Psychosomatic Medicine Faster Service from Outside North America
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Brown, K. W.
Right arrow Articles by Edgar, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Brown, K. W.
Right arrow Articles by Edgar, L.
Related Collections
Right arrow Cancer
Right arrow Depression
Right arrow Stress and Coping
Psychosomatic Medicine 65:636-643 (2003)
© 2003 American Psychosomatic Society

ORIGINAL ARTICLES

Psychological Distress and Cancer Survival: A Follow-Up 10 Years After Diagnosis

Kirk W. Brown, Adrian R. Levy, Zeev Rosberger and Linda Edgar

From Department of Clinical and Social Sciences in Psychology (K.W.B.), University of Rochester, Rochester, New York; Centre for Health Evaluation and Outcome Sciences (A.R.L.), St. Paul’s Hospital and Department of Health Care and Epidemiology, University of British Columbia, Vancouver, British Columbia; Division of Psychology (Z.R., L.E.), Sir Mortimer B. Davis-Jewish General Hospital, Montreal; and Departments of Psychology, Psychiatry, and Oncology (Z.R.), McGill University, Montreal, Quebec, Canada.

Kirk W. Brown, PhD, CSP, Meliora Hall, RC Box 270266, University of Rochester, Rochester, NY 14627-0266. E-mail: kirk{at}prodigal.psych rochester.edu

Received for publication May 10, 2002; revision received November 24, 2002.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 
OBJECTIVE: This study tested the predictive role of psychological distress in cancer survival, while attempting to overcome several important methodological and statistical limitations that have clouded the issue.

METHODS: Measures collected on a range of emotional and cognitive factors in the early postdiagnostic period and at 4-month intervals up to 15 months after diagnosis were used to predict survival time up to 10 years among 205 cancer patients heterogeneous in disease site, status, and progression.

RESULTS: With the use of both baseline and repeated measures, depressive symptomology was the most consistent psychological predictor of shortened survival time, after controlling for several known demographic and medical risk factors.

CONCLUSIONS: Given the importance of depressive symptoms to cancer survival, discussion focuses on the possible mechanisms mediating this relationship, the importance of psychological screening of cancer patients, and need for further research.

Key Words: depressive symptoms, • survival, • time dependence.

Abbreviations: CES-D = Center for Epidemiological Studies Depression scale;; CI = confidence interval;; HR = hazard ratio;; MI = multiple imputation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 
Cancer patients who outlive their prognoses often believe there is a direct relationship between their psychological states and their long-term survival (1). In the research community, considerable controversy surrounds the issue of whether cognitive and emotional states can influence long-term survival after cancer diagnosis. To date, research has explored the predictive role of both general psychological states and specific features. Measures of general emotional and/or cognitive condition have typically been shown to have little or no value in predicting length of disease-free and/or overall survival (2, 3). Several specific psychological features have been shown to influence cancer course, including depression (4) and "fighting spirit" (or helplessness/hopelessness) (5). Lack of social support has also been shown to play a predictive role (6). Several well-controlled studies have demonstrated that psychosocial intervention can lengthen survival in both early and advanced stage cancer patients (7, however see 8).

The role of the psyche in cancer course is lent support by such findings, but it is not difficult to find studies that contradict them, by showing null effects for the role of such specific factors as depression (9), hopelessness (10), and anxiety (11). Research has shown self-reported physical health to be a more important predictor of survival than psychological distress (12). Some have argued that current evidence for the role of psychological influences is inconclusive at best (13). Whether psychological factors affect cancer course has important implications, both for the role of psychological screening and treatment of cancer patients, and for our understanding of the role of the mind in the disease process.

Several methodological problems have contributed to the uncertainty surrounding the mind-cancer issue. Failure to control for important medical and demographic risk factors can inflate the importance of psychological factors in cancer survival. On the other hand, even when such control is in place, small sample sizes have limited the power of analyses to detect the effects of psychological factors, which may be small in comparison to powerful medical and demographic effects like cancer stage and gender. Furthermore, when samples are restricted to advanced stage patients, the role of psychological factors may be extremely difficult to detect, because the forward momentum of cancer-related biological processes is very strong.

As noted, measures of general psychological state have usually not predicted cancer course. But the use of a single measure of cognitive and/or emotional distress may mix specific features that have effects on cancer course with those that do not, thereby weakening or nullifying psychological influences. Measurement timing may also be important. Measures collected once, early in the postdiagnostic period, have been commonly used to predict survival years later (2, 11). High levels of distress are common in the postdiagnosis and cancer treatment periods (14). If measures of such distress largely reflect a patient’s short-term reaction to the stresses of diagnosis and medical treatment, they may be unlikely to have long-term prognostic value, given evidence that chronic psychological conditions are more likely to lead to physiological disruption and serious health problems than short-lived states (15). Longitudinal designs that track individual psychological change over time can assess the role of psychological states measured both in the short-term aftermath of diagnosis and later in time, once some degree of cognitive and emotional adaptation to the disease has been made (16). Longitudinal designs can also track the progression of a cancer (from lower to higher stages, for example) and its treatment so as to examine the relative predictive weight of medical and psychosocial factors as they change over time.

The present research attempted to overcome the methodological limitations of past research described here to obtain a more definitive picture of the role of psychological state in long-term cancer survival. A large sample of cancer patients, heterogeneous in both cancer stage and site, but homogeneous in time since diagnosis, were enrolled in a longitudinal study that assessed several demographic, medical, and psychological variables over a year-long interval after cancer diagnosis. Survival status was assessed 10 years after diagnosis. Given that, among the various psychological features studied, affective states have been most strongly implicated in cancer survival (4), this research included several indicators of emotional state and coping, including positive and negative mood, anxiety, and depressive symptoms. Other factors shown to relate to adjustment were also measured, including sense of control (17) and perceived stress (18). Given the demonstrated role of patient-reported physical health status as a predictor of survival (12), the predictive value of such a measure was also tested.


   METHODS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 
Participants
Patients were eligible for study participation if they met the following criteria: 1) 18 years of age or older; 2) contact permission had been granted by the primary physician; 3) able to speak and read English or French; 4) medical treatment accepted at the Jewish General Hospital; and 5) cancer diagnosis within the past 4 months. All patients were newly diagnosed. No inclusion conditions were placed on either site of cancer or stage at diagnosis. During the 2-year accrual period of the study, a total of 524 eligible patients were invited to participate, and 205 patients (39%) were enrolled. The study required regular participation over a 1-year period (see below), which may have influenced the enrollment rate. Participants were slightly younger than decliners (56.3 years vs. 60.8 years) and were more likely to be women (75% vs. 64%).

Each participant completed scaled questionnaires with one of two trained interviewers blind to the research purposes. A baseline interview occurred as soon as possible after diagnosis (M = 11 weeks), and follow-up interviews took place at 4, 8, and 12 months after baseline (M = 7, 11, and 15 months after diagnosis). Seventy-two individuals did not complete the entire year-long study, due to death (N = 38), deteriorating health, or other reasons. Because the time-dependent regression analyses used here accommodate such "missing data" (see Data Analysis), data from all 205 patients were used. Analyses comparing the 72 noncompleters with the rest of the sample on all baseline demographic, medical, and psychological variables found only one difference: Patients with data at all four time points were more likely to be diagnosed with breast cancer than with other cancers, {chi}2 (1) = 9.32, p < .002.

Measures
At study entry, the following information was collected: age, gender, marital status, cancer site, cancer stage, histopathological grade, number of positive lymph nodes, and cancer treatment protocol.1 Clinical information on tumor grade and nodal status was abstracted from hospital charts using a standardized protocol. A clinician naïve to the study purpose reviewed operative and pathology reports for cancer stage and tumor characteristics (size and histological grade when available), nodal status, and metastatic spread. Given the variety of cancer types in the sample, a local, regional, and metastatic system was used. Trained interviewers collected current treatment information and all psychological measures during each of the four interviews.

The following psychological measures assessed affective state and mood, impact of events, and sense of control. First, the 20-item CES-D (20) assessed depressive symptoms. This scale has been validated in both cancer and non-cancer populations (21). In this sample, coefficient {alpha} = .78. Second, the Impact of Events-Intrusion Subscale (22) was used. The Impact of Events scale assesses subjectively experienced stress related to an adverse life event. The 7-item intrusion subscale measures the degree to which intrusive or distracting thoughts increase distress, and has been more strongly linked with psychological distress than the avoidance subscale (18). In this sample, coefficient {alpha} for this subscale = .86. Third, the 9-item Lewis, Firich, and Parsell Anxiety Scale (23) was used, and has also been normed among cancer patients (sample {alpha} = .87). The 130-item Mood Adjective Check List (24) measures both pleasant and unpleasant mood. In the present research, the 12 major mood factors were used as items. Factor analysis of these items (25; cf., 26) showed that 6 items comprised a pleasant mood subscale, {alpha} = .75; four items formed an unpleasant mood subscale, {alpha} = .66. Two factor items, "serious" and "boastful," were dropped due to poor loadings. Fifth, the single-item Cantril’s Ladder (27) assessed sense of control over current life events on a 0 (no control) to 10 (most control possible) scale. Finally, a single-item self-report of physical health status [very poor = 1 to excellent = 5] was completed (28). Similar to the commonly used ECOG performance status measure (29), past research has found this report to correlate highly with physician’s assessments (30) and to predict mortality (28).

Outcome
For all patients, survival status, and, if deceased, the date and cause of death were sought from one of three sources: the hospital chart, hospital oncology clinic records, or provincial government health and social service agency records. Complete follow-up data were obtained for all but five patients, who were censored in survival analyses and the last hospital contact was recorded as the follow-up time (31).

Data Analyses
Multivariate Cox proportional hazards regression models were conducted using the SAS PHREG procedure (32). In these survival analyses, our interest was in the effect of the predictor variables on the time to death from cancer. As such, four individuals who died from a noncancer cause (three cardiac arrest victims and one acquired immune deficiency syndrome victim) were censored at the time of death.

The variables considered important to survival and used to form the core model were the following: age, gender, cancer site, cancer stage, treatment status, histopathological grade, and number of positive lymph nodes. The cancer site variable was dichotomized into breast vs. other diagnosis, given the predominance of breast patients and the heterogeneity of the other diagnoses. Given the small number of metastatic patients (N = 14), stage was coded as advanced (regional and metastatic) vs. localized to enhance analytic reliability (31). This basic "late" vs. "early" staging division is a regular predictor of survival (33). Treatment was coded as systemic (chemotherapy, radiation, the combination of the two, or either in combination with surgery) vs. local (surgery only or no treatment; only two patients were untreated).

Some patients had unknown values for histopathological status and lymph node status. To include all patients in the analyses, these values were replaced by MI with the use of NORM software (34). MI is considered a superior technique to deal with missing data than other, more commonly known and used methods (35), and is robust except when there is a very large proportion of missingness in the data set (36). In the present study, 14% of patients were missing only nuclear grade data, 12% were missing only lymph node data, and 20% were missing data on both variables. Preliminary analyses showed that the present data met the assumptions of MI (37). All independent and dependent variables used in the baseline models were used in imputation. Two other variables, the number of weeks since diagnosis and marital status, were also included.2,3

The distribution of imputed values for each missing datum was averaged before analysis, given both the large number of analyses performed and to obtain Cox regression model fit indices.

The data were shown to conform to the proportional hazards assumption of Cox regression analysis using the SAS LIFETEST procedure (32). The core baseline Cox model was fit, after which each psychological variable was entered separately. This tested whether each psychological variable contributed to survival over and above the contribution of all demographic and medical covariates. The fitting of separate models was done to preserve an acceptable events (deaths):predictors ratio, and to avoid problems caused by collinearity among the psychological variables (39). Psychological predictors found significant in these separate models were entered into a single, stepwise model.

Two sets of Cox models were constructed, each using the core and core + psychological variable fitting strategy: the first one used only baseline data, and the second, a time-dependent covariates analysis, assessed whether the most recent information available for each patient predicted survival. Because many patients show changes in medical and psychological status over time, time-dependent analyses can provide an improved estimation of prognosis over that provided by the use of baseline measures alone (40). Because the information that is "most recent" varies across patients, the predictive effects of a variable are not bound to a specific time point. This has both theoretical and clinical usefulness because it means that measurement of a variable has prognostic value when taken at various points in the prediction interval-up to 15 months postdiagnosis in the present case.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 
Patient Characteristics
Table 1 displays the demographic and medical characteristics of the sample, measured at baseline, arranged according to patient mortality status 10 years later. Of the 205 patients, 125 were alive at this time whereas 80 had died. The median survival time was 110 months (range = 3 to 131). Approximately 48% of the sample had breast cancer; the remainder had one of 14 other diagnoses, including lung, colon, head and neck, prostate, uterine, ovarian, and colon and rectal cancers. Compared with new cases in the North American population at the time of study accrual (mid-1980s) (41), this sample included a higher proportion of new breast cancer patients; the remaining diagnoses were sampled at a rate closely reflecting the proportion of new cancer cases.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Demographic, Medical, and Psychological Characteristics at Baseline, According to Vital Status at End of Follow-Up (n = 205)
 
Table 1 indicates a moderate degree of psychological distress in this sample (14, 42, 43), but patients reported fair to good physical health. Comparisons on the psychological variables at baseline between those alive vs. deceased generally show small differences in the expected direction. All psychological variables were significantly intercorrelated (p < .0001), ranging from 0.65 (depression and anxiety) to -0.21 (impact of events and perceived control).

Baseline Prediction of Survival
The left side of Table 2 presents HR and 95% CI from baseline data analyses. The core model using the 7 medical and demographic risk factors was highly significant, {chi}2 (7) = 58.72, p < .0001. When adjusted for other medical and demographic variables, cancer site (nonbreast), higher histopathological grade and nodal status, and higher cancer stage were significant predictors of survival time. Three baseline psychological variables were univariately significant predictors of survival after adjustment for the baseline covariates in the model-CES-D depressive symptoms, impact of events, and sense of control (see Table 2). With the use of a stepwise regression model that included all seven core risk factors before these three psychological variables were evaluated for entry, only depressive symptoms was a unique psychological predictor, HR = 1.03, CI = 1.00 to 1.06, p < .05 (model {chi}2 [8] = 62.22, p < .0001). In a stepwise model using median splits on the three psychological variables, only depressive symptoms significantly predicted survival time, HR = 1.62, CI = 1.02 to 2.56, p < .05.4


View this table:
[in this window]
[in a new window]
  		TABLE 2. HRs, CIs, and Significance of Variables from Cox Regression Models on Baseline and Time-Dependent Data
 
Figure 1 presents the survival curves for high and low CES-D score groups, based on a median split of the data, and after adjustment (45) for the most powerful medical risk factors from the baseline models: cancer site, histopathological grade, and nodal status. Patients with higher CES-D scores showed slightly lower survival rates from 5 months postdiagnosis. However, beginning at 15 to 25 months postdiagnosis, the survival rates of the two patient groups diverge substantially, suggesting a "time-delayed" effect of depressive symptoms on survival (46).



View larger version (14K):
[in this window]
[in a new window]
  		Figure 1. Overall survival according to level of depressive symptomology measured at baseline (3 months postdiagnosis). Curves have been adjusted for diagnostic category, histological grade, and nodal status.

 
Time-Dependent Prediction of Survival
The Cox regression model, including the risk factors measured both at baseline and over time (treatment status) was significant, {chi}2 (7) = 77.41, p < .0001. Cancer site and histopathological grade predicted survival, as did treatment status (see right side of Table 2). The significance of treatment in this analysis may reflect the fact that patients receiving systemic treatment up to 15 months after diagnosis were battling serious disease conditions.

Entry of each time-dependent psychological variable into separate models found four of prognostic significance: higher CES-D depressive symptoms, sense of control, impact of events, and subjective physical health (Table 2). In a stepwise regression model, only depressive symptoms emerged as a unique predictor, HR = 1.06, CI = 1.03 to 1.08, p < .0001 (model {chi}2 [8] = 81.03, p < .0001). With the use of median splits of the data in a stepwise model, depressive symptoms was the sole unique predictor of survival, median HR = 2.04, CI = 1.29 to 3.24, p < .005.

The prognostic importance of CES-D depressive symptoms in these time-dependent analyses suggests that this variable may reflect chronic levels of symptomology over time. This variable showed high stability over the four measurement points, {alpha} = 0.79. CES-D means across the four time points were 20.65 (SD = 8.95), 18.54 (SD = 8.12), 17.70 (SD = 6.83), and 17.94 (SD = 6.98), respectively. Repeated-measures ANOVAs on available patient data showed that depressive symptoms changed (dropped) significantly only between the first two measurement points [F(1,652) = 6.39, p < .01] and not thereafter (p values > .05).

It is possible that patients reporting more depressive symptomology had cancers with poorer prognoses or had more advanced disease. The t tests compared the level of depressive symptoms across levels of cancer site, stage, and treatment status at each measurement point. At baseline, breast cancer patients were more likely than nonbreast patients to have higher levels of depressive symptoms, p < .005. Yet breast patients showed better survival rates. No other differences were found (all p values > .05).

Supplementary Cox regression analyses confirmed that the effect of depressive symptoms on survival was not dependent on diagnostic category (through the use of an interaction term) in baseline and time-dependent models, both p values >.05. Neither were CES-D scores at each time point higher among patients receiving more toxic medical treatment (chemotherapy), all p values > .05.

Spearman correlations between depressive symptoms and both histological grade and nodal status showed no relations at any time point (all p values > .05). The possibility that the distress-survival relation was a "side effect" of advanced disease condition was not supported by survival analyses which excluded metastatic patients (N = 191), which again found a predictive role for depressive symptoms. Finally, the CES-D has seven somatic items (47), which, in medical populations, may tap symptoms of disease rather than depression (48). But survival analyses using nonsomatic CES-D scores (13 items) again found a significant predictive role for depressive symptoms in both univariate and multivariate models: in the baseline multivariate model, using continuous variables, HR = 1.06, CI = 1.01 to 1.10, p < .05; in the time-dependent multivariate model, HR = 1.05, CI = 1.01 to 1.10, p < .05. The full CES-D and nonsomatic CES-D scores were highly correlated at each time point (rrange = .87 to .91), indicating that full CES-D scores were not heavily influenced by responses to somatic items.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 
Past research into the role of psychological factors in long-term cancer prognosis has shown mixed results (13). Research in the area presents considerable challenges and the present study was designed to overcome a number of methodological and analytic problems that may have clouded results in this area. Using both early post-diagnostic and repeated measurements over a 15-month period, depressive symptomology was a consistent psychological predictor of shortened survival time in this heterogeneous sample of cancer patients. This was found after controlling for demographic and clinical risk factors.

Other research has linked depressive symptomology with reduced cancer survival (46, 49, 50). Watson et al. (50) found, as in the present research, both early postdiagnostic and time-dependent depressive symptom scores to predict survival time. Recent work has also linked this variable with shortened survival among heart disease patients (51) and with the incidence of stroke (52). The average CES-D score in this sample at baseline was higher than the oft-used clinical cut point, and the ability to detect a role for depressive symptoms in this sample may have been linked to these higher scores. This study tested the prognostic value of seven specific psychological factors, most of which have not often been measured in cancer survival research. Because depressive symptomology alone was generally shown to be of prognostic importance, past research, which found null results by using measures that combine specific states, may have underestimated the impact of such specific conditions as depression.

It must be noted that the psychological measures in this study were to some degree intercorrelated. Research has shown that depression may be a psychological concomitant of subjective stress (49), which the impact of events scale is designed to measure. Depressive symptoms and sense of control may also have a theoretical link. Taylor et al. (53) suggest that the effect of beliefs, including those of personal control, have their effect on health through emotional states such as depression. Thus, future attempts to converge on key psychological predictors of survival would still benefit by examining a number of potentially important psychological variables. Self-reported physical health status did not predict survival in this study. However, this relation has been found among the elderly (28) and those with advanced cancers, among whom physical concerns are more predominant (12).

Why is Depressive Symptomology Linked to Reduced Survival Time?
One explanation for the link between depressive symptoms and reduced survival is that distressed individuals believe that their lives are coming to an end, given either the generally poor prognosis for their particular cancer, the advanced stage of their disease, or both. If this explanation were true, distress would be a result of poor prognosis or shortened survival, not a predictor of it. Both of the Cox models showing depressive symptoms to have predictive value after adjustment for medical variables, and several supplementary analyses pointed to the independent importance of distress to survival.

Several mechanisms have been proposed to mediate or explain the distress-survival relation, including endocrinological and immunological pathways (49, 54, 55) and reduced medical compliance (54, however see 56, 57). If cancer precipitates distress (58) and distress affects cancer course, as the present results indicate, a positive feedback loop may be operating to further the disease. Specifically, cancer diagnosis and the physical effects of the disease may predispose to distress, which, if maintained over time, then enhances disease progression. Longitudinal research exploring the ongoing transaction between psychological and biological factors may enhance our understanding of such a process.

The present study had several limitations. In this sample, 75% were women, and they are known to evidence more depressive symptoms than men (59), a finding indirectly supported by the present results showing higher depressive symptoms among breast patients. Gender did not predict survival time, and breast cancer patients showed better survival rates, whereas patients with higher depressive symptomology had poorer survival rates. Although these findings indicate that the predictive effects of depressive symptoms were not "carried" by women and the breast cancer diagnosis, research in this area should use more balanced samples of men and women.

The nonbreast cancer group in the study sample included cancers with very different prognoses, and potentially different psychological consequences. Also, two clinical risk factors used as predictors here-nuclear grade and nodal status-may have different prognostic value for different cancer diagnoses. Clinical risk factors not included here, such as comorbidity (60), socioeconomic status (61), and smoking status (62) could potentially reduce the relation between psychological state and survival. Also, whereas the need to preserve a sizable and balanced grouping required the division of cancer treatment into systemic vs. local treatment, this may have masked mortality-relevant differences between, for example, those receiving chemotherapy and those receiving radiation. Relatedly, whereas treatment is administered according to the disease circumstances of each patient, future research should examine the effect of distress on medical treatment decision making.

Finally, patients in the study were self selected, and only 39% of available patients were enrolled. Study enrollment tends to be lower in more serious disease populations, and the average accrual into cancer clinical trials is <3% of the cancer patient population (63). However, self selection of a minority of a population raises the issue of generalizability. This issue has been considered most problematic in psycho-oncology research when the resulting sample is unrepresentative of the psychological state of the population in question. The present sample showed a broad spectrum of scores on all psychological variables and the levels of distress were comparable with those found in other samples of heterogeneous cancer patients. Aside from attending physician refusal, highly distressed patients may have been less likely than others to volunteer for this research (64). If so, the present results may represent a conservative estimate of the impact of depressive symptoms on survival. That said, these results cannot be generalized beyond the kind of patient who volunteers for psycho-oncology research.

To conclude, the present study found that both baseline and time-dependent measures had value in predicting long-term survival. Replication of these results is in order, and future research of this kind will be benefited by ever-tighter control over demographic, clinical risk, and endocrine and immune system factors, homogeneous diagnostic and treatment groups, and large patient samples with many events to allow for the testing of many predictors. However, this research suggests that screening for depressive and other distress symptoms could fruitfully be done at any point in the approximately year-long interval after cancer diagnosis. Levels of distress remained quite stable over this interval in this sample (65). Such screening could serve as a basis of referral for therapeutic intervention, given that psychotherapy has been found to reduce depressive symptoms (54). Such screening could not only assess current symptom levels, but also probe for prediagnostic history of distress (66). Research probing prediagnostic psychological history may be particularly important to understanding why depressive symptoms measured in the early postdiagnostic period can predict long-term survival, given the relative importance of enduring rather than short-lived psychological conditions in physical health (15).


   ACKNOWLEDGMENTS
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
REFERENCES
 
This research was supported by the National Cancer Institute of Canada with funds from the Canadian Cancer Society to the first author. Comments by Nancy Frasure-Smith, Elizabeth Maunsell, and Susan Scott resulted in significant improvements to the presentation of this research. We are grateful to Michal Abrahamowicz, Rhonda Amsel, Tim Carter, Thierry Ducruet, and Hubert Wong for statistical assistance. We especially thank Myriam Strukuska, Anna Abramovitch, Kim Davidman, Natasha Rossi, Yvon Papillon, Aly Wener, and Clarence White.


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
ACKNOWLEDGMENTS
 REFERENCES
 
1 For lymphoma, a histopathological grade was based on the Working Formulation (19). The number of positive lymph nodes was grouped into categories; the Ann Arbor classification (19) was used to facilitate scoring for several cases. Back

2 Preliminary analyses showed that both variables were nonsignificant predictors of survival. But variables not used in analyses can be used in MI to provide possible additional evidence for missing values on other variables (37). Back

3 When missing data on lymph node invasion were present for brain tumor patients, a value of zero was set, given that most brain cancers do not spread to regional lymph nodes (38). Back

4 In analyses using a median split at a score of 16 on the CES-D, this variable did not predict survival time. Although it is often used to demarcate clinical depression, a CES-D score of 16 may be an unreliable cut point (44). In the present sample, 66.8% (N = 137) had scores of 16 or above at baseline (M = 20.65; median = 19). Back


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 NOTES
 ACKNOWLEDGMENTS
 REFERENCES
 

  1. Roud PC. Psychosocial variables associated with exceptional survival of patients with advanced malignant disease. J Natl Med Assoc 1987; 79: 97–102.[Medline]
  2. Cassileth BR, Lusk EJ, Miller DS, Brown LL, Miller C. Psychosocial correlates of survival in advanced malignant disease. N Engl J Med 1985; 312: 1551–5.[Abstract]
  3. Tross S, Herndon J, Korzun A, Kornblith AB, Cella DF, Holland JF, Raich P, Johnson A, Kiang DT, Perloff M, Norton L, Wood W, Holland JC. Psychological symptoms and disease-free and overall survival on women with stage II breast cancer. J Natl Cancer Inst 1996; 88: 661–7.[Abstract/Free Full Text]
  4. Hislop T, Waxier N, Coldman A, Elwood J, Kan L. The prognostic significance of psychosocial factors in women with breast cancer. J Chron Dis 1987; 40: 729–35.[CrossRef][Medline]
  5. Greer S. Psychological response to cancer and survival. Psychol Med 1991; 21: 43–9.[Medline]
  6. Waxler-Morrison N, Hislop TG, Mears B, Kan L. Effects of social relationships on survival for women with breast cancer: a prospective study. Soc Sci Med 1991; 33: 177–83.
  7. Spiegel D, Bloom JR, Kraemer HC, Gottheil E. Effect of psychosocial treatment on survival of patients with metastatic breast cancer. Lancet 1989; 2: 888–91.[CrossRef][Medline]
  8. Goodwin PJ, Leszcz M, Ennis M, Koopmans J, Vincent L, Guther H, Drysdale E, Hundleby M, Chochinov HM, Navarro M, Speca M, Hunter J. The effect of group psychosocial support on survival in metastatic breast cancer. N Engl J Med 2001; 345: 1719–26.[Abstract/Free Full Text]
  9. Richardson JL, Zarnegar Z, Bisno B, Levine A. Psychosocial status at initiation of cancer treatment and survival. J Psychosom Res 1990; 34: 189–201.[CrossRef][Medline]
  10. Ringdal GI. Correlates of hopelessness in cancer patients. J Psychosoc Oncol 1995; 13: 47–66.
  11. Jamison R, Burish T, Wallston K. Psychogenic factors in predicting survival of breast cancer patients. J Clin Oncol 1987; 5: 768–72.[Abstract/Free Full Text]
  12. Chang VT, Thaler HT, Polyak TA, Kornblith AB, Lepore JM, Portenoy RK. Quality of life and survival: the role of multidimensional symptom assessment. Cancer 1998; 83: 173–79.[CrossRef][Medline]
  13. Fox BH. The role of psychological factors in cancer incidence and prognosis. Oncology 1995; 9: 245–53.[Medline]
  14. Portenoy RK, Thaler HT, Kornblith AB, Lepore J, Friedlanderklar H, Coyle N, Smartcurley T, Kemeny N, Norton L, Hoskinds W, Scher H. Symptom prevalence, characteristics and distress in a cancer population. Qual Life Res 1994; 3: 183–89.[CrossRef][Medline]
  15. Herbert TB, Cohen S. Stress and immunity in humans: a meta-analytic review. Psychosom Med 1993; 55: 364–79.[Abstract/Free Full Text]
  16. Edgar L, Rosberger Z, Nowlis D. Coping with cancer during the first year after diagnosis: assessment and intervention. Cancer 1992; 69: 817–28.[CrossRef][Medline]
  17. Taylor SE, Armor DA. Positive illusions and coping with adversity. J Pers 1996; 64: 873–98.[CrossRef][Medline]
  18. Baider L, DeNour AK. Psychological distress and intrusive thoughts in cancer patients. J Nerv Ment Dis 1997; 185: 346–48.[CrossRef][Medline]
  19. Molina A. Non-Hodgkin’s lymphoma. In: Pazdur R, editor. Cancer management: a multidisciplinary approach. 5th ed. Melville NY: PRR; 2001. p. 585–622.
  20. Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Measurement 1977; 1: 385–401.[CrossRef]
  21. Hann D, Winter K, Jacobsen P. Measurement of depressive symptoms in cancer patients: evaluation of the center for epidemiological studies depression scale (CES-D). J Psychosom Res 1999; 46: 437–43.[CrossRef][Medline]
  22. Horowitz MJ, Wilner N, Alvarez W. Impact of event scale: a measure of subjective stress. Psychosom Med 1979; 41: 209–18.[Abstract/Free Full Text]
  23. Lewis FM, Firsich SC, Parsell S. Clinical tool development for adult chemotherapy patients: process and content. Cancer Nurs 1979; 2: 99–108.[Medline]
  24. Nowlis V. Research with the mood adjective check list. In: Tompkins SS, Izard CE, editors. Affect, cognition and personality. New York: Springer; 1965. p. 352–89.
  25. Brown KW. Coping and the expression of emotion among cancer patients. Unpublished data. Montreal Quebec CA: McGill University 1996.
  26. Stone AA. The association between perceptions of daily experiences and self-and spouse-rated mood. J Res Pers 1981; 15: 510–22.[CrossRef]
  27. Cantril H. The pattern of human concerns. New Brunswick NJ: Rutgers University, 1965.
  28. Idler EL, Kasl S. Health perceptions and survival: do global evaluations of health status really predict mortality? J Gerontol 1991; 46: S55–65.[Medline]
  29. Zubrod CG, Schneiderman M, Frei E, Brindley C, Gold GL, Shnider B, Oviedo R, Gorman J, Jones R, Jonsson U, Colisky J, Chalmers T, Ferguson B, Dederick M, Holland J, Selaway O, Regelson W, Lasagna L, Owens AH. Appraisal of methods for the study of chemotherapy of cancer in man. J Chronic Dis 1960; 11: 7–33.[CrossRef]
  30. Conill A, Verger E, Salamero M. Performance status assessment in cancer patients. Cancer 1990; 65: 1864–66.[CrossRef][Medline]
  31. Parmar MKB, Machin D. Survival analysis: a practical approach. New York: Wiley, 1995.
  32. SAS Institute. SAS/STAT user’s guide, version 6. 4th ed. Cary NC: SAS Institute, 1990.
  33. American Joint Committee on Cancer. AJCC cancer staging manual. 5th ed. New York: Lippincott Williams & Wilkins, 1997.
  34. Little RJA, Rubin DB. Statistical analysis with missing data. New York: Wiley, 1987.
  35. King G, Honaker J, Joseph A, Scheve K. Analyzing incomplete political science data: an alternative algorithm for multiple imputation. Am Political Sci Rev 2001; 95: 49–69.
  36. Schafer JL, Olsen MK. Multiple imputation for multivariate missing-data problems: a data analyst’s perspective. Multivariate Behav Res 1998; 33: 545–71.[CrossRef]
  37. Schafer JL. NORM version 2.02 for Windows 95/98 NT. http://www/stat/psu.edu/-jls/misoftwa.html;7/21/01.
  38. DeAngelis LM, Loeffler JS, Namelak AN. Primary brain tumors. In: Pazdur R, editor. Cancer management: a multidisciplinary approach. 5th ed. Melville NY: PRR; p. 507–22.
  39. Markovitz JH, Matthews KA, Kannel WB, Cobb JL, D’Agostino RB. Psychological predictors of hypertension in the Framingham study. JAMA 1993; 270: 2439–43.[Abstract]
  40. Christensen E, Schlichting P, Andersen PK, Fauerholdt L, Schou G, Pedersen BV, Juhl E, Poulsen H, Tygstrup N. Copenhagen Study Group for Liver Diseases. Updating prognosis and therapeutic effect evaluation in cirrhosis with Cox’s multiple regression model for time-dependent variables. Scand J Gastroenterol 1986; 21: 163–74.[Medline]
  41. Silverberg E, Lubera J. Cancer statistics. CA Cancer J Clin 1986; 36: 9–25.[Free Full Text]
  42. Beeber LS, Shea J, McCorkle R. The center for epidemiologic studies depression scale as a measure of depressive symptoms in newly diagnosed patients. J Psychosoc Oncol 1998; 16: 1–20.
  43. Epping-Jordan JE, Compas BE, Howell DC. Predictors of cancer progression in young adult men and women: avoidance, intrusive thoughts, and psychological symptoms. Health Psychol 1994; 13: 539–47.[CrossRef][Medline]
  44. Santor DA, Zuroff DC, Ramsay JO, Cervantes J, Palacios J. Examining scale discriminability in the BDI and CES-D as a function of depression severity. Psychol Assess 1995; 7: 131–39.[CrossRef]
  45. Nieto FJ, Coresh J. Adjusting survival curves for confounders: a review and a new method. Am J Epidemiol 1996; 143: 1059–68.[Abstract/Free Full Text]
  46. Herrmann C, Brand-Driehorst S, Kaminsky B, Leibing E, Staats H, Rueger U. Diagnostic groups and depressed mood as predictors of 22-month mortality in medical inpatients. Psychosom Med 1998; 60: 570–77.[Abstract/Free Full Text]
  47. Sheehan TJ, Fifield J, Reisine S, Tennen H. The measurement structure of the Center for Epidemiologic Studies Depression scale. J Pers Assess 1995; 64: 507–21.[CrossRef][Medline]
  48. Devins GM, Orme CM, Costello CG, Binik YM, Frizzell B, Stam HJ, Pullin WM. Measuring depressive symptoms in illness populations: psychometric properties of the Center for Epidemiologic Studies Depression (CES-D) scale. Psychol Health 1988; 2: 139–56.
  49. McDaniel JS, Musselman DL, Nemeroff CB. Cancer and depression: theory and treatment. Psychiatr Ann 1997; 27: 360–64.
  50. Watson M, Haviland JS, Greer S, Davidson J, Bliss JM. Influence of psychological response on survival in breast cancer: a population-based cohort study. Lancet 1999; 354: 1331–36.[CrossRef][Medline]
  51. Lesperance F, Frasure-Smith N. Depression in patients with cardiac disease: a practical review. J Psychosom Res 2000; 48: 379–91.[CrossRef][Medline]
  52. Ostir GV, Markides KS, Peek MK, Goodwin JS. The association between emotional well-being and the incidence of stroke in older adults. Psychosom Med 2001; 63: 221–30.[Abstract/Free Full Text]
  53. Taylor SE, Kemeny ME, Reed GM, Bower JE, Gruenewald TL. Psychological resources, positive illusions, and health. Am Psychol 2000; 55: 99–109.[CrossRef][Medline]
  54. Spiegel D. Cancer and depression. Br J Psychiatry 1996; 168 (Suppl 30): 109–16.
  55. Allen-Mersh TG, Glover C, Fordy C, Henderson DC, Davies M. Relation between depression and circulating immune products in patients with advanced colorectal cancer. J R Soc Med 1998; 91: 408–13.[Abstract]
  56. Ayres A, Hoon PW, Franzoni JB, Matheny KB, Cotanch PH, Takayanagi S. Influence of mood and adjustment to cancer on compliance with chemotherapy among breast cancer patients. J Psychosom Res 1994; 38: 393–402.[CrossRef][Medline]
  57. Richardson JL, Marks G, Johnson CA, Graham JW, Chan KK, Selser JN, Kishbaugh C, Barranday Y, Levine AM. Path model of multidimensional compliance with cancer therapy. Health Psychol 1987; 6: 183–207.[CrossRef][Medline]
  58. Derogatis LR, Morrow GR, Fetting J, Penman D, Piasetsky S, Schmale AM, Henrichs M, Carnicke CLM. The prevalence of psychiatric disorders among cancer patients. JAMA 1983; 249: 751–7.[Abstract]
  59. Nolen-Hoeksema S. Sex differences in depression. Stanford CA: Stanford University Press, 1990.
  60. Repetto L, Venturino A, Vercelli M, Gianni W, Biancardi V, Casella C, Granetto C, Parodi S, Rosso R, Margliano V. Performance status and comorbidity in elderly cancer patients compared with young patients with neoplasia and elderly patients without neoplastic conditions. Cancer 1998; 82: 760–65.[CrossRef][Medline]
  61. Kunst AE, Groenhof F, Mackenbach JP, Health EW. Occupational class and cause specific mortality in middle aged men in 11 European countries: comparison of population based studies. BMJ 1998; 316: 1636–42.[Abstract/Free Full Text]
  62. American Cancer Society. Cancer facts and figures. Atlanta, GA: American Cancer Society, 1999.
  63. Gotay CC. Accrual to cancer clinical trials: directions from the research literature. Soc Sci Med 1991; 33: 569–77.
  64. Spiegel D. Psychological distress and disease course for women with breast cancer: one answer, many questions. J Natl Cancer Inst 1996; 88: 629–31.[Free Full Text]
  65. Hopwood P, Stephens RJ. Depression in patients with lung cancer: prevalence and risk factors derived from quality-of-life data. J Clin Oncol 2000; 18: 893–903.[Abstract/Free Full Text]
  66. Maunsell E, Brisson J, Deschenes L. Psychological distress after initial treatment of breast cancer. Cancer 1992; 70: 120–25.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 Psychosom. Med.Home page
K. L. Weihs, T. M. Enright, and S. J. Simmens
Close Relationships and Emotional Processing Predict Decreased Mortality in Women with Breast Cancer: Preliminary Evidence
Psychosom Med, January 1, 2008; 70(1): 117 - 124.
[Abstract] [Full Text] [PDF]

Home page
 Psychosom. Med.Home page
P. K. Myint, R. N. Luben, P. G. Surtees, N. W. J. Wainwright, A. A. Welch, S. A. Bingham, N. J. Wareham, R. D. Smith, I. M. Harvey, and K.-T. Khaw
Self-Reported Mental Health-Related Quality of Life and Mortality in Men and Women in the European Prospective Investigation into Cancer (EPIC-Norfolk): A Prospective Population Study
Psychosom Med, June 1, 2007; 69(5): 410 - 414.
[Abstract] [Full Text] [PDF]

Home page
 JCOHome page
J. M. Prieto, J. Atala, J. Blanch, E. Carreras, M. Rovira, E. Cirera, A. Espinal, and C. Gasto
Role of Depression As a Predictor of Mortality Among Cancer Patients After Stem-Cell Transplantation
J. Clin. Oncol., September 1, 2005; 23(25): 6063 - 6071.
[Abstract] [Full Text] [PDF]

Home page
 JCOHome page
K. Ell, K. Sanchez, B. Vourlekis, P.-J. Lee, M. Dwight-Johnson, I. Lagomasino, L. Muderspach, and C. Russell
Depression, Correlates of Depression, and Receipt of Depression Care Among Low-Income Women With Breast or Gynecologic Cancer
J. Clin. Oncol., May 1, 2005; 23(13): 3052 - 3060.
[Abstract] [Full Text] [PDF]

Home page
 PsychosomaticsHome page
L. Grassi, E. Rossi, S. Sabato, G. Cruciani, and M. Zambelli
Diagnostic Criteria for Psychosomatic Research and Psychosocial Variables in Breast Cancer Patients
Psychosomatics, December 1, 2004; 45(6): 483 - 491.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Brown, K. W.
Right arrow Articles by Edgar, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Brown, K. W.
Right arrow Articles by Edgar, L.
Related Collections
Right arrow Cancer
Right arrow Depression
Right arrow Stress and Coping

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS