Case Conceptualization: Interventions and Evaluation
cfscott
No-Show-TherapistRacial-EthnicDisparitiesinClientUnilateralTermination.pdf
BRIEF REPORT
‘No-Show’: Therapist Racial/Ethnic Disparities in Client Unilateral Termination
Jesse Owen University of Louisville
Zac Imel University of Utah
Jill Adelson University of Louisville
Emil Rodolfa University of California, Davis
In the present study, the authors examined the source of racial/ethnic minority (REM) disparities in unilateral termination (i.e., the client ending therapy without informing the therapist)—a form of dropout that is associated with poor alliance and outcome. First, the authors tested whether some therapists were more likely to have clients who reported unilaterally terminating as compared with other therapists. Next, the authors examined 2 competing hypotheses regarding the therapists role in termination disparities: (a) that racial/ethnic disparities in unilateral termination are similar across therapists and thus due to other components of the treatment process or (b) that racial/ethnic disparities in unilateral termination are specific to therapists, where some therapists are more likely, on average, to have higher rates of unilateral termination with REM clients as compared with White clients. The sample included 155 REM clients and 177 White clients who were treated by 44 therapists at a university counseling center. The results showed that therapists accounted for a significant proportion of the variation in clients’ unilateral termination, and REM clients were more likely to report they unilaterally terminated from therapy as compared with White clients. Furthermore, racial/ethnic disparities in clients’ report of unilateral termination varied across therapists’ caseloads. These results suggest that therapists have a central role in their clients’ unilateral termination and have implications for understanding racial/ethnic mental health disparities.
Keywords: unilateral termination, race/ethnicity, mental health disparities, therapist effects
Every therapist has had the experience of a client dropping out of psychotherapy. Unfortunately, this experience is extremely common. Approximately one third of adults who begin psycho- therapy do not return for a second session (Hamilton, Moore, Crane, & Payne, 2011; Simon & Ludman, 2010), and 40%– 60% of all clients drop out of psychotherapy (Clarkin & Levy, 2004; Wierzbicki & Pekarik, 1993). The situation is even more concern- ing for racial/ethnic minority (REM) clients 1 (e.g., Institute of Medicine, 2002; Kales, Blow, Bingham, Copeland, & Mellow, 2000; U.S. Surgeon General, 2001; van Ryn, 2002). Specifically, REM clients receive fewer services and drop out more often than White clients (Whaley & Davis, 2007; Wierzbicki & Pekarik,
1993). Several reasons for these disparities have been proposed, such as stigma associated with seeking mental health treatment and the failure of treatment to meet the unique needs of REM clients (Breaux & Ryujin, 1999; Leong, Wagner, & Tata, 1995). These mental health disparities have fueled the call for therapists to increase their cultural competence (American Psychological As- sociation, 2003; Arrendondo & Toporek, 2004).
To understand racial/ethnic disparities, it is important to clarify the definition of dropping out of psychotherapy (Hatchett & Parks, 2003). In clinical trials, the definition of dropout is typically defined by the failure to complete a prescribed dose of treatment (e.g., not attending at least five sessions or not completing the full treatment protocol). Although defining dropout based on the num- ber of sessions has intuitive appeal and avoids the need for sub- jective assessments (i.e., therapist or client judgment), this defini- tion may unintentionally classify clients who were treated
1 Ethnicity typically refers to a group of people who share a common cultural heritage, values, attitudes, and behaviors, whereas race is typically defined by physical attributes (e.g., skin color) that are shared by a group of people (Quintana, 2007). We have opted to use a hybrid definition that integrates both race and ethnicity, which appears to be most consistent with clients’ experience in therapy, as therapists are reacting to both the clients’ race and ethnicity.
This article was published Online First February 20, 2012. Jesse Owen, Department of Educational and Counseling Psychology,
College of Education and Human Development, University of Louisville; Zac Imel, Department of Educational Psychology, Counseling and Coun- seling Psychology Program, University of Utah; Jill Adelson, Department of Educational and Counseling Psychology, College of Education and Human Development, University of Louisville; Emil Rodolfa, Counseling and Psychological Services, University of California, Davis.
Correspondence concerning this article should be addressed to Jesse Owen, Department of Educational and Counseling Psychology, University of Louisville, Louisville, KY 40092. E-mail: [email protected]
Journal of Counseling Psychology © 2012 American Psychological Association 2012, Vol. 59, No. 2, 314 –320 0022-0167/12/$12.00 DOI: 10.1037/a0027091
314
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
successfully as failures (Lester, Resick, Young-Xu, & Artz, 2010). This definition is especially problematic in community settings, where the dose of treatment is not prescribed. For example, in a study comparing the dose-response model of psychotherapy (i.e., more treatment is better) with the good-enough model (i.e., clients participate in enough treatment to meet their needs and then stop), Baldwin et al. (2009) found that clients who received the lowest dose of treatment benefited the most and progressed most rapidly. In another study, clients who failed to return for second psycho- therapy session were just as likely to report symptom improve- ment, a quality working alliance, and treatment satisfaction as those who did return (Simon et al., 2010; see also Pekarik, 1992).
An alternative approach to defining dropout is to ask therapists or clients directly. Therapist ratings, although useful in many regards, can be biased, especially for clients who are not benefiting from therapy (e.g., Garb, 1998; Hannan et al., 2005). In studies relying on client perspective, therapy dropout typically has been defined as unilateral termination, or ending therapy without dis- cussing termination with the therapist (e.g., Callahan, Aubuchon- Endsley, Borja, & Swift, 2009; Masson, Perlman, Ross, & Gates, 2007; Owen, in press; Vandereycken & Vansteenkiste, 2009). Clients who unilaterally terminate report lower quality alliances with their therapists and have worse therapy outcomes, perhaps indicating that the end of treatment was unplanned or premature (Daughters et al., 2005; Masson et al., 2007; Owen, in press; Owen, Smith, & Rodolfa, 2009; Vandereycken & Vansteenkiste, 2009). In the present study, we used clients’ ratings of whether they ended therapy unilaterally.
Over the years, researchers have examined several factors as- sociated with unilateral termination, such as client diagnosis/ severity, client–therapist racial/ethnic match, and alliance (e.g., Maramba & Nagayama Hall, 2002; Sharf, Primavera, & Diener, 2010; Wierzbicki & Pekarik, 1993). However, there are no known studies examining whether therapists’ differ, on average, in their clients’ unilateral termination rates. Given that therapists have been found to account for approximately 3%–10% of the variance in their client outcomes (e.g., Baldwin & Imel, 2011; Wampold & Brown, 2005), it stands to reason that therapists would also vary in the extent to which their clients unilaterally terminate from ther- apy. Therapists’ variability in their clients’ unilateral termination provides an empirical estimation of therapists’ general competency based on actual therapy outcomes (Wampold & Brown, 2005) and, in this case, therapists’ general competency related to unilateral termination.
Extending this approach to racial/ethnic disparities, it may be that disparities in unilateral termination could be consistent across therapists (general disparity) or vary across therapists (therapist- specific disparity). For instance, although there may be differences between REM and White clients in rates of unilateral termination generally, these differences would be the same within each ther- apists’ caseload. This would be evidence of a general racial/ethnic disparity and would suggest that other aspects of treatment apart from therapist-specific factors, potentially including stigma for attending therapy, influence REM clients’ decision to stay in therapy as compared with White clients. In contrast, if disparities vary across therapists (e.g., some therapists have higher rates of REM clients terminating unilaterally than other therapists), then this would be evidence of a therapist-specific racial/ethnic dispar- ity. This finding may suggest that some therapists do not have the
skills to engage REM (or White) clients in the therapy process. Disentangling general versus therapist-specific effects for mental health, racial/ethnic disparities is an important step for building appropriate interventions to reduce these disparities.
Imel et al. (2011) examined general and therapist-specific dis- parities in therapy outcomes with a diverse sample of 582 adoles- cents who were treated for cannabis abuse/dependence by one of 13 therapists. They found that therapists differed both in their clients’ outcomes generally and that racial/ethnic disparities in outcomes varied across therapists. Specifically, some therapists were equally effective with White and REM clients, whereasothers were not. These results provide initial evidence that some compo- nent of therapist behavior contributes to mental health racial/ethnic disparities. These findings are in contrast to large correlations between clients’ reports of their therapists’ general and multicul- tural competencies (rs ranging from .50 to .80; Coleman, 1998; Constantine, 2002, 2007). Although there may be therapist- specific disparities in unilateral termination, we do not know of any studies in which this issue has been examined.
Therapists’ role in unilateral termination disparities may be understood within the context of forming a quality working alli- ance. In particular, a recent meta-analysis demonstrated that clients who have a strong working alliance with their therapists are less likely to unilaterally terminate (d � 0.55; Sharf et al., 2010). The alliance captures the degree to which clients and therapists are “engaged in collaborative, purposeful work” (Hatcher & Barends, 2006, p. 293). Accordingly, clients who are actively involved in the therapy process with a strong connection with their therapists may consequently be less likely to unilaterally terminate. Given the association between alliance and unilateral termination, it is important to consider the role of working alliance when testing therapists’ differences in racial/ethnic disparities in unilateral ter- mination.
In the present study, we examined the therapist as a source of racial/ethnic disparities in clients’ unilateral termination. First, we posited that therapists would differ in the rate of client unilateral termination (Hypothesis 1). That is, some therapists will be more likely to have clients who report they ended therapy unilaterally as compared with other therapists. Second, we predicted that REM clients would be more likely to unilaterally terminate therapy as compared with White clients (i.e., there is a health disparity; Hypothesis 2). Third, we expected that the disparity in unilateral termination between REM and White clients would vary across therapists (therapist-specific disparities), after controlling for alli- ance, psychological well-being, number of sessions, and the pro- portion of White to REM clients treated by the therapist (Hypoth- esis 3).
Method
Participants
Clients. The sample were 332 clients from a large university counseling center. All clients in this study had ended therapy. Of the 332 clients, 177 (53.3%) identified as White, and 155 (46.7%) identified as REM. Of those who identified as REM, 37.7% identified as Asian American, 22.4% identified as multiracial/ ethnic (of which 84.3% identified as White and Asian American), 21.9% identified as Hispanic/Latino(a), 9.3% identified as African
315TERMINATION STATUS AND RACIAL/ETHNIC DISPARITIES
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
American, 8.0% identified as Middle Eastern, and 0.7% identified as Native American. Given the low number of clients in some of the REM statuses, the decision was made to dichotomize racial/ ethnic status into White and REM. The majority of clients were women (n � 246, 73.4%). Clients reported their median number of sessions attended was 5.0 (M � 6.73, SD � 5.48, range � 1–21). Formal diagnoses were not provided at this counseling center, however, based on clients’ reports there were a wide range of presenting concerns, such as depression, disorder eating, anxiety, adjustment issues, anger, alcohol use, and relationship difficulties.
Therapists. Forty-four therapists treated the 332 clients in the present study, an average of 7.54 (range � 3–21) clients per therapist. For purpose of this study, only therapists who treated at least one White and one REM client were included. Thirteen of the therapists self-identified as REM, and 31 self-identified as White. Ten therapists were men and 34 were women. The therapists were predoctoral interns, postdoctoral fellows, staff psychologists, and staff therapists. In prior assessments at this counseling center, 100% of the therapists reported that they practiced some form of integrative therapy (e.g., psychodynamic/cognitive-behavioral, re- lational/systems/cultural; Owen, Quirk, Hilsenroth, & Rodolfa, 2011). This counseling center generally provides brief therapy (six to 10 sessions). Additionally, it is common practice at this coun- seling center for the therapist who conducted the intake to continue to see the client for therapy.
Measures
Client-defined termination status (Owen et al., 2009). Cli- ents were asked how therapy ended through a multiple-choice format, which resulted in five different categories: (a) client initi- ated the end of therapy without talking with the therapist (unilat- eral termination; n � 101, 30.4%), (b) client initiated the end of therapy after talking with the therapist (n � 64, 19.1%), (c) therapist initiated the end of therapy (n � 14, 4.2%), (d) the end of therapy was mutually decided (n � 103, 30.7%), and (e) ran out of allotted sessions (n � 50, 15.2%). These categories are similar with other studies in which client-rated termination is assessed (e.g., Callahan et al., 2009). Furthermore, this measure of termi- nation status has successfully differentiated clients’ alliance scores and treatment outcomes, with clients who ended therapy via uni- lateral termination reporting worse alliances and outcomes as compared with clients who ended in other ways (Owen, in press; Owen et al., 2009). Consistent with these findings and the previous literature regarding unilateral termination, clients were dichoto- mized into unilateral termination (30.4%, n � 101, coded 1) and other methods of termination (69.6%, n � 231, coded 0).
Schwartz Outcome Scale-10 (SOS-10; Blais et al., 1999). The SOS-10 is a measure of psychological well-being (over the past week), which has 10 items that are rated on a 7-point scale ranging from 1 (Never) to 7 (All the time or nearly all the time). Example items include “I am generally satisfied with my psycho- logical health” and “I feel hopeful about my future.” The SOS-10 has exhibited test–retest correlations and Cronbach’s alphas above .85 (e.g., Blais et al., 1999; Hilsenroth, Ackerman, & Blagys, 2001; Young, Waehler, Laux, McDaniel, & Hilsenroth, 2003). Convergent and discriminant validity has been supported in pre- vious studies with correlations in the predicted direction with a variety of clinical and psychological well-being scales (e.g., Out-
come Questionnaire-45, Beck’s Hopelessness Scale, the Positive and Negative Affect Schedule, and Personality Assessment Inven- tory), and reliably discriminated between clinical and nonclinical samples (see Owen & Imel, 2010, for a review). The Cronbach’s alpha for this sample was .95.
Working Alliance Inventory-Short Form Revised (WAI-SR; Hatcher & Gillaspy, 2006). The WAI-SR is a client-rated measure of working alliance that consists of 12 items that assess goals and tasks for therapy as well as the relational bond between the client–therapist. An example item is: “We agreed on what is important for me to work on.” (Note, the items were adjusted to reflect the past tense, because therapy had been completed for the present sample). Items were rated on a scale ranging from 1 (Never) to 7 (Always). The WAI-SR and the other variations of the instrument are commonly used in psychotherapy research, and the reliability and validity has been demonstrated in numerous studies comparing it with other working alliance and therapy outcome scales (Hatcher & Gillaspy, 2006; Horvath, Del Re, Flückiger, & Symonds, 2011). For purposes of this study, the total scale score was used, yielding a Cronbach’s alpha of .97.
Procedure
Participants were recruited from a large West Coast university counseling center. Clients were asked on their intake card(s) whether they would be willing to receive a survey about their therapy experience. All clients who agreed were sent an e-mail at the end of the academic quarter regardless of whether they were still in therapy or have ended therapy and were able to access the anonymous survey instruments online. For purposes of this study, clients were excluded if they were currently in therapy, endorsed multiple individual therapists (or no therapist), or they did not report their race/ethnicity. Clients initially completed an informed consent and then the outcome and process measures. For clients who were no longer in therapy, they were given measures wherein the items were adjusted to reflect the past tense. The participants from this study were drawn from previous published studies (Owen, Leach, Wampold, & Rodolfa, 2011; Owen, Quirk, et al., 2011; Owen, Tao, & Rodolfa, 2010; Owen, Wong, & Rodolfa, 2010), although the purpose of this study differs from those stud- ies. In the previous studies, the response rates ranged from 30% to 40% (37.4% for this study).
Data Analysis
Because clients were nested within therapist (i.e., therapists treated multiple clients), multilevel modeling (MLM) was used to account for the nonindependence of observations within therapists (Raudenbush & Bryk, 2002). MLM adjusts for the fact that clients of the same therapist have more similar outcomes than therapists of different therapists; that is, some therapists generally have better outcome than other therapists (Wampold & Serlin, 2000). There was a particular interested in the variability among therapists and the standard errors for that variability. However, maximum like- lihood estimates and their standard errors are biased with small to moderate number of therapists (Maas & Hox, 2005), making interval estimates of the parameters problematic (Draper, 2008). Additionally, maximum likelihood works best with balanced data (Raudenbush & Bryk, 2002). Due to the moderately small number
316 OWEN, IMEL, ADELSON, AND RODOLFA
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
of therapists and the unbalanced number of clients treated by each therapist, a Bayesian MLM was used. Because Bayesian models are not based on normality assumptions or asymptotic results (see Hamaker & Klugkist, 2011), this is recommended when the out- come variable is dichotomous (Draper, 2008), as in this study.
Two MLM models were conducted. In both models, the out- come variable was a dichotomous indicator of whether the client unilaterally terminated therapy or not. In the first model, client unilateral termination status was predicted by clients’ racial/ethnic status (1 � White, 0 � REM) at Level 1, therapists’ racial/ethnic status (1 � White, 0 � REM) at Level 2, and the cross-level interaction between client and therapist racial/ethnic status. This model is used to determine whether therapists vary in their clients’ termination status. Also, if REM clients were more likely to report that they ended therapy unilaterally as compared with White clients, then this would provide support for general mental health disparities. In the second model, clients’ racial/ethnic status was allowed to vary across therapists and included other predictors (alliance, psychological well-being, number of sessions, and pro- portion of REM to White clients for therapists’ caseload). If the random slope for clients’ racial/ethnic status is significant, it would provide support for the therapist-specific racial/ethnic disparity hypothesis.
Results
Preliminary Analyses
Descriptive statistics for the variables are provided in Table 1. Prior to conducting the main analyses, we first tested whether clients who unilaterally terminated had lower alliance and psycho- logical well-being scores as compared with clients who ended therapy in other ways. To do so, we conducted two MLMs, wherein WAI and SOS-10 scores were the dependent variables, respectively, and termination status was the only predictor variable at Level 1 (unilateral termination � 1, other termination � 0). The results showed that clients who unilaterally terminated had lower alliance scores (B � �0.55; 95% credible interval [CI] � �0.83, �0.24; p � .001; d � �0.46) and lower psychological well-being
(B � �0.40; 95% credible interval � �0.67, �0.13; p � .01; d � �0.36), compared with clients who ended therapy in alternative ways. Effect sizes were based on Cohen’sd, wherein a small-sized effect � 0.20, medium-sized effect � 0.50, and a large-sized effect � 0.80 (note that effect sizes were based on the coefficient divided by the standard deviation for the WAI and SOS-10, re- spectively). These results correspond to previous studies that have demonstrated that clients who reported ending therapy unilaterally also reported lower alliances and were more psychologically dis- tressed as compared with clients who terminated therapy in other ways (e.g., Owen et al., 2009; Sharf et al., 2010).
Primary Analyses
We tested our first hypothesis—that therapists would differ in their clients’ termination status— by examining the proportion of variance in unilateral termination status accounted for by therapists. The results demonstrated that the mode of the posterior distribution for the variance in the intercept in Model 1 was statistically significant (�therapists
2 was 0.26, 95% CI [0.05, 0.78]; see Table 2, random effects in Model 1). This finding suggests that clients’ termination status varied across therapists insofar that therapists accounted for 7.3% of variance in their clients’ termination status (supporting Hypothesis 1). The intraclass correlation for unilateral termination status was calcu- lated by: �therapists
2 /(�therapists 2 � �3) (Goldstein, Browne, & Rasbash,
2002). Additionally, as seen in the first model, REM clients were significantly more likely to unilaterally terminate as compared with White clients (�10 � �0.37, 95% CI [�0.70, �0.08]), which sup- ports our second hypothesis. Therapists’ racial/ethnic status and the interaction between clients’ and therapists’ racial/ethnic status were not statistically significant.
Next, we tested whether the association between clients’ race/ ethnicity and unilateral termination status varied across therapists (Hypothesis 3). We replicated Model 1, but we also controlled for number of sessions (Level 1, grand-mean centered), psychological well-being (Level 1, grand-mean centered), working alliance (Level 1, grand-mean centered), and proportion of REM to White clients for each therapist (Level 2, grand-mean centered). The association between clients’ race/ethnicity and unilateral termina- tion was allowed to vary across therapists, which will provide a test of our hypothesis. As seen in Model 2, the mode of the posterior distribution for the variance in clients’ racial/ethnic status was statistically significant, (�RE
2 � 0.07, 95% CI [0.02, 0.66]), supporting our third hypothesis. In other words, some therapists were more likely to have their REM clients report unilateral termination as compared with their White clients (and vice versa). Additionally, number of sessions was significantly related to uni- lateral termination, with clients who attended fewer sessions were more likely to unilaterally terminate. Also, clients who reported stronger alliances were less likely to unilaterally terminate.
For illustration, we randomly selected 10 therapists who treated at least four REM and four White clients each to illustrate the differences across therapists in their REM and White clients’ unilateral termination (see Figure 1). Therapist 2 had a similar proportion of White and REM clients unilaterally terminated (ap- proximately 25%), whereas Therapist 7 had approximately 75% and 80% of his or her REM and White clients unilaterally termi- nated, respectively. For Therapist 5, 75% of his or her REM clients unilaterally terminated and approximately 25% of his of her White
Table 1 Descriptive Information for Clients’ Unilateral Termination, Psychological Well-Being, Working Alliance, and Number of Sessions
Variable
Unilateral term Other term
N (%) N (%)
Client: REM 57 (36.8%) 98 (63.2%) Client: White 44 (24.9%) 133 (75.1%)
M (SD) M (SD)
SOS-10 4.94 (1.27) 5.36 (1.13) WAI 5.05 (1.38) 5.63 (1.22) Number of sessions 5.26 (4.71) 7.40 (5.69)
Note. Ns � 101 (Unilateral term.), and 231 (Other term.). term. � termination; REM � racial/ethnic minority; SOS-10 � Schwartz Outcome Scale-10 (possible range � 1–7); WAI � Working Alliance Inventory (possible range � 1–7).
317TERMINATION STATUS AND RACIAL/ETHNIC DISPARITIES
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
clients unilaterally terminated. In contrast, Therapist 10 had 20% of his or her REM clients unilaterally terminated and none of his or her White clients unilaterally terminated.
Discussion
Consistent with the therapist effect literature examining therapy outcomes (e.g., Baldwin & Imel, 2011; Wampold & Brown, 2005), we found that therapists accounted for approximately 7% of the variance in clients’ unilateral termination status. To our knowl- edge, the present study is the first in which therapist variability in clients’ unilaterally termination status or client dropout of any kind has been examined. This finding provides support for the notion that therapists difference in their general competence or ability, insofar that some therapists were more likely to have clients unilaterally terminate as compared with other therapists. Addition- ally, we found that REM clients were more likely to unilaterally terminate as compared with White clients, suggesting a mental health disparity, which is consistent with prior research (Wierz- bicki & Pekarik, 1993).
In addition, racial/ethnic disparities varied across therapists. That is, some therapists were more likely to have their REM clients report that they unilaterally terminated as compared with their White clients (and vice versa). Yet for some therapists, the rate of unilateral termination was consistent across their REM and White clients. These differences were evident even after accounting for other factors, such as clients’ psychological well-being, number of sessions, and working alliance. This variation across therapists supports the notion that therapists have an important role in racial/ ethnic disparities in unilateral termination and is consistent with Imel et al.’s (2011) findings on therapy outcome.
Therapist-specific racial/ethnic disparities should be framed within some absolute standard for effectiveness for unilateral ter- mination (e.g., less than 30% of clients). For instance, as seen in Figure 1, Therapist 7 had little difference in the percentage of his or her REM and White clients’ unilateral termination; however, this therapist had high rates of unilateral termination for both REM
and White clients (greater than 75%). Thus, examining racial/ ethnic disparities based solely on the discrepancy between REM and White clients in unilateral termination ignores the fact that some therapists may be generally ineffective at retaining their clients; such therapists would not be optimal for clients of any racial and ethnic background.
Although we cannot be sure as to why some therapists had higher disparities in unilateral termination, explanations for racial/ ethnic mental health disparities have typically focused on thera- pists’ cultural competence (Whaley & Davis, 2007). Lopez, Ko- pelowicz, and Canive (2002) noted that one core aspect of cultural competence is therapists’ ability to move between different cul- tural frameworks. Thus, some therapists may be more or less proficient at navigating the process of flexing between different cultural perspectives. However, if cultural competence is defined based on actual success with clients, therapists who have low retention rates for REM clients may be deemed culturally compe- tent in their work with REM clients by definition (e.g., Therapist 4 in Figure 1) and vice versa for therapists who have low retention rates for White clients (e.g., Therapist 10 in Figure 1).
However, as we did not have a measure of therapists’ cultural competence, therapist-specific racial/ethnic disparities in unilateral termination may be related to other therapeutic processes or ther- apist behaviors. For instance, clients in this study were not ran- domly assigned to therapists, thus some therapists’ caseloads may be easier to retain. Future research with random assignment of clients to therapists along with direct assessment of therapists’ cultural competence and clients’ treatment expectations/ preferences can help explore these possibilities.
Limitations and Implications
The present study adds to the racial/ethnic mental health dis- parities literature but is not without limitations. Our study is open to response bias as the survey response was approximately 37%. Thus, we do not know whether clients who did not respond differed significantly from those who participated (e.g., more likely to unilaterally terminate); however, our rate of unilateral termination is similar to previous studies (Hatchett & Parks, 2003). Additionally, clients who responded for any given therapist were not randomly selected. We can only assume that the types of client
Figure 1. Percentage of unilateral termination of White and REM clients for 10 selected therapists. REM � racial/ethnic minority.
Table 2 Summary of Multilevel Models With Bayesian Estimation
Variable Model 1:
Coeff [95% CI] Model 2:
Coeff [95% CI]
Fixed effects Intercept 0.31 [�0.13, 0.90] 0.08 [�0.68, 0.85] Client RE �0.37 [�0.70, �0.08] �0.48 [�1.10, 0.01] Therapist RE �0.37 [�1.34, 0.49] �0.17 [�0.69, 0.44] Therapist RE �
Clt RE 0.48 [�0.77, 1.97] �0.12 [�0.50, 0.57] Well-being — �0.11 [�0.27, 0.02] Sessions — �0.05 [�0.09, �0.03] Alliance — �0.16 [�0.30, �0.02] Proportion RE — 0.33 [�1.36, 2.19]
Random effects Intercept 0.26 [0.05, 0.78] 0.17 [0.01, 0.55] Race/ethnicity — 0.07 [0.02, 0.66]
Note. Coefficients are the mode of the posterior distribution. 95% CIs that do not include 0 in the range are statistically significant. CI � credible interval; RE � Race/ethnicity; Clt � Client. Dashes represent variables that were not included in the model.
318 OWEN, IMEL, ADELSON, AND RODOLFA
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
responses per therapist were consistent (i.e., the likelihood clients who responded that they unilaterally terminated were consistent across therapists). We also decided to use a client-defined termi- nation status as compared with other metrics for dropout. How- ever, in support of our client-defined termination status, we repli- cated results from previous studies on therapy dropout in many ways, such as prevalence rates and associated outcomes (e.g., lower alliance and well-being scores; Daughters et al., 2005; Lester et al., 2010; Owen et al., 2009; Wierzbicki & Pekarik, 1993). We also did not have information regarding clients’ pre- therapy functioning, which limited our ability to determine whether the clients who reported that they unilaterally terminated did so without making some gains in therapy. Although we had clients’ current psychological well-being scores, there could be other posttherapy factors that might have influenced these scores. We were not able to discern racial/ethnic differences within the REM group (e.g., differences in unilateral termination between African American clients vs. Asian American clients) due to limited sample sizes for some races/ethnicities. Future studies may want to examine therapists and clients’ racial/ethnic identity as it relates to unilateral termination. Lastly, although we had a large number of therapists, the number of REM and White therapists was relatively small. This might have hampered our ability to detect differences based on racial/ethnic similarities or differences (e.g., racial/ethnic matching). Yet, such tests have not revealed significant differences in past studies (see Maramba & Nagayama Hall, 2002).
Notwithstanding these limitations, our study has important im- plications for understanding racial/ethnic mental health dispari- ties—specifically, we provided evidence of therapist-specific ra- cial/ethnic disparities. Defining therapists’ competency based, in part, on their clients’ unilateral termination status could be infor- mative for current psychotherapy competency models that aim to assist the training and continual development of psychologists (e.g., Rodolfa et al., 2005; Wise et al., 2010). That is, therapists vary in their ability to retain their clients on the basis of clients’ racial/ethnic status, which may be an important indicator of their cultural competence. Consequently, therapists may want to track their clients’ rates of unilateral termination in general and, specif- ically, based on their clients’ racial/ethnic status. This information may provide useful feedback for therapists for their development.
References
American Psychological Association. (2003). Guidelines on multicultural education, training, research, practice, and organizational change for psychologists. American Psychologist, 58, 377– 402. doi:10.1037/0003- 066X.58.5.377
Arredondo, P., & Toporek, R. (2004). Multicultural counseling competen- cies � ethical practice. Journal of Mental Health Counseling, 26, 44 –55.
Baldwin, S. A., Berkeljon, A., Atkins, D. C., Olsen, J. A., & Nielsen, S. L. (2009). Rates of change in naturalistic psychotherapy: Contrasting does- effect and good-enough level models of change. Journal of Consulting and Clinical Psychology, 77, 203–211.
Baldwin, S. A., & Imel, Z. E. (2011). Therapist effects: Findings and methods. Manuscript in preparation.
Blais, M. A., Lenderking, W. R., Baer, L., deLorell, A., Peets, K., Leahy, L., & Burns, C. (1999). Development and initial validation of a brief mental health outcome measure. Journal of Personality Assessment, 73, 359 –373. doi:10.1207/S15327752JPA7303_5
Breaux, C., & Ryujin, D. H. (1999). Use of mental health services by ethnically diverse groups within the United States. Clinical Psycholo- gist, 52, 4 –15.
Callahan, J. L., Aubuchon-Endsley, N., Borja, S. E., & Swift, J. K. (2009). Pretreatment expectancies and premature termination in a training clinic environment. Training and Education in Professional Psychology, 3, 111–119. doi:10.1037/a0012901
Clarkin, J. F., & Levy, K. N. (2004). The influence of client variables on psychotherapy. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook of psychotherapy (pp. 194 –226). New York, NY: John Wiley and Sons.
Coleman, H. K. (1998). General and multicultural counseling competency: Apples and oranges? Journal of Multicultural Counseling and Develop- ment, 26, 147–156. doi:10.1002/j.2161-1912.1998.tb00194.x
Constantine, M. G. (2002). Predictors of satisfaction with counseling: Racial and ethnic minority clients’ attitudes toward counseling and ratings of their counselors’ general and multicultural counseling com- petence. Journal of Counseling Psychology, 49, 255–263. doi:10.1037/ 0022-0167.49.2.255
Constantine, M. G. (2007). Racial microaggressions against African Amer- ican clients in cross-racial counseling relationships. Journal of Counsel- ing Psychology, 54, 1–16. doi:10.1037/0022-0167.54.1.1
Daughters, S. B., Lejuez, C. W., Bornovalova, M. A., Kahler, C. W., Strong, D. R., & Brown, R. A. (2005). Distress tolerance as a predictor of early treatment dropout in a residential substance abuse treatment facility. Journal of Abnormal Psychology, 114, 729 –734. doi:10.1037/ 0021-843X.114.4.729
Draper, D. (2008). Bayesian multilevel analysis and MCMC. In J. de Leeuw & E. Meijer (Eds.), Handbook of multilevel analysis (pp. 77– 139). New York, NY: Springer. doi:10.1007/978-0-387-73186-5_2
Garb, H. N. (1998). Studying the clinician: Judgment research and psy- chological assessment. Washington, DC: American Psychological As- sociation. doi:10.1037/10299-000
Goldstein, H., Browne, W., & Rasbash, J. (2002). Partitioning variation in generalised linear multilevel models. Understanding Statistics, 1, 223– 232.
Hamaker, E. L., & Klugkist, I. (2011). Bayesian estimation of multilevel models. In J. J. Hox & J. K. Roberts (Eds.), Handbook of advanced multilevel analysis (pp. 137–161). New York, NY: Taylor & Francis Group.
Hamilton, S., Moore, A. M., Crane, R., & Payne, S. H. (2011). Psycho- therapy dropouts: Differences by modality, license, and DSM-IV diag- nosis. Journal of Marital & Family Therapy, 37, 333–343.
Hannan, C., Lambert, M. J., Harmon, C., Nielsen, S. L., Smart, D. W., Shimokawa, K., & Sutton, S. W. (2005). A lab test and algorithms for identifying clients at risk for treatment failure. Journal of Clinical Psychology, In Session, 61, 155–163.
Hatcher, R. L., & Barends, A. W. (2006). How a return to theory could help alliance research. Psychotherapy, 43, 292–299.
Hatcher, R. L., & Gillaspy, J. A. (2006). Development and validation of a revised short version of the Working Alliance Inventory. Psychotherapy Research, 16, 12–25. doi:10.1080/10503300500352500
Hatchett, G. T., & Parks, H. L. (2003). Comparison of four operational definitions of premature termination. Psychotherapy, 40, 226 –231. doi: 10.1037/0033-3204.40.3.226
Hilsenroth, M., Ackerman, S., & Blagys, M. (2001). Evaluating the phase model of change during short-term psychodynamic psychotherapy. Psy- chotherapy Research, 11, 29 – 47. doi:10.1080/713663851
Horvath, A. O., Del Re, A. C., Flückiger, C., & Symonds, D. (2011). Alliance in individual psychotherapy. Psychotherapy, 48, 9 –16. doi: 10.1037/a0022186
Imel, Z. E., Baldwin, S., Atkins, D. C., Owen, J., Baardseth, T., & Wampold, B. E. (2011). Racial/ethnic disparities in therapist effective- ness: A conceptualization and initial study of cultural competence. Journal of Counseling Psychology, 58, 290 –298. doi:10.1037/a0023284
319TERMINATION STATUS AND RACIAL/ETHNIC DISPARITIES
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
Institute of Medicine. (2002). Unequal treatment: Confronting racial and ethnic disparities in health care. Washington, DC: National Academy of Science.
Kales, H., Blow, F., Bingham, C., Copeland, L., & Mellow, A. (2000). Race and inpatient psychiatric diagnoses among elderly veterans. Psy- chiatric Services, 51, 795– 800. doi:10.1176/appi.ps.51.6.795
Leong, F. T. L., Wagner, N. S., & Tata, S. P. (1995). Racial and ethnic variations in help-seeking attitudes. In J. G. Ponterotto, M . J. Casas, L. A. Suzuki, & C. M. Alexander (Eds.), Handbook of multicultural counseling (pp. 415– 438). Thousand Oaks, CA: Sage Publications.
Lester, K., Resick, P. A., Young-Xu, Y., & Artz, C. (2010). Impact of race on early treatment termination and outcomes in posttraumatic stress disorder treatment. Journal of Consulting and Clinical Psychology, 78, 480 – 489. doi:10.1037/a0019551
Lopez, S. R., Kopelowicz, A., & Canive, J. M. (2002). Strategies in developing culturally congruent family interventions for schizophrenia: The case of Hispanics. In H. P. Lefley & D. L. Johnson (Eds.), Family interventions in mental illness: International perspectives (pp. 61–90). Westport, CT: Praeger.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 1, 85–91.
Maramba, G., & Nagayama Hall, G. (2002). Meta-analyses of ethnic match as a predictor of dropout, utilization, and level of functioning. Cultural Diversity and Ethnic Minority Psychology, 8, 290 –297. doi:10.1037/ 1099-9809.8.3.290
Masson, P. C., Perlman, C. M., Ross, S. A., & Gates, A. L. (2007). Premature termination of treatment in an inpatient eating disorder pro- gramme. European Eating Disorder Review, 15, 275–282. doi:10.1002/ erv.762
Owen, J. (in press). Systemic alliance in individual therapy: Factor analysis of the ITAS-SF and the relationship with therapy outcomes and termi- nation status. Journal of Marital and Family Therapy.
Owen, J. J., & Imel, Z. (2010). Utilizing rating scales in psychotherapy practice: Rationale and practical applications. In L. Baer & M. Blais (Eds.), Handbook of clinical rating scales and assessment in psychiatry and mental health (pp. 257–270). New York, NY: Humana Press.
Owen, J., Imel, Z., Tao, K., Wampold, B., Smith, A., & Rodolfa, E. (2011). Cultural ruptures in short-term therapy: Working alliance as a mediator between clients’ perceptions of microaggressions and therapy outcomes. Counselling and Psychotherapy Research, 11, 204 –212 doi:10.1080/ 14733145.2010.491551
Owen, J., Leach, M. M., Wampold, B., & Rodolfa, E. (2011). Client and therapist variability in clients’ perceptions of their therapists’ multicul- tural competencies. Journal of Counseling Psychology, 58, 1–9. doi: 10.1037/a0021496
Owen, J., Quirk, K., Hilsenroth, M. J., & Rodolfa, E. R. (2011). Working through: In-session processes that promote between-session thoughts and activities. Journal of Counseling Psychology. Advanced online publication. doi:10.1037/a0023616
Owen, J., Smith, A. A., & Rodolfa, E. (2009). Clients’ expected number of sessions, treatment effectiveness, and termination status: Using empiri- cal evidence to inform session limit policies. Journal of College Student Psychotherapy, 23, 118 –134. doi:10.1080/87568220902743660
Owen, J., Tao, K., & Rodolfa, E. (2010). Microaggressions against women in short-term psychotherapy: Initial evidence. The Counseling Psychol- ogist, 38, 923–946. doi:10.1177/0011000010376093
Owen, J., Wong, J. Y., & Rodolfa, E. (2010). Clients’ perceptions of helpful therapeutic processes: The relationship between clients’ confor- mity to masculine norms and common factors. Journal of Counseling Psychology, 57, 68 –78. doi:10.1037/a0017870
Pekarik, G. (1992). Relationship of clients reasons for dropping out of treatment to outcome and satisfaction. Journal of Clinical Psychology, 48, 91–98. doi: 10.1002/1097-4679(199201)48:1�91::AID-JCLP2270480113 3.0.CO;2-W
Quintana, S. M. (2007). Racial and ethnic identity: Developmental per- spectives and research. Journal of Counseling Psychology, 54, 259 –270.
Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models (2nd ed.). Newbury Park, CA: Sage.
Rodolfa, E., Bent, R., Eisman, E., Nelson, P., Rehm, L., & Ritchie, P. (2005). A cube model for competency development: Implications for psychology educators and regulators. Professional Psychology: Re- search and Practice, 36, 347–354.
Sharf, J., Primavera, L. H., & Diener, M. J. (2010). Dropout and thera- peutic alliance: A meta-analysis of adult individual psychotherapy. Psy- chotherapy, 47, 637– 645. doi:10.1037/a0021175
Simon, G. E., & Ludman, E. J. (2010). Predictors of early dropout from psychotherapy for depression in community practice. Psychiatry Ser- vice, 61, 684 – 689. doi:10.1176/appi.ps.61.7.684
U.S. Surgeon General. (2001). Mental health: Culture, race, and ethnicity. A supplement to mental health: A report of the Surgeon General. Rockville, MD: U.S. Department of Health Human Services.
Vandereycken, W., & Vansteenkiste, M. (2009). Let eating disorder pa- tients decide: Providing choice may reduce early drop-out from inpatient treatment. European Eating Disorders Review, 17, 177–183. doi: 10.1002/erv.917
van Ryn, M. (2002). Research on the provider contribution to race/ethnicity disparities in medical care. Medical Care, 40, I-140 –I-151. doi:10.1097/ 00005650-200201001-00015
Wampold, B. E., & Brown, G. S. (2005). Estimating variability in out- comes attributable to therapists: A naturalistic study of outcomes in managed care. Journal of Consulting and Clinical Psychology, 73, 914 –923. doi:10.1037/0022-006X.73.5.914
Wampold, B. E., & Serlin, R. C. (2000). The consequence of ignoring a nested factor on measures of effect size in analysis of variance. Psycho- logical Methods, 5, 425– 433. doi:10.1037/1082-989X.5.4.425
Whaley, A. L., & Davis, K. E. (2007). Cultural competence and evidence- based practice in mental health services. American Psychologist, 62, 563–574.
Wierzbicki, M., & Pekarik, G. (1993). A meta-analysis of psychotherapy dropout. Professional Psychology: Research and Practice, 24, 190 –195. doi:10.1037/0735-7028.24.2.190
Wise, E. H., Sturm, C. A., Nutt, R. L., Rodolfa, E., Shaffer, J. B., & Webb, C. (2010). Life-long learning for psychologists: Current status and a vision for the future. Professional Psychology: Research and Practice, 41, 288 –297.
Young, J. L., Waehler, C. A., Laux, J. M., McDaniel, P. S., & Hilsenroth, M. J. (2003). Four studies extending the utility of the Schwartz outcome scale (SOS-10). Journal of Personality Assessment, 80, 130 –138. doi: 10.1207/S15327752JPA8002_02
Received September 12, 2011 Revision received December 12, 2011
Accepted December 13, 2011 �
320 OWEN, IMEL, ADELSON, AND RODOLFA
T hi
s do
cu m
en t i
s co
py ri
gh te
d by
th e
A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. T
hi s
ar tic
le is
in te
nd ed
s ol
el y
fo r t
he p
er so
na l u
se o
f t he
in di
vi du
al u
se r a
nd is
n ot
to b
e di
ss em
in at
ed b
ro ad
ly .
