Fax: (617) 432-3416
Surviving colorectal cancer †
Patient-Reported Symptoms 4 Years After Diagnosis
Article first published online: 11 SEP 2007
Copyright © 2007 American Cancer Society
Volume 110, Issue 9, pages 2075–2082, 1 November 2007
How to Cite
Schneider, E. C., Malin, J. L., Kahn, K. L., Ko, C. Y., Adams, J. and Epstein, A. M. (2007), Surviving colorectal cancer . Cancer, 110: 2075–2082. doi: 10.1002/cncr.23021
The authors have received honoraria for presenting research results at conferences sponsored by the American Society of Clinical Oncology and the American College of Surgeons (ACoS). The assertions and opinions expressed herein are solely those of the authors and are not necessarily those of ASCO or its representatives.
- Issue published online: 18 OCT 2007
- Article first published online: 11 SEP 2007
- Manuscript Accepted: 1 JUN 2007
- Manuscript Revised: 30 MAY 2007
- Manuscript Received: 6 APR 2007
- American Society of Clinical Oncology (ASCO) and the Damon Runyon Cancer Research Foundation (to J.L.M.)
- colorectal cancer;
- cancer survivorship;
- quality of life;
- side effects of treatment
The number of long-term survivors after a cancer diagnosis is increasing, yet relatively few studies have addressed survivors' reported symptoms beyond the initial year after diagnosis.
The symptom reports of 474 survivors of colon and rectal cancer from 5 US metropolitan areas were collected during 2002-2003 as part a larger study of the quality of care for patients with cancer diagnosed in 1998. The relation between the prevalence of reported symptoms and prior treatments received was analyzed, adjusting statistically for other patient characteristics.
Sixty-nine percent of the survivors had colon cancer and 31% had rectal cancer. The most commonly reported symptoms were “fatigue” (23%), “negative feelings about body appearance” (14%), “diarrhea” (13%), and “constipation” (7%). Higher percentages of respondents attributed health effects to cancer or its treatment including “worry about health” (24%), “physical discomfort” (19%), and “activity limitations” (15%). In general, prior treatment was not associated with symptom prevalence. However, radiation therapy recipients and patients that received a diverting ostomy were more likely than others to report some of the symptoms we studied. Attribution of health effects to disease or treatment did not vary by prior treatment except that recipients of radiation therapy were more likely than others to report limitations in their activities (30% vs 10%; P = .003).
Among colorectal cancer survivors the prevalence of symptoms at 4 years was low and relatively comparable to published estimates for the general population, but some survivors continue to attribute health effects to cancer or its treatment. Cancer 2007. © 2007 American Cancer Society.
The number of Americans living with a prior diagnosis of cancer has increased from 3 million to over 10 million over the past 30 years.1 The Institute of Medicine recently recommended the establishment of cancer survivorship as a distinct phase of cancer care.2 The few studies that have examined the long-term patient-reported outcomes of cancer survivors have focused on women with prior breast cancer.3, 4 Such studies are important for several reasons. Knowledge of patient-reported outcomes may help patients select treatments. Long-term cancer survivors may have unique health problems and care needs.5 Clinicians may be able to anticipate and inquire about symptoms that patients may not otherwise volunteer. Effective symptom management may improve the quality of life of cancer survivors.
Colorectal cancer is the second most common cancer diagnosis affecting both men and women. Increasing numbers of patients with nonmetastatic colon or rectal cancer survive 5 years or more without a cancer recurrence.6 For long-term survivors of colorectal cancer the quality of life may be affected by both treatments received at the time of diagnosis (such as surgical interventions, chemotherapy, and radiation therapy) and after-effects of disease whether or not there is a cancer recurrence.5
This article describes the patient-reported symptoms of a cohort of patients who survived at least 4 years after a diagnosis of colorectal cancer. Specifically, we compared symptoms and health effects reported by patients that received and did not receive each of 3 commonly used treatments: chemotherapy, radiation therapy, and diverting colostomy (or ileostomy). In addition, we examined symptoms reported by patients that underwent reversal of diverting ostomy compared with those that retained the ostomy 4 years after diagnosis.7
MATERIALS AND METHODS
The data for this study were collected as part of the National Initiative on Cancer Care Quality (NICCQ) sponsored by the American Society for Clinical Oncology.8, 9 Eligibility and registry recruitment are described elsewhere.9 Institutional Review Boards at RAND, Harvard, and participating hospitals approved the study protocol.
Patient Sampling, Recruitment, and Survey Protocol
We used National Cancer Data Base (NCDB) data supplied by the American College of Surgeons to identify 1803 patients aged 18 to 80 years old with an initial diagnosis of colon or rectal cancer reported from 1 of 5 metropolitan areas (Atlanta, Cleveland, Houston, Kansas City, and Los Angeles) during 1998. After excluding 193 patients for whom the physician refused contact with the patient, the patient was out of country, or no valid contact address was available, we were able to confirm that 1077 patients were alive and eligible for survey (533 patients were deceased). Of the living patients, 676 (63%) completed the survey, which was administered approximately 4 years after diagnosis by mail with telephone follow-up (and administration) to nonrespondents. Among survey respondents, 478 (71%) consented to have medical records abstracted. Four patients were excluded because of inadequate data on stage of disease. Thus, the analytic cohort consisted of 474 patients with adequate information for analysis (44% of eligible patients). Three of the 474 patients reported a recurrence of colorectal cancer at the time of the survey and were retained in the analysis. To assess the potential effect of nonresponse, we defined a “nonincluded cohort” of otherwise eligible survivors that did not respond to the survey, did not consent to medical record review, or had incomplete data from medical records.
Each participating patient completed an 18-page survey, provided consent for medical record abstraction, and submitted a list of all providers seen since diagnosis. Photocopies of consenting patients' ambulatory medical records from all cancer providers including the hospital that reported the diagnosis to the registry were abstracted by trained nurses using a computer-based medical record abstraction instrument. The type of cancer (colon vs rectal) and stage of disease (II or III) were obtained from the NCDB and revised, if necessary, based on medical record data. Medical record abstraction was used to identify the presence of comorbid conditions at diagnosis (specifically, cardiovascular disease, peripheral vascular disease, hypertension, cerebrovascular disease, pulmonary disease, gastrointestinal disease, diabetes, diabetes with end-organ damage, connective tissue disease, other malignancy, human immunodeficiency virus, and mental health conditions) and treatments received including chemotherapy, radiation therapy, and diverting ostomy.
Survey data included patient's reports of the frequency of symptoms based on questions adapted from previous studies.10–13 The survey asked how often during the past 4 weeks the patient experienced each symptom: 1) stomachache, 2) pain, cramps or abdominal discomfort, 3) diarrhea, 4) constipation, 5) blood in stool, 6) pain in rectum, and 7) negative feelings about body appearance. Response categories included “very often,” “fairly often,” “a few times,” “once or twice,” or “not at all.” The survey included questions about attribution of health effects such as physical discomfort, worry, or activity limitations to cancer or its treatment. Response categories were: “a lot,” “some,” “a little,” or “none at all.” Patient demographic characteristics included age, sex, race, Hispanic ethnicity, educational attainment, insurance status at diagnosis, whether born in the US or not, and whether English was the primary language spoken at home. Survey data determined whether or not the patient had reversal of their ostomy during the 4 years subsequent to diagnosis. The survey questions underwent cognitive testing with 8 colorectal cancer patients recruited from local oncology practices. Mail and telephone instruments were pilot-tested among patients selected from the 5 metropolitan areas.
We used the NCDB registry data to compare the sociodemographic characteristics, disease stage, and treatments received by the analytic cohort versus the “nonincluded cohort.” We examined the sociodemographic characteristics of the study patients overall and by disease type (colon and rectal) and stage (stage II and III). We calculated the percentage of patients that received chemotherapy, radiation therapy, and diverting ostomy (ileostomy or colostomy), and among the latter group, the percentage of the patients that underwent reversal (or “take down”) of the ostomy. We stratified the cohort according to the 4 major treatments received (chemotherapy, radiation, diverting ostomy, and take-down of diverting ostomy) and summarized the demographic characteristics of patients who received and did not receive each treatment. We used a chi-square test to assess the statistical significance of comparisons.
We compared the prevalence of the reported symptoms and health effects among those who received and did not receive each of the 4 treatments, restricting the comparison of patients that had reversal of ostomy to the cohort that had received a diverting ostomy initially. To assess symptom frequency we calculated the percentage of patients that reported experiencing each symptom “very often” or “fairly often” during the past 4 weeks. To assess fatigue we used a 2-item version of the vitality subscale (Cronbach alpha 0.835) derived from the SF-36 and, as other studies have done, we counted those with a score of <0.5 on this scale as reporting fatigue.3 To compare our results with those of other studies, we also calculated the mean score on the vitality subscale. We summarized the percentage of patients reporting none, 1, and 2 or more symptoms. For the health effects, we calculated the percentage of respondents specifying either “a lot” or “some” to the question on health-related treatment effects. The reported prevalence of each symptom and health effect was adjusted for the sociodemographic variables listed above, disease type, and disease stage using a logistic regression model with the symptom or health effect as the dependent variable. Entering the treatment variable as an independent predictor, we used the t-test associated with the beta coefficient to assess the statistical significance of comparisons of the adjusted symptom prevalence. All analyses were performed using SUDAAN v. 9.0 (RTI International, Research Triangle Park, NC) and used sample weights that accounted for the selection of hospital registries and patients, and nonresponse at each stage of the data collection protocol.
Patients in the analytic cohort were similar to nonincluded patients with respect to the percentage that were <65 years of age (45% vs 42%; P = .44), female (49% vs 51%; P = .56), had private insurance (55% vs 47%; P = .06), received chemotherapy (56% vs 50%; P = .06), and received radiation (21% vs 23%; P = .43). Analytic cohort patients were more likely than nonincluded patients to be white (89% vs 77%; P ≤ .01). Among the analytic cohort 69% had colon cancer (37% with stage II and 32% with stage III colon cancer), and 31% had rectal cancer (13% with stage II and 18% with stage III rectal cancer). Table 1 reveals no significant sociodemographic differences by disease type and stage with the exception of race (on average, 1% of rectal cancer patients were African-American compared with an average of 9% of the colon cancer patients).
|Colon cancer||Rectal cancer|
|All N=474||Stage II n=179 (36.6%)||Stage III n=151 (32.0%)||Stage II n=66 (13.3%)||Stage III n=78 (18.2%)|
|High school graduate||31||35||26||32||31|
|Uninsured at diagnosis†||4||4||7||0||2||.1819|
|Born outside US||11||9||12||11||11||.8836|
|Language spoken at home is not English||4||4||5||3||4||.9178|
|Comorbid conditions at diagnosis||.0862|
|2 or more||28||35||27||29||14|
|Clinical treatments received‡|
|Ostomy reversed among patients who received ostomy||51||80||100||40||27||<.001|
As expected, treatments varied based on disease type and stage. Patients with rectal cancer were more likely than those with colon cancer to receive each treatment and patients with stage III cancer were more likely than those with stage II to receive chemotherapy. At the time of survey only 3 patients reported a recurrence of colorectal cancer; 2 had stage III colon cancer and 1 had rectal cancer. Two of the patients with recurrence had prior chemotherapy, 1 had prior radiation, and none had received an ostomy.
Table 2 compares characteristics of patients who received each treatment with those that did not. On average, chemotherapy recipients were younger (P = .003) than nonrecipients and were more likely to be uninsured at diagnosis (P < .001). Of note, there were no uninsured patients among the chemotherapy nonrecipients. Radiation therapy recipients included a smaller percentage of African-Americans than nonrecipients (consistent with the small proportion of African-Americans in the rectal cancer cohort).
|Chemotherapy||Radiation therapy||Ostomy†||Ostomy reversed†|
|Characteristic||N=364 (76%)||N=108 (24%)||N=116 (23%)||N=358 (77%)||N=78 (15%)||N=396 (85%)||N=48 (51%)||N=30 (49%)|
|High school graduate||31||30||33||31||29||31||27||32|
|Uninsured at diagnosis‡||5||0||2||5||3||4||4||0|
|Born outside US||12||9||7||12||12||11||6||20|
|Language spoken at home is not English||4||6||5||4||7||4||1||13|
|Comorbid conditions at diagnosis§|
|2 or more||25||36||20||30||23||29||15||30|
Table 3 summarizes the symptom results overall. Four years after diagnosis, relatively small percentages of survivors reported frequent symptoms in the prior 4 weeks. The most commonly reported symptoms were fatigue (23%), “negative feelings about body appearance” (14%), “diarrhea” (13%), and “constipation” (7%). Although fatigue was the most prevalent symptom, the mean vitality score (from which fatigue prevalence was derived) was 59.9 (close to the US population norm of 60.9).14 Many respondents attributed health effects to cancer or its treatment including “worry about health” (24%), “physical discomfort” (19%), and “activity limitations” (15%).
|ALL N=474||Yes n=364 (76%)||No n=108 (24%)||Yes n=116 (23%)||No n=358 (77%)|
|Adjusted %||P†||Adjusted %||P†|
|Symptom frequency (% reporting “fairly often” or “very often” past 4 wks)|
|Negative feelings about body appearance||14||13||16||.49||19||13||.26|
|Pain, cramps, or abdominal discomfort||7||7||7||.96||9||6||.27|
|Pain in rectum||4||4||2||.49||5||3||.40|
|Blood in stool||1||1||0||NA||1||1||.31|
|No. of symptoms reported (“fairly often” or ”very often” past 4 weeks)||.75||.54|
|3 or more symptoms||8||8||9||—||10||8||—|
|Health effects attributed to cancer or treatment (% reporting “a lot” or “some”)|
|Worry about health||24||21||26||.52||29||21||.31|
Tables 3 and 4 show that, with few exceptions, treatment was not associated with symptom prevalence after controlling for disease and patient characteristics. Nor was there an association between the number of symptoms reported by patients who received treatments and those who did not. Table 3 shows that chemotherapy recipients were less likely to report “stomachache” (3% vs 11%; P = .03) and radiation therapy recipients were more likely than nonrecipients to report diarrhea (24% vs 10%; P = .04). Table 4 shows that patients who received a diverting ostomy were more likely than others to report “negative feelings about body appearance” (25% vs 12%; P = .02), “pain, cramps, or abdominal discomfort” (17% vs 5%; P = .003), and “stomachache” (12% vs 4%; P = .02). Among ostomy recipients, those who had the ostomy reversed were less likely to report constipation (3% vs 21%; P = .02).
|Yes n=78 (15%)||No n=396 (85%)||Yes n=48 (51%)||No n=30 (49%)|
|Adjusted %||P‡||Adjusted %||P‡|
|Symptom frequency (% reporting “fairly often” or “very often” in past 4 weeks)|
|Negative feelings about body appearance||25||12||.02||13||30||.13|
|Pain, cramps, or abdominal discomfort||17||5||.003||15||8||.35|
|Pain in rectum||6||3||.18||5||8||.54|
|Blood in stool||1||1||.51||2||0||NA|
|Number of symptoms reported (“fairly often” or “very often” past 4 weeks)||.17||.16|
|3 or more symptoms||12||8||21||9|
|Health effects attributed to cancer or treatment (% reporting “a lot” or “some”)|
|Worry about health||24||22||.83||18||29||.39|
Attribution of health effects to cancer or its treatment did not differ among treatments except that recipients of radiation therapy were more likely to report limitations in their activities (30% vs 10%; P = .003) and more likely to report physical discomfort (30% vs 15%; P = .06). The latter comparison did not meet the conventionally defined level of statistical significance.
Four years after diagnosis and treatment for colorectal cancer, relatively low percentages of survivors reported each of the symptoms we inquired about. Nearly two-thirds of survivors reported none of these symptoms during the 4 weeks before the survey. Fewer than 1 in 10 patients reported more than 2 symptoms. To place our results in context, the prevalence of each symptom reported by the colorectal survivors was relatively similar to published estimates for other survivors and for the general population. For example, the prevalence of fatigue, the most commonly reported symptom, was lower among our cohort than the 34% prevalence of fatigue reported by long-term survivors of breast cancer.3 The level of fatigue reported by colorectal cancer survivors (corresponding to an adjusted mean vitality subscale score of 59.9 where a lower score indicates more fatigue) was not substantially different from the average vitality scores of the general adult population (60.9).14 On average, the colorectal cancer survivors reported less fatigue than patients with other noncancer “acute and chronic” conditions like past myocardial infarction.14
As near as we could discern, the prevalence of symptoms other than fatigue were also fairly similar to published estimates for the US adult population.16–18 For example, approximately 7% of colorectal survivors reported abdominal pain or discomfort “fairly often” or “very often” in the prior 4 weeks, whereas a prior survey of the general population found that approximately 8% of adults reported 4 or more days of abdominal pain or discomfort in the previous month.17 Similarly, whereas 13% of the colorectal cancer survivors experienced diarrhea fairly often or very often in the prior 4 weeks, approximately 8% of adults in a US population survey reported at least 1 episode of diarrhea in the past 4 weeks and 10% of the adults in a separate US population survey reported 4 or more days of diarrhea in the prior month.17, 18 Seven percent of the colorectal survivors reported constipation very often or fairly often in the previous 4 weeks. The prevalence of chronic constipation is estimated to range between 3% and 17% in the general population.16 Fewer than 1% of colorectal survivors reported rectal bleeding very often or fairly often in the prior 4 weeks. In a survey of Olmstead County adults, approximately 14% reported rectal bleeding during the prior year.19
Both prior radiation therapy and receipt of an ostomy were associated with diarrhea, more frequent thoughts about dying, and physical discomfort and activity limitations attributed to cancer or its treatment. Diarrhea is a well-known complication of both treatments. The higher prevalence of thoughts about dying among patients that received radiation or an ostomy might be related to the higher disease risk associated with the initial diagnosis and treatments (although we controlled for these factors statistically). Alternatively, chronic symptoms like diarrhea or the experience of the ostomy may serve as a constant “reminder” of mortality.
One prior study has suggested that ostomy and nonostomy patients have similar quality of life, but experience different types of symptoms.7 Our results suggest a tradeoff between symptoms such as constipation and diarrhea. In our cohort, there was not a consistent association between ostomy reversal and the prevalence of symptoms we studied. On the contrary, compared with the group that retained an ostomy, those that had undergone reversal were more likely to report symptoms.
Our results are consistent with those of prior studies that examined the quality of life of colorectal cancer survivors over shorter time frames. These studies suggest that the emotional, cognitive, and social functioning domains of the quality of life may pose the greatest difficulty for colorectal cancer survivors.20–24 In our study, as many as 1 in 4 respondents reported health effects that they attributed to cancer or treatment such as worry, physical discomfort, and activity limitations. Our study shows that these self-reported health effects are quite persistent over time. Furthermore, the percentage of patients attributing health effects to cancer or treatment was higher than the prevalence of specific symptoms. These reported health effects may be an important target for future interventions designed to improve the quality of life for cancer survivors.
Our study is among the first to gather information about the quality of life of a population-based cohort of cancer survivors 4 years after diagnosis. It includes a large sample of colorectal cancer survivors from 5 diverse geographic areas of the US. We had detailed information about treatments received based on review of medical records documented at the time of diagnosis and treatment. Nevertheless, the study also has limitations. The timing of the follow-up survey at 4 years was determined by data availability and logistical constraints. Whereas a 5-year follow-up would be more typical in oncology research, we would expect very little difference in the results had we instead conducted a 5-year follow-up survey. An additional delay raises the potential for bias related to nonresponse of survivors. The direction and magnitude of this bias is uncertain. If nonrespondents are more prone to symptoms than respondents, then our estimates are too low (and vice versa). We selected diverse metropolitan areas, but the proportion of minority patients at each stage of the sampling was lower than anticipated. This may limit the generalizability of our results to minority populations. Although we were able to adjust the symptom prevalence for the sociodemographic characteristics of the patients, disease type, and disease stage, our observational study design does not enable us to conclude that differences in the prevalence of reported symptoms are caused by the treatments received. Selection for treatment may be influenced by other unmeasured factors that are independently related to symptoms. We also cannot say with certainty whether the psychological impact of reported symptoms is the same for cancer survivors and other groups of patients. Relatively minor symptoms may cause greater distress among cancer survivors than others, and may appropriately prompt greater alarm among clinicians concerned about potential recurrence.
We believe the results can be useful to clinicians and patients in 2 ways. First, the results offer general prognostic guidance about the long-term quality of life that cancer survivors may experience after treatment. The majority of colorectal cancer survivors can reasonably expect a probability of physical symptoms that does not differ substantially from the general population, with the exception of a slightly greater potential to experience fatigue and, among patients who received radiation, a greater potential to experience diarrhea. Second, our results suggest a need for additional studies of the psychological consequences of symptoms in the context of a past cancer diagnosis and the interventions that might ameliorate these consequences. Identifying the cohort of patients that continue to experience psychological symptoms during survivorship may be an important initial step.
In summary, we found that 4 years after diagnosis the quality of life of survivors of stage II and III colorectal cancer appears generally comparable to that of the general population and did not vary substantially based on prior treatments received. As oncology care becomes more effective and the numbers of long-term survivors of cancer treatment increase, studies of long-term survivors of other types of cancer and detailed studies of the relation between treatments and long-term quality of life will become increasingly important.
- 1Cancer survivorship—United States, 1971–2001. MMWR Morb Mortal Wkly Rep. 2004; 53: 526–529.
- 2Institute of Medicine. From Cancer Patient to Cancer Survivor — Lost in Transition. Washington, DC: National Academies Press; 2005.
- 16Harrison's Principles of Internal Medicine. New York: McGraw-Hill; 2005., , , et al.
- 19Self-reported rectal bleeding in a United States community: prevalence, risk factors, and health care seeking. Am J Gastroenterol. 1998; 93: 2179–2183., .Direct Link:
- 24Quality of life in long term survivors of colorectal cancer. Am J Gastroenterol. 2002; 97: 1228–1234., , , , .Direct Link: