It is well established that cancer and its treatment bring with it pain and fatigue.1 This is certainly the case during the period immediately after diagnosis and throughout treatment. However, the degree to which these symptoms persist over time and their effects on functioning are less well known. Because research on cancer has only recently turned attention to long-term survivorship2 (defined as 5 or more years after diagnosis/treatment), there is little empirical evidence on the persistence of cancer-related pain and fatigue among long-term survivors. Moreover, with more individuals surviving for longer periods of time, there has been an increase in the proportion of survivors who are older adults. With increasing age, survivors are likely to experience a range of comorbidities that are capable of causing pain and fatigue. Thus, the question arises as to the relative importance of cancer-related pain and fatigue compared with pain and fatigue that result from other health problems common among older adults.

The purpose of this research is to examine the degree to which pain and fatigue are reported by a sample of older-adult (>=60), long-term (5 years) survivors of breast, prostate, and colorectal cancer. These 3 types of cancer are the most common survivable types of cancer affecting older adults. The analyses also examine the relative importance of cancer-related and age-related factors as predictors of pain and fatigue. Finally, this research documents the role that pain and fatigue play in the functional difficulties reported by long-term survivors.

Long-term Survivorship

Considerable research has established that many cancer survivors experience diminished quality of life including physiological late effects, psychological distress, and social life disruption.3 However, the long-term effects of cancer later in the life course have received much less attention due to the relative recency of the improvement in cancer survival rates.4 Thus, research on survivors has typically concentrated on its effects during the period of diagnostic testing,5 during treatment,6 or the first year after treatment.7 Research on pain and fatigue similarly has not focus on long-term survivorship.

Research on long-term survivorship has identified a number of continuing effects of cancer that could be related to pain and/or fatigue. For example, Schultz and colleagues found that the most reported health effects were arthritis and osteoporosis. These are 2 conditions that have special significance for the pain and functional decline of older adults. They also found that these 2 health conditions increased with the passage of time.8 Other research has found that decreases in general functioning may persist for up to 20 years after cancer treatment.9

Cancer and Aging

The significance of cancer for older adults is derived from its increased incidence with advancing age. More than half of all cancers diagnosed are among individuals older than 65 years.10 Furthermore, the incidence rate for all cancer sites triples in the age group of 60 to 79 years, when compared with individuals aged 40 to 59 years.10 Moreover, the occurrence of the most common survivable cancers, those that are the focus of this research (breast, colorectal, and prostate), increases dramatically after age 60 years.10

Older adults are a particularly vulnerable group of survivors because other health problems are typically associated with aging. Research has shown that among older cancer survivors, comorbidities are a significant issue11 and are present in nearly 70% of cancer patients. Nearly one third of all survivors report having 2 or more comorbid conditions.12 Given and colleagues13 clearly document the important role that comorbidities play in producing pain and fatigue among those diagnosed and treated for cancer. Furthermore, studies that compare cancer survivors with adults who have never had cancer document substantially higher levels of comorbidity among survivors. It is these other health conditions that have the potential to increase functional limitations.14 Havlick et al11 note that although the presence of comorbidities has been well documented among older adult survivors, their effects have not been well examined.

Cancer-Related Pain and Fatigue

Cancer-related symptoms may persist for years after treatment ends. Schroevers and colleagues15 found that 8-year survivors report significantly higher levels of physical symptoms over time when compared with a healthy reference group. That study did not examine pain and fatigue as specific symptoms, however. Dow and colleagues16 were among the first to identify both pain and fatigue as important symptoms reported by longer-term survivors.

Bernebei and colleagues17^(p1877) noted, "Pain is one of cancer's most frequent and disturbing symptoms." They suggest that it may be both underreported and undertreated, especially among older and minority patients.17 Moreover, Hwang and colleagues18 found that pain is one of the strongest predictors of fatigue among those previously treated for cancer. As with many survivorship studies, the research on pain has used samples consisted of short-term survivors of breast cancer. For example, Katz and colleagues19 documented the correlates of acute pain and persisting acute pain among patients with breast cancer during the first month after surgery. These correlates included being younger and unmarried and having had more invasive surgeries. Ganz20 noted that past systemic adjuvant therapies (ie, chemotherapy or hormone therapy) are predictors of pain in long-term survivors. In one of the few studies of other cancers, Ramsey and colleagues21 found that for colorectal cancer survivors who were on average 3 years posttreatment, pain was a substantial problem, one that did not dissipate over time. Bosompra and colleagues22 found that pain is an important part of the "disablement process" among more recent survivors, and although intensity and frequency levels were low, they predicted functional status.

In summary, research conducted to date has shown that pain persists beyond the treatment period and is linked to poorer general functioning. However, because most research has focused on short-term survivors in general and breast cancer survivors specifically, there is a great deal that we do not know about the role that pain plays in other survivorship contexts.

Research also provides clear evidence that cancer produces fatigue during the period surrounding treatment.18 However, research has also identified fatigue as an important effect of cancer and its treatment that may persist over an extended period of time.20 In fact, Jacobsen and Stein23^(p256) noted that fatigue is, "the most common and distressing symptom experienced by breast cancer survivors." Similarly, Servaes and colleagues24 found that 38% of breast cancer survivors who were on average over 2 years posttreatment reported severe fatigue compared with 11% in a matched comparison sample. In a study of 1- to 6-year breast cancer survivors, Berglund and colleagues25 found that fatigue was present in more than three fourths of patients, and its severity did not decrease over time. Mast,26 however, found that low to moderate levels of fatigue that persisted were, in part, related to comorbidities. However, even with the effects of other illnesses controlled, fatigue was found to be related to having had chemotherapy.26

Furthermore, Jacobsen and Stein's23 review of the literature in this area found that chemotherapy and bone marrow transplantation produced fatigue years after treatment, but surgery and radiation did not. Fossa and colleagues27 compared long-term testicular cancer survivors with a sample of Hodgkin disease survivors and the general public. Chronic fatigue was present among 10% of the general public, 16% of testicular cancer survivors, and 25% of Hodgkin disease survivors. Moreover, they found that differences between these groups increased with age.

Other research found contrasting results in the persistence of fatigue over time. Hann and colleagues28 in a study of breast cancer survivors found that fatigue does not persist after radiation treatment. Similarly, Bower and colleagues29 found minimal differences in fatigue in short-term breast cancer survivors compared with the general population. However, fatigue was related to pain, depression, and sleep disturbances and affects quality of life when it does persist.29 Bower30^(p8281) noted that "given the adverse effects of fatigue on quality of life, this type of research is critical for maintaining the physical and emotional well-being in the growing population of cancer survivors."

In summary, research on fatigue presents mixed findings on the prevalence of this symptom. However, when it is found to persist, it seems to have important implications on the quality of life of survivors. Again, given the limitations of the research conducted to date, much is not known about the effects of fatigue on long-term survivors.

Linking Cancer/Treatment-Related Symptoms With Pain, Fatigue and Functioning

Prior research has found that poorer physical functioning in long-term survivors can be linked to specific treatments and their attendant effects,20 and continuing cancer-related symptoms have also been found to be related to overall physical functioning.31,32 Looking at the symptoms associated with specific types of cancer, for breast cancer survivors, lymphedema and shoulder morbidity are typical continuing symptoms.33 Moreover, these may progress in severity over time. These symptoms have the potential to generate considerable pain, which may in turn affect overall functioning.34 Among colon cancer survivors, pain has been shown to persist along with functional difficulty.21 For prostate cancer survivors, the late effects of radiation may generate chronic pain along with other symptoms such as difficulty with or frequency of urination, diarrhea,35 and rectal complications.36 These symptoms can continue well into the second decade after treatment37 and represent a lifelong risk for functional decline.38

Research Questions

The above research provides substantial documentation for the persistence of cancer sequelae that have the potential to cause pain and fatigue as well as affect functioning in later life. Therefore, the aim of this research is to document the relative importance of cancer-related factors, compared with other age-related factors as predictors of pain, fatigue (as measured by weakness and energy level), and functional difficulty. This research will address the following research questions:

Research question 1: To what degree are pain and fatigue reported by long-term survivors?

Research question 2: What are the age-related and cancer-related correlates of pain and fatigue among these survivors?

Research question 3: What are the best predictors of pain and fatigue among older-adult, long-term cancer survivors and how do pain and fatigue contribute to basic functional difficulty?

The Figure displays an analytic model that allows us to address these issues in a systematic way. It is based on the prior work of Nagi39 and expanded on by Verbrugge and Jette40 and previously applied to research on pain and functioning by Bosompra and colleagues.22 In our research, this model focuses on the role that cancer-related factors play when compared with age-related factors in determining the pain and fatigue reported by long-term survivors.

Figure Conceptual model.

Methods

Data Source

The cross-sectional data analyzed in this research are derived from the first wave of interviews from an ongoing, longitudinal study funded by the National Cancer Institute. Six waves of data will be collected during a 10-year period (1998-2008). The sample was selected from the cancer tumor registry of the Ireland Cancer Center at University Hospitals Health System in Cleveland, Ohio, using a stratified random sample design. Oversampling African Americans resulted in nearly 40% of the final sample being from this minority group.

The sample was further stratified by sex and cancer type. The 3 most common survivable types of cancer among older adults, breast, colorectal, and prostate were selected for inclusion. Lung cancer was excluded because of the generally poor survival rates that limited the number of long-term survivors in the registry. The results reported here are derived from the analysis of data obtained from the initial in-person interviews with 295 of the 321 older adult (older than 60 years), long-term survivors (5 or more years from the end of active treatment, 5-34 years postdiagnosis) from wave 1 of our data. This group of 295 respondents represents the subset of the all respondents interviewed during wave 1 for whom data are available on all of the variables relevant to the analysis presented here.

SAMPLE ACQUISITION

Sample acquisition for the study began in March 1999. Based on the study's inclusion criteria, the tumor registry provided a sampling frame of 2,129 cancer survivors. The study randomly selected from among these individuals to fill the study cells related to race, sex, and cancer type as discussed above. Three hundred twenty-one survivors completed the initial wave 1 interview, and 295 respondents had complete information on the variables on which the current research is based.

The study was reviewed by an Ireland Cancer Center research review committee, and approval was obtained from the University Hospitals' institutional review board. Before beginning the face-to-face interviews, all respondents read and signed written informed consents. Interviews took, on average, 2 hours to complete and were conducted in respondents' homes.

SAMPLE CHARACTERISTICS

The demographic, health, and cancer-related characteristics of the sample are provided in Tables 1 and 2. Fifty-nine percent of the sample is female, and the mean age of the survivors is 72.4 years. Our design strategy of over sampling African Americans resulted in approximately 35% of the sample being from this racial group, the remainder being white. There were too few other racial or ethnic minorities in this tumor registry from which we could sample, so none are included in the sample.

Table 1 Personal Characteristics (n = 295)

Table 2 Cancer and Health Characteristics (n = 295)

Table 2 continued

With regard to cancer type, the proportion of the total sample with breast cancer was 42%, prostate cancer survivors comprised approximately 29% of the sample, and colorectal survivors also comprised 29%. The mean age of our survivors at diagnosis was 62.3, and the period of time since diagnosis, on average, was 10.1 years.

The tumor registry provided information on the stage of the cancer at diagnosis. Most survivors in the sample had in situ (6.8%) or localized cancer (56.6%) at diagnosis. However, nearly 30% had more advanced disease, either regional (27.5%) or distal (1.7%). Focusing next on cancer treatment patterns, the average number of types of treatment received was 1.6. The majority of the sample (42.7%) received surgery as the only type of treatment, reflecting the localized nature of the disease. The remainder of the sample received combined therapies, including radiation, chemotherapy, and hormone therapy. Approximately 12% received both surgery and radiation therapy, and approximately 10% had surgery and chemotherapy. Less than 5% of the sample was treated with a combination of surgery, radiation, and chemotherapy; however, approximately 22% had other combinations of therapies.

INSTRUMENTATION AND MEASURES

In-person interviews were conducted with older adult cancer survivors by experienced interviewers who had received extensive training in administering the structured interview instrument. On average, it took approximately 2 hours to conduct each interview. The specific measures used to operationalize the variables in this analysis are detailed below.

Survivor demographic/personal characteristics.

Information on age, sex, and race were documented in the tumor registry and verified at the time of the interview. The respondents' education was coded in number of years of formal education completed.

One other personal characteristic of survivors that has been included in our research is dispositional optimism. This trait-like characteristic has been included as a covariate in our prior research, which has shown this to be correlated with many of our self-report measures.41,42 Optimism was measured using the subscale of the Life Orientation Test.43 Optimism is generally viewed as a stable dispositional characteristic of the individual.44 This 8-item scale included items such as "In uncertain times, I usually expect the best," with responses coded on a 5-point continuum ranging from "strongly disagree" to "strongly agree." Scores on this index had a potential range of 8 to 40, and the scale had an alpha reliability of .78 in our total wave-1 sample. The mean levels of optimism in this sample was 30.2, with an SD of 4.4.

Cancer/treatment measures.

The type of cancer and stage at diagnosis were based on tumor registry information. For statistical analysis, the type of cancer was binary coded for breast, colorectal, or prostate cancers (0 = did not have this type of cancer, 1= had this type of cancer). The staging of the cancer was coded by the tumor registry as "in situ," "localized," "regional," or "distal." These were given numeric codes ranging from 1 to 4 for statistical analysis. The length of survivorship, measured as the number of years survived since diagnosis, was based on tumor registry data and confirmed during the interview. The period of time ranged from 3 to 34 years.

Three measures were used to operationalize the nature and extent of cancer treatment. The first was the reported number of treatment types the individuals received for their cancer (ie, surgery, radiation, chemotherapy, hormone therapy, or other). Fifty-one percent reported receiving only 1 form of treatment (typically surgery), 31.5% received 2 forms of treatment, and 14.6% received 3 or more types of treatment. The mean number of treatment types was 1.6, with an SD of 0.8. To identify the effects of specific types of treatment, surgery, radiation, or chemotherapy was coded as dichotomous variable (0 = did not receive, 1 = did receive). In this sample, 88% of the survivors had surgery, 32% received radiation therapy, and 19% received chemotherapy (not shown in the table).

Reported symptoms and health conditions.

Three symptom indices were constructed using a list of 21 possible symptoms such as nausea, vomiting, swelling, impaired immunity, loss of balance, numbness, burns, and others. The presence of each symptom was totaled to create a simple count with a potential range from 0 to 21. Note that pain and fatigue were not included in this measure and operationalized as noted below. Respondents were asked which of these symptoms they experienced during cancer treatment, and their responses were used to construct an index of Symptoms During Treatment. The mean number of symptoms reported was 1.5 with an SD of 2.0.

Using the same list of symptoms, an index was constructed documenting the number of Current Symptoms Not Related to Cancer. The mean number of current symptoms not related to cancer was 2.1, with an SD of 2.1. A second index documented Current Cancer-Related Symptoms. On average, respondents reported less than 1 (mean = 0.7, SD = 1.3) continuing cancer-related symptom.

The presence of pain was measured with 2 additional items, one which assessed the frequency and one that assessed severity. The first item asks respondents, "In a typical week, how often do you have pain." Response categories ranged from "never" scored as "0" and to "always" scored as "4." The mean score on this item was 1.7, with an SD of 1.2. The second item asked, "Please rate your pain on a scale of from 1 to 10 with '1' being no pain and '10' being unbearable pain." The mean score on this item was 4.7, with an SD of 2.2. The z scores for these 2 items were added to form a 2-item scale. The mean for resulting pain scale was 0.6, with an SD of 1.9. The [alpha] reliability was .82.

Fatigue was measured using 2 items that tap different aspects of this construct: weakness and energy level. The first item asked respondents, "In a typical week, to what extent do you feel weak?" Response categories range from "Not at all" scored as "0" to "Very much" scored as "3." The mean score on this item was 0.5, with an SD of 0.7. The second item asked, "In a typical week to what extent do you feel energetic?" Responses categories ranged "Not at all" scored as "0" to "Very much" scored as "3." The mean score on this item was 1.7, with an SD of 1.7.

A single index of comorbid health conditions was used to assess the current health difficulties reported by the survivors (Table 3). The Health Conditions Index is a measure of comorbidity that is a sum of the number of health conditions (other than cancer) that the survivor reported from a list of 27 possible conditions based on the Older Americans Resources and Services.45 The possible range of health conditions was 0 to 27, with an actual range of 0 to 11. The mean number of health conditions reported was 3.7, with an SD of 2.4. Nearly 20% of the sample reported 4 or more health conditions. The psychometric properties of this index have been well established for use with older adults by Fillenbaum,46 who provides validity and reliability information on the Older Americans Resources and Services methodology from which this index is derived.

Table 3 Univariate Statistics for Weakness, Energy, and Pain (n = 295)*

The measure of physical functioning used in this research was an adaptation of an index developed by Nagi.39 This Functional Difficulty Scale evaluates tasks that relate primarily to the World Health Organization/International Classification of Functioning "activity limitations" category.47 It is a self-report measure that assesses the problems survivors have with 11 specific motions or movements, such as standing, lifting or carrying objects, moving limbs, stooping/bending or kneeling, or buttoning a shirt. Responses to each item were scored on a 4-point continuum from "0" (no difficulty) to "3" (unable to perform). The possible range of scores is 0 to 44, with an actual range of 0 to 25. The mean for total sample was 5.2, with an SD of 5.4, indicating a modest level of impairment.

ANALYSIS STRATEGY

The first step in the data analysis was to document the pain and fatigue as reported by the cancer survivors (research question 1). These data are reported as frequencies and percentages for specific answer categories of pain, energy, and weakness. Data are also provided as mean scores by cancer type and sex. F ratios and P values are provided from analysis of variance.

The next step in the analysis was to examine the bivariate correlation coefficients for all variables included in this analysis (research question 2).

Finally, ordinary least squares regression analysis was used to first identify the independent (net) effects of demographic, age, and cancer-related factors on pain and fatigue and, second, to identify the effects of pain and fatigue on functional difficulty (research question 3). This portion of the analysis is guided by the analytic model portrayed in the Figure.

Results

Pain and Fatigue Among Long-term Survivors: Cancer Type and Sex

In addressing the first research question, data presented in Table 3 show that 42.4% of survivors report feeling weakness during the past week. Most reported only "a little" weakness (33%), and less than 10% reported "quite a bit or very much" weakness. In terms of energy, approximately 5.4% reported not feeling energetic at all, with about a third reporting feeling "a little" energetic and approximately 45% feeling "quite a bit" energetic. Moreover, of those who experienced either aspect of fatigue (weakness or a lack of energy), about one fourth attributed it to cancer (not shown).

With regard to pain, nearly half of the sample report that during the past week, they rarely or never had pain. However, about one fourth reported that they frequently or always had pain during the past week. Of those who reported having any pain, the mean level of pain reported was 4.7 on a scale from 1 to 10 (unbearable), indicating moderate severity. Finally, of those reporting pain during the past week, approximately 12% attributed it at least in part to cancer or its treatment (not shown).

Comparing pain, energy, and weakness by cancer type and sex, one significant difference was found (Table 4). Breast cancer survivors, specifically, and women in general, had higher scores on our pain index. These same groups also reported higher levels of pain severity and greater frequency of pain (not shown). Although there were no statistically significant differences by sex and/or cancer type in weakness or energy level, the general pattern observed for pain was repeated. Breast cancer survivors and women in general reported higher levels of weakness and lower levels of energy when compared with colorectal and prostate cancer survivors and men in general.

Table 4 Pain, Weakness, and Energy (n = 295)

Correlates of Pain and Fatigue

Before the regression analysis, the bivariate correlation coefficients were examined for all of the variables included in the conceptual model (research question 2). They are provided in Table 5 because zero-correlation coefficients are often helpful to future researchers conducing meta-analyses. These Pearson r coefficients and the P values are obtained from 2-tailed t test. However, for the sake of parsimony, only statistically significant coefficients are reported in the narrative section that follows. These are presented because they provide context for the multivariate analyses that follow, where we see that some bivariate relationships that are substantial and achieve significance are reduced and do not achieve significance in regression where the effects of covariates are controlled.

Table 5 Inter-item Correlations (n = 295)

Looking first at the correlates of pain using the 2-item index, race and sex were significant correlates of pain with African Americans and women having the highest index scores (r = 0.15 and 0.29, respectively). Among the age-related factors, the strongest correlates of pain were the number of health conditions (r = 0.36), functional difficulty (r = 0.47), and current symptoms not attributed to cancer (r = 0.36). Among the cancer-related factors, symptoms during treatment and current symptoms attributed to cancer were also significant, albeit weaker correlates of pain (r = 0.21 and 0.14, respectively). As shown earlier, breast cancer is also a significant correlate of pain (r = 0.28) compared with the other 2 forms of cancer included in this study, whereas prostate cancer was associated with lower scores on the pain index (r = -0.22). When compared with those in our sample with the other types of cancer, colorectal cancer was not significantly correlated with pain. Finally, pain correlated 0.27 with weakness and -0.23 with energy.

Turning to the correlates of energy and weakness, the 2 fatigue-relevant measures, similar patterns are evident. First, being African Americans and female were each associated with lower levels of energy (r = -0.25 and -0.14, respectively), and more educated survivors reported higher levels of energy (r = 0.24). Functional difficulty and the number of health conditions are significant correlates of lower levels of energy (r = -0.41 and -0.18, respectively), and both of these are correlated with race and sex. It is not surprising that age is a significant correlate of energy, with older survivors reporting lower levels of energy (r = -0.17). Symptoms not attributed to cancer (r = -0.22) were modestly correlated with energy level. Of the cancer-related factors, only breast cancer (compared with having either of the other 2 types of cancer included in this study) was associated with lower levels of energy (r = -0.15). Optimism, a trait-like characteristic that may dispose a survivor to report higher levels of energy, was significantly correlated (r = 0.27) with reported energy level. Given this correlation, it is important to consider it as a covariate in the multivariate analysis so that the impact of this dispositional factor can be statistically controlled.

Looking next at the correlates of survivor's reports of weakness during the prior week, a number of the patterns of correlation noted above are replicated. First, being female is associated with higher levels of weakness (r = 0.12), although this coefficient is quite small and barely achieves statistical significance. And although race is not a significant correlate of weakness, education is, which is a surrogate measure for socioeconomic status (r = -0.16). More educated respondents reported lower levels of weakness. Optimism is also a correlate of weakness, with more optimistic individuals less likely to report weakness (r = -0.15).

Health conditions and functioning were among the strongest correlates of weakness. The number of comorbid health conditions and noncancer symptoms along with functional difficulty all demonstrated relatively stronger levels of association (r = 0.30, 0.30, and 0.43, respectively). Of the cancer-related factors, the number of symptoms during treatment is a relatively strong predictor of weakness (r = 0.30) as are current symptoms attributed to cancer (r = 0.27). Having prostate cancer is associated with lower levels of weakness (r = -0.14). No other cancer-related factors were statistically significant predictors of weakness.

Regression Analyses

With the correlational analysis as background, the analyses moved to the multivariate level. The model portrayed in the Figure suggests a number of relationships that can be more fully evaluated in a multivariate context (research question 3). Ordinary least squares regression allows us to examine the net or independent effect of the demographic/personal characteristics, age-related and cancer-related predictors on the pain, and energy and weakness levels reported by cancer survivors. It also provides an indicator of the additive effects of these factors (adjusted R2). As portrayed in the model, regression will also be used to examine the impact of all of the above factors on functional difficulty. For comparison of the relative influence of predictors, only statistically significant standardized regression coefficients ([beta]) will be reported in the text. The unstandardized coefficient (B) and exact P values are provided in the accompanying tables (Tables 6 and 7). R2 and the adjusted R2 indicate the total variance explained by all predictors in each equation.

Table 6 Regression Analyses of Weakness, Energy, and Pain (n = 295)

Table 7 Regression Analyses of Functional Difficulty (n = 295)

PREDICTORS OF PAIN

Looking first at the predictors of the pain index, sex and breast cancer, which were significant correlates of this outcome in the bivariate analysis, are no longer significant. This change is largely the result of the relative strength of the health conditions measure ([beta] = .21) and noncancer symptoms ([beta] = .22) in predicting pain. Importantly, chemotherapy is a significant cancer-related predictor of pain ([beta] = -.24). This indicates that compared with those who did not have chemotherapy, those who had this type of treatment reported lower scores on the pain index. The only other significant predictor, albeit a relatively weak predictor, is the number of years since cancer diagnosis ([beta] = -.13). This indicates that with greater temporal distance from cancer, pain diminishes for these older adult survivors. All predictors taken together explain 29% of the variance (adjusted R2) in pain reported by survivors. The survivors' demographic/personal characteristics explain 8% of the variance, with age-related factors explaining the majority of the variance (14%). Cancer-related factors explain only 2% additional variance (not shown in the table).

PREDICTORS OF ENERGY AND WEAKNESS

The second regression analysis examines energy, 1 of the 2 fatigue-relevant indicators. The strongest predictor of energy is optimism ([beta] = .21), indicating that reports of energy are at least in part a function of this dispositional trait. However, after controlling for the effect of this factor, a number of other predictors remain statistically significant. Among these, race continues to be a statistically significant, albeit modest, predictor ([beta] = -.19), along with age ([beta] = -.16).

Only 1 cancer-related factor, having received chemotherapy as a form of treatment, was a statistically significant predictor of energy level ([beta] = .18). Survivors who had chemotherapy reported higher levels of energy than those who did not have this type of treatment; the direction of relationship was opposite to what might be expected based on logic and prior research. All predictors taken together explain 19% of the variance in survivors' reports of their energy level. The majority of this variance is explained by the personal characteristics of the survivor (13%), primarily their optimism. Age-related factors add an additional 4% to the variance explained, whereas cancer-related factors add less than 2% (not shown in the table).

The third regression analysis examines the survivor's report of weakness. None of the demographic/personal characteristics were significant predictors of weakness. However, 2 age-related factors that continued to be statistically significant predictors were health conditions and noncancer symptoms ([beta] = .17 and 18, respectively). One important cancer-related factor emerged, the current level of symptoms being attributed to cancer ([beta] = .16). Thus, both cancer-related and noncancer symptoms have similar impact on the weakness reported by older adult survivors. All predictors taken together explain 19% of the variance in weakness. The demographic and personal characteristics account for approximately 4% of this variance. Age-related factors explain an additional 9% of the variance, whereas cancer-related factors account for the remaining 6% of the total variance (not shown in the table).

PREDICTORS OF FUNCTIONAL DIFFICULTY

After having examined the predictors of pain, energy level and weakness regression analysis was used to identify the key predictors of functional difficulty (Table 7). In this analysis, each of the personal characteristics and age-related and cancer-related predictors was entered in the equation (see model 1). However, for this outcome, a second equation is estimated (model 2). In this second equation, pain, energy, and weakness were added as predictors in last step.

Looking first at model 1, a large number of predictors were statistically significant. Among the personal characteristics, race, sex, and marital status were all statistically significant ([beta] = .14, .18, and .13, respectively). Of the age-related factors, the survivors' current age, number of health conditions, and noncancer symptoms were significant ([beta] = .12, .22, and .28, respectively). Three cancer-related factors were statistically significant predictors: cancer stage ([beta] = .13), current symptoms attributed to cancer ([beta] = .14), and chemotherapy ([beta] = -.16). All predictors taken together explain 43% of the variance, with personal characteristics explaining 16%, age-related characteristics explaining an additional 21%, and cancer-related characteristics explaining 6% (not shown in the table).

In model 2, when pain, energy, and weakness are added, each is a significant predictor ([beta] = .20, .11 and -.16, respectively). However, a number of predictors significant in model 1 no longer reach statistical significance. These include race, stage at diagnosis, current symptoms attributed to cancer, and chemotherapy. This indicates that a portion of the explanatory power of these predictors is shared with the pain and fatigue predictors. Finally, including the pain and fatigue predictors in the model adds an additional 7% to the variance in functional difficulty that is explained.

Discussion

Summary of Findings

From the data presented, it is clear that older adult cancer survivors do experience pain, and, to a lesser extent, they report weakness and a lack of energy. Furthermore, the pain, energy, and weakness levels they report are empirically linked to functional difficulty. Because this research was not designed to compare cancer survivors to the general population, we cannot answer the question of whether having cancer (as compared with not having ever had it) produces additional pain or fatigue symptoms and functional difficulty among older adults. Addressing these issues would require a larger epidemiological design with a matched comparison group.

The present research, however, can address the important question of whether these symptoms and functional difficulties, when they do occur, are empirically linked to cancer-related factors. Our multivariate approach also allows us to assess whether cancer-related factors are more or less important than other factors associated with aging. The answers produced by our analysis are not unequivocal. For the most part, age-related and personal characteristics seem to play a larger role in the pain, fatigue, and functioning of these older adult survivors. And yet, there are specific instances where cancer-related factors are worth noting.

In terms of pain, it is clear that other health problems and symptoms not attributed to cancer are relatively strong predictors of pain. However, the number of years since diagnosis is negatively related to the pain index. This suggests that any cancer-pain link that does exist may diminish with time. Given that our sample is, on average, a decades past treatment, it may be that our lack of significant cancer-related effects is the result of the passage of an extended period of time. The only other cancer-related factor that was statistically significant was having had chemotherapy. The direction of this relationship indicates that those who had chemotherapy report lower levels of pain. This may be because those not treated with chemotherapy were more likely to be treated by either surgery or radiation. These are 2 forms of treatment that may be more likely to produce pain that persists years after treatment, whereas the effects of chemotherapy on pain may be more transient.

In terms of the 2 fatigue-relevant measures, once again we see the relatively more powerful effects of other health problems and symptoms. However, the number of current cancer-related symptoms is a significant predictor of weakness, and its strength as a predictor is comparable to noncancer symptoms. Similarly, cancer survivor's reports of their energy level are largely a function of other health conditions and age. Of the cancer-related factors, only chemotherapy is a significant predictor; however, it is in the opposite direction, with those having had chemotherapy reporting higher levels of energy, compared with those who had other forms of treatment. This may indicate a "rebound" or "contrast" effect, whereas those who had chemotherapy report higher levels of energy now in part because they compare their energy level now to the relatively low levels of energy they had during treatment.

Looking at functional difficulties experience by survivors, it is clear that other health problems and symptoms play the larger role in diminished capacity compared with cancer-related symptoms. Of the cancer-related factors, only breast cancer was a significant predictor, but the direction of the relationship was opposite to the anticipated direction. Breast cancer survivors, compared with the reference group (colorectal cancer survivors) and with the effects of sex statistically controlled, reported lower levels of functional difficulty.

It is also clear from the analysis that pain, weakness, and lower levels of energy make a substantial contribution to the explanation of functional difficulty. Individuals who report more pain and weakness and less energy also report more difficulty such as reaching, bending, walking, and others. Of course, this finding that is not unique to cancer survivorship and reflects issues that are part of "normal" aging.

Limitations

This research presented here has a number of limitations that grow out of the study's original intent and design. First among these is the measurement of the pain and fatigue outcomes. The broader nature of the study that assessed a large range of quality of life issues limited our opportunity to assess pain and fatigue in more detail. However, our approach using single-item indicators for weakness and energy is similar to that used by other cancer researchers such as Given et al13 and is not uncommon in research on survivorship. The second limitation is the lack of a comparison group of older adults who have never had cancer. However, as noted above, the focus of this research is on the relative impact of cancer-related versus age-related factors among survivors, not on comparing survivors with those who have never had cancer. The conceptual model and analytic approach taken allow us to address important issues for clinicians working with older adults with a cancer history.

Clinical Implications

For nurses and others who treat long-term survivors in primary care or geriatric settings, this research point to the fact that both cancer and noncancer factors contribute to the pain, energy level, and weakness that older adults report. As such, the presence of a history of cancer and its treatment is important part of the assessment process. However, it is not unusual for those providing healthcare for older adults to be unaware of their cancer history including the types of treatments they may have had, especially if that cancer occurred decades in the past. Furthermore, older adults may be unaware of the potential for their prior cancer to be a source of the pain and fatigue they experience. As a result, the role that cancer may have played in these reports of pain and fatigue may go unreported.

Furthermore, as our prior research has shown, cancer-related worries persist for decades after treatment, including fears about current symptoms.46 Importantly, these worries are related to broader aspects of psychosocial distress in terms of both anxiety and depression. Thus, not only is physical health quality of life potentially compromised by symptoms such as pain and fatigue, but psychosocial quality of life may also be affected. This is more likely to be the case if ambiguous symptoms are interpreted as possibly cancer related. Clinicians working with cancer survivors can play an important role in helping the survivor sort through their symptoms in terms of the likelihood that they are the result of either a new or recurrent cancer.

All of the above suggest that clinicians have the opportunity to help older adult, long-term survivors as they face the challenges that both cancer and aging present.

References