second therapist for the next patient. It should be noted that patients were offered ...... although they are only partly similar. Volun- tary second opinion plans have.
On What Bases Their Therapists?
Do Patients
B.
Pii.D.
LESLIE
ALEXANDER,
P.
JACQUES
LESTER
Piu.D.
nity
to have therapists
ongoing
two sessions and
then
treatment.
whom
they
better,
graphic tions
were given
outpatients
ent
liked
not
similarity
reported
the o’/iportu-
with
as more
helpful
on the basis
of the patients
for
and
of demoexpecta-
qualities.
the opportunity reasons
differ-
the thera-
or pretreatment
liking
two
one for chose
the therapist’s
definite 75%
each choose
Patients viewed
but
about
stated
M.D.
AtlIBAchI.
Forty-four
ever,
Ph.D.
CRITS-CIIRIST0PI1,
ARTIJUR
pist
Pt-iD.
BARBER.
Ltl3oIsKY,
PAtTI,
Choose
Patients
to choose their chose
therapist.
Possible
explanations
ponderant
choice
of the second
and
A
choosing largely
search remains,
on the basis even though
own
therapist
are
discussed.
their anecdotal.
many private and among therapist.”2 The aim
of
choosing patients
therapists have reThis paucity of refor patients’ choices potential patients are
patients therapists
of the
seeking seeking
present
study
a thertheir
was
to de-
would view as own therapist. that they liked
better Would
more that
or that they they choose
similarities the process
Received cepted
June
8,
helpful? they per-
Pennsylvania reprint
or complementarities and outcome of psycho-
1992;
December
of Social Work, vania, and the
Social
felt were therapists
as similar to themselves? There is a history in psychotherapy research of to determine which, if any, patient-
therapist facilitate
10,
revised 1992.
Bryn Mawr Department School
requests
AND
December the
3, 1992; Graduate
College, Bryn of Psychiatry,
Alexander,
Work,
PRACTICE
From
of Medicine,
to Dr.
Bryn Mawr Mawr, PA 19010. Copyright © 1993
OF PSYCHOTHERAPY
indications use in criteria
termine what criteria patients important in choosing their Would they choose therapists
ceived long trying
JOURNAL
some
encouraged in the popular press and news media to shop for a therapist.3 Such “therapist shopping” is believed to be common
for
the pre-
are
use in mained
among apist4
How-
there
the criteria therapists their own therapists,”2 the
the second
choices.
lthough
PennsylUniversity of
Mawr,
Philadelphia. Graduate
College,
300 Airdale
American
Psychiatric
RESEARCH
ac-
School
Address School Road, Press,
of Bryn Inc.
136
oi’ THERAPIST
ChOIcE
therapy. A wide range pist matching variables
of these patient-therahave been used,
cluding
demographic,
personality matching
variables!7 literature has
findings-with variable
cognitive, Even yielded
no single
and
though the inconsistent
important
emerging-matching
evident in many patient-therapist lems in research,
in-
patients ers had
matching effects
remain
aged
terviews
that
may
prove
fruitful,
and
would
have
similar
not
patients
the
match
with
hunch intake
certain early
We therefore proposed that approach to matching, where choose between two therapists
in in-
patients
we reasoned
that
provide new matching beneficial to
patients’ treatment. There was the hope as well that findings about this approach to matching might have some practical utility for patients seeking a therapist. In tients
the
Penn
whose
with them patients
Psychotherapy
therapists
had more assigned
Study,9
had
chosen
pa-
of the
also that
consistent exercising
helping with choice
the
patient facilitate
alliance.
proposal
the popular is beneficial
and
it
VOLUME
ideal
therapist, as
understanding,
a therapist.
by their therapists
it outwere
and
Patients
desir-
were
rated
chosen therapists than on treatment desirability
attraction,
suggesting
and liking. A definite order effect favoring the last therapist. chose the therapist, though
is
conviction to overall
analogue
choices, and these choices
Chosen therapists were rated signifihigher on the following factors: perattraction, likability, possession of qualities, competence, match with
personal
the opportuthe forma-
This
unambiguous bases on which
by the and
a mutuality
of
choice
tients a chance to choose their therapists was beneficial to the 1,12 Most of the earlier data were limited, however, in that nonpopulations
the higher other
well-being’#{176} and that shopping for a therapist should be encouraged. At the time the study was started, there was some empirical evidence that giving pa-
clinical
made. cantly sonal admired
favorable outcomes than to therapists randomly.
ment to its formation. In the present study was proposed that the counterpart of random
tion
and made lined the
ability
to therapists with the helpas an impedi-
et al.’8 gave 194 outto have three halfdifferent therapists
and then choose the one they wished to continue with in ongoing therapy. This study revealed that patients liked having a choice
to work
Random assignment of patients was also negatively correlated ing alliance, probably serving
assignment-giving nity to choose-would
on the basis of secondhand information (actual videotapes of the therapists with patients or intake coordinators’ descriptions of the therapists). Hollander-Goldfein patients the opportunity hour sessions with three
hunches.
a more direct patients could after two ther-
apy sessions with each, would insights about the patient-therapist process and would also be
found that in some instances choosmore favorable outcomes on the basis
of two therapy sessions,’5 were more likely to show up for the first treatment session,’6 and liked the opportunity to choose.’3”7 But in these last three studies patients made choices
studies.8 Likewise, although matching may pose probit is often quite well man-
in clinical practice.8 A therapist may have a strong pretherapy telephone or in-person
were the norm.’3 Studies also varied in terms of empirical rigor and often resulted in nonsignificant differences between choosers and nonchoosers.’4 Nevertheless, studies using
studies
2 #{149} NUMBER
third and
these
evenly
results
were
significant,
among
were were
interview
first Al-
further
revealed that this of the frequency
individual therapists and less chosen therapists distributed
detected, percent
therapist, 32% chose the 22% chose the second.
case-by-case analysis effect was independent which More
was also Forty-six
order with
chosen. roughly positions.
Further, if one therapist in the triad was clearly preferred over the other two, he or she was chosen irrespective of position. If all three able,
therapists were rated as equally preferthe therapist seen last was most fre-
quently
2 #{149} SPRING
chosen.
1993
ALEXANDER
STUDY
The
137
ET14L.
be a model,
HYPOTHESES
main
study
hypotheses
and
were:
completed 1.Patients
would
choose
with whom they helping alliance 2.
3.
the
tionship. Patients would choose their on the basis of demographic and pretreatment expectations
whom
they
minded their
qualities. choose
liked
them
and
of an admired
rein
structs are reliable predictors of favorable outcomes: 1) the therapeutic alIiance;’2’ 2) matching between patients and therapists on
apist;9 therapist
and
ature from
3) (see
a review).
and
expectations
A more
detailed
on predictors Dr. Alexander. M
of
T
review
II
0
2.
ther-
his or her et al.’8 for of the
of outcome
l’
the
liter-
That
parallel completed sures
1) S
each
4) for
1.
that patients.
administered therapist
each
patient.
after (i.e.,
after
The
following
measures
used: Apfelbaum’s
Q-Sort.22
This
60-item
questionnaire taps three types of patient expectations about the qualities of the therapist: that the therapist will give nurturance, that the therapist will
JOURNAl.
OF
PSYCHOTHERAPY
is, each
3.
pre-
subject
had
for Therapist 2.
two 1 and
of children,
religion,
and
activity. These were checked
defor
each similar characterispoint to the index. This outcome in the Penn Study.9 (Indicators of
style and institutional affiliain the Penn Psychotherapy
were
omitted
in this
study;
gen-
der and race were added.) The Mutual Attraction Questionnaire.
forms
2 and
the
item rattherapist
of religious variables
two
sessions
between
amount mographic
This
sessions
Q.Sort
number
therapists These mea-
the
The
the two seswith the pa-
Demographic Similarity. The developed to summarize the similarity between patherapist.9 Items included gender, race, marital status,
cognitive tion used
measures having to do of therapists and the
of measures about the
were
with were
set
expected
Index of index was demographic tient and were age,
Study on the choice
correlation
similarity, with tic adding one index predicted Psychotherapy
is available
M ea s tires We report with patients’
to be like.
correlations, one one for Therapist
on the basis
about
patient liking Hollander-Goldfein
will
treat-
they
treatment and the postsession ings for each subject for each pairing.
characteristics
before
what
therapist
a Pearson
person
Hypotheses 1, 2, and 3 were based on repeated findings that the following con-
of pretreatment
therapist
tients rating the items on the basis of their two-session actual experience with each therapist. An index of similarity between pretreatment and postsession sort items was derived by computing
therapist who
Q-Sort
indicating
their
past.
demographic
the
was then repeated after sions with each therapist,
therapist similarity about
the
best
the
ment,
therapist
developed a stronger and therapeutic rela-
the therapist’s Patients would
that
be a critic. We used an abbreviated version containing 47 items. The patients
questionnaire,
oped
for
this
used
bility23),
study for
has
which
was devel-
(adapted
from
teacher-student
a patient
form
compati(20
items)
and a therapist form (13 items) and was used to assess whether patients and therapists liked each other. Using a 5point scale format (1 = not at all to 5 = very much), both patients and therapists rated the degree to which they saw each
other
tive,
interesting,
PRACTICE
AND
as likable,
RESEARCH
physically
understandable,
attraceasy
138
CII0IcE
to talk ing
to, similar
them
liked
or disliked,
with. The Penn
4.
to themselves,
of someone and
Helping
naire.24’25 measures
and The
assessor, who provided study protocol and
they
to work
Alliance
This 11-item the helping
formed
Question-
and comprises of the therapist understanding Relationship
This questionnaire26’27 to assess Rogers’s
consent
participate. As in
questionnaire alliance between
change. Barrett-Lennard
ventory. veloped
past
desirable
patient and therapist24 two factors: helpfulness and attainment of new 5.
remind-
in their
who
first
with
initiated
patients
deof
this procedure was discontinued tients had begun the protocol. made
each
a choice
therapist
and
before
after 14 paOf the 14, 8
6 dropped
out
they
and
is completed
therapists. have
by
Gurman shown that
the patients’ early perceptions to 5 sessions) of the therapeutic tionship are related to positive following a course of treatment.
(after relachange
saw for two sessions
therapist At the
When
patients
pointment at the at the University worker worker)
telephoned
(an experienced presented brief
study
criteria:
to patients were
not
for
who
in need
met
program was described work with two different each
and
then
the
study
during
telephone
At that
this to before time,
the they
emer-
impaired; and had no The special
to choose
with in expressed
come
following
as an opportunity therapists for
to continue Patients who
asked to appointment
social about
of immediate
wished ment.
ap-
clinic intake
clinical information
gency care; were not organically were not clearly schizophrenic; significant cognitive impairment.
sessions
an
outpatient psychiatric of Pennsylvania, the
to two
the one ongoing interest
they treatin the
contact
were
clinic for a lfhour their first therapy sesmet
with
the
VOLUME
paired, be the second should
#{149} NUMBER
were
trans-
whom they also end of the fourth
previously,
therapists
were
alternating so that one of them would first therapist for one patient and the therapist for the next patient. It be noted that patients were offered a
limited
choice,
available specifically
i.e.,
therapists. wanted
neither
of the
patients were the study and
ticipate. because
a choice
between
two
For example, if patients a female therapist and
female, entered
available
therapist told could
pair
was
a
this before they elect not to par-
No patients declined participation of the gender or any other back-
ground characteristic There were 20 ings;
the
any
pairing
average
of the different
number
was
2. One
therapists. therapist
of patients therapist
pairseen
pair
saw
by 6
patients; 1 therapist pair saw 4 patients; 2 therapist pairs saw 3 patients each; 12 therapist pairs saw 2 patients each; and 4 therapist pairs saw 1 patient each.
clinical
2
then
patients whom
session, the patient chose his or her therapist for ongoing treatment. This modified procedure was favorably received by patients, resulting in fewer dropouts. As mentioned
Procedure
before
Patients and therapists alike six sessions before making a
ferred to a second saw for two sessions.
2
three
choosing,
Thereafter to a therapist
and
al.’8
had
was
choice was too many. were randomly assigned
congruence)
et
study sessions
underof regard,
to
would be choice if with two lasting when the
of positive standing,
regard, empathic unconditionality
wished
Hollander-Goldfein
In-
was concept
the in-
study, we reasoned that patients better able to make an informed they had actual therapy experience therapists and had an experience more than one session. Although
making a choice. complained that
both patients and and Barrett-Lennard27
sion.
those
the necessary conditions of therapy and the therapeutic relationship. The inventory assesses four variables (level
and
the
more detail about obtained written
from
the
OF THERAPIST
The
2
#{149} SPRING
same
1993
assessor
met
with
every
patient
ALEXANDER
139
ETAL.
after sessions nation. In
2 and addition
4 and to
quantitative
measures
at treatment administering
related
termithe
to choice,
Pennsylvania. in the study
the
assessor asked patients how things were going
after sessions 2 and 4 with the just-seen ther-
apist
at the
and
asked
them
end
of session
4
why they chose a particular therapist. At treatment termination, patients were further asked to reflect on the opportunity to choose. Therapists completed sures related to choice of each
patient
a parallel set after sessions
variable,
position.
therapists
could
a mean
E S IT L
S’rtDY
followed
(SD
=
30.2).
Po11LATI0N
of children, of religious (77%) and Therapists
30.8
patients
Patients
somewhat
and
amount patients were female. white (25;
fewer
Nineteen divorced,
patients
therapists separated
and 6 had never married. the majority of patients married; 9 were divorced,
On the other (64%) had separated, or
dropped from subsequent analyses because of organicity, which was attributed to chronic alcoholism. Based on a clinical interview by
(67%) 1 had
were M.D.s, a combined
8 (30%) M.D.-Ph.D.
patients
an experienced tients who made
training college
clinical a choice
assessor, received
diagnoses:
major
(15), adjustment disorders disorders (9), and anxiety Twenty-five percent of these some type of prior therapy, cases
it was
limited
to no
the the
was
44 pafollow-
depression
(13), personality disorders (7). patients had had but in almost all more
than
one
(21%)
6 were
were or wid-
As expected, education higher for therapists:
made
only
were
the study. were much
patients who in treatment
and
and
characnumber
entered levels
DSM-III-R
made
± 3.8 years.
widowed;
but one of the and continued
who
did not differ sigaveraged 29.9 ± 9.6
education, religion, activity. Thirty-four 12 therapists (44%) were overwhelmingly
whereas
owed; hand, never
of Therapists
a
ing
45
if
a
choice, choice
were
pairing, pairing
differed on the following gender, race, marital status,
white (34; 77%). married; 2 were
L S
Patients There
and therapists patients
therapists
therapists teristics:
open-ended psyThe number of choice was quite
of 26.1
of each
to another
Characteristics Patients
93%),
R
At the end
therasecond
chose.
years, provided
each pairing, the first and
commit
Patients and nificantly in age:
psychotherapy
with
therapist
of mea2 and 4
pairing.
Luborsky’s3#{176} guidelines for chodynamic psychotherapy. treatment sessions following
volunteered to be to see a pair of
patients at a time. With pists alternated between
they
Treatment The
All therapists and committed
had
or a graduate graduates,
married
had
either
when
Ph.D.s, degree.
some
degree, 14 (32%)
they 19
and Nine
graduate
11(25%) were had some col-
lege, 1 had some high school, and 2 had completed some grade school. Finally, there were disparities in terms of religious preference
and
amount
of religious
activity.
Attrition
or
two sessions. Of the Therapists Therapists residents (n
=
were fourth-year psychiatric 19) or Ph.D. psychologists (n
8) in a postdoctoral tient
psychiatric
internship clinic
89 patients
nity to participate failed to come
of
in the the
JOURNAL
University
ment, =
outpaof
OF PSYCHOTHERAPY
than
a no-show the
rate
estimated
at the time (25%). than the assessment
PRACTICE
who
in the for the
AND RESEARCH
had
the
study, only first clinic
that no-show
was
opportu18 (20%) appoint-
somewhat
rate
This rate was of scattered
at the
less clinic
also better studies of
140
CHOICE
clinic refusers reviewed revealed no-show rates
by Garfield,6 ranging from
which 24% in
two =
studies to over 40% in five studies. Of those who actually began the study (N 71), 24 (27%) discontinued before actually
making sessions. viewed
a choice-i.e., had less than four (Five of these 24 patients were by their therapists as too ill to con-
tinue in the study making a choice. the
protocol
and After
from
were removed learning more
the
clinical
before about
assessor,
elected not to participate for usual clinic services.)
Although problematic,
even figures
with, those
of 27%, compare
that data published on dropouts in the 1950s, indicating that 42.7% of patients had less than five treatment interviews, were still typi-
in general
also
that
higher
for
open-ended
are
dropout
met the criteria for a minimum of four
site
what
you’re
being
assigned
number
for making R
E
S
U
study inclusion sessions (the
and requi-
a choice).
Before about
we
the
attractiveness Program
For
the
present
the
three
main
qualitative of the
apist
gave
eral, greater
study
evidence program.
who
made
the
For
those
opportunity
it was useful
would
for someone
with
for
doing
so. In gen-
vs.
Stayers
returned
to the
all also reported so. Most simply
first
ther-
sensible felt more
Therapist 1. For example: like [about Therapist 2]-I him if I didn’t have anyone
with
her-uh, I’m not exactly I think I can talk more with her than with Dr. for what I don’t know; Ijust feel more comfort-
freely reason
-,
with her.” Several other swappers indicated tive experience with both, but one first therapist might cut down the
a choice,
that
to see react
(33)
or “‘Cause
treatment.
Another
per felt that my life have tant to me
the
75%
their improvement and and satisfaction in working
who
comfortable with “Nothing I didn’t could go on with
gave
individuals
session,
the second therapist to the first therapist. with the second ther-
therapist.
apist (swappers), reasons for doing
me-I out
me
just
Choice
reasons
Swappers
from things
ferent
fourth
second
them to choose their therapist was generally appealing. Some typical responses were: “It
in a program
not
able
Attractiveness
44 patients
of being
the
good
was
idea
for
they cited comfort
apist
the
Reasons
comfortable
EVIDENCE
addressing
into-you’re
anyone.”
elected to remain with rather than to go back All patients who stayed
sure why”;
hypotheses,
dif-
else”; “[Therapist 1] makes me feel comfortable and not embarrassed to talk with himwe work well”; or “I wasn’t quite as
U ‘1’ 5
QUALI’rA’rlvE
getting
treat-
offered in this study, than for preset termination date and Further analysis of dropouts be included in a subsequent we concentrate on the patients
article. Here
to see
rates
ment, which was treatment with a brief duration in our study will who had
a chance
question, the choice was valuable though extremely difficult to make”; and “I think it was real helpful. I think I did better; you know
with be noted
you
ferent styles”; “Well, I was glad that I had the choice”; “I think it was real helpful”; “It’s a very good idea-makes you feel like you have a little control over something”; “Without
Following
is certainly favorably
and are in fact lower on average than, cited by Garfield.6 Garfield concluded
cal today. It should
gives
2 ad-
ditional patients and were referred attrition, our
in therapy-it
OF THERAPIST
how
to me”;
allowed
two
dif-
“I think
no experience
VOLUME
2 #{149} NUMBER
2 #{149} SPRING
not
“eliciting
felt
that a lot
the
a posifelt the time in
second
ther-
of information
didn’t think she was like drawing of me particularly.” Another swap-
1993
“Since the dominant people in been women, it’s sort of importo be able to talk to a male”
ALEXANDER
141
ETAL.
[Therapist
1 was
male,
Therapist
2 female].
problems
I came
Finally, one swapper revealed at the end of treatment that he had chosen the first therapist because he wanted to avoid discussing his
now with
homosexuality, had addressed
pist,” did included
which
the
second
therapist
directly.
item,
“I feel
QUANTITATIVE
I can
Because
of the
many
measures
included
two
were first run for each a separate hypothewere then followed by
for
the
two
the
and
The
factors of the for the first
scores
for first
versus
group
factor
mean differences were the chosen and unchosen
Measures
The within-group factor was an order effect. The
therapist
stronger lationship
Patients
with
would
whom
they
helping alliance as measured
ance Questionnaire and Relationship Inventory. from
The Helping this sample
choose developed
the
a
reAlli-
Barrett-Lennard
standing includes
and the
change). following
therapist
is helping
ment is helping understands me,” me to achieve together with and
“I believe
nature
Questionnaires VA-Penn Psycho-
standing have have ready
loaded Factor Factor items:
on two 2: new
factors under-
1, helpfulness, “I believe
rated
pist. For significant
us
to
assess
found betherapist.
tested whether there interaction of choice
the the
=
chosen
Helping interaction
0.82,
and
unchosen
Alliance was
F= 4.44,
thera-
factors, only a found (Wilks’s
df=
2,41,
my goals,” “I feel I am working the therapist in ajoint effort,” similar ideas about the Factor 2, new underincludes these items: “I
obtained some new understanding,” “I been feeling better recently,” “I can alsee that I will eventually work out the
JOURNAL
OF
was no main effect. Univariate repeated analyses showed that these interactions cant for the first factor-helpfulness--of Helping Alliance Questionnaire 1,42, therapist
=
Further my
me,” “I believe the treatme,” “I feel the therapist “I feel the therapist wants
we have problems.” and change,
of my
allowed
P< 0.02);
there
Alliance and the
The items 1: helpfulness;
tially
lambda
therapy Sample9 were subjected to a principal components factor analysis with a varimax rotation. (Factor
(choice)
on
and therapist (scores for the first and second therapist) assessed whether patients differen-
the
and therapeutic by the Helping
of
factor. scores on
therapist
whether tween
1:
Helping
as a repeatedThe between-
The
Hypothesis
to be
therapist
second
each measure were analyzed measure factor (therapist).
set of measures. as a
thera-
Helping Alliance quesand second therapists.
for
of Choice
deal 11th
in a repeatedpatients’ choice
analyses.
Basis
presented
of the second
separately
Patients’
are
the
Therapist 1 or 2 as a between-group The dependent measures were the the two tionnaire
“I feel
and the
enough
factors
first
and
upon
univariate
each
results
depend
were analyzed simultaneously measures MANOVA with
or more subscales, many statistical analyses were required. In order to minimize capitalization on chance, multivariate analyses of variance (MANOVAs) measure that addressed sis. Significant results
on
Alliance
C 0 M P A R I S 0 N
for,” myself Note that
not correlate strongly in either factor.
Scores RESULTS:
to treatment
that I can understand myself on my own.”
PSYCHOTHERAPY
of variance were signifithe (F= 8.22, df
P= 0.006). That is, patients chose the who was relatively more helpful. 1-tests
indicated
nificantly higher scores factor of the Helping naires for the therapist Although the interaction
that
patients
had
sig-
on the helpfulness Alliance Questionthey chose (Table 1). was not significant
for the second factor (understanding and change) of the Helping Alliance Questionnaire, we also present mean scores in Table 1 in order to better explain the results. Thus, as expected, patients chose the therapist they viewed there action
PRACTICE
as more
helpful
was no main effect and the subsequent
AND
RESEARCH
(Factor for order, 1-tests
1). Because the interin Table 1
142
CHoICE
indicate
that
the
pists on the independently
basis
patients
Barrett-Lennard No
chose
Relationship
significant
findings
repeated-measures multaneously
their
thera-
of the helping alliance of the order effect.
and
1,42,
obtained
MANOVA the four subscales
2:
Patients
a
therapist on the basis larity and pretreatment
choose
the therapist’s qualities Index of Demographic
by the the Q-
findings
neously
these
addressed
Hypothesis
3:
Patients
would
past.
was subjected
by the
choose
Mutual
with
the
lambda=
0.80,
and a significant = 0.75, F= 6.82,
Mean
on the
pro-
more
factors
the
than
did in the
unchosen
scores.
nificant
results.
thera-
interactions was more
tioning
that
that
also
No
the
trend
therapists were 0.05) scores
were
MANOVAs for
run
performed
of the Helping
Alliance
Therapist
on
the
an
Helping
as a function Mean (Ti)
Score Second
of patient’s
Therapist
yielded
Ti
29.3
27.0
who
had
chosen
T2
27.1
29.6
Alliance
scales
1 (r= 0.29, P< P< 0.01). The Relationship
choice
(T2)
t
2.27’ -
2.78**
0.61
change)
Patients
who
had
chosen
Ti
15.5
16.2
-
who
had
chosen
T2
15.5
16.6
-
0.005.
VOLUME
2
sig-
(P