METHODS
Background
This study is part of a larger investigation of the
psychological aspects of breast cancer (18). Data on
demographic characteristics of the patient populations
as well as on their families, family and social
role functioning, and measures of the impact of the
intervention programs on patients with primary
cancer are reported elsewhere (19). In this paper, the
outcome of the group support program for women
with metastatic disease is reported.
Study Design and Procedure
A number of oncologists agreed to participate in
the study. When patients with metastatic breast
cancer consulted them for medical treatment, the
oncologists presented the study and solicited their
permission to be contacted. Those patients who
agreed were called by our research interviewer, who
told them about the details and purpose of the study
and invited them to participate. After written, informed
consent was obtained, a battery of psychological
instruments was administered. The subjects
were then randomly assigned to either the treatment
or control condition, and follow-up testing was done
at 4-month intervals for a total period of one year.
The patients in the treatment condition were assigned
to one of two cancer support groups which
met once weekly for 90 minutes; group treatment
was offered at no charge to the patient. Control
patients continued their regular oncological treatment
but had no additional psychological support.
There was no communication between the experimental
team and the physicians providing routine
oncological and analgesic care.
Sample
Only subjects with primary carcinoma of the
breast and documented metastases were included.
One hundred and nine women were referred to the
study by their oncologists. Of these, 86 completed
the first questionnaire; 18 refused to participate, and
5 died prior to contact. More subjects were randomly
assigned to the treatment (50) than to the control
condition (36) in order to provide sufficient members
for two functioning groups. Fourteen of the
subjects assigned to the treatment condition were too
weak or too ill at the time of the initial interview to
participate (in fact, 6 died before the groups commenced);
2 subjects moved away. Twelve subjects
were lost from the control sample: 4 were too ill to
participate and 2 died; 4 refused to participate, and
we were unable to contact 2,
The final treatment sample consisted of 34
women, and the final control sample consisted of 24
women. The average age of the treatment sample
was 54, and that of the control sample 55. Two
members of the treatment and 3 of the control sample
lived alone. The average length of time since diagnosis
of recurrence was 23.8 months among the
treatment population, and 26.7 months among the
control population. This difference was not statistically
significant (t = 0.59). Members of the treatment
and the control samples received equivalent
amounts of chemotherapy during the period of
study. Fifteen treatment and 11 control patients died
during the 1-year course of the study. Thus, the two
populations appeared equivalent in severity of
illness.
Analgesic and psychoactive medication use patterns,
shown in Table 1, were quite similar in the
control and treatment samples. At entry into the
study, only three members of both the control and
treatment populations utilized prescription analgesics,
including propoxyphene and opiate derivatives.
There was comparatively little use of analgesic
and psychotropic medications in the total sample,
and this use was comparably distributed in the
treatment and control samples at baseline. A slight
increase in the use of analgesics over time was
observed in the control but not the experimental
population. Therefore, differences in medication use
cannot account for any treatment effect observed in
this study.
At the onset of the study, statistically significant
differences were found between the treatment and
control samples, presumably by chance, on only one
variable—social status: members of the treatment
sample were of higher social status (51.7 ± 11.5 on
the Hollingshead scale) (20) than were the members
of the control sample (43.3 ± 14.9; p < 0.03). This
difference enters a potential bias into subsequent
analyses, as others have found that, in the face of