported historically. To the extent that subjects conditions vary across time, the lack of strictly concurrent measures may create spurious associations or conceal real ones. Again, it is reassuring that from 69% to 92% of subjects who ever had assistance with money reported it as occurring at present or within the last 6 months, mitigating (though not entirely eliminating) the problem of nonconcurrent measures. uture studies might emphasize concurrent assessment of leverage use, with additional measures beyond self-report (e.g., observational methods), a broader focus on uses of leverage other than to promote treatment adherence, and ascertainment of how leverage is applied and what makes it effective. These data should stimulate discussion about a prevalent but little-discussed phenomenon. Whether it is fair to use control of patients money to leverage adherence (or for other purposes) is a normative question that cannot be answered by empirical data. But the data do indicate that a more impaired, less functional segment of the patient population is targeted for money leverage, suggesting at a minimum that the technique is not being used indiscriminately. Those who have experienced money leverage attest to its effectiveness, although they also make clear its cost in perceptions of coercion and reduced autonomy. amily members seem to evoke fewer negative reactions from patients when they serve as money managers, but whether they are as effective in that role as unrelated persons is unknown. Also, the prevalence of money leverage once more evokes the query regarding whether more systematic protection of patients due process is warranted before control of their funds is handed over to another person. ACKNOWLEDGMENTS The authors thank Pamela Robbins and Roumen Vessilinov for assistance with data analysis and members of the MacArthur oundation Research Network on Mandatory Outpatient Treatment for their comments. REERENCES American Psychiatric Association (1994) Diagnostic and Statistical Manual of Mental Disorders (4th ed). Washington DC: American Psychiatric Association. Brotman AW, Muller JJ (1990) The therapist as representative payee. Hosp Community Psychiatry. 41:167 171. Cogswell SH (1996) Entitlements, payees and coercion. In DL Dennis and JL Monahan (Eds), Coercion and Aggressive Community Treatment: A New rontier in Mental Health Law (pp 115 125) New York: Plenum Press. Congress EP, Chernesky RH (1993) Representative payee programs for the elderly: administrative, clinical and ethical issues. J Gerontol Soc Work 1993. 21:77 93. Conrad KJ, Matters MD, Hanrahan P, Luchins DJ, Savage C, Daugherty B (1998) Characteristics of persons with mental illness in a representative payee program. Psychiatr Serv. 49:1223 1225. Conrad KJ, Matters MD, Hanrahan P, Luchins DJ, Savage C, Daugherty B, Shinderman M (1999) Representative payee for individuals with severe mental illness at community counseling centers of Chicago. Alcohol Treat Q. 17:169 186. Dixon L, Turner J, Krauss N, Scott J, McNary S (1999) Case managers and clients perspectives on a representative payee program. Psychiatr Serv. 50:781 786. Elbogen E, Soriano C, Van Dorn R, Swartz M, Swanson J (2005) Consumer perspectives on representative payee use of disability funds to leverage adherence to mental health services. Psychiatr Serv. 56:45 49. Elbogen EB, Swanson JW, Swartz MS (2003a) Effects of legal mechanisms on perceived coercion and treatment adherence among persons with severe mental illness. J Nerv Ment Dis. 191:629 637. Elbogen EB, Swanson JW, Swartz MS (2003b) Psychiatric disability, the use of financial leverage and perceived coercion in mental health services. Int J oren Ment Health. 2:119 127. Elbogen EB, Swanson JW, Swartz MS, Wagner HR (2003c) Characteristics of third-party money management for persons with psychiatric disabilities. Psychiatr Serv. 54:1136 1141. Gallmeier CP, Levy JA (1998) Holding the check hostage: Representative payees, clients and the underground economy. J Contemp Ethnogr. 26: 395 425. Ganju V (2000) The MHSIP Consumer Survey. Available at http://www. mhsip.org/documents/mhsipconsumersurvey.pdf. Accessed January 22, 2005. Gardner W, Hoge S, Bennett N, Roth L, Lidz C, Monahan J, Mulvey E (1993) Two scales for measuring patients perceptions of coercion during hospital admission. Behav Sci Law. 20:307 320. Greener MR (1975) The Social Security Administration s representative payee program: An act of benevolence or cruelty? Cardozo Law Rev. 12:2025 2057. Hanrahan P, Luchins D, Savage C, Patrick G, Roberts D, Conrad KJ (2002) 2006 Lippincott Williams & Wilkins 301

Appelbaum and Redlich The Journal of Nervous and Mental Disease Volume 194, Number 4, April 2006 Representative payee programs for persons with mental illness in Illinois. Psychiatr Serv. 53:190 194. LogXact (2002) Software for exact logistic regression. Cambridge (MA): Cytel Software Corp. Luchins DJ, Hanrahan P, Conrad KJ, Savage C, Matters MD, Shinderman M (1998) An agency-based representative payee program and improved community tenure of persons with mental illness. Psychiatr Serv. 49: 1218 1222. Luchins DJ, Roberts DL, Hanrahan P (2003) Representative payeeship and mental illness: A review. Admin Pol Ment Health. 30:341 353. Lucksted A, Coursey RD (1995) Consumer perceptions of pressure and force in psychiatric treatments. Psychiatr Serv. 46:146 152. McEvoy J, Apperson LJ, Appelbaum P, Ortlip P, Brecosky J, Hammill K, Geller JL, Roth LH (1989) Insight into schizophrenia: Its relationship to acute psychopathology. J Nerv Ment Dis. 177:43 47. Monahan J, Redlich AD, Swanson J, Robbins PC, Appelbaum PS, Petrila J, Steadman HJ, Swartz M, Angell B, McNiel D (2005) Use of leverage to improve adherence to psychiatric treatment in the community. Psychiatr Serv. 56:37 44. Neale MS, Rosenheck RA (2000) Therapeutic limit setting in an assertive community treatment program. Psychiatr Serv. 51:499 505. Rao J, Scott A (1992) A simple method for the analysis of clustered binary data. Biometrics. 577 585. Ries RK, Comtois KA (1997) Managing disability benefits as part of treatment for persons with severe mental illness and comorbid drug/ alcohol disorders. Am J Addict. 6:330 338. Ries RK, Dyck DG (1997) Representative payee practices of community mental health centers in Washington state. Psychiatr Serv. 48:811 814. Rosen MI, Bailey M, Rosenheck RR (2003a) Principles of money management as a therapy for addiction. Psychiatr Serv. 54:171 173. Rosen MI, Desai R, Bailey M, Davidson L, Rosenheck R (2003b) Consumer experience with payeeship provided by a community mental health center. Psychiatr Rehab J. 25:190 195. Rosen MI, Rosenheck RA, Shaner AS, Eckman TA, Gamache GR, Krebs CW (2002a) Substance abuse and the need for money management assistance among psychiatric inpatients. Drug Alcohol Depend. 67:331 334. Rosen MI, Rosenheck RA, Shaner AS, Eckman TA, Gamache GR, Krebs CW (2002b) Veterans who may need a payee to prevent misuse of funds for drugs. Psychiatr Serv. 53:995 1000. Rosen MI, Rosenheck R, Shaner A, Eckman T, Gamache G, Krebs C (2003c) Payee relationships: Institutional payees versus personal acquaintances. Psychiatr Rehab J. 26:262 267. Selzer ML (1971) The Michigan Alcoholism Screening Test: The quest for a new diagnostic instrument. Am J Psychiatry. 127:1653 1658. Skinner HA (1982) The Drug Abuse Screening Test. Addict Behav. 7:363 371. Social Security Administration (2003) SSI Annual Statistical Report. Available at http://www.ssa.gov/policy/docs/statcomps/ssi_asr/2003/table27.html. Accessed January 22, 2005. Social Security Administration (2001) A Guide for Representative Payees. Available at http://www.ssa.gov/pubs/10076.html. Accessed January 22, 2005. StatXact (2003) Statistical Software for Exact Nonparametric Inference. Cambridge (MA): Cytel Software Corp. Swartz MS, Swanson JW, Monahan J (2003) Endorsement of personal benefit of outpatient commitment among persons with severe mental illness. Psychol Pub Pol Law. 9:70 93. Woerner MG, Mannuzza S, Kane JM (1988) Anchoring the BPRS: An aid to improved reliability. Psychopharmacol Bull. 24:112 117. Zelen M (1971) The analysis of several 2 2 contingency tables. Biometrika. 58:129 137. 302 2006 Lippincott Williams & Wilkins