Research interest in psychosocial predictors of the onset and course of can
cer has been active since the 1950s. However, results have been contradicto
ry and the literature is noted for methodological weaknesses. In this prosp
ective study we aimed to systematically obtain data on psychosocial factors
thought to be associated with cancer outcome in early-stage breast cancer.
Consecutive patients with early-stage breast cancer seen for the first tim
e for ongoing treatment at a major teaching hospital in Sydney participated
in the study. Exclusion criteria included inability to speak English and c
o-existing psychiatric illness. Participants completed a series of question
naires every 3 months for up to 2 years following their diagnosis. Domains
measured included cognitive appraisal of threat, coping, psychological adju
stment, social support, quality of life and perceived aim of treatment. Dis
ease and demographic data were obtained from medical records. including age
. marital status. education and occupational status, treatment received, me
nopausal status. tumour size and grade. 233 patients agreed to participate
in the study and 211 (88%) had sufficient data for analysis. Patients were
a mean age of 51 years; 84% of the sample were Australian-born; 25% of the
sample had a University education and 39% were in managerial or professiona
l occupations. 59% had chemotherapy following surgery and 54% were taking t
amoxifen. Women entered the study a median of 91 days since diagnosis. 30 (
14%) of the 211 women died within the study period and 36 (17%) relapsed. M
ultivariate analysis of baseline measures only, used the Cox proportional h
azards model. Two psychosocial factors independently predicted both surviva
l and time to relapse. Apart from disease variables. three non-medical fact
ors remained in both of the final models: appetite, concern about dependenc
e and concern about the disease. Patients who had high levels of concern ab
out the disease and low levels of concern about dependence survived for a s
ignificantly shorter time (Hazard Ratio (HR) 1.04. 95% CI = 1.01-1.06: HR =
0.96, 95% Cl = 0.94-0.99) and relapsed quicker (HR 1.03, 95% Cl = 1.0 1-1.
06; HR = 0.96 95% CI = 0.93-0.98).
Concern about the disease was also predictive of outcome in primary melanom
a patients, but the directional effect was opposite: that is. patients with
a higher concern about their disease did better. Perhaps patients with ear
ly melanoma are mobilized by their concern to take active seif-protective m
easures which can have a real impact on their disease (such as staying out
of the sun), whereas similar self-protective strategies are not available t
o patients with early breast cancer. Alternatively, concern about the disea
se may reflect medical factors in breast cancer which we did not record. In
terestingly, the coping response of minimization, which was significantly p
redictive of outcome in our metastatic breast cancer sample. was not signif
icant in this population. It is likely that different coping styles are sal
ient and predictive of outcome at different stages of disease. There is sti
ll much to learn about the potential relationships between psychological we
llbeing, human behaviours and cancer outcome. Research in this area needs t
o clarify the psychological processes themselves as well as understand the
biological and/or behavioural mechanisms that may link them to outcome. (C)
2001 Harcourt Publishers Ltd.