95

Psychology and Psychotherapy: Theory, Research and Practice (2005), 78, 95111
q 2005 The British Psychological Society

The
British
Psychological
Society
www.bpsjournals.co.uk

The relationship between social deprivation

and unilateral termination (attrition) from
psychotherapy at various stages of the health care
pathway
Roland Self *, Paula Oates, Tia Pinnock-Hamilton and Chris Leach
South West Yorkshire Mental Health NHS Trust, UK
The relationship between social deprivation and attrition from psychotherapy was
examined at various stages of the health care pathway. Data providing information on
service users discharge status gave a measure of attrition at different stages along this
pathway. On the basis of their postcode, service users were allocated a Townsend
deprivation score, which is a measure of social deprivation. Of the sample, 60%
terminated therapy prior to agreed discharge at various pathway stages. Service users
who failed to attend their first appointment had significantly higher social deprivation
scores than those who completed therapy. Early terminators, who stopped attending
before their fifth session, had a significantly higher level of social deprivation than the late
terminators, and those who completed therapy. The late terminators did not significantly
differ from completers on social deprivation. These results support earlier findings
showing that socio-economic status influences attrition from therapy. However, socioeconomic status only affects two stages of the pathway attendance at the first
appointment, and the early stage of attending therapy. It does not affect earlier or later
stages of the pathway. Applying the health action process approach (HAPA) model to the
results, the action/maintenance stage can be represented by attendance for therapy.
Socio-economic status influences this stage of the model, because social support and
resources are important determinants for compliance. Earlier stages, characterized by
HAPA as a decisional/motivation stage thought to be influenced by beliefs, was not
affected by social deprivation. It is concluded that attrition from therapy should be
studied separately for each pathway stage. Earlier conflicting findings on causes of
attrition may have resulted from studying different combinations of pathway stages.

* Correspondence should be addressed to Roland Self, Department of Clinical Psychology, Dewsbury and District Hospital,
Halifax Road, Dewsbury, West Yorkshire, WF13 4HS, UK (e-mail: roland.self@midyorks.nhs.uk).
DOI:10.1348/147608305X39491

96

Roland Self et al.

Furthermore, high rates of attrition related to socio-economic status raises serious
questions about the external validity of therapy trials where these factors are not
routinely reported, and the ecological validity of current service delivery models.

Background
The effectiveness of psychological therapies has received increasing attention in recent
years. The introduction of clinical governance (Department of Health, 1997) has
promoted the increasing use of evidence-based philosophy in all areas of health care in
the National Health Service (NHS). Recent publications, such as, What works for whom
(Roth & Fonagy, 1996), have given considerable reassurance for the efficacy of
psychological therapies based on research trials.
Despite the confident optimism arising from such publications, criticism of
psychotherapy continues to come from many sources, including from the study of
attrition, or drop-out, from therapy in routine clinical practice. High levels of attrition
seem to have been overlooked by those who point to the outcome of clinical trials as
support for the efficacy of psychotherapy, with up to 60% drop-out being recorded in
routine clinical practice. As observed by Hunt and Andrews (1992), The finding that
dropouts are ubiquitous in psychotherapy is very damaging, for if patients do not stay
for treatment then there is little point in developing effective treatment.

Attrition in clinical practice

Attrition is a major problem for psychotherapy services, with levels reaching as high as
60% in everyday service delivery systems. For example, reviews of psychotherapy
literature (Baekeland & Lundwall, 1975; Garfield, 1986) indicate that between 30% and
60% of psychotherapy out-patients terminate their treatment prematurely. In a more
recent meta-analysis of 125 psychotherapy drop-out studies, Wierbicki and Pekarik
(1993) found a mean rate of 47% ending therapy early.
Relatively few studies have investigated treatment outcome for people who drop out
of therapy, but those that have, typically report a pattern of poor outcome (Pekarik,
1992) and low client satisfaction (Lebar, 1982), especially when drop-out occurs within
the first few sessions. Psychotherapy drop outs pose substantial problems for the
delivery system itself, with subsequent effects on therapy. Several significant issues are:
morale problems for mental health professionals, reduced treatment efficacy, and
decreased cost effectiveness (Garfield, 1986; Pekarik, 1985b).

Attrition in psychotherapy research

While most psychotherapy research focuses on the microstructure of particular therapy
interventions, few highlight, or even report, attrition adequately. In a study of attrition
from psychotherapy research, Flick (1988) analysed 26 clinical intervention studies in
the Journal of Consulting and Clinical Psychology. Of these studies, seven did not
report any attrition information. Six of the studies reported attrition rates for the

Social deprivation and unilateral termination

97

complete sample, but this does not provide important information about biases that
attrition might produce by unbalancing the groups. Thirteen studies reported attrition
for each group separately, but in most cases the rate of attrition was simply reported, and
no effort was made to explain how it might affect the results.

Causes of attrition
Whereas the high rate of drop-out from therapy has been well documented, information
on the causes is quite weak. Several researchers (Brandt, 1965; Garfield, 1986; Pekarik,
1985a, 1985b) have noted that the attrition literature is replete with conflicting findings,
replication failures, and generally small differences between drop-outs and completers.
Despite the multitude of variables reported, which include client, therapist, and
programme-related factors, very few variables emerge as significant predictors of
attrition when attrition studies are aggregated. The only consistent findings relate to
socio-economic variables. In their meta-analysis of 125 studies of attrition from
psychotherapy, Wierbicki and Pekarik (1993) looked at 32 variables, but found
significant size effects for just three: racial status, education, and income. This
supported an earlier meta-analysis by Garfield, who concluded that drop-out was most
consistently related to client social class variables, such as low socio-economic status,
low levels of education, and minority racial status.
Authors in the field of general medicine have long noted the relationship between
low socio-economic status and inequality of access to health care provision (e.g.
Townsend & Davidson, 1992). Despite this, research into attrition from psychotherapy
continues in something of a vacuum. This was highlighted by Harris (1998) in her
extensive review of the literature, when she noted, Consequently, despite considerable
empirical substantiation of the relationship between premature termination from
treatment and client race, education and socio-economic status, the precise causal
mechanisms driving minorities, persons with low education and individuals in poverty
to leave treatment early has yet to be determined. Harris goes on to conclude, : : : it is
necessary for investigators to move beyond research on correlations of attrition to
propose and test theoretical models with clearer implications for preventing attrition.

Defining attrition
One of the main problems in attrition research is the lack of a uniform definition of
attrition. Variations in reported attrition rates and associated variables may arise as a
result of differing methods used to define dropping out. In many studies, failure to
attend a specified number of sessions is the criterion (Pekarik, 1985a). Unfortunately,
researchers have used different parameters, with the consequence that patients
considered drop-outs in one study are viewed as continuers in another (Garfield, 1994).
Other methods used include therapist judgment, and failure to keep the last scheduled
appointment, both of which yield different results again (Pekarik, 1985b). Other
problems arise over which stage of dropping out is included in the attrition research.

98

Roland Self et al.

For example, Gould, Shaffer, and Kaplan (1985) reported a drop-out rate of 11%,
referring to those who attended an initial screening interview but failed to attend a later
evaluation interview. In contrast Novick, Benson, and Rembar (1981) reported a rate of
85.4%, obtained by including all cases of non-agreed termination at any stage of the
referral/treatment uptake process. Cottrell, Hill, Walk, Dearnaley, and Ierotheou (1988)
reported a drop-out rate of 53%, by including all non-agreed termination prior to agreed
discharge, and early and late treatment drop-outs. As pointed out by Morton (1995) one
might imagine that the reasons for dropping out before the first appointment would be
different to those of someone who dropped out after this first appointment, and
different again for a person who unilaterally terminates after several months of therapy.
However, studies do not routinely distinguish between these groups, or seek to explain
their relationship to one another. Thus, attrition needs to be defined in terms of the
stage at which it occurs, and each stage should be studied separately.

Health care pathways

The failure within the psychotherapy literature to place attrition within a theoretical
framework contrasts with health care utilization literature, where theories abound as to
why some people, but not others, make successful use of health services. Theories
include, for example, the health belief model (Rosenstock, 1966), the theory of planned
behaviour (Ajzen 1988, 1991), and the health action process approach (HAPA;
Schwartzer, 1992). Within this literature, the concept of health care pathways
occasionally appears. The perception of help seeking as a pathway has a long history in
medical sociology and anthropology (e.g. Friedson, 1970; Parsons 1951; Suchman
1964). More recently Pescosolido (1991) presented an elegant version of the pathways
approach based on the earlier work of Suchman (1972), and Twaddle and Hessler
(1987). Here, the pathway is conceptualized in terms of five decision points, which
represent critical stages in the process of coping with illness:
Recognition: represents the decision that something is wrong. Labelling the individual
as sick starts in the community, and is made by the individual or by powerful others.
Utilization: refers to the decision to enter the patient role, and has traditionally
referred to contact with formal medical services.
Initial compliance: involves the willingness to accept the authority and control of the
practitioner. A variety of types of compliance might include taking medication,
keeping appointments, or changing lifestyles.
Outcome: represents the point where there is a definable outcome. For some this may
mean recovery, while for others it might mean death. Many others may enter a career
of permanent disability or chronicity.
Secondary compliance: is the choice open to this latter group of patients. It is the
option of whether to follow long-term treatment regimes and, if they do so, to what
degree and for how long.

Social deprivation and unilateral termination

99

This conceptualization of the illness career forces us to take a dynamic view of the
health care pathway that sees decision points as key role entrances and exits in the
social process of illness. Within this view, attrition is seen in the context of a series of
decisions on whether to continue or discontinue on a particular heath care pathway,
with the attrition literature pointing to a key role for socio-economic variables in
determining the outcome. A similar pathway can be envisaged for people referred for
psychotherapy, which puts us in a position to look at the stages independently.

Inequalities and attrition

The pathways whereby inequalities impact on health are many and complex. Schofield
(1964) long ago pointed to the attraction to and for psychotherapists of Yarvis (young,
attractive, verbal, intelligent, and successful) clients. However, Hagan and Smail (1997)
argue that the real significance of this has been neither fully assimilated, nor elaborated.
it is not just that well-resourced, more educated and middle class clients are likely to be
able better to make use of therapy than those less privileged, but that they have
available to them powers and resources which make it possible for them to operate on
their proximal environment. In making this argument, Hagan and Smail (1997) are
claiming that peoples use of psychotherapy services is affected by the very same
variables that they, and others, argue give rise to psychological distress in the first place.
Apart from the direct effects of socio-economic deprivation on health, members of
marginalized groups often lack the material and/or symbolic resources to deal with
health-damaging stress (Campbell & Jovchelovitch, 2000). Furthermore, people who
lack the power to shape their life course in significant ways are less likely to believe that
they can take control of their health, and thus less likely to engage in health-promoting
behaviours (Bandura, 1996).

Pathways, attrition and inequalities

Placing attrition within the context of a health care pathway allows us to see attrition as
a phenomenon that will occur at a number of stages in the process, from primary care
consultation, to agreed discharge. This phenomenon can be seen as a unilateral decision
on the part of the service user to terminate the pathway. Thus, the study becomes that of
the unilateral termination of the psychotherapy pathway at its various stages. The
decision to unilaterally terminate will at least in part be shaped by a persons social
position. These ideas are explored in the analyses that follow, where socio-economic
status is examined for various stages of the health care pathway. According to the
inequalities literature, people of low socio-economic status will be doubly
disadvantaged in relation to attending psychotherapy, and therefore we might expect
that the lower a persons socio-economic status, the earlier the stage at which they will
drop out on the pathway. Other reasons for termination may affect pathway stages
differently, but these are not explored here.

100

Roland Self et al.

Method
The sample consisted of 676 adults discharged from Dewsbury and District Hospital
Clinical Psychology Service between April 1995 and March 2001. Dewsbury is the
principal town of the metropolitan district of North Kirklees, UK, which consists of a
number of small former industrial towns. The hospital is situated approximately 1 mile
(1.5 km) from the town centre, and no town is further than 9 miles (15 km) from the
hospital. Depending on the reason for referral, service users were allocated to cognitivebehavioural therapy, psychodynamic psychotherapy, and integrative psychotherapy or
counselling. There was considerable staff turnover during this period with clinical
psychologists, trainee clinical psychologists, clinical nurse therapists, and counsellors
providing therapies.
Measures
Data were collected from the Adult Clinical Psychology Services database that is
maintained for resource management and audit purposes. This source provided
information on the service users age, gender, reason for referral, waiting time, total
numbers of appointments, attendances, cancellations, non-attendances, and discharge
status. All service users included in the study had a valid UK postcode and lived within
the metropolitan district of Kirklees. On the basis of their postcode, service users were
allocated a Townsend deprivation score. These scores were obtained from Calderdale
and Kirklees Health Authority for all the local postcodes. Townsend scores are
calculated using the percentage of unemployment, the percentage of households with
no car, the percentage of overcrowded households, and the percentage of households
not owned by their occupier within each enumeration district (i.e. the area covering
approximately 500 people for census purposes). Various statistical techniques are then
used to standardize these percentages into z scores. A positive figure shows an area is
relatively deprived, whereas a negative figure shows an area is relatively better off.
Clearly not all people living in such an area will be identical in relation to social
deprivation, but composite area-based measures have been shown to be sufficiently
sensitive to identify groups that are disadvantaged in relation to health (Morgan, 1983),
so were thought adequate for the current study.

Pathway stages
Various pathway stages are defined by administrative practice; that is, discharging
patients who fail to opt-in, or fail to attend their first appointment. Therapists, using
local criteria, established at discharge subsequent unilateral termination. Unilateral
termination was defined as, when a service user that the therapist determines has not
received the full benefit of therapy fails to attend an agreed appointment, or refuses to
take an offered appointment and fails to make or keep any subsequently arranged
appointments. Service users who agreed termination with their therapist were
classified as completers. Therapist judgment was used to define a service user

Social deprivation and unilateral termination

101

as completing therapy where they had failed to attend an agreed final appointment or
follow-up appointment. A further stage of termination is determined below in order to
differentiate between people who stop attending therapy soon after it begins, and those
who stop attending later in therapy.
Service users
Service users were an average age of 36 years old at referral, with a higher percentage of
female (65.6%) compared with male (34.4%). Regarding reason for referral, 52% of
people were described as anxious or depressed. The remaining 48% consisted of posttraumatic stress disorder, obsessive-compulsive, eating disorder, and other referrals.

Analyses
A one-way analysis is of variance was used to examine the relationship between attrition
at various pathway stages and levels of social deprivation. Post hoc analysis using the
least significant difference (LSD) test allowed for comment on differences in social
deprivation between the ANOVA groups. Independent t tests were used to find the point
where early attrition could be separated from late attrition. The point was located by
looking for significant differences in social deprivation between the two groups.

Results
Comparison of complete and incomplete data samples
The database covered 755 case records. Of these cases, 79 were eliminated from any
analyses as the data were incomplete, and could not be completed by searching through
microfiche slides. Complete and incomplete data were compared on a number of
variables, including age, deprivation score, total number of appointments, total
attendances, total cancellations (a service user contacted the department to cancel and
make another appointment), and total DNAs (a service user failed to attend a scheduled
appointment without first cancelling it). No significant differences were found for any
of these variables. Hence, it was concluded that the complete data were representative
of the service users in this sample.

Attrition in therapy
Figure 1 demonstrates that only 40% of service users referred to the Clinical Psychology
Department actually completed therapy; 60% of service users dropped out of therapy at
earlier pathway stages; nearly 23% failed to respond to a letter inviting them to opt-in to
the service following referral; 13% failed to attend their first appointment (and failed to
make or keep a further arranged appointment); and finally, 24% dropped out at some
stage following their first attendance.

102

Roland Self et al.

Figure 1. Percentage of clients terminating at each stage

Deprivation and attrition

The following analyses examine the relationship between social deprivation and
unilateral decisions to terminate the psychotherapy pathway.
The relationship between social deprivation and attrition described in the literature
was confirmed here. The mean level of social deprivation was higher for those service
users who dropped out (1.12), than for those who completed (0.51). This difference
was small, but significant, t 22:29, df 674, p :02; a 95% confidence interval for
the difference in means is 0.08, 1.14.

Defining early and late attrition

Early and late attrition in therapy was examined in relation to deprivation, to see
whether this would support the assertion that people drop out at different stages of
therapy for different reasons. This was based on both the observations of Morton
(1995), quoted earlier, that we might expect different reasons for dropping out at an
earlier rather than later stage, and also repeated reports that the majority of people who
unilaterally terminate and stop attending at an early stage. Baekeland and Lundwall
(1975) reported in their comprehensive review that 31% to 56% of clients attended four
or fewer sessions, and Garfield (1994) reported that 23% to 49% of clients failed to
attend following an intake interview, and two-thirds terminated before 10 sessions.
The point at which service users dropped out of therapy is shown in Fig. 2, which
illustrates that attrition was most frequently observed during the first 10 sessions. The
percentage of service users unilaterally terminating therapy by 10 sessions was 86.7% of
the sample. Hence, these 10 sessions were examined in greater detail.
The logic of the following analyses is based on the arguments put forward above, that
people who drop out of therapy once they have started are made up of two populations.
In particular, if people who terminate later on in therapy are viewed as people who have
essentially completed therapy, and would soon be discharged by their therapist anyway,
then we would not expect them to be affected by social deprivation in the same way

Social deprivation and unilateral termination

103

Figure 2. Decay of attendance in clients dropping out of therapy.

as those who terminate at an earlier stage. Thus, we should be able to demonstrate a

clear difference between early and late terminators in relation to social deprivation.
In order to determine whether these two groups could be defined and examined
further, deprivation scores were analysed for people who dropped out of therapy, before
or after a given session, from 1 to 10. Thus, where the denoted number of sessions was
one, all those people who dropped out immediately following the first session were
classified as early drop-outs, and those that dropped out from the second session onwards
were classified as late terminators. The respective mean deprivation indexes were
calculated for the early drop-outs and late terminators. Similarly, where the denoted
number of sessions was two, those service users who dropped out on the first or second
session were classified as early drop-outs, and those who dropped out from the third
session onwards were classified as late terminators. Again the mean deprivation index
was calculated for the early and late groups. This process was repeated for all 10 sessions.
Where the denoted number of sessions was 10, all those service users who dropped out in
the first 10 sessions were classified as early drop outs, whereas those who dropped out
from the 11th session onwards were classified as late terminators, and the final mean
deprivation indexes for the early and late groups were calculated.
Inspection of Table 1 demonstrates that, after the first session, early and late
terminators do differ in relation to social deprivation, with late terminators having lower
deprivation scores. This supports the view argued above that people drop out of
therapy for different reasons, and that socio-economic variables affect various pathway
stages differently. To incorporate this finding into the subsequent pathway analyses, it
was necessary to determine a point that optimally defined early and late termination. In
order to find the point with the greatest difference between the deprivation indexes for
early drop-outs and late terminators, independent samples t tests were used to
investigate the observed differences at each of the 10 sessions. The results were used to
divide early drop-outs from late terminators, and hence, early attrition from late attrition.
From these results, late attrition was classified as incorporating those service users
who dropped out of therapy after five or more sessions, and early attrition as those
service users who terminated following one of the first four sessions. Aubrey, Self,

104

Roland Self et al.

Table 1. Differences between early and late attrition at sessions 1 10

Mean deprivation index
Denoted session number
1
2
3
4
5
6
7
8
9
10

Early

Late

Significance ( p value)

0.627
1.971
1.481
1.701
1.813
1.683
1.618
1.574
1.541
1.415

1.142
0.875
1.012
0.648
0.404
0.467
0.309
0.250
0.224
0.615

.482
.060
.395
.057
.013*
.039*
.040*
.050*
.063
.301

*Significant at a :05.

and Halstead (1999) determined a similar cut-off point when investigating the predictive
effects of early non-attendance on subsequent non-attendance and attrition.
The relationship between social deprivation and attrition at various pathway stages
Having established these two stages of unilateral termination with therapy underway,
they were combined with stages defined by local practice to describe the complete
pathway to be used in subsequent analyses. Definitions of the stages are described
below:
Unseen (U): service users who failed to respond to a request to opt-in following
referral;
Non-engagers (N): service users who opted in but who failed to attend their first and
any subsequently arranged appointments;
Early drop-outs (E): service users who attended for at least one appointment but who
unilaterally stopped attending prior to their fifth attendance;
Late terminators (L): service users who attended for at least four appointments, but
who unilaterally stopped attending at various points after this;
Completers (C): patients who agreed termination with their therapist (the therapists
judgment was used where patients failed to attend an agreed final or follow-up
session).
Figure 3 shows notched box plots of the Townsend scores for each group, with
groups ordered by pathway stage. The filled circle in the middle of the box is the median
score for the group, the boxes enclose the middle 50% of the data and the whiskers
show the upper and lower 25% for each group. The notches show 90% confidence

Social deprivation and unilateral termination

105

Figure 3. Notched box-plots of deprivation scores for terminators at the five pathway stages.
(Notches show approximate 90% confidence intervals around the medians [black dots] for each group,
allowing approximate 5% significance tests in comparing pairs of medians.)

intervals for the medians, which allows approximate 5% significance levels when
comparing pairs of medians (McGill, Tukey, & Larson, 1978).
An ANOVA shows that the five group means differ significantly, F4; 671 3:77,
p :005. Post hoc analyses using the LSD test show significant differences between
groups E and C ( p :001), E and L ( p :010), E and U ( p :013), and N and C
(p :025). All other comparisons yielded non-significant results. The results of the post
hoc tests can be summarized as follows, where the lines join groups that do not differ
significantly. The results of the post hoc tests can be summarized as follows, where the
lines join groups that do not differ significantly:
ENULC
In order to examine the strength of the effect of social deprivation on attrition at
each pathway stage, data were allocated to either a high, medium, or low deprivation
group. The high deprivation group consisted of people whose deprivation score was
one standard deviation above the mean, and the low deprivation group included all
whose deprivation score was one standard deviation below the mean. Those in between
were classed as the medium deprivation group. This latter group is omitted from Fig. 4
for the sake of clarity but were consistently midway between the high and low groups.
The percentages of people in each group who terminated at each pathway stage were
compared. Figure 4 shows that there was an observable and statistically significant
difference between the high and low groups, x 2 12:5, df 4, p :014, with the
main contribution to chi-squared coming from the large differences in proportions of

106

Roland Self et al.

Figure 4. Percentage of clients with high and low Townsend scores terminating at each pathway stage

clients with high and low Townsend scores at the early drop out and completion stages.
People in the high deprivation group are twice as likely to drop out at an early stage of
attendance, as compared to people in the low deprivation group. Thus, the effect
of socio-economic variables on drop-out from therapy is pronounced for the first and
next few appointments, and accounts for virtually all the large difference at completion.

Discussion
The review of the literature in the Introduction shows that the study of people who stop
attending therapy is flawed by a lack of consistent definitions and theoretical basis. The
current study suggests that concepts from related fields provide the basis for a solution
to these problems. By incorporating decision making and pathways from heath care
utilization literature, the study is now defined as unilateral termination of
a psychotherapy pathway at its various stages.
The importance of defining stages separately is demonstrated in the current study.
The single most robust finding in the literature that unilateral termination is related to
socio-economic status was shown to be true for only some pathway stages. Unlike the
prediction outlined in the Introduction, that the lower a persons socio-economic status
the earlier they will drop out, this study demonstrates a complex relationship between
socio-economic status, and termination of a psychotherapy health care pathway.
People who unilaterally stop attending therapy in the first four sessions (early dropouts), had significantly higher deprivation scores than completers and people who
terminated therapy at a later stage. Similarly, people who failed to attend their first
appointment (and any subsequently arranged appointments) had a significantly higher
deprivation score than completers, thus demonstrating the effect of social deprivation

Social deprivation and unilateral termination

107

on initial attendance for therapy. In contrast, people who failed to opt-in to therapy
only weeks after a referral was made had a deprivation score similar to completers and
late terminators, suggesting that social deprivation does not have a direct impact on this
stage of the pathway. The lack of social difference between late terminators and
completers lends support to the idea that late terminators are in fact unilateral
appropriate terminators.
In relation to Pescosolidos (1991) model outlined in the Introduction, social
deprivation is seen only to affect the compliance stage, and does not appear to affect
earlier or later stages. This suggests that there are different mechanisms at work at the
various stages. Clearly, the most obvious difference is that initial compliance, which
requires actual attendance for therapy, involves a much greater behavioural and
resource commitment than the earlier stages. People who drop out at a later stage can
be viewed as appropriate terminators, and socio-economic status has no effect on this.
This description is consistent with the HAPA (Schwarzer, 1992) model of health care
utilization. According to this model, there is a distinction to be made between a decision
making/motivational stage and an action/maintenance stage.
The decision/motivational stage is made up of three components:
Threat appraisal: beliefs around the severity of the problem and ones vulnerability;
Self-efficacy: beliefs about ones ability to carry out the required action; and
Outcome expectancies: beliefs about the efficacy of the proposed course of action.
The end result of this stage is the formation of an intention to act.
The action/maintenance stage is brought about by a sufficient intention to act but is
then further influenced by two components:
Self-efficacy: as applied to the earlier stage; and
Barriers and resources: factors such as social support, financial or practical
problems, and so forth.
The results of this study suggest that socio-economic status (SES) influences the
action/maintenance stage of this model, where this is represented by actual attendance
for therapy. It is social support, and resources such as skills, money, and practical help
that are important in determining compliance within this stage of the health care
pathway. The lack of a SES effect on the earlier opt-in stage of the pathway would
suggest that SES does not influence expectancy beliefs.
These findings demonstrate the differential impact of socio-economic status on the
various pathway stages of therapy attendance. It follows from this that attrition should
only be viewed within the context of a pathway. Unilateral termination at different
stages represents different phenomena that require different, if related, explanations.
Beyond these findings that attrition should only be studied in relation to pathway
stages, the confirmation of high rates of attrition in clinical practice raises serious
questions about the theoretical basis of modern psychotherapy, as well as about the way

108

Roland Self et al.

it is delivered. Considerable emphasis is currently placed upon the need for evidencebased practice, with evidence accumulated through therapy trials, and research journals
reporting the outcomes. However, while the internal validity of these trials is explicitly
reported, the effects of client selection and attrition on external validity are rarely
mentioned.
If we are to understand the relevance of clinical research to clinical practice, then
research needs to be more open and transparent with regard to the population reported
upon. At the very least each study should report upon the SES of the original sample, and
the sample completing therapy. This will ensure that research populations reflect the
population as a whole, and demonstrate that interventions are equally effective for all
social groups, or, where they are not, alternative interventions suggested and
investigated. Furthermore, studies should report upon the original sample size and
terminators at each pathway stage, whether through non-inclusion or unilateral
termination. Under these circumstances unilateral termination will become an outcome
measure in itself.
Although some may protest that this is an unrealistic goal, reporting as suggested will
encourage researchers to be more proactive in ensuring the external validity of their
studies, and devise ways of making their intervention more relevant to a wider
population. This, on the evidence, would not be a bad thing for the practice of
psychotherapy. If we take a broad view of the therapeutic process, we see that up to
50% of people who have an identified need for therapy unilaterally terminate before any
significant benefit can accrue. This suggests that the social perspective needs to be
brought back fully into play alongside the individual when determining effective
interventions.
In terms of theories and therapies, this means that we may no longer turn the
understandable distress experienced by an isolated unmarried mother in a run-down
tower-block into depression, and offer her anti-depressants and individual cognitive
therapy, but rather interventions based on social actions. In terms of health care
pathways, if we accept that some people can only get off work with great difficulty, or at
prohibitive cost, then we might have to consider working unsocial hours, or situating
our clinics nearer to where people actually are during the day. As people often work in a
different health district from that in which they live, this will mean changes in referral
practice, and perhaps methods of funding services.
Although those engaged in counselling and psychotherapy in the British NHS may
feel that this has little to do with them, they will not be able to ignore the wider
government agenda indefinitely. The inequality of access to health care has not gone
unnoticed, and considerable government attention in recent years has focused upon
health inequalities and socio-economic status. The Independent Inquiry into Inequalities in Health (Acheson, 1998), brings together the research evidence setting out the
main influences on health inequalities such as poverty, housing, and education. The
governments response, Reducing health inequalities: An action report (Department of
Health, 1999), and the White Paper, Saving lives: Our healthier nation (Department of

Social deprivation and unilateral termination

109

Health, 1999), sets out the governments commitment to improving the health of the
worst off in society. The NHS Plan (Department of Health, 2000) gives an
unprecedented focus on the inequalities agenda within the NHS, and concentrates on
closing the gap: setting local targets to reduce health inequalities (NHS Health
Development Agency, 2001), emphasizing, among other things, inequality of access as a
target for action.

Conclusion
The high rate of attrition found in this study is influenced by socio-economic status.
However, socio-economic status appears only to impact on the initial compliance and
compliance stages of the pathway as described by Pescosolido (1991) and the
action/maintenance stage of the HAPA model (Schwarzer, 1992). This may be because
these stages require a much greater behavioural and resource component than earlier
stages. Hence, people who have less social support and resources, will be less likely to
comply, and therefore less likely to attend therapy. Earlier and later stages of the
pathway do not appear to be influenced by socio-economic status. These results
demonstrate that attrition at the various pathway stages should be studied separately,
and that earlier conflicting findings may have resulted from studying different
combinations of pathway stages.
However, perhaps more importantly, the external validity of current psychotherapy
research findings is brought into question by the demonstration that unilateral
termination from therapy is not randomly distributed, but is heavily influenced by an
individuals socio-economic status. The findings also bring into question the ecological
validity of current service delivery models.

References
Acheson, D. (1998). Independent inquiry into inequalities in health. London: Stationery Office.
Ajzen, I. (1988). Attitudes, personality and behaviour. Milton Keynes: Open University Press.
Ajzen, I. (1991). The theory of planned behaviour. Organizational Behaviour and Human
Decision Processes, 50, 179 211.
Aubrey, R., Self, R., & Halstead, J.E (1999). Early non-attendance as a predictor of continued nonattendance and subsequent attrition from therapy. Unpublished. Dewsbury Health Care
NHS Trust.
Baekeland, F., & Lundwall, L. (1975). Dropping out of treatment: A critical review. Psychological
Bulletin, 82(5), 738 783.
Bandura, A. (1996). Self-efficacy in changing societies. Cambridge: Cambridge University Press.
Brandt, L. (1965). Studies of dropout patients in psychotherapy: A review of findings.
Psychotherapy: Theory, Research and Practice, 12, 6 12.
Campbell, C., & Jovchelovitch, S. (2000). Health, community and development: towards a social
psychology of participation. Journal of Community and Applied Social Psychology, 10,
255 270.

110

Roland Self et al.

Cottrell, D., Hill, P., Walk, D., Dearnaley, J., & Ierotheou, A. (1988). Factors influencing nonattendance at child psychiatry out-patient appointments. British Journal of Psychiatry, 152,
201 204.
Department of Health (1997). The new NHS: modern, dependable. London: HMSO.
Department of Health (1999). Reducing health inequalities: An action report. London: HMSO.
Department of Health (1999). Saving lives: Our healthier nation. London: HMSO.
Department of Health (2000). The NHS plan: A plan for investment. A plan for reform. London:
HMSO
Flick, S. N. (1988). Managing attrition in clinical research. Clinical Psychology Review, 8,
499 515.
Friedson, E. (1970). Profession of medicine. New York: Dodd Mead.
Garfield, S. L. (1986). Research on client variables in psychotherapy. In S. L. Garfield & A. E. Bergin
(Eds.), Handbook of psychotherapy and behaviour change. New York: Wiley.
Garfield, S. L. (1994). Research on client variables in psychotherapy. In S. L. Garfield & A. E. Bergin
(Eds.), Handbook of psychotherapy and behaviour change. New York: Wiley.
Gould, M. S., Shaffer, D., & Kaplan, D. (1985). The characteristics of dropouts from a child
psychiatry clinic. Journal of the American Academy of Child Psychiatry, 24, 316 328.
Hagan, T., & Smail, D. (1997). Power-mapping. 1. Background and basic methodology. Journal of
Community and Applied Social Psychology, 7, 257 267.
Harris, P. M. (1998). Attrition revisited. American Journal of Evaluation, 19(3), 293 305.
Hunt, C., & Andrews, G. (1992). Drop-out rate as a performance indicator in psychotherapy. Acta
Psychiatrica Scandinavica, 85, 275 278.
Lebar, J. L. (1982). Consumer satisfaction with mental health treatment. Psychological Bulletin,
91, 244 259.
McGill, R., Tukey, J. W., & Larsen, W. A. (1978). Variations of boxplots. American Statistician, 32,
12 16.
Morgan, M. (1983). Measuring Social inequality: Occupational classifications and their
alternatives. Community Medicine, 5, 116 124.
Morton, A. (1995). The enigma of non-attendance: A study of clients who do not turn up for their
appointment. Therapeutic Communities, 16(2), 117 132.
NHS Health Development Agency. (2001). Closing the gap: setting local targets to reduce health
inequalities. London: HMSO.
Novick, J., Benson, R., & Rembar, J. (1981). Patterns of termination in an outpatient clinic for
children and adolescents. Journal of the American Academy of Child Psychiatry, 20,
834 844.
Parsons, T. (1951). The social system. Glencoe, IL: Free Press.
Pekarik, G. (1985a). Coping with dropouts. Professional Psychology: Research and Practice, 16,
114 123.
Pekarik, G. (1985b). The effect of employing different termination classification criteria in dropout research. Psychotherapy, 22, 86 91.
Pekarik, G. (1992). Relationship of clients reasons for dropping out of treatment to outcome and
satisfaction. Journal of Clinical Psychology, 48(1), 91 98.
Pescosolido, B.A. (1991). Illness careers and network ties: A conceptual model of utilization and
compliance. In G. Albrech & J. Levy (Eds.), Advances in Medical Sociology, 2, 161 184.

Social deprivation and unilateral termination

111

Rosenstock, I. M. (1966). Why people use health services. Millbank Memorial Fund Quarterly,
44, 94 124.
Roth, A. D., & Fonagy, P. (1996). What works for whom? A critical review of psychotherapy
research. New York: Guilford.
Schofield, W. (1964). Psychotherapy: the purchase of friendship. Hemel Hempstead:
Prentice-Hall.
Schwarzer, R. (1992). Self-efficacy in the adoption and maintenance of health behaviours:
Theoretical approaches and a new model. In R. Schwarzer (Ed.), Self-efficacy: Thought control
of action (pp. 217 243). London: Hemisphere.
Suchman, E. (1964). Sociomedical variations among ethnic groups. American Journal of
Sociology, 70, 319 331.
Suchman, E. (1972). Social patterns of illness and medical care. In E. G. Jaco (Ed.), Patients
physicians and illness (pp. 262 279). New York: Free Press.
Townsend, P., & Davidson, N. (1992). The Black Report. In P. Townsend, M. Whitehead, &
N. Davidson (Eds.), Inequalities in health: The Black Report and the health divide. London:
Penguin.
Twaddle, A. C., & Hessler, R. M. (1987). A sociology of health (2nd ed.). New York: Macmillan.
Wierbicki, M., & Pekarik, G. (1993). A meta-analysis of psychotherapy dropout. Professional
Psychology: Research and Practice, 24(2), 190 195.
Received 14 May 2003; revised version received 5 May 2004