Long-term adjustment of survivors of early-stage breast carcinoma, 20 years after adjuvant chemotherapy

Abstract

BACKGROUND

The long-term impact of breast carcinoma and its treatment was assessed in 153 breast carcinoma survivors previously treated on a Phase III randomized trial (Cancer and Leukemia Group B [CALGB 7581]) a median of 20 years after entry to CALGB 7581.

METHODS

Survivors were interviewed by telephone using the following standardized measures: Brief Symptom Inventory (BSI), PostTraumatic Stress Disorder Checklist with the trauma defined as survivors' response to having had cancer (PCL-C), Conditioned Nausea, Vomiting and Distress, European Organization for Research and Treatment of Cancer QLQ-C30 (quality of life), Life Experience Survey (stressful events), MOS Social Support Survey, comorbid conditions (Older Americans Resources and Services Questionnaire), and items developed to assess long-term breast carcinoma treatment side effects and their interference with functioning.

RESULTS

Only 5% of survivors had scores that were suggestive of clinical levels of distress (BSI), 15% reported 2 or more posttraumatic stress disorder (PTSD) symptoms (PCL-C) that were moderately to extremely bothersome, 1–6% reported conditioned nausea, emesis, and distress as a consequence of sights, smells, and tastes triggered by reminders of their treatment, 29% reported sexual problems attributed to having had cancer, 39% reported lymphedema, and 33%, reported numbness. Survivors who reported greater lymphedema and numbness that interfered with functioning had significantly worse PTSD (PCL-C; P = 0.008) com- pared with survivors who reported less lymphedema and numbness. Survivors with a lower level of education (P = 0.026), less adequate social support (P = 0.0033), more severe negative life events (P = 0.0098), and greater dissatisfaction with their medical care (P = 0.037) had worse PTSD compared with other survivors.

CONCLUSIONS

Twenty years after the initial treatment, the impact of breast carcinoma on survivors' adjustment was minimal. However, the higher prevalence of PTSD symptoms in response to having had cancer is indicative of continuing psychologic sequelae long after treatment completion. Findings related to lymphedema and numbness and continued symptoms of PTSD suggest that the long-term psychologic and medical sequelae on adjustment may be underrecognized. To establish in more detail whether survivors' overall psychologic state is any different from that of individuals without cancer, a population of community residents without cancer would need to be studied. Cancer 2003;98:679–89. © 2003 American Cancer Society.

DOI 10.1002/cncr.11531

The majority of psychosocial studies of women with breast carcinoma have been conducted in the first 2 years after treatment completion. To our knowledge, only a few studies have assessed survivors 5–15 years after treatment completion.1 In general, studies have shown that at 1–2 years posttreatment, approximately 75% of patients with cancer adjusted well, with this percentage varying by disease site and length of time since treatment completion. Among breast carcinoma survivors, the prevalence of psychiatric disorders ranged from 22% to 47%.1–3 For patients who had breast- conserving treatment compared with mastectomy, only small to modest advantages in survivors' adjustment were found 18 months after initial treatment. However, significantly better body image and sexual functioning continued over time for patients who received breast-conserving treatment.4–6

Without a longer-term follow-up, the impact of breast carcinoma on survivors' psychosocial adjustment is unknown. Consequently, the major objective of the current study was to describe the long-term adaptation of a cohort of breast carcinoma survivors who were treated 20 years ago on a Phase III adjuvant Cancer and Leukemia Group B chemotherapy trial (CALGB 7581). The vulnerability model was used as the broad theoretic model for understanding cancer survivors' adaptation. In this model, the impact of cancer and its treatment is experienced through physical functioning, which itself is affected by the presence of comorbid conditions. Physical functioning may then exacerbate or ameliorate the stress of having been treated for cancer by a range of mediating factors, including social support, economic resources, other stressors in their lives (e.g., deaths in their family), the relationship with their physicians, and individual characteristics (e.g., personality factors, preexisting psychiatric history, age). By applying this model, a number of mediating and moderating factors were evaluated with respect to the patients' psychosocial adjustment to having been treated for cancer.1

MATERIALS AND METHODS

Sample

All patients eligible to participate in this study had been treated on CALGB 7581,7 which was open to accrual between May 1975 and June 1980. Patients had been randomized to one of three adjuvant regimens: 1) cyclophosphamide, methotrexate, 5-fluorouracil (5-FU), vincristine, and prednisone (CMFVP); 2) cyclophosphamide, methotrexate, and 5-FU; or 3) cyclophosphamide, methotrexate, 5-FU, and the methanol extraction residue of bacillus Calmette-Guerin (BCG).7 Eligibility criteria for the current study were as follows: 1) no evidence of breast carcinoma; 2) completion of all cancer treatment 1 year or more before the phone interview; and 3) no evidence of a major psychiatric disorder or cognitive deficit that would preclude consent and participation.

Of 194 eligible survivors,16.5% (n = 32) refused participation, 3.6% (n = 7) were too ill to participate, and less than 1% (n = 2) did not participate for other reasons. Therefore, 153 survivors participated in the study, representing 78.9% of the eligible participants. Sample bias was tested in 153 survivors who were interviewed versus 41 survivors who had been eligible for this study but had not been interviewed, in terms of current age, age at diagnosis, and race/ethnicity. Survivors who were not interviewed were more likely to be older by 6.3 years (P = 0.0004), to be older at diagnosis by 7.7 years (P = 0.0006), and to be African American (21.9% [9 of 41] vs. 8.5% [13 of 153]).

Research Procedures

The institutional clinical research associate (CRA) confirmed survivors' eligibility status using the CALGB list of patients who had participated in CALGB 7581. Upon obtaining the treating physician's approval to contact the patient, the CRA sent a letter by the treating physician to the survivor. The letter introduced the study and informed the patient that she would be called by a research interviewer to discuss the study in more detail. If the treating physician had left the institution, the letter to the patient was sent by the CALGB institution's principal investigator. If the treating physician had either retired, died, or relocated to an institution that was no longer a member of the CALGB, Dr. Weiss, the Co-Chair of CALGB 7581, called the survivor and requested her participation. All survivors accrued from the University of Pretoria, South Africa, were called directly by the research interviewer there (M. M.). Survivors were assessed only after approval was obtained by the local human investigation committee with an assurance filed with and approved by the Department of Health and Human Services.

After consent was obtained, the research interviewer scheduled an appointment to conduct the telephone interview, mailed a packet of questionnaires to the patient, and requested that she complete them before the interview. The interview lasted 60–90 minutes, during which time the research interviewer clarified questions and obtained the responder's answers. All interviews were conducted by three individuals (E. L. Z., S. R., M. M.). Based on our research in the CALGB with cancer patients,8–11 telephone interviewing has proven to be a feasible and valid method for conducting psychosocial research within cooperative clinical trials and has improved the quantity and quality of the data.8

Measures

The psychosocial adjustment battery consisted of measures to assess survivors' adjustment to having had breast carcinoma, as well as potential mediating and moderating factors of adaptation. These included physical symptoms, stressful life events occurring in the past year, treatment sequelae of breast carcinoma treatment, social support, satisfaction with medical care, obtaining help from mental health services, and sociodemographic characteristics.

European Organization for Research and Treatment of Cancer QLQ-C30

The European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 (Version 1.0), a measure of quality of life, consists of 30 items that address general physical symptoms and functioning, fatigue/malaise, and social and emotional functioning.12 Higher scores on functional scales represented a better level of functioning, whereas higher scores on symptom scales represented worsening symptoms.

Brief Symptom Inventory

The Brief Symptom Inventory (BSI)13 consists of 53 symptoms of psychopathology and 9 subscales: somatization, obsessive-compulsive disorder, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. Norms have been developed separately for men and women for healthy adults and adolescents and for psychiatric outpatients. The cutoff score is 1.5 standard deviations (SD) above the norm on the BSI General Severity Index (GSI) score, suggestive of a possible psychiatric disorder requiring further evaluation.

Posttraumatic Stress Disorder Checklist-Civilian

The Posttraumatic Stress Disorder Checklist-Civilian (PCL-C)14 is a measure of posttraumatic stress disorder (PTSD) that is applicable to any traumatic event. In the current study, the trauma was the cancer diagnosis. The PCL-C consists of 17 items that correspond to the criteria for a diagnosis of PTSD in the 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV): persistently reexperiencing the event, persistent avoidance of stimuli associated with the trauma and numbing of general responsiveness, and persistent symptoms of increased arousal.15 PTSD symptomatology will be presented as symptoms under each type of the PTSD' symptom groups. The original scoring system of the PCL-C was designed to reflect a diagnosis of PTSD that involved the presence of one or more symptoms in each group of symptoms, as listed above. However, the PCL-C score indicative of a PTSD diagnosis does not include the DSM-IV diagnostic criterion that PTSD symptoms had to be of a sufficient intensity to impair patient's functioning. Consequently, for purposes of the current study, the prevalence of the diagnostic levels of PTSD was determined by survivors who meet the PCL-C' diagnostic score, as based on its original scoring,14 and who also score above the BSI Global Severity cutoff score suggestive of a psychiatric disorder. This will be referred to in the current study as the adjusted level of a PTSD diagnosis.

Conditioned Nausea and Vomiting

Conditioned Nausea and Vomiting16 consisted of six items rated from 0–10 that concerned any sights, smells, or tastes that triggered nausea, emesis, or psychologic distress in cancer survivors within the past year. The Conditioned Nausea, Vomiting, and Distress Indices are a summation of separate ratings of nausea, emesis, and distress across sight, smell, and taste. The scores for each index represent the sum of the ratings (range, 0–90), with higher scores indicating worse symptoms.

Employment and insurance difficulties attributed to cancer

Survivors attributed 11 employment and 10 insurance problems to having had cancer. Total scores included the total number of employment problems and the total number of insurance problems that the survivor attributed to having had cancer. All items were developed for survivor studies conducted in the CALGB.11, 17

Sexual problems attributed to cancer

Problems with current sexual functioning that survivors attributed to having had cancer were measured by six questions that addressed sexual interest, sexual activity, sexual attractiveness, sexual acceptance by one's partner, painful intercourse, and interest in having a child. The index was a measure of the total number of sexual problems the survivor attributed to having had cancer. This scale has also been used in previous CALGB survivor studies.11, 17

Comorbid conditions

A slightly modified version of the Older American Resources and Services Questionnaire (OARS)18 was used to assess common comorbid conditions among older persons. For each condition reported, survivors were asked to rate the degree to which it interfered with their activities: not at all, a little, or a great deal. Total scores included the number of comorbid conditions experienced by survivors and the mean of the activity interference scores across all comorbid conditions.

Utilization of mental health and psychosocial services

Several items from both the OARS and the Caregiver Burden Interview were used.19 These items addressed past and current use of mental health services. The total number of mental health services used by each patient was calculated.

Medical Outcomes Study (MOS) Social Support Survey

Twenty items concerning perceived availability of social support are grouped into four subscales: emotional/informational, tangible, affectionate, and positive social interactions.20 Higher scores indicate better social support on each of the dimensions and total score.

Life Experience Survey

The Life Experience Survey (LES)21 is a 57-item measure of stressful life events survivors' occurred within the past year. It also includes survivors' perception of its impact on their lives, rated on a 7 point scale, from extremely negative to extremely positive (−3 to +3). The following LES scores were created: the total number of negative events, the total number of positive events, and the magnitude of the impact of positive and negative events scores on survivors' lives.

Survivors' satisfaction with the medical team

The Technical Quality, Communication, Interpersonal Care, and Outcomes subscales of the Group Health Association of America Consumer Satisfaction Survey22 consisted of 14 items. They were used to assess patients' satisfaction with their oncology care. A higher score across subscales indicated a patient's greater satisfaction with their care.

Sociodemographic characteristics

The CALGB background information form was used to assess basic sociodemographic characteristics.23 Age, gender, and ethnicity were obtained on the registration form at the time of the accrual of survivors' to the study. Survivors were also asked to indicate whether they believed that their current education, income, and employment status were attributed to having had cancer and to indicate what they believed their socioeconmoc status would have been if they had not had cancer. A Perceived Negative Socioeconomic Impact Attributed to Cancer Index was created from these variables, based on a current lower socioeconomic status compared with what surviors believed their status would have been if they had not had cancer. The Perceived Negative Socioeconomic Impact Attributed to Cancer Index total score ranged from 0 to 3.

Medical characteristics

Age at diagnosis, disease stage, and treatment arm of CALGB 7581 were obtained from the CALGB 7581 medical file. The interview provided information on the type of mastectomy, additional treatments received for breast carcinoma, recurrent disease and re-treatment subsequent to survivors' treatment on CALGB 7581, breast reconstruction, and treatment-related sequelae (e.g., lymphedema, numbness in hands, chest, or feet). Information on treatment-related sequelae were rated in terms of whether survivors ever had these symptoms, and if so, how long they persisted, how distressing they were, how much they interfered with their functioning, and if they were still present.

Statistical Methods

Pearson product moment correlations were calculated for all continuous medical, psychosocial, and sociodemographic variables. The percentage of survivors who scored above the established cutoff scores for the BSI and revised diagnostic PCL-C scores were calculated. Hierarchical multiple regression analyses were conducted for the BSI, PCL-C, and the EORTC Emotional Functioning scales. Predefined sets of predictors were entered into the model in the following order: sociodemographic and cancer treatment characteristics, current physical status, and psychosocial variables.

RESULTS

Sample Characteristics

Breast carcinoma survivors' characteristics are presented in Table 1.

Table 1. Survivors' Sociodemographic and Medical Characteristics (n = 153)

Characteristic	%
Current age (yrs)
40–49	5
50–59	25
60–69	43
70+	27
Mean = 65
Range = 41–87
Age at Diagnosis (yrs)
Mean = 45.0
Range = 23–66
Length of time since last treatment (yrs)
Mean = 18
Range = 15–23
Type of mastectomy
Radical	36
Modified radical	64
Ethnicity
White (USA)	78
White (South Africa)	13
Black	6
Hispanic	3
Breast reconstruction
Yes	24
Educational level
Less than high school	11
High school graduate	30
Some college/junior college	31
Bachelor's degree or higher	29
Marital status
Married	66
Separated/divorced	11
Widowed	18
Single	5
Household composition
Lives alone	21
Employment status
Full/part-time	35
Retired	39
Homemaker/unemployed	22
Unknown	4
Recurrent and Re-treated
Yes	22
Mean = 10.2 yrs (n = 25)
Range = 2–19 yrs
No. of comorbid conditions
0	12
1	22
2–3	44
4+	22
Mean = 2.5

General Distress and Cancer-Related Psychologic, Sexual, and Socioeconomic Problems

The BSI was used as an indicator of survivors who were experiencing severe psychologic distress. The cutoff score was 1.5 SD above the norm for nonpsychiatric patients. Only 5% (95% confidence interval [CI], 2–10%) of survivors scored at a level indicative of a possible psychiatric disorder13 (Table 2). The median EORTC Emotional Functioning score was 91.5, which reflects a substantial improvement in emotional functioning from those in active treatment (EORTC score = 74), comparable to a difference of 0.8 SD between mean scores.24 The average survivor had a median of two other negative stressful life events and one positive event in the year before being interviewed (LES).

Table 2. Impact of Cancer on Survivors Psychologically, Sexually, and Socioeconomically

Measure	% (95% CI)
CI: confidence interval; SD: standard deviation. a Mean number of symptoms = 21.3 (SD, 6.2).
General psychologic distress (i.e., non–cancer-specific)
Brief Symptom Inventory (1.5 S/D above norms, indicative of psychiatric disorder)	5 (2–10)
Cancer-related psychologic, sexual, and socioeconomic problems
Posttraumatic Stress Disorder Checklist-Civilian	18 (12–25)a
Conditioned nausea, vomiting, or distress due to:
Smells	8 (4–13)
Sights	17 (11–24)
Taste	7 (3–12)
Sexual problems attributed to cancer (one or more)	29 (22–37)
Employment problems attributed to cancer (one or more)	13 (8–20)
Insurance problems attributed to cancer (one or more)	26 (20–34)
Negative socioeconomic impact attributed to cancer (one or more)	7 (3–12)
Cancer-related medical sequelae
Lymphedema
Had Lymphedema	39 (31–47)
Had ≥ 10 yrs	25 (18–33)
Somewhat distressing to extremely distressing	18 (12–25)
Interference with functioning—‘Somewhat’ to ‘a great deal’	22 (15–29)
Numbness of hands, chest, or feet
Had numbness	33 (26–41)
Had ≥ 10 yrs	19 (13–26)
Somewhat distressing to extremely distressing	9 (5–15)
Interference with functioning—‘Somewhat’ to ‘a great deal’	9 (5–15)

Cancer-Related Psychologic Issues

Only 8% (95% CI, 4–13%), 17% (95% CI, 8–20%), and 7% (95% CI, 3–12%) responded that in the past 2 weeks they had experienced reminders of their treatment in terms of smell, sight, or taste, respectively. In contrast to these low levels of general psychologic distress and cancer-related psychologic problems, PTSD (PCL-C) was more prevalent, with 15% (24 of 152; 95% CI, 10–23%) rating 2 or more PTSD symptoms as occurring moderately to extremely often. Among items scored as occurring moderately to extremely often, 12% (95% CI, 7–18%) reported 1 or more symptoms of reexperienced events, 16%, (95% CI, 10–23%) reported 1 or more symptoms of avoidance-numbing, and 24% (95% CI, 18–32%) reported 1 or more symptoms of increased arousal. The prevalence rate of PTSD among our survivors was 4.6% (7 of 153) using our adjusted rate for achieving a diagnostic score of PTSD.

Twenty-nine percent of survivors (95% CI, 22–37%) reported 1 or more sexual problems occurring since diagnosis, which they attributed to having been treated for cancer. The most prevalent problems reported were decreased interest in sexual activities (13%, n = 20; 95% CI, 8–20%), decreased sexual activity (12.5%, n = 19; 95% CI, 8–19%), and feelings of sexual unattractiveness (14.5%, n = 22; 95% CI, 9–21%). It was hypothesized that survivors might currently feel less sexually attractive if they were younger when initially diagnosed (pre-menopausal), compared with those who had been menopausal at the time of diagnosis. However, no significant differences in feeling sexually unattractive were found between age groups at the time of diagnosis: 16.5% (95% CI, 9–26%) for 20–44-year-olds, 11.5% (95% CI, 4–23%) for 44–55-year-olds, and 14.3% (95% CI, 3–36%) for survivors older than 55 years old. No significant differences regarding other sexual problems were found between survivors who were premenopausal and postmenopausal at the time of diagnosis.

The prevalence of survivors' perceived negative socioeconomic impact attributed to cancer since diagnosis varied in frequency by the type of problem. Thirteen percent of survivors (95% CI, 8–20%) reported 1 or more employment problems. Only 2% (95% CI, 0–6%) reported incidents of gross discrimination such as being fired or demoted. Twenty-six percent (95% CI, 20–34%) reported 1 or more insurance problems, the most frequent being denial of life insurance due to cancer (13%, n = 20; 95% CI, 9–20%), and the feeling that they could not change their jobs for fear of losing their health insurance (9%, n = 13; 95% CI, 5–14%). Only 7% (n = 10; 95% CI, 3–12%) of survivors attributed a negative socioeconomic impact due to having had cancer occurring from 1 year before diagnosis to the time of the interview.

As a consequence of having had a mastectomy and axillary lymph node dissection (ALND), survivors' primary long-term medical sequelae were lymphedema and numbness in their hands or chest on the operated side (Table 2). Oncologists attributed numbness in survivors' feet to vincristine in the CMFVP arm of CALGB 7581. A sizable subset of survivors reported having had lymphedema and numbness in their hands, feet, or chest: 39%; (95% CI, 31–47%) and 33% (95% CI, 26–41%), respectively. Twenty-five percent (n = 59; 95% CI, 13–26%) of those with lymphedema reported it as a problem for 10 or more years, with 22% (95% CI, 13–26%) reporting that it interfered with functioning ‘somewhat’ to ‘a great deal’. Of survivors reporting numbness, 19% (95% CI, 13–46%) stated that it continued for 10 years or more. No significant correlation was found between the type of mastectomy (radical vs. modified radical) and the length of time that survivors experienced lymphedema or numbness in their hands, feet, or chest (1–9 years vs. 10 years or longer), distress experienced due to these medical sequelae (somewhat to extremely bothersome), or any of the specific perceived sexual problems attributed to having had cancer.

As expected, other stressful life events occurred in survivors' lives in the past year. The mean number of negative life events was 2.3, which were attributed mostly to the death of a close family member (29%, n = 44; 95% CI, 22–37%), serious illness/injury of a close family member (37.5%, n = 57; 95% CI, 30–46%), major personal illness or injury (12%, n = 18; 95% CI, 7–18%), or the death of a close friend (12%, n = 18; 95% CI, 7–18%). Survivors overwhelmingly reported excellent social support, with medians of 87.5–100 for each of the subscales and total support (possible range, 0–100), comparable to ratings of having adequate support ‘most’ to ‘all of the time’.

Profile of Cancer Survivors Reporting Psychologic Distress

Pearson product correlations were calculated to identify which variables of distress (BSI GSI, the PCL-C, EORTC Emotional Functioning subscale) were at a sufficient level (P < 0.20) to be included in the hierarchic regression analyses (Table 3). These analyses revealed that less adequate social support (MOS) and more severe negative life events (LES) were related significantly to worse distress (BSI; P = 0.0032 and P < 0.05, respectively) and worse PTSD (PCL-C; P = 0.0033 and P = 0.0098, respectively) (Table 4). Lymphedema and numbness in the hands, feet, or chest that interfered with functioning were related significantly only to greater PTSD (PCL-C; P = 0.0081). Clinical trial treatment arm, having recurrent disease and being re-treated, and breast reconstruction were not significant predictors of distress.

Table 3. Correlations of Predictor Variables with Measures of Distress

Predictor variables	Worse distress (BSI General Severity Index)a	Worse posttraumatic stress disorder (PCL-C)a	Better emotional functioning (EORTC)b
BSI: Brief Symptom Inventory; PCL-C: Posttraumatic Stress Disorder Checklist-Civilian; GSI: General Severity Index; EORTC: European Organization for Research and Treatment of Cancer; LES: Life Experience Survey; MOS: Medical Outcomes Study. a Higher scores indicate worse distress or physical symptoms. b Higher scores indicate better functioning, social support, greater age or higher educational level. c P < 0.05. d P < 0.01. e P < 0.001. f P < 0.0001.
Higher distress
BSI-GSI	—	0.74f	−0.58f
BSI-Anxiety	—	0.58f	−0.54f
BSI-Depression	—	0.51f	−0.59f
EORTC
Cancer symptomsa	0.59f	0.38f	—
Physical functioningb	−0.47f	−0.25d	—
Role functioningb	−0.39f	−0.17	—
Social functioningb	−0.67f	−0.29d	—
Lymphedema and numbness interfering with functioninga	0.32f	0.33f	−0.24d
Greater no. of negative stressful life events (LES)a	0.45f	0.24d	−0.50f
Less adequate social support (MOS Social Support)b	−0.34f	−0.29e	0.31f
Conditioned distress due to reminders of treatmenta	0.09	0.24d	0.03
No. of sexual problems attributed to cancera	0.12	0.23d	−0.18c
Negative socioeconomic impacta	−.07	0.07	0.08
Ageb	−0.04	−0.05	0.08
Educationb	−0.12	0.07	0.13

Table 4. Significant Predictors of Global Distress (Brief Symptom Inventory) and Posttraumatic Stress Disorder (Posttraumatic Stress Disorder Checklist-Civilian): Hierarchical Regression Analyses

Predictor	BSI Global Severity Index P valuea	PTSD (PCL-C) P valueb
BSI: Brief Symptom Inventory; PTSD: posttraumatic stress disorder; PCL-C: Posttraumatic Stress Disorder Checklist-Civilian; EORTC: European Organization for Research and Treatment of Cancer; ns = nonsignificant. a Higher scores indicate better functioning, social support, patient satisfaction with medical care, greater age or higher education level. b Higher scores indicate worse symptoms and severity of negative life events.
Age	0.017	ns
Education levela	ns	0.026
EORTC Physical Functiona	0.013	ns
EORTC Physical Symptomsb	0.0015	0.071
Lymphedema and numbness interfering with functionb	ns	0.0081
MOS Social Supporta	0.0032	0.0033
Severity of negative life eventsb	0.045	0.0098
Patient satisfaction with medical carea	ns	0.037
R^2	0.51	0.39

Comparison of Survivors of Breast Carcinoma, Hodgkin Disease, and Adult Leukemia

Hodgkin disease, leukemia, and breast carcinoma survivors' adjustment has been studied in our prior studies.11, 17 To make comparisons of physical and psychosocial concerns among these 3 cohorts more valid, 3 subsets were created, comprising only women and those currently younger than the retirement age of 65 years. The subset of breast carcinoma survivors age < 65 years still was significantly older (P < 0.0001) and less likely to have disease recurrence and to have been re-treated (P < 0.0001) than the subset of younger female survivors of Hodgkin disease or leukemia (Table 5). Breast carcinoma survivors younger than age 65 years had significantly less depression, anxiety, conditioned distress, and perceived negative socioeconomic impact on their lives attributed to cancer when compared with younger female survivors of Hodgkin disease or leukemia (P < 0.002, P < 0.0001, P < 0.0001, and P < 0.0001, respectively). The breast carcinoma subset also had significantly less conditioned nausea triggered by cancer treatment-related stimuli than Hodgkin disease survivors (P < 0.0001). No statistically significant differences were found among the three subsets in terms of those scoring at suggested levels of psychiatric disorder (BSI).

Table 5. Comparison of Generic and Cancer-Related Problems Experienced by Survivors of Breast Carcinoma, Hodgkin Disease, and Adult Leukemia

Measure	Breast carcinoma		Hodgkin disease		Leukemia
	All patients (n = 153)	Women ages 18–64 yrs (n = 79)	All patientsa (n = 273)	Women ages 18–64 yrs (n = 107)	All patientsb (n = 206)	Women ages 18–64 yrs (n = 95)
CI: confidence interval; SD: standard deviation. a Data from Kornblith AB, Herndon J, Zuckerman E, et al. Comparison of long-term psychosocial adaptation of advanced stage Hodgkin's disease and acute leukemia survivors. Ann Oncol. 1998;9:297–306.9 b Data from Greenberg DB, Kornblith AB, Herndon JE, et al. Quality of life of adult leukemia survivors treated on clinical trials of the Cancer and Leukemia Group B from 1971–1988: predictors for later psychological distress. Cancer. 1997;80:1936–1944.11 c P < 0.0001 for comparison with value for women ages 18–64 years with breast carcinoma. d P < 0.002 for comparison with value for women ages 18–64 years with breast carcinoma.
Median age at interview (yrs)	64	58	37	34c	41	39c
Marital status (% [95% CI])
Married	66	73 (62–83)	64	61 (51–70)	69	69 (59–78)
Separated/divorced/widowed	29	22 (13–32)	16	20 (13–28)	14	14 (8–23)
Never married	5	5 (1–12)	20	20 (13–28)	16	16 (10–25)
Recurrence and re-treatment (% [95% CI])	22	25.3 (16–36)	29	36.5c (27–46)	7	5.2c (2–12)
BSI
High distress (% [95% CI])	5	8 (3–16)	21	14 (8–22)	14	14 (7–22)
Mean anxiety score (SD)	0.3 (0.47)	0.21 (0.42)	0.52	0.69c (0.68)	0.45	0.21c (0.42)
Mean depression score (SD)	0.2 (0.42)	0.27 (0.46)	0.56	0.60d (0.72)	0.35	0.61d (0.66)
Sexual problems (% [95% CI])
Decreased interest in sexual activity	13	15 (8–25)	17	21 (13–29)	21	19 (11–28)
Decreased actual sexual activity	12.5	13 (6–22)	20	23 (16–33)	29	27 (18–37)
Feelings of sexual unattractiveness	14.5	16.5 (9–26)	12	16 (10–24)	17	22 (14–31)
Conditioned…(%)
Nausea	3.5 (n = 144)	9 (95% CI, 3–16; n = 76)	35 (n = 225)	41c (95% CI, 30–51; n = 92)	21 (n = 199)	20 (95% CI, 12–29; n = 94)
Vomiting	1 (n = 143)	1 (95% CI, 0–5; n = 75)	3 (n = 225)	4 (95% CI, 1–9; n = 92)	2.5 (n = 200)	2 (95% CI, 0–6; n = 94)
Distress	6 (n = 144)	8 (95% CI, 3–16; n = 76)	41 (n = 225)	46c (95% CI, 35–56; n = 92)	42 (n = 198)	51c (95% CI, 40–60; n = 94)
Negative socioeconomic impact of 1+ problems (% [95% CI])	7	9 (4–17)	36	37c (28–47)	39	44c (34–55)

DISCUSSION

Adjustment of Survivors of Breast Carcinoma

There were many areas in which cancer had a minimal effect on survivors' adjustment 20 years after diagnosis: global psychologic distress, smells and tastes associated with cancer treatment that triggered conditioned responses of nausea and emesis, and a perceived negative socioeconomic impact attributed to cancer. These results were upheld when breast carcinoma survivors younger than age 65 years were compared with female Hodgkin disease survivors and leukemia survivors who were younger than 65 years old, controlling for gender and retirement age. Breast carcinoma survivors' who reported their social support as excellent may have experienced to offset any lasting impact of cancer. Other stressful life events may have overtaken breast carcinoma in their importance in survivors' distress. For example, 29% reported the death of a close family member and 37.5% reported that the serious illness of a close family member, along with other negative events in the year before the interview, were stressful life events. When other sociodemographic, medical, and psychosocial variables were statistically controlled, the severity of negative, stressful events remained as a significant predictor of both global psychologic distress (BSI: P < 0.05) and PTSD (PCL-C: P = 0.0098).

Although the majority of breast carcinoma survivors demonstrated a remarkable recovery from their cancer diagnosis and treatment, a subset of survivors reported a range of cancer-related emotional problems that included PTSD, sexual and insurance problems attributed to cancer, and long-term medical sequelae of lymphedema and numbness in the hands, feet, or chest. Univariate analyses revealed a significant correlation between global psychologic distress (BSI) and lymphedema and numbness that interfered with functioning at some point after the cancer diagnosis. These medical sequelae were retained in multivariate analyses involving PTSD (PCL-C) but not in those involving the BSI. The report by Kwan et al.25 ‘supported this finding. Kwan et al. found a highly significant relationship between the interference of lymphedema’ with functioning and social functioning, relative to the social functioning of survivors who did not have lymphedema (EORTC: P = 0.007).

In the current study, 25% of survivors experienced lymphedema for 10 years or longer. These figures were quite similar to those in another cohort of breast carcinoma survivors treated 20 years ago. In that study,26 30% of survivors reported moderate to severe lymphedema (arm enlargement of 0.5 in. [1.27 cm] or greater compared with the contralateral side). In another study,27 ALND affected numbness in the arm in 63% of survivors from 6 months to 10 years post-ALND. This prevalence was much higher than in the current study, which reported numbness in the hands, chest (due to primary surgery), or feet (due to vincristine) in 19% of survivors who had the symptom for 10 years or longer. These 2 sets of findings suggest that, perhaps, numbness decreases over time but still persists in a subset of cancer survivors 20 years after their diagnosis.

It was counterintuitive to find that disease recurrence was not related significantly to psychologic distress. It is likely that with the interview occurring a mean of 10.2 years after disease recurrence and re-treatment, survivors already had adapted well to recurrence and re-treatment. Similarly, studies of survivors of Hodgkin disease and leukemia11, 17 who had completed treatment an average of 6 years earlier showed no correlation between disease recurrence and psychologic distress.

PTSD involves highly traumatizing events that are associated with numerous stimuli related to survivors' cancer treatment long after event occurred.15 PTSD symptoms were correlated most significantly with frequency of lymphedema, numbness in the hands, feet, or chest, and other physical problems. These breast carcinoma–related medical sequelae, along with current physical symptoms (EORTC), may have become the trigger for reexperiencing the traumatic event.

In our sample, 4.6% (7 of 153) of survivors scored at both a suggested PTSD diagnostic level based on its original scoring14 and above cutoff levels of the psychiatric disorder (BSI). Using the adjusted rate of PTSD, these findings suggest that this represented the true prevalence of survivors with a psychiatric diagnosis of PTSD. In several studies involving national rates of psychiatric disorders determined by a structured clinical interview, the prevalence of PTSD ranged from 0.4% to 13%, considering both a PTSD diagnosis within the past 1–6 months and lifetime rates.28 The prevalence of survivors with PTSD ranged from 3.5% to 16% for those who had experienced any type of trauma.28 Using our adjusted method for determining survivors with a clinical diagnosis of PTSD, our reported PTSD prevalence of 4.6% is consistent with the lower PTSD rates of survivors exposed to any type of trauma.

Sexual problems attributed to cancer were common. Twenty-nine percent attributed one or more current sexual problems to having had cancer, primarily due to feeling sexually unattractive and to decreased sexual interest and activity. The continued sense of feeling sexually unattractive is a telling response across age groups at diagnosis and likely is related to survivors' original mastectomy and ALND, as well as to any continued medical sequelae. Similar prevalence of sexual problems in survivors approximately 7. 5 years posttreatment has been noted in other studies: breast carcinoma, 33%;29 Hodgkin disease, 18–25%; leukemia, 21–29%; and testicular carcinoma, 35%.1 However, in the current study, as well as in the study by Ganz et al.,30 breast carcinoma survivors' sexual problems attributed to cancer have not been correlated strongly with either global psychologic distress or PTSD. Problems with sexual functioning may not be viewed as a source of distress in the lives of survivors, with other stressful negative life events potentially assuming greater importance.

Methodologic Limitations

In the current study, the degree to which survivors' adjustment differed from that of a community sample without chronic disease or psychiatric disorders was not tested. There is no doubt that having a community sample for a comparison group would have enhanced the findings of our study with respect to similar problems (such as psychologic distress) that were unrelated to cancer. Because the cost of obtaining a comparison group was prohibitive, the normative data from the BSI and the reports from other studies concerning PTSD were used as comparisons for our findings. It also should be noted that comparison of survivors' reports of problems attributed to having had cancer (e.g., sexual problems) with similar types of problems in community residents may not be valid, as the attributed cause is not identical. Without the ability to have identical items for both cancer survivors and community residents, it would not be meaningful to compare the findings from these two groups.

A portion of the questions in the questionnaires were retrospective. Survivors were asked to recall their sociodemographic status; cancer-related physical symptoms; and sexual, insurance, and employment problems from 1 year before their diagnosis up to the current interview, 20 years postdiagnosis. It is possible that survivors excessively attributed their problems to cancer. Although an objective evaluation of problems attributed to cancer would be preferred, such evaluations rarely are performed because their results are very difficult, if not impossible, to assess.

As in all cross-sectional studies, including the current one, with the exception of most sociodemographic and medical/treatment factors, the causal relationship between social support and stressful life events predicting distress cannot be established definitively. Only longitudinal studies can more unequivocally determine these relationships. However, the current study presents a profile of characteristics that may co-occur with psychologic distress. Statistical analyses were guided by the Vulnerability Model to test possible predictors of distress. A case in point is previous research that involved a cross-sectional sample to test causal relationships involving three theoretic models regarding the relationship between social support and stressful life events and distress.31

All survivors who participated in the current study had been treated on one clinical trial. As 2.5% of all patients with cancer are enrolled in clinical trials,32 the current sample represented only a small proportion of patients treated in the mid-1970s who would have become cancer survivors 20 years later. Our ability to generalize the results of the study is limited further by having had all survivors participate in just one clinical trial, the low accrual of minorities, and the likelihood that survivors had higher incomes and better health at the beginning of treatment than those who did not meet the medical eligibility criteria of this treatment trial. In addition, because survivors with major psychiatric disorders, which could have been related to their cancer, were excluded, the results may have underestimated the prevalence of psychiatric problems. Our approach allowed us to obtain a homogeneous sample in terms of disease stage, gender, treatment received, and length of time since diagnosis, which helped to illuminate the impact of cancer on survivors' lives.

Because the sample size involved 153 survivors, the 95% CIs (Tables 2, 5) showed a wide range of values for many of the variables, which most likely would have been reduced to more precise figures had the sample size been larger. However, the current study represented survivors who had been treated on this single trial.

Conclusions and Clinical Implications

Many of the results concerning survivors' adaptation suggested a minimal impact of cancer on survivors' lives 20 years after diagnosis. Other negative, stressful events and adequacy of social support assumed a greater importance and affected their psychologic distress. However, the impact of cancer was not entirely forgotten, as there was also a subset of survivors who continued to experience PTSD symptoms, lymphedema, and numbness in the hands and chest due to breast carcinoma surgery, and numbness in their feet due to vincristine. The impact of these medical sequelae on psychologic distress might be underrecognized. By detecting these cancer-related medical and psychologic sequelae earlier in survivors' follow-up care, those more vulnerable to subsequent adjustment problems could be identified to receive mental health treatment.

Acknowledgements

The authors thank all the breast carcinoma survivors who so graciously elected to participate in this study and who shared their private thoughts and feelings about their experiences with cancer.