- Email: [email protected]
Psychosocial status at initiation of cancer treatment and survival
Journal Printed
of Psychosomatic in Great Bntain
Research.
Vol.
34, No.
2. pp.
189-20
I, 1990. 0
0022-3999/90 $3.00 + .oO 1990 Pergamon Press plc
PSYCHOSOCIAL STATUS AT INITIATION OF CANCER TREATMENT AND SURVIVAL JEAN L. RICHARDSON,*~ ZOHREHZARNEGAR,~ BETSYBISNO~ and ALEXANDRA LEVINE~ (Received
11 May 1989;accepted in revised form 7 September 1989)
Abstract-Ninety-two newly diagnosed patients with hematologic malignancies treated with chemotherapy, and 47 patients with rectal cancer treated with abdominal-perineal resection were prospectively studied to assess the relationship between mood and survival. Hematology patients were measured within one week of diagnosis and were remeasured at six months. Rectal cancer patients were measured within three months of surgery and remeasured six months later. Medical records were abstracted to obtain data on treatment given, disease characteristics and outcome of treatment. On univariate analyses using Cox regression, we examined the effect of depression, coping style and locus of control on survival. None of these variables were found to be significantly related to survival whether assessed at intake or six months later. Furthermore there were no statistically significant correlations between these factors and subsequent survival at two years or with long-term survival. Biological prognostic variables including extent of disease for rectal cancer patients and severity of specific type of hematologic cancer were significantly related to survival. Although psychosocial adjustment is important for the quality of life experienced by cancer patients it was not related to length of survival in this study. Further exploration of this issue should be conducted using patients with a single site and preferrably an early stage of disease.
INTRODUCTION
Is SURVIVAL after the diagnosis of cancer influenced solely by the biological characteristics of the disease or might psychological factors such as depression, coping style, locus of control, or a pessimistic outlook influence survival as well? Physicians, psychologists and lay people have anecdotally expressed a belief that people they have known who were diagnosed with cancer survived for a longer period because they had a ‘fighting spirit’ or survived for a shorter time because they ‘just gave up’. Although conventional wisdom suggests there may be a relationship between psychological factors and survival, studies of this relationship have been inconsistent and inconclusive when applied to malignant disease. Several articles that review the literature relating psychosocial factors to the course of malignant disease indicate that the results are often contradictory and noncomparable [l&4]. In fact, the study of psychosocial factors as predictors of cancer survival has been fraught with methodological difficulties that have contributed to the lack of coherence in construct relationships [3]. The difficulties include (1) *Department of Preventive Medicine, USC School of Medicine. TInstitute for Health Promotion Disease Prevention Research, USC School of Medicine. IVeterans Administration, Outpatient Center, Los Angeles. §Department of Medicine, USC School of Medicine. This work was supported by Grant number CA 31151 ‘Assessment and Enhancement of Compliance with Chemotherapy’ DHHS, NCI; Grant number T32 CA09492 ‘Cancer Control Research Training Grant’; and American Cancer Society California Division Postdoctoral Fellowship Grant. The authors acknowledge the assistance of Dr Mark Krailo for consultation concerning statistical procedures. Requests for reprints should be sent to: Dr Jean Richardson, 1420 San Pablo St, PMB A-301, Los Angeles, CA 90033, U.S.A. 189
190
J. L. RICHARDSONet al.
misguided attempts to treat cancer as a single disease and to expect homogeneity in the role of psychosocial factors in survival, (2) failure to measure the same psychosocial phenomena across studies, (3) failure to conduct separate analyses for different sites of disease and treatments for disease, (4) failure to analyze data using lifetable analyses rather than median or criterion lengths of survival, and (5) failure to differentiate whether etiology versus progression of disease is being studied. Several authors have indicated that a coping style categorized as ‘fighting spirit’ as opposed to ‘passive and hopeless’, is correlated with longer survival [558]. Rogentine et al. [9] found that those who had more difficulty adjusting to their diagnosis were less likely to relapse than those who reported less difficulty. Depression, which is a similar though not identical construct to hopelessness, may reduce survival time. although some studies have indicated no such relationship between depression and survival [lo], and still others have found that more depressed patients survive longer [l I]. Many authors have failed to substantiate a relationship between psychological factors and survival [12-161. Of particular note in this regard is a study by Cassileth et al. [12], of 204 advanced unresectable cancers of various sites, and 155 early cases of melanoma (Stage I or II) or breast cancer (Stage II). None of the psychological factors studied (hopelessness, depressive affect or adjustment to diagnosis) related to survival for either patient group. A study by Jamison et al. [ 131 of 49 metastatic breast cancer patients indicated that results of a variety of scales (depression, self esteem, locus of control) did not relate to survival. A study by Hislop et al. [17], provided partial support for a relationship between expression of emotion and survival but not for the role of locus of control, depression, coping style, self-esteem or a number of other psychological factors on survival. It is apparent that the evidence for a relationship between psychological processes and survival is less than definite although mechanisms for a behavioral or biological relationship have been hypothesized. For example, a more assertive patient might demand more thorough medical evaluation, exhibit greater compliance, or maintain better self-care than might occur with a more passive patient. Furthermore, mood states could influence complex endocrinological or immunological pathways which might alter the growth rate of certain tumors. We therefore initiated a study to examine prospectively, the role of psychological variables in subsequent patient survival. The constructs examined were similar to those previously examined i.e. locus of control, depression and coping style. The study design addressed many of the shortcomings mentioned above. METHODS
Data were obtained from two groups of cancer patients: 92 patients with hematologic and 47 patients with rectal cancer. Questionnaires and procedures were comparable although between the groups allowing us to examine the same constructs within each group. interviewed privately by a trained interviewer. Each group was analyzed separately. This compare the specificity or robustness of relationships between psychological factors separately for the two groups of patients.
malignancies, not identical Patients were allowed us to and survival
Rrctcrl catneT scm~plr. Patients were accrued from 32 private practtce surgeons. and two large cancer hospitals in the Los Angeles area. All subjects had a permanent abdominal perineal resection (permanent
Survival
and mood
191
colostomy) performed due to a diagnosis of rectal cancer, were over 18 years of age, spoke English, and were cognitively capable of understanding and completing a consent form and measures. Potential subjects were sent letters and contacted by telephone to request their participation in the study. Of those meeting the selection criteria, 70% agreed to participate. Those who declined typically mentioned that they did not want any extra burdens on them at this time or that they did not want anyone probing into their personal lives or both. All subjects were interviewed within one to three months of surgery and reinterviewed six months later. Patients who survived less than 90 days were not included in the analysis because coping style and depression might be unreliably measured due to imminent death. Fifty subjects were recruited, of these three were eliminated for this reason. Hematologic cancer. Patients who were newly and consecutively diagnosed with hematologic malignancies were accrued at the Los Angeles County-University of Southern California Medical Center. All patients were treated with combination chemotherapy. Participants were over 18 years of age, spoke English, or Spanish and were cognitively capable of understanding and completing a consent form and measures. Of those meeting the selection criteria 80% agreed to participate. Hematology patients were interviewed within one to two weeks of diagnosis and reinterviewed six months later. Of the 108 patients followed, 14 were not included in this analysis because they were previously treated for cancer and two were eliminated because they died within 90 days of recruitment. Measures Demographics. Information concerning gender, ethnicity, marital status, education, income and age was collected from respondents in both studies. Severity. Rectal cancer: medical severity was determined from a medical abstract completed by the physician. This abstract contained information on the cancer diagnosis, other medical problems, extent of disease (local, adjacent, regional or distant), length of hospitalization, complications and adjuvant use of chemotherapy or radiation. Hematologic malignancy: the patients were grouped into three disease severity or prognostic levels according to risk of death [18]. The high severity group (37%) had aggressive malignancies, requiring intensive multi-agent chemotherapy. Although initial response to therapy in these illnesses may be good, the average survival for affected patients is generally less than two years. The moderate severity group (24%) consisted of patients with malignancies that generally demonstrate good response to therapy, and average survival of two to five years. The remaining 36% of the patients had typically slow progressing diseases with expected average survival of five years or more.* Most regimens for the treatment of hematologic malignancies use multiple chemotherapeutic agents. In the present study, these regimens were categorized as either ‘easy’ or ‘complex’ based on the difficulty of tolerating the treatment and the number of chemotherapeutic agents utilized. Thirty-two per cent of patients took ‘easy’ regimens and 68% underwent ‘complex’ regimens.? Depression. Rectal cancer: depression was assessed using the Beck depression inventory (BDI) [19]. This is a 21 item inventory with items scored from 0 to 3 and total scores from 0 to 63. Item scores are summed to create a total, with higher scores indicating a greater degree of depression. In addition to the total BDI, we also examined a physical subscale and a psychological subscale. Hematologic malignancy: the 20 item Zung self-rating depression scale (SDS) [2&23] was orginally divided into 3 subscales of pervasive affect, physiological and psychological components. Each item is scored from 1 to 4 and using the following algorithm to create the final score (sum raw store/80*100) [21]. Although we preferred to use the BDI with this group as well, the BDI requires more reading and caused difficulty for the less well educated population at Los Angeles County-USC Medical Center. A depression scale was administered to each sample at intake and at six month follow-up. *The disease severity categories for hematologic malignancy are as follows: High = (acute myeloblastic leukemia, acute lymphocytic leukemia, small non-cleaved lymphoma, lymphoblastic T-cell lymphoma, immunoblastic sarcoma (lymphoma)); Moderate = (large cleaved lymphoma, large non-cleaved lymphoma, multiple myeloma); Low = (chronic granulocytic leukemia, chronic lymphocytic leukemia, Hodgkin’s disease, plasmacytoid lymphocytic lymphoma, small cleaved lymphoma). tThe chemotherapy regimen complexity categories of patients with hematologic malignancy are as follows: Easy = (1) melphelan and prednisone; (2) cyclophosphamide, vincristine, and prednisone (CVP) and (3) daily oral chlorambucil, with or without prednisone. Complex = (1) nitrogen mustard, vincristine, procarbazine, and prednisone (MOPP); (2) hydroxyldaunamycin, bleomycin, vincristine, and DTIC (ABVD); (3) hydroxyldaunamycin, cyclophosphamide, vincristine, and prednisone (CHOP); (4) bleomycin, hydroxyldaunamycin, cyclophosphamide, vincristine, and prednisone (BACOP); and (5) methotrexate, bleomycin, hydroxyldaunamycin, cyclophosphamide, vincristine, and dexamethasone (M-BACOD).
192
J. L. RICHARDSON et al
Locus ofconrrol. Beliefs about the factors controlling one’s health were measured with nine items from the Multidimensional Health Locus of Control Scale [24-251. We used three self-control (internal items. three doctor control items, and three chance control items selected on the basis of the highest item-subscale correlations (see Appendix). This was administered at the intake interview for both samples. Moos coping SC&. For this study we used a coping assessment which was part of the health and daily living form measure developed by Moos et al. [26] and the internal consistency alpha has been reported. Respondents are asked to rate their frequency of use (four-point scale from 0 to 3) of 32 different coping responses. The coping scale is divided into method of coping (active behavioral (alpha = 0.74), cognitive (alpha = 0.62), and avoidance (alpha = 0.60)) and focus of coping (logical analysis (alpha = 0.58). information seeking (alpha = 0.64). problem solving (alpha = 0.63). aflective regulation (alpha = 0.5 1) and emotional discharge (alpha = 0.54)). For colostomy patients, this scale was administered on the first interview and for hematology patients at the follow-up interview. SurtGoalmeasures. For patients with hematologic malignancy. study enrolment was opened in April 1982 and closed in June 1985. Patients were evaluated for survival status in June 1987. For rectal cancer patients, enrolment was opened in February 1983 and closed in August 1984. Patients were evaluated for survival status in December 1987. Patients who survived less than 90 days were not included in the analysis because coping style and depression might be unreliably measured due to imminent death. Two hematology subjects and three rectal cancer subjects were eliminated for this reason. Sturistical nwthod,v
Survival time in days was taken to be the time from study enrolment to last contact for hematology patients and from date of surgery to last contact for rectal cancer patients. Death. regardless of cause, was considered an event; in all other cases (e.g. lost to follow-up, or alive at termination of the study) the individual was considered censored at last contact. A patient who was alive at last contact was considered to be censored with respect to treatment failure. Analyses were repeated for short term survival. ccnsorinp all cases who survived 24 months or more. Plots of the estimated survivor functions were constructed using the method of Kaplan and Meier [27]. Univariate prognostic factor evaluations were carried out by means of the proportional hazards model of Cox [27]. The risk ratios for each variable and 95% confidence intervals were computed from the beta coefficients and their standard errors. The relative risk is a ratio of the dealh rate in a given group over that of the baseline group.
RESULTS
Subject
charucteristics
Descriptive socioeconomic characteristics and specific diagnosis of the patients are shown in Tables I and II. The sample of subjects with hematologic malignancy was younger than the rectal cancer sample because these diseases affect a younger age group. The hematology sample was also of lower SES and more likely to be Hispanic reflecting the fact that they were accrued from the Los Angeles County-USC Medical Center which serves pritnarily indigent, minority patients. The rectal cancer patients were referred from the private medical community and were more likely to be white and middle class.
We wondered whether patients with poor prognosis might experience more psychological distress than those with better prognosis. We thus initially tested the correlation between the measures of severity of disease and the psychological variables studied. No significant relationships (all ps > 0.10) were found between the psychological measures for each sample (depression, locus of control. or coping) and severity of illness (hematologic) or extent of disease (rectal cancer).
Descriptive and comparative statistical analyses were calculated. In each comparative analysis at least two categories of predictor variable for each sample was tested
Survival
and mood
193
TABLE I.-DESCRIPTIVE CHARACTERISTICS OF SAMPLE Colostomy patients N = 41
Gender Male Female Ethnic group Caucasian Black Hispanic/Latin0 Asian/Other Marital status Single Married Divorced Widowed Other Education completed Graduate of professional school College or university Partial college High school graduate Partial high school Junior high school (7-9) Under 7 years of schooling Total ho&hold income Under $5.000
$5,00&$9,999 $10,00~_$14,999 $15,OO(r$l9,999 $20,00&$24,999 %25,00&_$39,999 %40,000or over Don’t know or refused to answer Age Under 20 years old 20-29 yrs 30-39 yrs 4049 yrs 50.-59yrs 6069 yrs 70-79 yrs 2 80 yrs
Hematology patients N = 92
33 (70.2%) 14 (29.8%)
57 (62.0%) 35 (38.0%)
40 (85.1%) 4 (8.5%) 2 (4.3%) 1 (2.1%)
16 (17.4%) 22 (23.9%) 51 (55.4%) 3 (3.3%)
3 (6.4%) 31 (65.9%) 6 (12.8%) 6 (12.8%) 1 (2.1%)
31 (33.7%) 33 (35.9%) 7 (7.6%) 14 (15.2%) 7 (7.6%)
3 (6.4%) 6 (12.7%) 10 (21.3%) 15 (31.9%) 7 (14.9%) 4 (8.5%) 2 (4.3%)
1 4 17 12 20 9 29
6 (12.8%) 5 (10.6%) 5 (10.6%) 2 (4.3%) 5 (10.6%) 11 (23.5%) 5 (10.6%) 8 (17.0%)
15 (16.3%) 18 (19.6%) 4 (4.4%) 4 (4.4%) 1 (1.1%) 1 (1.1%) 1 (1.1%) 48 (52.2%)
0 (0.0%) 0 (0.0%) 1 (2.1%) 3 (6.4%) 10 (21.3%) 17 (36.2%) 13 (27.6%) 3 (6.4%)
3 25 16 14 13 15 5 1
(1.1%) (4.4%) (18.5%) (13.0%) (21.7%) (9.8%) (31.5%)
(3.3%) (27.2%) (17.4%) (15.2%) (14.1%) (16.3%) (5.4%) (1.1%)
using Kaplan Mier plots and the logrank statistic and a Cox proportional hazards regression. These two analyses confirm one another and the Cox regressions are reported in Table IV. Descriptive The Kaplan Mier survival curves for rectal cancer and hematologic cancer patients are presented in Fig. 1. The curves indicated that hematologic cancer patients had a poorer survival outlook, however, the curves were not statistically different from each other (Log rank x2 = 1.516, p = 0.9970). Table III presents the number and per cent of patients in each group who were dead and censored during each year of follow-up. Overall, the per cent confirmed dead at the termination of the study was remarkably similar, 50% for the rectal cancer patients and 44% for the hematology
J. L. RICHARDSON
194 TABLE
el al.
II.-SEVERITY OF DISEASE FOR EACH SAMPLE
N Hematology Disease severity High Moderate Low Regimen complexity Easy Complex Colostomy Extent of disease Local Adjacent Regional Distant (5 missing) Radiation No Yes (4 missing) Chemotherapy No Yes (7 missing) Hospital stay 14 days or less 15-21 days 22~ 30 days More than 30 days (2 missing)
%
33 23 36
35.9% 25.0% 39.1%
29 63
32.0% 68.0%
8 17 6 II
19.0~/” 40.5% 14.3% 26.2%
26 17
60.5% 39.5%
23 17
S7.SV0 42.5%
20 I6 4 5
44.4% 35.6% 8.9% 11.1%
cancer patients. More of the hematology patients were censored earlier. which we attribute to the inherent diff’erences in the subjects and the sources of subject accrual. The purpose of the Figure and Table III is not specifically to compare the two groups but rather to provide the distribution of the dependent variable of survival outcomes.
-----
HMdTOLOGIC
8.75-
e.ee: 1
B
FIG. 1.-Survival
curves
for patients
4
3
2
5
YEARS
with hematologic
cancer
and with rectal cancer.
Survival TABLE III.-SURVIVOR
195
and mood
STATUSFOR COLOSTOMYAND HEMATOLOGYPATIENTS
Hematologic cancer N = 92
Rectal cancer N = 47 -___-.
Censored
Dead No. End of End of End of End of End of (4yr
year 1 year 2 year 3 year 4 follow-up 10mo)
5 6 5 6
I 23
%
cum. %
No.
11% 13% 11% 13% 2%
11% 24% 35% 48% 50%
0 2 3 9 10
50%
24
Dead
%
cum. %
No.
0% 4% 6% 19% 21%
0% 4% 10% 29% 50%
17 18 2 2 2
50%
41
Censored
%
cum. %
No.
%
cum. %
18% 20% 2?/0 2% 2%
18% 38% 40% 42% 44%
12 7 8 17 7
13% 8% 9% 18% 8%
13% 21% 30% 48% 56%
44%
51
56%
Demographics
Seven demographic variables were analyzed for their relationship to survival (Table IV). These included age, gender, marital status, ethnicity, education, income and occupation. Only gender for hematology patients (p = 0.11) and age for colostomy patients (p = 0.09) were even marginally related to survival. None of the variables measuring socioeconomic status were predictive of survival. Disease/treatment
Severity of specific disease was predictive of survival among hematology patients. Both extent of disease and hospital stay were predictive of survival among rectal cancer patients. Treatment characteristics were not predictive of survival in either group. Thus, regimen complexity was not predictive of survival for hematology patients and use of radiation and/or chemotherapy were not predictive of survival among rectal cancer patients. Psychological
processes
The relationship between survival and depression, coping style, and locus of control are presented in Table IV. These data indicate that none of these variables were related to length of survival for either sample of patients. In addition to the total depression scales presented, we divided the Beck and the Zung self-rating depression scales into psychological and physical subscales which were also not significantly related to survival (see Appendix). We also computed the relationship between the six-month follow-up measures of depression for each sample and survival and again found no relationship. We also examined the effect of these variables on two-year survival and found that the relationship between survival and these factors was again not significant. DISCUSSION
The analyses of the relationship between psychosocial factors and survival for each of these two groups failed to produce evidence of any significant relationship between the two. Tests we conducted for long and short-term survival. There were no significant associations between the psychosocial variables and survival for either
Disease severity Low Moderate High
Disease/treatment
Age 1741 over 41 Marital Status Single Married Other Ethnicity Black Hispanic Other Occupation Low Medium High Income Under %lOK $IOKp$20K Over $20K Education Under 7 yrs I yrs to H.S.
Sex Male Female
Demographics
Variable
1.0 0.67 0.89 1.0 1.11 1.16 I.0 0.45 I .06 I.0 1.15
22 51 19 53 25 6 33 8 3 29 41
1.0 3.78 1.85
1.0 0.85 0.68
31 33 28
36 23 33
1.0 1.48
0.57
35
46 46
1.0
RR
51
No.
RISK
1.68, 8.54 0.84, 4.09
0.53, 2.45
0.13, 1.55 0.24, 4.64
0.55. 2.23 0.34, 3.89
0.32, 1.41 0.38. 2.08
0.41, 1.77 0.31, 1.47
0.77. 2.85
0.29, 1.13
95% CI
Hematology
TABLE IV.-RELATIVE
10.001 0.13
0.73
0.21 0.94
0.77 0.81
0.30 0.78
0.67 0.32
0.24
0.11
P
OF DEATHFOR
8 17 6 II
28 19
Education H.S. or less More than H.S.
Extent Local Adjacent Regional Distant
18 21
7 12 24
Occupation Low Medium High Income Under $20K $20K or more
40 7
Ethnicity Caucasian Other
22 25
33 14
No.
1.0 4.08 6.39 14.74
1.0 0.8 1
1.0 1.49
1.0 0.65 0.58
1.04
0.49, 34.19 0.29, 142.88 1.86, 117.09
0.34. 1.94
0.59. 3.80
0.18, 2.31 0.18. 1.85
0.31, 3.53
0.42. 2.37
1.0 1.0
1.0
0.19. 1.06
0.46. 2.54
95% CI
1.0 0.45
1.0 1.08
RR
Colostomy
HEMATOLOGYPATIENTS
31 16
AND
Age 64 or less over 64 Marital Status Married Other
Sex Male Female
Variable
COLOSTOMY
0.19 0.11 0.01
0.64
0.40
0.50 0.36
0.95
0.99
0.07
0.87
P
High Express emotion (Moos) Low High Logical reasoning (Moos) Low High Affective (Moos) Low High
Zung depression scale 59 or under Over 59 Behavioral coping (Moos) Low High Cognitive coping (Moos) Low High Avoidance coping (Moos) Low High Se;ci$ormation (Moos)
Psychosocial
Regimen complexity Difficult Easy
1.0 1.0 1.0 0.94 1.0 1.15 1.0 1.08 1.0 1.02 1.0 0.68 1.0 1.43 1.0 0.62
32 31 32 32 28 35 30 34 29 33 28 35 31 32
0.82
1.0
II 13
63 29
0.28, 1.39
0.62, 3.27
0.30, 1.53
0.45, 2.36
0.48, 2.44
0.52, 2.58
0.42, 2.10
0.42, 2.41
0.42, 0.62
0.24
0.40
0.35
0.96
0.85
0.73
0.87
0.99
0.57
Beck depression scale 13 or under Over 13 Behavioral coping (Moos) Low High Cognitive coping (Moos) Low High Avoidance coping (Moos) Low High Seek information (Moos) Low High Express emotion (Moos) Low High Logical reasoning (Moos) Low High Affective (Moos) Low High
Hospital stay I 14 days 15-21 days 222 days Radiation No Yes Chemotherapy No Yes
0.82, 4.31
0.34, 1.79
1.0 1.88 1.0 0.78
27 20 22 25
0.53
0.56
0.14
0.79
0.42
0.91
0.76
0.17
0.19
0.90
0.66 0.01
(Continued over).
0.34, 1.75
0.49, 2.57
1.0 1.12 23 24
1.0 0.77
0.32, 1.62
1.0 0.71 21 26
22 25
0.46, 2.42
1.0 1.05
0.38, 2.01
22 25
1.0 0.88
24 23
0.73, 6.37
0.74. 4.54
1.0 1.84
23 17
1.0 2.15
0.43, 2.60
1.0 1.06
26 17
41 6
0.45, 3.46 1.34. 12.32
1.0 1.25 4.05
20 16 9
Problem Low High Locus of Low High Locus of Low High Locus of Low High
Variable
control+zloctor
control-self
control--luck
solving (Moos)
1.0 1.12
I.0 1.09
RR
46 1.0 44 1.09 (Walston) 48 1.0 42 1.16
(WaktOI,)
32 31 (Walston) 40 50
No.
0.61. 2.19
0.58, 2.04
0.60, 2.11
0.48, 2.46
95% CI
Hematology
0.65
0.80
0.73
0.83
P
TABLE
Problem Low High Locus of Low High Locus of Low High Locus of Low High
Variable
IV.---Conhued
controldoctor
control-self
control-luck
solving (Moos)
RR
21 1.0 26 0.74 (Walston) 26 1.0 21 1.80 (WaIston) 21 I.0 26 1.44 (Walston) 23 1.0 24 1.16
No.
Colostomy
0.51. 2.63
0.62, 3.32
0.79, 4.09
0.33. 1.70
95% CI
0.73
0.40
0.16
0.48
P
Survival
and mood
199
patient population even though we tested each variable within two different disease and treatment combinations. The hematology patients were treated with chemotherapy, were younger, and were of lower SES whereas the rectal cancer patients were treated with surgery, were older, and were of middle SES. The findings, or lack of findings, were robust across both samples and are confirmatory of several previous studies [12, 13, 171. Despite these results, the role of psychological adaptation in cancer patient survival deserves further study. First, it is possible that psychological factors are important only for certain types of cancers. Those with a known or suspected endocrine etiology, for example, may lend themselves more easily to a hypothesized biological relationship between mood, hormone levels, and survival. Furthermore, a hypothesized psychological effect on survival could occur only when the disease is in the early stages. Subsequent studies could be designed to test groups that are homogeneous not solely on disease site but on stage as well. Second, a major limitation in any study of this issue is the sensitivity of available measures. Although the scales chosen represent standard measures of the constructs involved and have been used before in research relevant to these questions, they may nevertheless be inadequate to tap the constructs under study. Finally, with the rectal cancer sample there is a possibility that surgeons preselected more well adjusted patients even though they were specifically requested to not do so. This could over-balance our population to a more well adjusted group with a smaller variance on the independent variables possibly reducing the computed relationship between the independent variable and survival. Such issues of selection bias can best be addressed if full surgery rosters are available and minimal data were collected on pre-screened patients and refusers to assure representative samples. Further a rating by the physician or nurse of all patients could provide additional data on physician selection or patient refusal criteria that may serve as a proxy for the independent variables for patients who cannot be recruited. Nevertheless, these results should not be taken to mean that patients themselves have no role in determining their duration of survival after the diagnosis of cancer. Our previous research has shown that compliance, the amount of medication taken by the patient and the number of appointments kept, are important predictors of survival [28]. However, it is probable that any hypothesized relationship between psychological status and survival after cancer diagnosis would constitute a small component of a very complex causal chain dominated by the biological basis of the disease. Thus a relationship may exist but be small in magnitude relative to more conventional predictors of survival such as severity and extent of disease. Further the effect may be apparent when the disease is diagnosed at an early stage but may have little effect after the disease has become regional or metastatic. These issues need to be explored empirically and further attention needs to be given to the biological basis underlying these hypotheses.
REFERENCES 1. 2.
Fox BH. Premorbid psychological factors as related to cancer incidence. J Behau Med 1978; 1: 2941. Fox BH. Current theory of psychogenic effects on cancer incidence and prognosis. J Psychosoc Oncol 1983; 1: 17-31.
J. L. RICHARDSON et al.
200 3.
4. 5. 6. 7. 8.
9.
TEMOSHOKL, HELLER BW. On comparing apples, oranges and fruit salad: a methodological overview of medical outcome studies in psychosocial oncology. In Psychosocial Sfress and Cancer (Edited by COOPER CL). Ch 10, pp. 231-260. New York: John Wiley, 1984. TEMOSHOKL. Personality, Coping Style, Emotion and Cancer: Towards an Integrative Model. Cancer Sutz 1987; 6: 545-567. PETTINGALE KW, MORRIS T, GREER S. HAYBITTLE J. Mental attitudes to cancer: an additional prognostic factor. Lancer 1985; 1: 750. GREER S. MORRIS T. Psychological attitudes of women who develop breast cancer: a controlled study. Jpsychosom Res 1975; 19: 147-153. GREER S, MORRIS T, PETTINGALE KW. Psychological response to breast cancer: effect on outcome. Lancer 1979; 2: 785-787. TEMOSHOK L, HELLER BW, SAGEBIELW, BLOIS MS, %VEET DM, DICLEMENTE RJ, GOLD ML. The Relationship of Psychosocial Factors to Prognostic Indicators in Cutaneous Malignant Melanoma. Jpsychosom Res 1985; 29: 139-153. ROCEN~INE G, VAN KAMMEN D, Fox B, DOCHER~Y JR, ROSENBLAT SE. BOYD SC, BUNNEY WE. Psychological factors in the prognosis of malignant melanoma: A prospective study. Psychosom Med 1979; 41: 647455.
10.
I I. 12. 13. 14.
LEIGH H, PERCARPIO B, OPSAHL C, UNGERER J. Psychological predictors of survival in cancer patients undergoing radiation therapy. Psychother Psychosom 1987; 47: 65-~73. DEROGATISLR, ABELOFF MD, MELISARATOSN. Psychological coping mechanisms and survival time in metastatic breast cancer. JAMA 1979; 242: 15041508. CASSILETH BR, LUSK EJ, MILI.ER DS, BROWN LL, MILLER C. Psychosocial correlates of survival in advanced malignant disease. N Engl J Med 1985; 312: 1551Ll555. JAMI~ONRN, BURISH TG, WALLSTON KA. Psychogenic factors in predicting survival of breast cancer patients. J Clin Oncol 1987; 5: 768-772. ACHTE’ K, VAUHKONEN M. Cancer and Psyche. Part I: Psychic Factors in Cancer, and Part II: Psychological Determinants of Cancer. Monographs from the Psychiatric Clinic qf /he Helsinki University
15.
16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
Central
Hospital
1: I-35,
1970.
ABSE DW, WILKINS MM, VON DE CASTLE RL, BUXTON WD, DEMARS JP, BROWN RS, KIRSCHNER LG. Personality and behavioral characteristics of lung cancer patients. J psychosam Res 1974; IS: 101~113. DREHER H. Psychosocial factors and breast cancer survival: Recent developments. Mind Body Health Digest 1987; 1: I-2. HISLOP TG, WAXLER NE, ANDREW EW, COLDMAN AJ, ELWO<)D JM, KAN 1~. The prognostic significance of psychosocial factors in women with breast cancer. J Chron Dis 1987; 40 (7): 729-735. DE VITA VT, HELLMAN S, ROSENBERGSA. Cancer: Principles and Practice of Oncology. Philadelphia: Lippincott, 1982. BECK AT, WARD CH. MENUELSONM. MOCK J, ERBAUGH J. An inventory for measuring depression. Arch gen Psychiat 1961; 4: 561. ZUNG WWK, DURHAM NA. Self-Rating Depression Scale. Arch gen Pswhiat 1965; 12: 63--70. ZUNG WWK, DURHAM NC. Factors influencing the Self-Rating Depression Scale. Arch gen P.yychia/ 1967; 16: 543-547. ZUNG WWK, DURHAM NC. From art to science. Arch gm Psychiat 1973; 29: 328-337. ZUNG WWK, DURHAM NC, RICHARDS CB, GABLES FLA, SHORT MJ. Self-Rating Depression Scale in an outpatient clinic. Arch gen Psych& 1965; 13: 508-515. WALLSTON KA, WALLSTON BS. Health locus of control scales. In Research bt,ith rhe Locu.v qf Control Consfrucr (Edited by LEFCOUR~ HM), Vol 1, pp. 189-243 (1981). WALLSTON KA, WALLSTON BS, DEVELLIS R. Development of the multidimensional health locus of control (MHLC) scales. Health Education Monogruphs 1978; 6: 160~170. Moos RH, CRONKITE PC, BILLINGS AC, FINNEY JW. Healrh and Daily Living Form Mutual. Palo Alto, CA. Social Ecology Laboratory, Standford University Medical Center, 1983. KALBFLEISHJD, PRENTICERL. The Statistical Analysis of Failure Time Data. New York: John Wiley and Sons, 1980. RICHARDSON JL, SHELTON DR, KRAILO M, LEVINE AM. The effect of compliance with treatment on survival among patients with hematologic malignancy. J C/in One. In Press. 1990.
APPENDIX Health Luck.
locus of control
subscales
Most things that affect my health happen to me by accident. Luck plays a big part in determining how soon I will recover from an illness. My good health is largely a matter of good fortune.
Survival
and mood
201
SelJ: I am in control of my health. The main thing which affects my health is what I myself do. If I take the right actions, 1 can stay healthy. Doctor. Having regular contact with my physician is the best way for me to avoid illness. Health professionals control my health. Regarding my health, I can only do what my doctor tells me to do. Depression subscales Zung self rating depression scale-psychological subscale. I feel down-hearted, blue and sad. I have crying spells or feel like it. My mind is as clear as it used to be. I find it easy to do the things I used to. I am restless and can’t keep still. I feel hopeful about the future. I am more irritable than usual. I find it easy to make decisions. I feel that 1 am useful and needed. My life is pretty full. I feel that others would be better off if I were dead. I still enjoy the things I used to do. Zung se,f rating depression scale-physical subscale. Morning is when I feel the best. I have trouble sleeping through the night. I eat as much as I used to. I enjoy looking at, talking to and being with attractive women/men. I notice that I am losing weight. I have trouble with constipation. My heart beats faster than usual. I get tired for no reason. Beck depression inventory-psychological subscale. I am so sad or unhappy that 1 can’t stand it. I feel the future is hopeless and that things cannot improve. I feel I am a complete failure as a person. I am dissatisfied or bored with everything. I feel guilty all of the time. I feel I am being punished. I hate myself. I blame myself for everything bad that happens. I would kill myself if I had the chance. I used to be able to cry, but now I can’t cry even though I want to. I feel irritated all the time now. I have lost all of my interest in other people. I can’t make decisions at all anymore. Beck depression inomfory-physical subsrale. I believe that I look ugly. I can’t do any work at all. I wake up several hours earlier than I used to and cannot get back to sleep. I am too tired to do anything. I have no appetite at all anymore. I have lost more than 15 pounds. I am so worried about my physical problems that I cannot think about anything else. I have lost interest in sex completely.
of Psychosomatic in Great Bntain
Research.
Vol.
34, No.
2. pp.
189-20
I, 1990. 0
0022-3999/90 $3.00 + .oO 1990 Pergamon Press plc
PSYCHOSOCIAL STATUS AT INITIATION OF CANCER TREATMENT AND SURVIVAL JEAN L. RICHARDSON,*~ ZOHREHZARNEGAR,~ BETSYBISNO~ and ALEXANDRA LEVINE~ (Received
11 May 1989;accepted in revised form 7 September 1989)
Abstract-Ninety-two newly diagnosed patients with hematologic malignancies treated with chemotherapy, and 47 patients with rectal cancer treated with abdominal-perineal resection were prospectively studied to assess the relationship between mood and survival. Hematology patients were measured within one week of diagnosis and were remeasured at six months. Rectal cancer patients were measured within three months of surgery and remeasured six months later. Medical records were abstracted to obtain data on treatment given, disease characteristics and outcome of treatment. On univariate analyses using Cox regression, we examined the effect of depression, coping style and locus of control on survival. None of these variables were found to be significantly related to survival whether assessed at intake or six months later. Furthermore there were no statistically significant correlations between these factors and subsequent survival at two years or with long-term survival. Biological prognostic variables including extent of disease for rectal cancer patients and severity of specific type of hematologic cancer were significantly related to survival. Although psychosocial adjustment is important for the quality of life experienced by cancer patients it was not related to length of survival in this study. Further exploration of this issue should be conducted using patients with a single site and preferrably an early stage of disease.
INTRODUCTION
Is SURVIVAL after the diagnosis of cancer influenced solely by the biological characteristics of the disease or might psychological factors such as depression, coping style, locus of control, or a pessimistic outlook influence survival as well? Physicians, psychologists and lay people have anecdotally expressed a belief that people they have known who were diagnosed with cancer survived for a longer period because they had a ‘fighting spirit’ or survived for a shorter time because they ‘just gave up’. Although conventional wisdom suggests there may be a relationship between psychological factors and survival, studies of this relationship have been inconsistent and inconclusive when applied to malignant disease. Several articles that review the literature relating psychosocial factors to the course of malignant disease indicate that the results are often contradictory and noncomparable [l&4]. In fact, the study of psychosocial factors as predictors of cancer survival has been fraught with methodological difficulties that have contributed to the lack of coherence in construct relationships [3]. The difficulties include (1) *Department of Preventive Medicine, USC School of Medicine. TInstitute for Health Promotion Disease Prevention Research, USC School of Medicine. IVeterans Administration, Outpatient Center, Los Angeles. §Department of Medicine, USC School of Medicine. This work was supported by Grant number CA 31151 ‘Assessment and Enhancement of Compliance with Chemotherapy’ DHHS, NCI; Grant number T32 CA09492 ‘Cancer Control Research Training Grant’; and American Cancer Society California Division Postdoctoral Fellowship Grant. The authors acknowledge the assistance of Dr Mark Krailo for consultation concerning statistical procedures. Requests for reprints should be sent to: Dr Jean Richardson, 1420 San Pablo St, PMB A-301, Los Angeles, CA 90033, U.S.A. 189
190
J. L. RICHARDSONet al.
misguided attempts to treat cancer as a single disease and to expect homogeneity in the role of psychosocial factors in survival, (2) failure to measure the same psychosocial phenomena across studies, (3) failure to conduct separate analyses for different sites of disease and treatments for disease, (4) failure to analyze data using lifetable analyses rather than median or criterion lengths of survival, and (5) failure to differentiate whether etiology versus progression of disease is being studied. Several authors have indicated that a coping style categorized as ‘fighting spirit’ as opposed to ‘passive and hopeless’, is correlated with longer survival [558]. Rogentine et al. [9] found that those who had more difficulty adjusting to their diagnosis were less likely to relapse than those who reported less difficulty. Depression, which is a similar though not identical construct to hopelessness, may reduce survival time. although some studies have indicated no such relationship between depression and survival [lo], and still others have found that more depressed patients survive longer [l I]. Many authors have failed to substantiate a relationship between psychological factors and survival [12-161. Of particular note in this regard is a study by Cassileth et al. [12], of 204 advanced unresectable cancers of various sites, and 155 early cases of melanoma (Stage I or II) or breast cancer (Stage II). None of the psychological factors studied (hopelessness, depressive affect or adjustment to diagnosis) related to survival for either patient group. A study by Jamison et al. [ 131 of 49 metastatic breast cancer patients indicated that results of a variety of scales (depression, self esteem, locus of control) did not relate to survival. A study by Hislop et al. [17], provided partial support for a relationship between expression of emotion and survival but not for the role of locus of control, depression, coping style, self-esteem or a number of other psychological factors on survival. It is apparent that the evidence for a relationship between psychological processes and survival is less than definite although mechanisms for a behavioral or biological relationship have been hypothesized. For example, a more assertive patient might demand more thorough medical evaluation, exhibit greater compliance, or maintain better self-care than might occur with a more passive patient. Furthermore, mood states could influence complex endocrinological or immunological pathways which might alter the growth rate of certain tumors. We therefore initiated a study to examine prospectively, the role of psychological variables in subsequent patient survival. The constructs examined were similar to those previously examined i.e. locus of control, depression and coping style. The study design addressed many of the shortcomings mentioned above. METHODS
Data were obtained from two groups of cancer patients: 92 patients with hematologic and 47 patients with rectal cancer. Questionnaires and procedures were comparable although between the groups allowing us to examine the same constructs within each group. interviewed privately by a trained interviewer. Each group was analyzed separately. This compare the specificity or robustness of relationships between psychological factors separately for the two groups of patients.
malignancies, not identical Patients were allowed us to and survival
Rrctcrl catneT scm~plr. Patients were accrued from 32 private practtce surgeons. and two large cancer hospitals in the Los Angeles area. All subjects had a permanent abdominal perineal resection (permanent
Survival
and mood
191
colostomy) performed due to a diagnosis of rectal cancer, were over 18 years of age, spoke English, and were cognitively capable of understanding and completing a consent form and measures. Potential subjects were sent letters and contacted by telephone to request their participation in the study. Of those meeting the selection criteria, 70% agreed to participate. Those who declined typically mentioned that they did not want any extra burdens on them at this time or that they did not want anyone probing into their personal lives or both. All subjects were interviewed within one to three months of surgery and reinterviewed six months later. Patients who survived less than 90 days were not included in the analysis because coping style and depression might be unreliably measured due to imminent death. Fifty subjects were recruited, of these three were eliminated for this reason. Hematologic cancer. Patients who were newly and consecutively diagnosed with hematologic malignancies were accrued at the Los Angeles County-University of Southern California Medical Center. All patients were treated with combination chemotherapy. Participants were over 18 years of age, spoke English, or Spanish and were cognitively capable of understanding and completing a consent form and measures. Of those meeting the selection criteria 80% agreed to participate. Hematology patients were interviewed within one to two weeks of diagnosis and reinterviewed six months later. Of the 108 patients followed, 14 were not included in this analysis because they were previously treated for cancer and two were eliminated because they died within 90 days of recruitment. Measures Demographics. Information concerning gender, ethnicity, marital status, education, income and age was collected from respondents in both studies. Severity. Rectal cancer: medical severity was determined from a medical abstract completed by the physician. This abstract contained information on the cancer diagnosis, other medical problems, extent of disease (local, adjacent, regional or distant), length of hospitalization, complications and adjuvant use of chemotherapy or radiation. Hematologic malignancy: the patients were grouped into three disease severity or prognostic levels according to risk of death [18]. The high severity group (37%) had aggressive malignancies, requiring intensive multi-agent chemotherapy. Although initial response to therapy in these illnesses may be good, the average survival for affected patients is generally less than two years. The moderate severity group (24%) consisted of patients with malignancies that generally demonstrate good response to therapy, and average survival of two to five years. The remaining 36% of the patients had typically slow progressing diseases with expected average survival of five years or more.* Most regimens for the treatment of hematologic malignancies use multiple chemotherapeutic agents. In the present study, these regimens were categorized as either ‘easy’ or ‘complex’ based on the difficulty of tolerating the treatment and the number of chemotherapeutic agents utilized. Thirty-two per cent of patients took ‘easy’ regimens and 68% underwent ‘complex’ regimens.? Depression. Rectal cancer: depression was assessed using the Beck depression inventory (BDI) [19]. This is a 21 item inventory with items scored from 0 to 3 and total scores from 0 to 63. Item scores are summed to create a total, with higher scores indicating a greater degree of depression. In addition to the total BDI, we also examined a physical subscale and a psychological subscale. Hematologic malignancy: the 20 item Zung self-rating depression scale (SDS) [2&23] was orginally divided into 3 subscales of pervasive affect, physiological and psychological components. Each item is scored from 1 to 4 and using the following algorithm to create the final score (sum raw store/80*100) [21]. Although we preferred to use the BDI with this group as well, the BDI requires more reading and caused difficulty for the less well educated population at Los Angeles County-USC Medical Center. A depression scale was administered to each sample at intake and at six month follow-up. *The disease severity categories for hematologic malignancy are as follows: High = (acute myeloblastic leukemia, acute lymphocytic leukemia, small non-cleaved lymphoma, lymphoblastic T-cell lymphoma, immunoblastic sarcoma (lymphoma)); Moderate = (large cleaved lymphoma, large non-cleaved lymphoma, multiple myeloma); Low = (chronic granulocytic leukemia, chronic lymphocytic leukemia, Hodgkin’s disease, plasmacytoid lymphocytic lymphoma, small cleaved lymphoma). tThe chemotherapy regimen complexity categories of patients with hematologic malignancy are as follows: Easy = (1) melphelan and prednisone; (2) cyclophosphamide, vincristine, and prednisone (CVP) and (3) daily oral chlorambucil, with or without prednisone. Complex = (1) nitrogen mustard, vincristine, procarbazine, and prednisone (MOPP); (2) hydroxyldaunamycin, bleomycin, vincristine, and DTIC (ABVD); (3) hydroxyldaunamycin, cyclophosphamide, vincristine, and prednisone (CHOP); (4) bleomycin, hydroxyldaunamycin, cyclophosphamide, vincristine, and prednisone (BACOP); and (5) methotrexate, bleomycin, hydroxyldaunamycin, cyclophosphamide, vincristine, and dexamethasone (M-BACOD).
192
J. L. RICHARDSON et al
Locus ofconrrol. Beliefs about the factors controlling one’s health were measured with nine items from the Multidimensional Health Locus of Control Scale [24-251. We used three self-control (internal items. three doctor control items, and three chance control items selected on the basis of the highest item-subscale correlations (see Appendix). This was administered at the intake interview for both samples. Moos coping SC&. For this study we used a coping assessment which was part of the health and daily living form measure developed by Moos et al. [26] and the internal consistency alpha has been reported. Respondents are asked to rate their frequency of use (four-point scale from 0 to 3) of 32 different coping responses. The coping scale is divided into method of coping (active behavioral (alpha = 0.74), cognitive (alpha = 0.62), and avoidance (alpha = 0.60)) and focus of coping (logical analysis (alpha = 0.58). information seeking (alpha = 0.64). problem solving (alpha = 0.63). aflective regulation (alpha = 0.5 1) and emotional discharge (alpha = 0.54)). For colostomy patients, this scale was administered on the first interview and for hematology patients at the follow-up interview. SurtGoalmeasures. For patients with hematologic malignancy. study enrolment was opened in April 1982 and closed in June 1985. Patients were evaluated for survival status in June 1987. For rectal cancer patients, enrolment was opened in February 1983 and closed in August 1984. Patients were evaluated for survival status in December 1987. Patients who survived less than 90 days were not included in the analysis because coping style and depression might be unreliably measured due to imminent death. Two hematology subjects and three rectal cancer subjects were eliminated for this reason. Sturistical nwthod,v
Survival time in days was taken to be the time from study enrolment to last contact for hematology patients and from date of surgery to last contact for rectal cancer patients. Death. regardless of cause, was considered an event; in all other cases (e.g. lost to follow-up, or alive at termination of the study) the individual was considered censored at last contact. A patient who was alive at last contact was considered to be censored with respect to treatment failure. Analyses were repeated for short term survival. ccnsorinp all cases who survived 24 months or more. Plots of the estimated survivor functions were constructed using the method of Kaplan and Meier [27]. Univariate prognostic factor evaluations were carried out by means of the proportional hazards model of Cox [27]. The risk ratios for each variable and 95% confidence intervals were computed from the beta coefficients and their standard errors. The relative risk is a ratio of the dealh rate in a given group over that of the baseline group.
RESULTS
Subject
charucteristics
Descriptive socioeconomic characteristics and specific diagnosis of the patients are shown in Tables I and II. The sample of subjects with hematologic malignancy was younger than the rectal cancer sample because these diseases affect a younger age group. The hematology sample was also of lower SES and more likely to be Hispanic reflecting the fact that they were accrued from the Los Angeles County-USC Medical Center which serves pritnarily indigent, minority patients. The rectal cancer patients were referred from the private medical community and were more likely to be white and middle class.
We wondered whether patients with poor prognosis might experience more psychological distress than those with better prognosis. We thus initially tested the correlation between the measures of severity of disease and the psychological variables studied. No significant relationships (all ps > 0.10) were found between the psychological measures for each sample (depression, locus of control. or coping) and severity of illness (hematologic) or extent of disease (rectal cancer).
Descriptive and comparative statistical analyses were calculated. In each comparative analysis at least two categories of predictor variable for each sample was tested
Survival
and mood
193
TABLE I.-DESCRIPTIVE CHARACTERISTICS OF SAMPLE Colostomy patients N = 41
Gender Male Female Ethnic group Caucasian Black Hispanic/Latin0 Asian/Other Marital status Single Married Divorced Widowed Other Education completed Graduate of professional school College or university Partial college High school graduate Partial high school Junior high school (7-9) Under 7 years of schooling Total ho&hold income Under $5.000
$5,00&$9,999 $10,00~_$14,999 $15,OO(r$l9,999 $20,00&$24,999 %25,00&_$39,999 %40,000or over Don’t know or refused to answer Age Under 20 years old 20-29 yrs 30-39 yrs 4049 yrs 50.-59yrs 6069 yrs 70-79 yrs 2 80 yrs
Hematology patients N = 92
33 (70.2%) 14 (29.8%)
57 (62.0%) 35 (38.0%)
40 (85.1%) 4 (8.5%) 2 (4.3%) 1 (2.1%)
16 (17.4%) 22 (23.9%) 51 (55.4%) 3 (3.3%)
3 (6.4%) 31 (65.9%) 6 (12.8%) 6 (12.8%) 1 (2.1%)
31 (33.7%) 33 (35.9%) 7 (7.6%) 14 (15.2%) 7 (7.6%)
3 (6.4%) 6 (12.7%) 10 (21.3%) 15 (31.9%) 7 (14.9%) 4 (8.5%) 2 (4.3%)
1 4 17 12 20 9 29
6 (12.8%) 5 (10.6%) 5 (10.6%) 2 (4.3%) 5 (10.6%) 11 (23.5%) 5 (10.6%) 8 (17.0%)
15 (16.3%) 18 (19.6%) 4 (4.4%) 4 (4.4%) 1 (1.1%) 1 (1.1%) 1 (1.1%) 48 (52.2%)
0 (0.0%) 0 (0.0%) 1 (2.1%) 3 (6.4%) 10 (21.3%) 17 (36.2%) 13 (27.6%) 3 (6.4%)
3 25 16 14 13 15 5 1
(1.1%) (4.4%) (18.5%) (13.0%) (21.7%) (9.8%) (31.5%)
(3.3%) (27.2%) (17.4%) (15.2%) (14.1%) (16.3%) (5.4%) (1.1%)
using Kaplan Mier plots and the logrank statistic and a Cox proportional hazards regression. These two analyses confirm one another and the Cox regressions are reported in Table IV. Descriptive The Kaplan Mier survival curves for rectal cancer and hematologic cancer patients are presented in Fig. 1. The curves indicated that hematologic cancer patients had a poorer survival outlook, however, the curves were not statistically different from each other (Log rank x2 = 1.516, p = 0.9970). Table III presents the number and per cent of patients in each group who were dead and censored during each year of follow-up. Overall, the per cent confirmed dead at the termination of the study was remarkably similar, 50% for the rectal cancer patients and 44% for the hematology
J. L. RICHARDSON
194 TABLE
el al.
II.-SEVERITY OF DISEASE FOR EACH SAMPLE
N Hematology Disease severity High Moderate Low Regimen complexity Easy Complex Colostomy Extent of disease Local Adjacent Regional Distant (5 missing) Radiation No Yes (4 missing) Chemotherapy No Yes (7 missing) Hospital stay 14 days or less 15-21 days 22~ 30 days More than 30 days (2 missing)
%
33 23 36
35.9% 25.0% 39.1%
29 63
32.0% 68.0%
8 17 6 II
19.0~/” 40.5% 14.3% 26.2%
26 17
60.5% 39.5%
23 17
S7.SV0 42.5%
20 I6 4 5
44.4% 35.6% 8.9% 11.1%
cancer patients. More of the hematology patients were censored earlier. which we attribute to the inherent diff’erences in the subjects and the sources of subject accrual. The purpose of the Figure and Table III is not specifically to compare the two groups but rather to provide the distribution of the dependent variable of survival outcomes.
-----
HMdTOLOGIC
8.75-
e.ee: 1
B
FIG. 1.-Survival
curves
for patients
4
3
2
5
YEARS
with hematologic
cancer
and with rectal cancer.
Survival TABLE III.-SURVIVOR
195
and mood
STATUSFOR COLOSTOMYAND HEMATOLOGYPATIENTS
Hematologic cancer N = 92
Rectal cancer N = 47 -___-.
Censored
Dead No. End of End of End of End of End of (4yr
year 1 year 2 year 3 year 4 follow-up 10mo)
5 6 5 6
I 23
%
cum. %
No.
11% 13% 11% 13% 2%
11% 24% 35% 48% 50%
0 2 3 9 10
50%
24
Dead
%
cum. %
No.
0% 4% 6% 19% 21%
0% 4% 10% 29% 50%
17 18 2 2 2
50%
41
Censored
%
cum. %
No.
%
cum. %
18% 20% 2?/0 2% 2%
18% 38% 40% 42% 44%
12 7 8 17 7
13% 8% 9% 18% 8%
13% 21% 30% 48% 56%
44%
51
56%
Demographics
Seven demographic variables were analyzed for their relationship to survival (Table IV). These included age, gender, marital status, ethnicity, education, income and occupation. Only gender for hematology patients (p = 0.11) and age for colostomy patients (p = 0.09) were even marginally related to survival. None of the variables measuring socioeconomic status were predictive of survival. Disease/treatment
Severity of specific disease was predictive of survival among hematology patients. Both extent of disease and hospital stay were predictive of survival among rectal cancer patients. Treatment characteristics were not predictive of survival in either group. Thus, regimen complexity was not predictive of survival for hematology patients and use of radiation and/or chemotherapy were not predictive of survival among rectal cancer patients. Psychological
processes
The relationship between survival and depression, coping style, and locus of control are presented in Table IV. These data indicate that none of these variables were related to length of survival for either sample of patients. In addition to the total depression scales presented, we divided the Beck and the Zung self-rating depression scales into psychological and physical subscales which were also not significantly related to survival (see Appendix). We also computed the relationship between the six-month follow-up measures of depression for each sample and survival and again found no relationship. We also examined the effect of these variables on two-year survival and found that the relationship between survival and these factors was again not significant. DISCUSSION
The analyses of the relationship between psychosocial factors and survival for each of these two groups failed to produce evidence of any significant relationship between the two. Tests we conducted for long and short-term survival. There were no significant associations between the psychosocial variables and survival for either
Disease severity Low Moderate High
Disease/treatment
Age 1741 over 41 Marital Status Single Married Other Ethnicity Black Hispanic Other Occupation Low Medium High Income Under %lOK $IOKp$20K Over $20K Education Under 7 yrs I yrs to H.S.
Sex Male Female
Demographics
Variable
1.0 0.67 0.89 1.0 1.11 1.16 I.0 0.45 I .06 I.0 1.15
22 51 19 53 25 6 33 8 3 29 41
1.0 3.78 1.85
1.0 0.85 0.68
31 33 28
36 23 33
1.0 1.48
0.57
35
46 46
1.0
RR
51
No.
RISK
1.68, 8.54 0.84, 4.09
0.53, 2.45
0.13, 1.55 0.24, 4.64
0.55. 2.23 0.34, 3.89
0.32, 1.41 0.38. 2.08
0.41, 1.77 0.31, 1.47
0.77. 2.85
0.29, 1.13
95% CI
Hematology
TABLE IV.-RELATIVE
10.001 0.13
0.73
0.21 0.94
0.77 0.81
0.30 0.78
0.67 0.32
0.24
0.11
P
OF DEATHFOR
8 17 6 II
28 19
Education H.S. or less More than H.S.
Extent Local Adjacent Regional Distant
18 21
7 12 24
Occupation Low Medium High Income Under $20K $20K or more
40 7
Ethnicity Caucasian Other
22 25
33 14
No.
1.0 4.08 6.39 14.74
1.0 0.8 1
1.0 1.49
1.0 0.65 0.58
1.04
0.49, 34.19 0.29, 142.88 1.86, 117.09
0.34. 1.94
0.59. 3.80
0.18, 2.31 0.18. 1.85
0.31, 3.53
0.42. 2.37
1.0 1.0
1.0
0.19. 1.06
0.46. 2.54
95% CI
1.0 0.45
1.0 1.08
RR
Colostomy
HEMATOLOGYPATIENTS
31 16
AND
Age 64 or less over 64 Marital Status Married Other
Sex Male Female
Variable
COLOSTOMY
0.19 0.11 0.01
0.64
0.40
0.50 0.36
0.95
0.99
0.07
0.87
P
High Express emotion (Moos) Low High Logical reasoning (Moos) Low High Affective (Moos) Low High
Zung depression scale 59 or under Over 59 Behavioral coping (Moos) Low High Cognitive coping (Moos) Low High Avoidance coping (Moos) Low High Se;ci$ormation (Moos)
Psychosocial
Regimen complexity Difficult Easy
1.0 1.0 1.0 0.94 1.0 1.15 1.0 1.08 1.0 1.02 1.0 0.68 1.0 1.43 1.0 0.62
32 31 32 32 28 35 30 34 29 33 28 35 31 32
0.82
1.0
II 13
63 29
0.28, 1.39
0.62, 3.27
0.30, 1.53
0.45, 2.36
0.48, 2.44
0.52, 2.58
0.42, 2.10
0.42, 2.41
0.42, 0.62
0.24
0.40
0.35
0.96
0.85
0.73
0.87
0.99
0.57
Beck depression scale 13 or under Over 13 Behavioral coping (Moos) Low High Cognitive coping (Moos) Low High Avoidance coping (Moos) Low High Seek information (Moos) Low High Express emotion (Moos) Low High Logical reasoning (Moos) Low High Affective (Moos) Low High
Hospital stay I 14 days 15-21 days 222 days Radiation No Yes Chemotherapy No Yes
0.82, 4.31
0.34, 1.79
1.0 1.88 1.0 0.78
27 20 22 25
0.53
0.56
0.14
0.79
0.42
0.91
0.76
0.17
0.19
0.90
0.66 0.01
(Continued over).
0.34, 1.75
0.49, 2.57
1.0 1.12 23 24
1.0 0.77
0.32, 1.62
1.0 0.71 21 26
22 25
0.46, 2.42
1.0 1.05
0.38, 2.01
22 25
1.0 0.88
24 23
0.73, 6.37
0.74. 4.54
1.0 1.84
23 17
1.0 2.15
0.43, 2.60
1.0 1.06
26 17
41 6
0.45, 3.46 1.34. 12.32
1.0 1.25 4.05
20 16 9
Problem Low High Locus of Low High Locus of Low High Locus of Low High
Variable
control+zloctor
control-self
control--luck
solving (Moos)
1.0 1.12
I.0 1.09
RR
46 1.0 44 1.09 (Walston) 48 1.0 42 1.16
(WaktOI,)
32 31 (Walston) 40 50
No.
0.61. 2.19
0.58, 2.04
0.60, 2.11
0.48, 2.46
95% CI
Hematology
0.65
0.80
0.73
0.83
P
TABLE
Problem Low High Locus of Low High Locus of Low High Locus of Low High
Variable
IV.---Conhued
controldoctor
control-self
control-luck
solving (Moos)
RR
21 1.0 26 0.74 (Walston) 26 1.0 21 1.80 (WaIston) 21 I.0 26 1.44 (Walston) 23 1.0 24 1.16
No.
Colostomy
0.51. 2.63
0.62, 3.32
0.79, 4.09
0.33. 1.70
95% CI
0.73
0.40
0.16
0.48
P
Survival
and mood
199
patient population even though we tested each variable within two different disease and treatment combinations. The hematology patients were treated with chemotherapy, were younger, and were of lower SES whereas the rectal cancer patients were treated with surgery, were older, and were of middle SES. The findings, or lack of findings, were robust across both samples and are confirmatory of several previous studies [12, 13, 171. Despite these results, the role of psychological adaptation in cancer patient survival deserves further study. First, it is possible that psychological factors are important only for certain types of cancers. Those with a known or suspected endocrine etiology, for example, may lend themselves more easily to a hypothesized biological relationship between mood, hormone levels, and survival. Furthermore, a hypothesized psychological effect on survival could occur only when the disease is in the early stages. Subsequent studies could be designed to test groups that are homogeneous not solely on disease site but on stage as well. Second, a major limitation in any study of this issue is the sensitivity of available measures. Although the scales chosen represent standard measures of the constructs involved and have been used before in research relevant to these questions, they may nevertheless be inadequate to tap the constructs under study. Finally, with the rectal cancer sample there is a possibility that surgeons preselected more well adjusted patients even though they were specifically requested to not do so. This could over-balance our population to a more well adjusted group with a smaller variance on the independent variables possibly reducing the computed relationship between the independent variable and survival. Such issues of selection bias can best be addressed if full surgery rosters are available and minimal data were collected on pre-screened patients and refusers to assure representative samples. Further a rating by the physician or nurse of all patients could provide additional data on physician selection or patient refusal criteria that may serve as a proxy for the independent variables for patients who cannot be recruited. Nevertheless, these results should not be taken to mean that patients themselves have no role in determining their duration of survival after the diagnosis of cancer. Our previous research has shown that compliance, the amount of medication taken by the patient and the number of appointments kept, are important predictors of survival [28]. However, it is probable that any hypothesized relationship between psychological status and survival after cancer diagnosis would constitute a small component of a very complex causal chain dominated by the biological basis of the disease. Thus a relationship may exist but be small in magnitude relative to more conventional predictors of survival such as severity and extent of disease. Further the effect may be apparent when the disease is diagnosed at an early stage but may have little effect after the disease has become regional or metastatic. These issues need to be explored empirically and further attention needs to be given to the biological basis underlying these hypotheses.
REFERENCES 1. 2.
Fox BH. Premorbid psychological factors as related to cancer incidence. J Behau Med 1978; 1: 2941. Fox BH. Current theory of psychogenic effects on cancer incidence and prognosis. J Psychosoc Oncol 1983; 1: 17-31.
J. L. RICHARDSON et al.
200 3.
4. 5. 6. 7. 8.
9.
TEMOSHOKL, HELLER BW. On comparing apples, oranges and fruit salad: a methodological overview of medical outcome studies in psychosocial oncology. In Psychosocial Sfress and Cancer (Edited by COOPER CL). Ch 10, pp. 231-260. New York: John Wiley, 1984. TEMOSHOKL. Personality, Coping Style, Emotion and Cancer: Towards an Integrative Model. Cancer Sutz 1987; 6: 545-567. PETTINGALE KW, MORRIS T, GREER S. HAYBITTLE J. Mental attitudes to cancer: an additional prognostic factor. Lancer 1985; 1: 750. GREER S. MORRIS T. Psychological attitudes of women who develop breast cancer: a controlled study. Jpsychosom Res 1975; 19: 147-153. GREER S, MORRIS T, PETTINGALE KW. Psychological response to breast cancer: effect on outcome. Lancer 1979; 2: 785-787. TEMOSHOK L, HELLER BW, SAGEBIELW, BLOIS MS, %VEET DM, DICLEMENTE RJ, GOLD ML. The Relationship of Psychosocial Factors to Prognostic Indicators in Cutaneous Malignant Melanoma. Jpsychosom Res 1985; 29: 139-153. ROCEN~INE G, VAN KAMMEN D, Fox B, DOCHER~Y JR, ROSENBLAT SE. BOYD SC, BUNNEY WE. Psychological factors in the prognosis of malignant melanoma: A prospective study. Psychosom Med 1979; 41: 647455.
10.
I I. 12. 13. 14.
LEIGH H, PERCARPIO B, OPSAHL C, UNGERER J. Psychological predictors of survival in cancer patients undergoing radiation therapy. Psychother Psychosom 1987; 47: 65-~73. DEROGATISLR, ABELOFF MD, MELISARATOSN. Psychological coping mechanisms and survival time in metastatic breast cancer. JAMA 1979; 242: 15041508. CASSILETH BR, LUSK EJ, MILI.ER DS, BROWN LL, MILLER C. Psychosocial correlates of survival in advanced malignant disease. N Engl J Med 1985; 312: 1551Ll555. JAMI~ONRN, BURISH TG, WALLSTON KA. Psychogenic factors in predicting survival of breast cancer patients. J Clin Oncol 1987; 5: 768-772. ACHTE’ K, VAUHKONEN M. Cancer and Psyche. Part I: Psychic Factors in Cancer, and Part II: Psychological Determinants of Cancer. Monographs from the Psychiatric Clinic qf /he Helsinki University
15.
16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
Central
Hospital
1: I-35,
1970.
ABSE DW, WILKINS MM, VON DE CASTLE RL, BUXTON WD, DEMARS JP, BROWN RS, KIRSCHNER LG. Personality and behavioral characteristics of lung cancer patients. J psychosam Res 1974; IS: 101~113. DREHER H. Psychosocial factors and breast cancer survival: Recent developments. Mind Body Health Digest 1987; 1: I-2. HISLOP TG, WAXLER NE, ANDREW EW, COLDMAN AJ, ELWO<)D JM, KAN 1~. The prognostic significance of psychosocial factors in women with breast cancer. J Chron Dis 1987; 40 (7): 729-735. DE VITA VT, HELLMAN S, ROSENBERGSA. Cancer: Principles and Practice of Oncology. Philadelphia: Lippincott, 1982. BECK AT, WARD CH. MENUELSONM. MOCK J, ERBAUGH J. An inventory for measuring depression. Arch gen Psychiat 1961; 4: 561. ZUNG WWK, DURHAM NA. Self-Rating Depression Scale. Arch gen Pswhiat 1965; 12: 63--70. ZUNG WWK, DURHAM NC. Factors influencing the Self-Rating Depression Scale. Arch gen P.yychia/ 1967; 16: 543-547. ZUNG WWK, DURHAM NC. From art to science. Arch gm Psychiat 1973; 29: 328-337. ZUNG WWK, DURHAM NC, RICHARDS CB, GABLES FLA, SHORT MJ. Self-Rating Depression Scale in an outpatient clinic. Arch gen Psych& 1965; 13: 508-515. WALLSTON KA, WALLSTON BS. Health locus of control scales. In Research bt,ith rhe Locu.v qf Control Consfrucr (Edited by LEFCOUR~ HM), Vol 1, pp. 189-243 (1981). WALLSTON KA, WALLSTON BS, DEVELLIS R. Development of the multidimensional health locus of control (MHLC) scales. Health Education Monogruphs 1978; 6: 160~170. Moos RH, CRONKITE PC, BILLINGS AC, FINNEY JW. Healrh and Daily Living Form Mutual. Palo Alto, CA. Social Ecology Laboratory, Standford University Medical Center, 1983. KALBFLEISHJD, PRENTICERL. The Statistical Analysis of Failure Time Data. New York: John Wiley and Sons, 1980. RICHARDSON JL, SHELTON DR, KRAILO M, LEVINE AM. The effect of compliance with treatment on survival among patients with hematologic malignancy. J C/in One. In Press. 1990.
APPENDIX Health Luck.
locus of control
subscales
Most things that affect my health happen to me by accident. Luck plays a big part in determining how soon I will recover from an illness. My good health is largely a matter of good fortune.
Survival
and mood
201
SelJ: I am in control of my health. The main thing which affects my health is what I myself do. If I take the right actions, 1 can stay healthy. Doctor. Having regular contact with my physician is the best way for me to avoid illness. Health professionals control my health. Regarding my health, I can only do what my doctor tells me to do. Depression subscales Zung self rating depression scale-psychological subscale. I feel down-hearted, blue and sad. I have crying spells or feel like it. My mind is as clear as it used to be. I find it easy to do the things I used to. I am restless and can’t keep still. I feel hopeful about the future. I am more irritable than usual. I find it easy to make decisions. I feel that 1 am useful and needed. My life is pretty full. I feel that others would be better off if I were dead. I still enjoy the things I used to do. Zung se,f rating depression scale-physical subscale. Morning is when I feel the best. I have trouble sleeping through the night. I eat as much as I used to. I enjoy looking at, talking to and being with attractive women/men. I notice that I am losing weight. I have trouble with constipation. My heart beats faster than usual. I get tired for no reason. Beck depression inventory-psychological subscale. I am so sad or unhappy that 1 can’t stand it. I feel the future is hopeless and that things cannot improve. I feel I am a complete failure as a person. I am dissatisfied or bored with everything. I feel guilty all of the time. I feel I am being punished. I hate myself. I blame myself for everything bad that happens. I would kill myself if I had the chance. I used to be able to cry, but now I can’t cry even though I want to. I feel irritated all the time now. I have lost all of my interest in other people. I can’t make decisions at all anymore. Beck depression inomfory-physical subsrale. I believe that I look ugly. I can’t do any work at all. I wake up several hours earlier than I used to and cannot get back to sleep. I am too tired to do anything. I have no appetite at all anymore. I have lost more than 15 pounds. I am so worried about my physical problems that I cannot think about anything else. I have lost interest in sex completely.