Abstract
Background
Few studies have examined risk for severe symptoms during early cancer survivorship. Using baseline data from the American Cancer Society’s Study of Cancer Survivors-I, we examined cancer survivors with high symptom burden, identified risk factors associated with high symptom burden, and evaluated the impact of high symptom burden on health-related quality of life (HRQoL) 1 year post-diagnosis.
Methods
Participants were enrolled from 11 state cancer registries approximately 1 year after diagnosis and surveyed by telephone or mail. Outcomes measures were the Modified Rotterdam Symptom Checklist and Profile of Mood States-37 (to assess symptom burden) and the Satisfaction with Life Domains Scale-Cancer (to assess HRQoL).
Results
Of 4903 survivors, 4512 (92%) reported symptoms related to their cancer and/or its treatment. Two-step clustering yielded 2 sub-groups, one with low symptom burden (n=3113) and one with high symptom burden (n=1399). Variables associated with high symptom burden included metastatic cancer (odds ratio [OR], 2.05), number of comorbid conditions (OR, 1.76), remaining on active chemotherapy (OR, 1.93), younger age (OR, 2.31), lacking insurance/being underinsured (OR, 1.57), having lower income (OR, 1.61), being unemployed (OR, 1.27), or being less educated (OR, 1.29). Depression, fatigue, and pain had the greatest impact on HRQoL in survivors with high symptom burden, who also had lower HRQoL (P < .0001).
Conclusions
More than 1 in 4 cancer survivors had high symptom burden 1 year post-diagnosis, even after treatment termination. These results indicate a need for continued symptom monitoring and management in early posttreatment survivorship, especially for the underserved.
Keywords: cancer survivorship, symptom burden, late effects, risk factors, quality of life, cluster analysis
INTRODUCTION
A cancer survivor is any person who has received a diagnosis of cancer; survivorship begins at diagnosis and continues through the balance of the patient’s life.1 Owing to advances in cancer detection and treatment, the number of individuals living years after their cancer diagnosis has increased steadily. The 5-year survival rate for all U.S. cancer patients diagnosed between 1996 and 2004 has increased to 66%, and there are more than 11 million cancer survivors in the United States.2
The benefit of a longer life is offset for many survivors, however, by multiple persistent symptoms, including fatigue, distress, pain, and cognitive impairment. Some survivors remain on anticancer treatment and thus continue to experience treatment-related symptoms. Others who have completed treatment will experience residual symptoms of both the disease and the treatment, compounded by lack of the close medical monitoring that accompanies active therapy. In either case, symptoms cause a significant burden that diminishes survivors’ quality of life.3–6
The 12 months after diagnosis may be especially stressful for patients,7 and risk for symptom burden induced by cancer and its treatment remains high during this period. As patients move into the posttreatment period, individual variations in the presence and severity of physical and psychological morbidities are common.8 The severity of symptoms is critical, as higher levels of symptom burden disproportionately impair function and quality of life.9,10
Therefore, it is important to identify specific groups of cancer survivors who are at greater risk for severe symptoms. Historically, however, in most symptom research the determination of variables associated with high symptom burden has been limited by a number of factors, including lack of statistical power due to small sample size11; lack of the assessment of symptom severity, an understanding of which would allow identification of the most symptomatic survivors12; confounding because of variability in the length of time as a survivor13; and a tendency to consider symptoms individually, regardless of the co-occurrence of multiple symptoms.14
To address these limitations, we conducted an analysis of population-based data from the American Cancer Society’s Study of Cancer Survivors-I (SCS-I). All subjects had received their cancer diagnosis approximately 1 year prior to enrollment. We aimed to identify a subgroup with higher symptom burden in the early stages of cancer survivorship and to document possible factors that could contribute to having more-severe symptoms in early survivorship. We also investigated the pattern of multiple symptoms in this high-symptom group and identified symptoms that had the greatest impact on survivors’ ratings of health-related quality of life (HRQoL).
METHODS
Study population
The SCS-I study was designed to identify longitudinal patterns of change in cancer survivors’ HRQoL over an 11-year period. Cancer survivors representing diverse diagnoses, geographic regions, and demographic groups were randomly recruited from the cancer registries of 11 states. All survivors were at least 18 years old, diagnosed with one of the 10 most highly incident cancers (non-Hodgkin lymphoma, skin melanoma, or prostate, female breast, lung, colorectal, urinary bladder, kidney, ovarian, or uterine cancer), and approximately 1 year postdiagnosis at the time of recruitment and initial assessment. Sampling and investigation details for the SCS-I study have been described.15 Overall approval for the SCS-I was obtained from the Institutional Review Board (IRB) of Emory University. Additional IRB and/or regulatory approvals were obtained for each state registry.15
To ensure the reliability of our analyses, we selected only those survivors whose missing-response rate on the study’s baseline symptom assessment was less than 20%.
Measurement
Medical, demographic, and socioeconomic data in the SCS-I database was obtained from the state cancer registries or via a self-report survey. Patient-reported HRQoL and symptom severity were collected via mailed questionnaires. The Modified Rotterdam Checklist (RSCL-M) was used to assess physical symptoms: patients indicated the extent to which they have been bothered by each of 30 symptoms during the past week (not at all, a little, quite a bit, or very much).16 The short-form Profile of Mood States (POMS-37) was used to assess psychological symptoms: patients indicated for 37 adjectives from 6 subscales (tension, depression, confusion, fatigue, anger, and vigor) the extent to which they had been feeling that way during the past 2 weeks (not at all, a little, moderately, quite a bit, or extremely).17 The Satisfaction with Life Domains Scale-Cancer (SLDS-C) was used to assess HRQoL: patients indicated their current satisfaction with 17 life domains relevant to HRQoL by choosing 1 of 7 faces, ranging from a “terrible” face with a deep, downturned frown to a “delighted” face with a large smile.18
Statistical analysis
Because the current study focused on symptomatic cancer survivors, respondents who reported “not at all” for all RSCL-M and all POMS-37 items were removed from analysis. For missed symptom items, mean imputation was used for both scales according to the instrument’s manual. Symptom scores were linearly transformed to a 0–100 metric to facilitate interpretation, with higher scores indicating more severe symptoms. Because of their inverse scoring, 6 symptoms belonging to the POMS-37 vigor subscale were reverse scored for a consistent interpretation with other symptoms.
To identify survivors with high symptom burden, we applied a 2-step cluster analysis to divide cancer survivors into groups with distinct levels of symptom severity (high versus low). With the 67 symptoms from RSCL-M and POMS-37 as variables, the clustering solution was obtained using log-likelihood estimation, and the optimal number of groups was defined as that with the lowest Bayesian Information Criterion (BIC). To confirm that 2-step clustering effectively differentiated survivors according to symptom burden, we used Student’s t test and Cohen’s d effect size to compare the severity of individual symptoms between the high-symptom and low-symptom groups. The magnitude of effect size was based on the value of d: 0.2 as small, 0.5 as medium, and 0.8 as large.19 We expected the 2-step clustering to generate subgroups of survivors with distinct levels of symptoms with at least a medium effect size. We used the total score of the SLDS-C to represent HRQoL, with a lower score indicating poorer HRQoL; t tests were used to compare SLDS-C scores among groups with different symptom levels.
To estimate the associations between demographic and clinical variables and symptoms, we conducted logistic modeling with group membership as the dependent variable. Independent variables were all demographic, socioeconomic, and disease characteristics, including age, gender, race, education level, annual household income, marital status, current health insurance status, current employment status, cancer stage, number of comorbid conditions, cancer type, and type of therapy (surgery, chemotherapy, and/or radiotherapy). Odds ratios (OR) and 95% confidence intervals (CI) were calculated to estimate the effect of each risk factor. Only survivors with complete data for all variables were used for this analysis.
To explore the pattern of physical symptoms in the high-symptom group, we conducted exploratory factor analysis of the RSCL-M items using principal axis factoring as the extraction method. A factor loading of 0.30 or greater was deemed as significant.20 Factor-based component scores were computed by averaging items loading principally on a particular factor. We used regression analyses to compare RSCL-M factor scores and POMS-37 subscale scores between survivors actively undergoing chemotherapy or radiotherapy and survivors who had already completed therapy.
To evaluate the impact of cancer-related symptoms on HRQoL in the high-symptom group, we employed a linear regression model of the factor-based composite scores of the RSCL-M and 4 subscale scores of the POMS-37 (depression, tension, anger, and confusion). POMS-37 fatigue and vigor subscales were excluded to avoid redundancy with the RSCL-M variables (which also measured fatigue). The most distressing symptoms were selected on the basis of the standardized regression coefficient, a parameter used to identify the independent variable that has a greater unique effect upon the criterion (SLDS-C total score). The larger the standardized coefficient value, the greater the effect.21
We used SPSS version 16.0 statistical software (Chicago, IL) to perform the 2-step cluster analysis and SAS version 9.2 statistical software (Cary, NC) to perform all other analyses. All statistical tests were 2-sided; P values <.05 were considered statistically significant.
RESULTS
Patient characteristics
The SCS-I database contained 5277 cancer survivors whose self-reported cancer site was consistent with information in the cancer registry. From these, we identified 4903 (93%) whose missing-response rate on the study’s baseline symptom assessment was less than 20%. Demographic and clinical characteristics are shown in Table 1. Of the 4903 survivors with adequate symptom data, 391 (8%) were nonsymptomatic (ie, reported “not at all” for all RSCL-M and POMS-37 items) and were removed from further analysis, resulting in a sample of 4512 survivors with some level of symptom burden.
Table 1.
Cancer Survivor Characteristics (N=4903)
| Characteristic | No. of Survivors (%) | P | |
|---|---|---|---|
| Symptomatic Survivors (n=4512) | Asymptomatic Survivors (n=391) | ||
| Gender | |||
| Women | 2633 (58.4) | 188 (48.1) | <.0001 |
| Men | 1879 (41.6) | 203 (51.9) | |
| Age | |||
| <55 y | 2200 (48.8) | 192 (49.1) | .896 |
| ≥55 y | 2312 (51.2) | 199 (50.9) | |
| Race | |||
| Non-Hispanic White | 3957 (87.7) | 349 (89.3) | |
| Black | 310 (6.9) | 17 (4.3) | |
| Hispanic | 153 (3.4) | 13 (3.3) | |
| Other | 92 (2.0) | 12 (3.1) | |
| Married | |||
| Yes | 3302 (73.6) | 305 (78.0) | .064 |
| No | 1183 (26.4) | 86 (22.0) | |
| Education | |||
| College or higher | 2676 (60.2) | 263 (68.1) | .007 |
| Other | 1766 (39.8) | 128 (32.9) | |
| Annual household income | |||
| < $5,000 | 56 (1.3) | 1 (0.3) | .149 |
| $5,000 – $9,999 | 170 (3.9) | 5 (1.3) | |
| $10,000 – $19,999 | 437 (9.9) | 17 (1.3) | |
| $20,000 – $39,999 | 1001 (22.9) | 79 (20.2) | |
| $40,000 – $74,999 | 1251 (28.6) | 118 (30.2) | |
| ≥$75,000 | 1005 (23.0) | 120 (30.7) | |
| Currently employed | |||
| Yes | 2342 (51.9) | 233 (59.6) | .004 |
| No | 2170 (48.1) | 158 (40.4) | |
| Health insurance | |||
| Employer paid | 2427 (54.7) | 244 (66.8) | .004 |
| A plan that you or someone else buys on your own | 250 (5.6) | 22 (5.6) | |
| Military, Champus, TriCare, Veterans Administration | 51 (1.2) | 1 (0.3) | |
| Medicare | 1117 (25.2) | 92 (23.5) | |
| Medicaid or medical assistance | 122 (2.8) | 4 (1.0) | |
| Other source | 80 (1.8) | 2 (0.5) | |
| None | 97 (2.2) | 2 (0.5) | |
| Cancer type | |||
| Breast | 1098 (24.3) | 81 (20.7) | <.0001 |
| Prostate | 822 (18.2) | 81 (20.7) | |
| Colorectal | 676 (15.0) | 63 (16.1) | |
| Bladder | 179 (4.0) | 16 (4.1) | |
| Uterine | 222 (4.9) | 20 (5.1) | |
| Melanoma | 210 (4.6) | 61 (15.6) | |
| Non-Hodgkin lymphoma | 287 (6.4) | 21 (5.4) | |
| Kidney | 269 (6.0) | 27 (6.9) | |
| Lung | 463 (10.3) | 7 (1.8) | |
| Ovarian | 286 (6.3) | 14 (3.6) | |
| Cancer stage | |||
| In situ | 100 (2.2) | 10 (2.6) | <.0001 |
| Localized | 2616 (58.0) | 275 (70.3) | |
| Regional | 1245 (27.6) | 81 (20.7) | |
| Distant | 551 (12.2) | 25 (6.4) | |
| Comorbid conditions | |||
| None | 1977 (43.8) | 226 (57.8) | <.0001 |
| One | 1251 (27.7) | 116 (29.7) | |
| Two or more | 1284 (28.5) | 49 (12.5) | |
| Anticancer treatments | |||
| Previous surgery | 3684 (83.9) | 333 (85.2) | .083 |
| Chemotherapy-completed | 1594 (36.8) | 107 (27.4) | .231 |
| Chemotherapy-active | 512 (11.8) | 43 (11.0) | |
| Radiotherapy-completed | 1520 (35.3) | 104 (26.6) | .253 |
| Radiotherapy-active | 393 (9.1) | 34 (8.7) | |
Survivors with High Symptom Burden
Two-step clustering generated 2 groups: a low-symptom group (n=3113, 69%) and a high-symptom group with significantly higher mean symptom severity (n=1399, 31%). Except for weight gain and problems controlling urine, effect sizes (Cohen’s d) of differences between the high-symptom and low-symptom groups for all symptoms equaled or exceeded 0.5, indicating medium-to-large magnitude of effect (Appendix 1; appendices online only). We compared HRQoL in the high-symptom and low-symptom groups, using the SLDS-C total score as an index of overall HRQoL. The high-symptom group reported significantly lower mean SLDS-C scores than did the low-symptom group (73.72 (95%CI, 72.74–74.70) vs. 98.24 (95%CI 97.79–98.69); P < .0001).
Variables Associated with Symptom Burden
Multivariate logistic regression modeling revealed that survivors who were younger than 55 years (OR, 2.31; 95% CI, 1.91–2.80), had an annual household income below $40,000 (OR, 1.61; 95% CI, 1.34–1.94), were currently unemployed (OR, 1.27; 95% CI, 1.05–1.53), had no more than a high school education (OR, 1.29; 95% CI, 1.09–1.53), or were uninsured/underinsured (ie, on Medicaid or medical assistance) (OR, 1.57; 95% CI, 1.10–2.24) were more likely to experience severe symptoms (Table 2). Clinical characteristics for severe symptoms included having lung cancer (OR, 2.27; 95% CI, 1.76–2.94), having distant metastases (OR, 2.05; 95% CI, 1.60–2.62), or undergoing active chemotherapy (OR, 1.93; 95% CI, 1.47–2.54). Further, the likelihood of severe symptoms increased as the number of comorbidities increased (OR, 1.76; 95% CI, 1.59–1.95). Because survivors with lung cancer were more likely to report severe symptoms, we repeated the logistic regression analyses separately in survivors with lung cancer and those with other cancers. Survivors with other cancers were similar to the entire sample in the associations between symptom burden and multiple variables. In lung cancer survivors, younger age, low income, unemployment, minimal health insurance, more comorbidities, and metastasis remained significantly associated with high symptom burden (data not shown).
Table 2.
Variables Associated with Symptom Burden
| High-Symptom Group (n=1054) a | Low-Symptom Group (n=2338)a | OR (95% CI) | P | |
|---|---|---|---|---|
| No. Patients (%) | No. Patients (%) | |||
| Age | ||||
| ≥55 y | 433 (27.7) | 1132 (72.3) | 1.00 | |
| <55 y | 621 (34.0) | 1206 (66.0) | 2.31 (1.91 – 2.80) | <.0001 |
| Sex | ||||
| Male | 410 (29.3) | 989 (70.7) | 1.00 | |
| Female | 644 (32.3) | 1349 (67.7) | 1.05 (0.88 – 1.25) | .592 |
| Race | ||||
| Non-Hispanic white | 904 (30.0) | 2106 (70.0) | 1.00 | |
| Other | 150 (39.3) | 232 (60.7) | 1.07 (0.84 – 1.36) | .597 |
| Married | ||||
| Yes | 730 (28.9) | 1794 (71.1) | 1.00 | |
| No | 324 (37.3) | 544 (62.7) | 1.18 (0.98 – 1.42) | .090 |
| Education level | ||||
| College or higher | 595 (27.4) | 1576 (72.6) | 1.00 | |
| High school or lower | 459 (37.6) | 762 (62.4) | 1.29 (1.09 – 1.53) | .003 |
| Annual household income | ||||
| ≥$40,000 | 525 (25.3) | 1548 (74.7) | 1.00 | |
| < $40,000 | 529 (40.1) | 790 (59.9) | 1.61 (1.34 – 1.94) | <.0001 |
| Currently employed | ||||
| Yes | 542 (27.6) | 1419 (72.4) | 1.00 | |
| No | 512 (35.8) | 919 (64.2) | 1.27 (1.05 – 1.53) | .012 |
| Health insurance | ||||
| Yes | 954 (29.7) | 2263 (70.3) | 1.00 | |
| None/Medicaid | 100 (57.1) | 75 (54.3) | 1.57 (1.10 – 2.24) | .012 |
| Metastasis | ||||
| No | 865 (29.0) | 2113 (71.0) | 1.00 | |
| Yes | 189 (45.7) | 225 (54.3) | 2.05 (1.60 – 2.62) | <.0001 |
| Comorbidities | 1.76 (1.59 – 1.95) | <.0001 | ||
| None | 369 (24.0) | 1168 (76.0) | 1.00 | |
| One | 283 (29.1) | 688 (70.9) | 1.52 (1.26 – 1.83) | <.0001 |
| Two or more | 402 (45.5) | 482 (54.5) | 3.22 (2.65 – 3.91) | <.0001 |
| Lung cancer | ||||
| No | 900 (29.3) | 2175 (70.7) | 1.00 | |
| Yes | 154 (48.6) | 163 (51.4) | 2.27 (1.76 – 2.94) | <.0001 |
| Surgery | ||||
| No | 183 (35.7) | 330 (64.3) | 1.00 | |
| Yes | 871 (30.3) | 2008 (69.7) | 0.96 (0.76 – 1.21) | .748 |
| Chemotherapy | ||||
| No therapy | 471 (27.1) | 1268 (72.9) | 1.00 | |
| Therapy completed | 417 (33.2) | 840 (66.8) | 1.23 (1.02 – 1.49) | .031 |
| Active therapy | 166 (41.9) | 230 (58.1) | 1.93 (1.47 – 2.54) | <.0001 |
| Active therapy vs. therapy completed | 1.45 (1.11 – 1.90) | <.0001 | ||
| Radiotherapy | ||||
| No therapy | 593 (31.0) | 1318 (69.0) | 1.00 | |
| Therapy completed | 367 (30.5) | 837 (69.5) | 1.02 (0.85 – 1.21) | .863 |
| Active therapy | 94 (33.9) | 183 (66.1) | 0.82 (0.85 – 1.21) | .211 |
| Active therapy vs. therapy completed | 0.83 (0.60 – 1.15) | .627 |
OR indicates odds ratio; CI, confidence interval.
Only survivors with complete data in all variables (n=3392) were included in the logistic regression model.
Symptom profiles in the high-symptom group
Exploratory factor analysis of the RSCL-M in the high-symptom group revealed 7 factors (Table 3): pain, treatment-related symptoms, fatigue, nausea, bowel/bladder control problems, weight change, and lung symptoms, which accounted for 7%, 6%, 5%, 4%, 4%, and 3%, respectively, of the total variance among the RSCL-M items. The pain factor comprised 5 items that loaded principally upon it: pain (other than low back), sore muscles, low back pain, difficulty sleeping, and headaches. The treatment-related symptoms factor included dry mouth, burning/sore eyes, loss of hair, sore mouth/pain when swallowing, tingling hands/feet, skin irritation, swelling of arms/legs, and dizziness. The fatigue factor was composed of tiredness and lack of energy. The nausea factor included nausea and vomiting. The control-problems factor included problems controlling bowels, diarrhea, and abdominal aches. The weight-change factor was composed of weight loss, lack of appetite, and weight gain, which exhibited a negative loading. The lung-symptoms factor included coughing and shortness of breath. Five items did not belong to any factor because their factor loadings were lower than 0.30: decreased sexual interest, shivering, constipation, heartburn/belching, and problems controlling urine.
Table 3.
Factor Structure for RSCL-M in Cancer Survivors with High Symptom Burden (n=1399)
| Factor Loading | |||||||
|---|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | |
| Pain-related | |||||||
| Pain (other than low back pain) | 0.691 | ||||||
| Sore muscles | 0.674 | ||||||
| Low back pain | 0.591 | ||||||
| Difficulty sleeping | 0.339 | ||||||
| Headaches | 0.328 | ||||||
| Decreased sexual interest | 0.158 | ||||||
| Treatment-related | |||||||
| Dry mouth | 0.495 | ||||||
| Burning/sore eyes | 0.408 | ||||||
| Loss of hair | 0.393 | ||||||
| Sore mouth/pain when swallowing | 0.392 | ||||||
| Tingling hands or feet | 0.372 | ||||||
| Skin irritation | 0.370 | ||||||
| Swelling of arms or legs | 0.365 | ||||||
| Dizziness | 0.300 | ||||||
| Shivering | 0.293 | ||||||
| Constipation | 0.286 | ||||||
| Heartburn/belching | 0.239 | ||||||
| Fatigue | |||||||
| Tiredness | 0.871 | ||||||
| Lack of energy | 0.818 | ||||||
| Nausea | |||||||
| Nausea | 0.807 | ||||||
| Vomiting | 0.675 | ||||||
| Controlling problem | |||||||
| Problems controlling your bowels | 0.799 | ||||||
| Diarrhea | 0.634 | ||||||
| Abdominal aches | 0.360 | ||||||
| Problems controlling your urine | 0.291 | ||||||
| Weight change | |||||||
| Weight loss | 0.645 | ||||||
| Lack of appetite | 0.606 | ||||||
| Weight gain | −0.505 | ||||||
| Lung symptoms | |||||||
| Coughing | 0.664 | ||||||
| Shortness of breath | 0.453 | ||||||
RSCL-M indicates Modified Rotterdam Checklist.
To derive a component score for each RSCL-M factor or POMS-37 subscale in the high-symptom group, we averaged the ratings for all items loading on a particular factor or subscale. The 3 highest RSCL-M symptom factor scores were for fatigue, pain, and lung symptoms, whereas the top-scoring POMS-37 subscales were vigor, fatigue, and tension (Appendix 2; appendices online only).
Given that currently receiving chemotherapy contributed to membership in the high-symptom group, we examined whether completion of chemotherapy or radiotherapy affected symptom severity in the high-symptom group, of whom 523 (37%) had completed chemotherapy and 212 (15%) were still undergoing chemotherapy, and 460 (33%) had completed radiotherapy and 137 (10%) were still undergoing radiotherapy. No significant differences in symptom severity between survivors undergoing active chemotherapy and survivors who had completed chemotherapy were found for any symptom other than nausea (higher in the active treatment group; P = .005). No significant differences in any symptoms were found between survivors who had completed radiation and those still receiving treatment (Appendix 2).
Impact of symptoms on HRQoL
We estimated effects of symptoms on HRQoL in the high-symptom group using a linear regression model, with the total score of SLDS-C as the dependent variable. The POMS-37 depression subscale, RSCL-M fatigue factor, and RSCL-M pain factor had the strongest standardized coefficients (−0.41, −0.22, and −0.09, respectively), suggesting that among all 67 symptom items, depression, fatigue, and pain affected a survivor’s HRQoL most (Table 4).
Table 4.
Regression Model Analyses for the Impact of Symptoms on Health-Related Quality of Life
| Variable | Estimate | Standard Error | t | P | Standardized Estimate |
|---|---|---|---|---|---|
| RSCL-M factors | |||||
| Fatigue | −0.146 | 0.011 | −13.83 | <.0001 | −0.222 |
| Pain | −0.084 | 0.015 | −5.75 | <.0001 | −0.092 |
| Weight change | −0.085 | 0.019 | −4.37 | <.0001 | −0.060 |
| Lung symptoms | −0.044 | 0.013 | −3.39 | .0007 | −0.049 |
| Nausea | −0.037 | 0.018 | −1.98 | .048 | −0.026 |
| Bowel/bladder control problems | −0.025 | 0.015 | −1.67 | .095 | −0.021 |
| Treatment-related symptoms | −0.032 | 0.024 | −1.30 | .192 | −0.020 |
| POMS-37 sub-scales | |||||
| Depression | −0.409 | 0.022 | −18.19 | <.0001 | −0.409 |
| Tension | 0.072 | 0.020 | 3.65 | .0003 | 0.075 |
| Confusion | −0.044 | 0.021 | −2.06 | .040 | −0.040 |
| Anger | −0.012 | 0.020 | −0.60 | .545 | −0.011 |
RSCL-M indicates Modified Rotterdam Checklist; POMS-37, short form of the Profile of Mood States.
DISCUSSION
In this analysis of 1-year cancer survivors drawn from nationwide, population-based survey data, more than 1 in 4 survivors were categorized into a high-symptom group via 2-step cluster analysis. Survivors in this high-symptom group had significantly higher mean symptom severity on 2 assessment measures (RSCL-M and POMS-37) and lower HRQoL ratings than did survivors in the low-symptom group. Characteristics associated with high-symptom group membership included being younger than 55 years, having low socioeconomic status (no better than a high-school education, annual household income below $40,000, no health insurance [or Medicaid/medical assistance], and being unemployed), having lung and/or metastatic cancer, having comorbid conditions, and current chemotherapy. To control the effect of cancer site, we stratified the entire sample by lung cancer vs. other cancers and found similar results, suggesting that risk factors for high symptom burden are common in all cancer survivors. In survivors with high symptom burden, exploratory factor analysis of RSCL-M symptom items revealed 7 factors, of which fatigue and pain contained the items with the highest severity scores. For the high-symptom group, symptom severity was similar whether or not therapy had been completed. Fatigue, pain, and depression most negatively affected survivors’ ratings of HRQoL.
Previous studies of symptom burden in cancer survivors have compared the mean symptom burden of the entire cancer survivor population with the healthy population. In the current study we identified a subgroup of survivors with high symptom burden, which should enable us to identify survivors who may require additional symptom monitoring and management after termination of therapy. The transition from active cancer therapy to follow-up has been identified as a period of disrupted adjustment during survivorship.7 Residual symptoms from cancer and cancer treatment are often not monitored as closely during follow-up as during active cancer therapy, which may contribute significantly to this disrupted adjustment. Cancer survivors in our study were approximately 1 year postdiagnosis, and more than 80% had completed anticancer treatment. Similar levels of major symptoms (eg, fatigue, pain, and depression) were reported in survivors who completed their therapy and those who were still in active therapy, implying that symptom burden in the early period of survivorship may persist whether or not cancer treatment has been completed.
Several of the socioeconomic variables we identified as associated with membership in the high-symptom group are often used to define “medically underserved” persons, including 4 characteristics representing low socioeconomic status: low annual income, low education level, unemployment, and minimal health insurance. It is generally accepted that socioeconomic status may affect the adequacy of symptom control and therefore the level of symptom burden, with results from numerous studies indicating that symptom control may be comparatively poorer for medically underserved populations. For example, patients with low socioeconomic status have been found to be unable to access appropriate health care services for their disease22 or to communicate with clinicians who could aid in managing symptoms after treatment has ended.23 Socioeconomic disparity in cancer survivors has been associated with poor survival rates, more-rapid disease progression,24 poor HRQoL,25 and serious psychological distress.26 However, whether or not low income or unemployment caused high symptom burden could not be determined from the available data, because these variables were collected only once, 1 year after diagnosis. High symptom burden could contribute to change in a survivor’s income or employment status—or vice versa. Studies tracking longitudinal changes in survivor income and employment over time would help to characterize the association between symptom burden and socioeconomic status. Finally, our sample tended to be well-educated and well-insured. Better representation from medically underserved populations might have resulted in a higher percentage of survivors with greater symptom burden. Nonetheless, even in our well-educated and well-insured sample we found a subset of cancer survivors with higher symptom burden.
The finding of greater risk for higher symptom burden in younger survivors is consistent with a report from the 2002 National Health Interview Survey,12 which evaluated the contribution of age to the presence of multiple symptoms in 1904 cancer survivors, and with a 2005 report from Baker et al., who found that younger survivors were more likely than older survivors to reportpsychosocial problems as a result of their disease and treatment.27 Younger cancer survivors may have higher symptom burden because they are more likelyto receive aggressive cancer therapies.28 In addition, younger patients may have higher expectations for resuming full vocational and family obligations, which may exacerbate symptoms.29
Fatigue, pain, and depression were identified in the NIH State-of-the-Science Statement on symptom management in cancer30 as the most common symptoms of which clinicians and researchers need to be aware. Not surprisingly, our analyses revealed that fatigue, pain, and depression had the greatest negative impact on HRQoL. These symptoms are highly correlated with poor HRQoL in patients with cancer,31,32 affecting their daily activities and ability to work and socialize; pain has been shown to predict disease recurrence and survival.33,34 In addition, depressive symptoms usually occur with fatigue, pain, and sleep disturbance.35 The NIH report suggests that these symptoms are frequently undertreated in patients with cancer, even though guidelines for their management have been published by the National Comprehensive Cancer Network36 and other groups. Our findings suggest that these critical symptoms should be monitored well beyond the end of curative treatment into the early stage of survivorship, when survivors often attempt to return to normal roles.
The current study had several potential limitations. For example, the physical and psychological aspects of symptom burden were measured by different instruments with different response formats. In addition, our study attempted to providean overview of symptom burden among U.S. cancer survivors by collapsingheterogeneous cancer types into 1 broad category. Certain types of cancers, such as lung cancer, are associated withhigher symptom burden, and symptom patterns may vary by diagnosis or anticancer treatment. Future studies need to describe how such variables as disease site and treatment contribute to survivor symptom burden. As with most studies of symptoms in cancer survivors, the cross-sectional nature of the data limited our abilityto describe developmental trajectories of symptom burden. However, the SCS-I’s longitudinal design provides cancer survivors the opportunity to contribute information about the persistence of symptoms throughout their survival.
Because they are derived from a population-based nationwide survey, our findings have implications for the clinical care of cancer survivors. The fact that one fourth of the cancer survivors in the current study experienced high symptom burden highlights the need for routineposttreatmentsymptom assessment and management, especially in the first year after diagnosis. Symptom assessment and routine medical monitoring are likely to be infrequent after active therapy has ended, yet some severe symptoms reported by our sample could possibly have been addressed, leading to better functioning and less distress.7 In addition, the significantly higher burden experienced by uninsured and underinsured cancer survivors suggests another area of disparity in health care delivery that needs to be addressed at a policy level.
Supplementary Material
Acknowledgments
The American Cancer Society wishes to acknowledge the cooperation and efforts of the cancer registries and public health departments from the states of Alabama, Connecticut, Iowa, Idaho, Minnesota, New Jersey, Ohio, Pennsylvania, Rhode Island, South Carolina, and Wyoming. We also thank the staff of the hundreds of hospitals, including Stamford Hospital, who reported cases to the participating cancer registries. Last, we are grateful to the thousands of cancer survivors, their physicians, and their loved ones who contributed to the collection of these data.
The authors acknowledge Jeanie F. Woodruff, ELS, of the Department of Symptom Research at MD Anderson, for editorial assistance, and Dr. Xianglin Du of The University of Texas School of Public Health for review and comment on the manuscript.
Footnotes
Financial disclosures: None reported.
Previous Presentation
These results were presented as a poster at the Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, June 4–8, 2010.
Financial Disclosure and Funding Support
The American Cancer Society Studies of Cancer Survivors (SCS) were funded as an intramural program of research conducted by the American Cancer Society’s Behavioral Research Center. MD Anderson research support was provided by funds from the Hawn Foundation, Dallas, Texas (Shi), the National Cancer Institute (R01 CA026582; Shi and Cleeland), and MD Anderson’s Department of Symptom Research (Michonski). The funding organizations played no role in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication.
Contributor Information
Qiuling Shi, Email: qshi@mdanderson.org, Department of Symptom Research, The University of Texas MD Anderson Cancer Center, Houston, TX.
Tenbroeck G. Smith, Email: Tenbroeck.Smith@cancer.org, Behavioral Research Center, American Cancer Society, Atlanta, GA 30303.
Jared D. Michonski, Email: jdmichonski@gmail.com, Department of Symptom Research, The University of Texas MD Anderson Cancer Center, Houston, TX Department of Psychology, University of Houston, Houston, TX.
Kevin D. Stein, Email: Kevin.Stein@cancer.org, Behavioral Research Center, American Cancer Society, Atlanta, GA.
Chiew Kwei Kaw, Email: chiewkwei.kaw@cancer.org, Behavioral Research Center, American Cancer Society, Atlanta, GA.
Charles S. Cleeland, Email: ccleeland@mdanderson.org, Department of Symptom Research, The University of Texas MD Anderson Cancer Center, Houston, TX.
References
- 1.Office of Cancer Survivorship. [accessed February 1, 2011];About Cancer Survivorship Research: Survivorship Definitions. Available from URL: http://cancercontrol.cancer.gov/ocs/definitions.html.
- 2.American Cancer Society. [accessed December 2, 2010];Cancer facts & figures. 2009 Available from URL: http://www.cancer.org/acs/groups/content/@nho/documents/document/500809webpdf.pdf.
- 3.Kuhnt S, Ernst J, Singer S, et al. Fatigue in cancer survivors--prevalence and correlates. Onkologie. 2009;32(6):312–317. doi: 10.1159/000215943. [DOI] [PubMed] [Google Scholar]
- 4.Fu OS, Crew KD, Jacobson JS, et al. Ethnicity and persistent symptom burden in breast cancer survivors. J Cancer Surviv. 2009;3(4):241–250. doi: 10.1007/s11764-009-0100-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Auchincloss SS. After treatment. Psychosocial issues in gynecologic cancer survivorship. Cancer. 1995;76(10 Suppl):2117–2124. doi: 10.1002/1097-0142(19951115)76:10+<2117::aid-cncr2820761335>3.0.co;2-7. [DOI] [PubMed] [Google Scholar]
- 6.Hurt GJ, McQuellon RP, Barrett RJ. After treatment ends: neutral time. Cancer Pract. 1994;2(6):417–420. [PubMed] [Google Scholar]
- 7.Hewitt M, Ganz PA. From Cancer Patient to Cancer Survivor: Lost in Transition. Washington, DC: The National Academies Press; 2006. [Google Scholar]
- 8.Millar K, Purushotham AD, McLatchie E, George WD, Murray GD. A 1-year prospective study of individual variation in distress, and illness perceptions, after treatment for breast cancer. J Psychosom Res. 2005;58(4):335–342. doi: 10.1016/j.jpsychores.2004.10.005. [DOI] [PubMed] [Google Scholar]
- 9.Given B, Given CW, Sikorskii A, et al. Establishing mild, moderate, and severe scores for cancer-related symptoms: how consistent and clinically meaningful are interference-based severity cut-points? J Pain Symptom Manage. 2008;35(2):126–135. doi: 10.1016/j.jpainsymman.2007.03.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Serlin RC, Mendoza TR, Nakamura Y, Edwards KR, Cleeland CS. When is cancer pain mild, moderate or severe? Grading pain severity by its interference with function. Pain. 1995;61(2):277–284. doi: 10.1016/0304-3959(94)00178-H. [DOI] [PubMed] [Google Scholar]
- 11.Broeckel JA, Thors CL, Jacobsen PB, Small M, Cox CE. Sexual functioning in long-term breast cancer survivors treated with adjuvant chemotherapy. Breast Cancer Res Treat. 2002;75(3):241–248. doi: 10.1023/a:1019953027596. [DOI] [PubMed] [Google Scholar]
- 12.Mao JJ, Armstrong K, Bowman MA, Xie SX, Kadakia R, Farrar JT. Symptom burden among cancer survivors: impact of age and comorbidity. J Am Board Fam Med. 2007;20(5):434–443. doi: 10.3122/jabfm.2007.05.060225. [DOI] [PubMed] [Google Scholar]
- 13.Kim SH, Son BH, Hwang SY, et al. Fatigue and depression in disease-free breast cancer survivors: prevalence, correlates, and association with quality of life. J Pain Symptom Manage. 2008;35(6):644–655. doi: 10.1016/j.jpainsymman.2007.08.012. [DOI] [PubMed] [Google Scholar]
- 14.Cleeland CS, Bennett GJ, Dantzer R, et al. Are the symptoms of cancer and cancer treatment due to a shared biologic mechanism? A cytokine-immunologic model of cancer symptoms. Cancer. 2003;97(11):2919–2925. doi: 10.1002/cncr.11382. [DOI] [PubMed] [Google Scholar]
- 15.Smith T, Stein KD, Mehta CC, et al. The rationale, design, and implementation of the American Cancer Society’s studies of cancer survivors. Cancer. 2007;109(1):1–12. doi: 10.1002/cncr.22387. [DOI] [PubMed] [Google Scholar]
- 16.Stein KD, Denniston M, Baker F, et al. Validation of a modified Rotterdam Symptom Checklist for use with cancer patients in the United States. J Pain Symptom Manage. 2003;26(5):975–989. doi: 10.1016/s0885-3924(03)00321-x. [DOI] [PubMed] [Google Scholar]
- 17.Baker F, Denniston M, Zabora J, Polland A, Dudley WN. A POMS short form for cancer patients: psychometric and structural evaluation. Psychooncology. 2002;11(4):273–281. doi: 10.1002/pon.564. [DOI] [PubMed] [Google Scholar]
- 18.Baker F, Denniston M, Hann D, et al. Factor structure and concurrent validity of the Satisfaction with Life Domains Scale for Cancer (SLDS-C) J Psychosoc Oncol. 2007;25(2):1–17. doi: 10.1300/j077v25n02_01. [DOI] [PubMed] [Google Scholar]
- 19.Cohen J. The statistical power of abnormal-social psychological research: a review. J Abnorm Soc Psychol. 1962;65:145–153. doi: 10.1037/h0045186. [DOI] [PubMed] [Google Scholar]
- 20.Floyd FJ, Widaman KF. Factor analysis in the development and refinement of clinical assessment instruments. Psychological Assessment. 1995;7(3):286–299. [Google Scholar]
- 21.Freund RJ, Littell RC. SAS System for Regression. 3. SAS; 2000. [Google Scholar]
- 22.Darby K, Davis C, Likes W, Bell J. Exploring the financial impact of breast cancer for African American medically underserved women: a qualitative study. J Health Care Poor Underserved. 2009;20(3):721–728. doi: 10.1353/hpu.0.0176. [DOI] [PubMed] [Google Scholar]
- 23.Ok H, Marks R, Allegrante JP. Perceptions of health care provider communication activity among American cancer survivors and Adults Without Cancer Histories: an analysis of the 2003 Health Information Trends Survey (HINTS) Data. J Health Commun. 2008;13(7):637–653. doi: 10.1080/10810730802412172. [DOI] [PubMed] [Google Scholar]
- 24.Winer E, Gralow J, Diller L, et al. Clinical cancer advances 2008: major research advances in cancer treatment, prevention, and screening--a report from the American Society of Clinical Oncology. J Clin Oncol. 2009;27(5):812–826. doi: 10.1200/JCO.2008.21.2134. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ozturk A, Sarihan S, Ercan I, Karadag M. Evaluating quality of life and pulmonary function of long-term survivors of non-small cell lung cancer treated with radical or postoperative radiotherapy. Am J Clin Oncol. 2009;32(1):65–72. doi: 10.1097/COC.0b013e31817e6ec2. [DOI] [PubMed] [Google Scholar]
- 26.Hoffman KE, McCarthy EP, Recklitis CJ, Ng AK. Psychological distress in long-term survivors of adult-onset cancer: results from a national survey. Arch Intern Med. 2009;169(14):1274–1281. doi: 10.1001/archinternmed.2009.179. [DOI] [PubMed] [Google Scholar]
- 27.Baker F, Denniston M, Smith T, West MM. Adult cancer survivors: how are they faring? Cancer. 2005;104(11 Suppl):2565–2576. doi: 10.1002/cncr.21488. [DOI] [PubMed] [Google Scholar]
- 28.Yancik R, Wesley MN, Ries LA, Havlik RJ, Edwards BK, Yates JW. Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. JAMA. 2001;285(7):885–892. doi: 10.1001/jama.285.7.885. [DOI] [PubMed] [Google Scholar]
- 29.Hensel M, Egerer G, Schneeweiss A, Goldschmidt H, Ho AD. Quality of life and rehabilitation in social and professional life after autologous stem cell transplantation. Ann Oncol. 2002;13(2):209–217. doi: 10.1093/annonc/mdf031. [DOI] [PubMed] [Google Scholar]
- 30.NIH State-of-the-Science Statement on symptom management in cancer: pain, depression, and fatigue. NIH Consens State Sci Statements. 2002;19(4):1–29. [PubMed] [Google Scholar]
- 31.Breivik H, Cherny N, Collett B, et al. Cancer-related pain: a pan-European survey of prevalence, treatment, and patient attitudes. Ann Oncol. 2009;20(8):1420–1433. doi: 10.1093/annonc/mdp001. [DOI] [PubMed] [Google Scholar]
- 32.Cleeland CS, Nakamura Y, Mendoza TR, Edwards KR, Douglas J, Serlin RC. Dimensions of the impact of cancer pain in a four country sample: new information from multidimensional scaling. Pain. 1996;67(2–3):267–273. doi: 10.1016/0304-3959(96)03131-4. [DOI] [PubMed] [Google Scholar]
- 33.Halabi S, Vogelzang NJ, Kornblith AB, et al. Pain predicts overall survival in men with metastatic castration-refractory prostate cancer. J Clin Oncol. 2008;26(15):2544–2549. doi: 10.1200/JCO.2007.15.0367. [DOI] [PubMed] [Google Scholar]
- 34.Scharpf J, Karnell LH, Christensen AJ, Funk GF. The role of pain in head and neck cancer recurrence and survivorship. Arch Otolaryngol Head Neck Surg. 2009;135(8):789–794. doi: 10.1001/archoto.2009.107. [DOI] [PubMed] [Google Scholar]
- 35.Miaskowski C, Cooper BA, Paul SM, et al. Subgroups of patients with cancer with different symptom experiences and quality-of-life outcomes: a cluster analysis. Oncol Nurs Forum. 2006;33(5):E79–E89. doi: 10.1188/06.ONF.E79-E89. [DOI] [PubMed] [Google Scholar]
- 36.National Comprehensive Cancer Network. [accessed December 2, 2010];NCCN Guidelines for Supportive Care. Available from URL: http://www.nccn.org/professionals/physician_gls/f_guidelines.asp#supportive.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.