Factors Affecting Perceptions of Disability and Self-Rated Health Among Older Adult,Long-Term Cancer Survivors

Abstract

Objective: This article examines the relative importance of cancer-related and noncancer illness factors as they predict the health quality of life among older adult, long-term cancer survivors. Specifically, it examines the effects that continuing cancer symptoms and comorbidities have on functional difficulties and how they in turn affect perceptions of disability and self-rated health. Method: Data from an National Cancer Institute (NCI)-funded, tumor registry–based 10-year study of 321 older adult (age 60+), long-term (5+years post diagnosis) survivors of breast colorectal and prostate cancer are examined using regression analyses. Results: The analyses documented the independent effects of both cancer-related and age-related health factors as they contribute to explaining functional difficulties, perceptions of disability and self-rated health. Gender and racial differences in health quality of life were also identified. Discussion: The findings suggest that geriatricians, geriatric nurses, and clinical gerontologists who work with cancer survivorsneed to be aware of the ways in which both cancer and noncancer illness factors work together in producing threats to health quality of life through the extent and nature of functional impairments.

Keywords

cancer and aging comorbidities physical function disability health quality of life

Introduction: The Intersection of Aging and Cancer

The purpose of this research is to examine the relative importance of cancer-related factors compared with other health issue experiences by older adult cancer survivors on subjective health quality of life. More specifically, it tests a multivariate model of the relative importance of cancer-related factors and those related to other comorbid health conditions on the physical functioning of older adult survivors and how each of these contribute to older cancer survivors’ perceptions of disability and overall health.

Cancer has been called a disease of the elderly because most cancers occur in individuals above 65 years of age (Ershler, 2006; Krok-Schoen, Palmer-Wackerly, Dailey, Wojno, & Krieger, 2017). According to American Cancer Society (2002), the incidence rate triples in the age group 60 to 79 years, compared with individuals age 40 to 59. As such, a history of cancer earlier in life is not unusual in later life. Because of the potential for long-term and late effects of the disease and its treatment among older adults, it can affect health quality of life even decades after successful treatment.

Considerable research has already documented that the long-term and late effects of cancer and its treatment persist into later life for long-term survivors (Bowman, Rose, Deimling, Kypriotakis, & O’Toole, 2010; Deimling, Kahana, & Bowman, 2012; Deimling, Kahana, Bowman, & Schaefer, 2002; Porter, 2013). The importance of studying how cancer influences the lives of older survivors has been emphasized by Rowland and Yancik (2006) at the National Cancer Institute (NCI) and the National Institute on Aging, respectively. They noted that “ . . . understanding the contribution of cancer to the older individual’s functional health status and how one chronic health condition can influence another is critical if we are to reduce the human and social burden of cancer” (p. 504). They asserted that “new knowledge must be generated at the cancer-aging interface and applied for optimal results for our aging and older cancer survivors” (p. 505).

The Role of Comorbid Health Conditions

Moreover, studies have found out that older adult survivors not only suffer from cancer-triggered health problems but also encounter other age-related health problems similar to other older adults who have not had cancer (Braithwaite et al., 2010; Piccirillo, 2000; Yancik, 2001). For example, population-based studies have documented that comorbidities are present in nearly 70% of older cancer patients (Havlik, Yancik, Long, Ries, & Edwards, 1994) and increase with advancing age. Among cancer survivors, nearly one third report having two or more comorbid health conditions (Ogle, Swanson, Woods, & Azzouz, 2000) which have been linked to increased functional limitations (Catalá-López et al., 2014; Hewitt, Rowland, & Yancik, 2003). Ganz (2003) pointed out that this interaction of cancer-related factors and comorbidities is an essential element of geriatric oncology.

Functional Difficulties and Perceptions of Disability

Importantly, different type of cancer such as those commonly experienced in later life have been shown to differentiate the types of symptoms and functional problems experienced (Ramsey et al., 2000). Both the gerontology and oncology literatures suggest that physical functioning is a key factor in understanding health-related quality of life across the life course (Litwin, 2003; Romoren & Blekeseaune, 2003; Sergeant, Ekerdt, & Chapin, 2010) and is essential in evaluating the relative impact of cancer on older survivors (Piccirillo, 2000; Yancik, 2001).

As central as functioning is to understanding the health quality of life of older cancer survivors, some research has shown that individuals with functional difficulties do not always view themselves as disabled. This may be because perceived disability involves a good deal of individual interpretation and other contextual issues (Kelley-Moore, Schumacher, Kahana, & Kahana, 2006; Land & Yang, 2011). For example, decades ago, Zola (1993) discussed the complicated language of disability and that the perception of disability involves labeling and identity aspects as part of this interpretive process. More recently, Darling and Heckert (2010) also discussed how people determine their “disability orientation” (p. 132) and how self-identifying as disabled is only one part of their disability orientation. Interestingly, Kelley-Moore and colleagues (2006) did not find a relationship between self-reported functional limitations (Activities of Daily Living-ADL and Instrumental Activites of Daily Living-IADL) and perceived disability. In an article about disability over the life course, Langlois and colleagues (1996) found that over half of older adults who could not perform or who had problems performing one ADL did not consider themselves disabled. Similarly, in a study by Iezzoni, McCarthy, Davis, and Siebens (2000) that focused on mobility difficulties, the researchers found that while most of the respondents with mobility difficulties perceived themselves as disabled, some did not—including some respondents who used wheelchairs. In addition, they also found that women, racial minorities, and Hispanic respondents were less likely to perceive themselves as disabled, whereas low-income persons were more likely to hold this self-perception.

All of the above suggest the importance of treating functional difficulties and perceptions of disability as separate factors, both conceptually and empirically. It also speaks to the importance of examining the factors that may predict functional difficulties and perceptions of disability separately. To our knowledge, there has been no research on how functional difficulties contribute to older adult cancer survivors feeling “disabled.” Prior research on functioning and disability were concerned primarily with work-related disability and have not used concrete measure of functioning and/or perceived disability (Greenwald et al., 1989; Short, Vasey, & BeLue, 2008). Nor has prior research examined the relative importance of cancer-related factors compared with noncancer health/illness on disability perceptions.

Functional Difficulties and Self-Rated Health

A widely used, broader indicator of health quality of life is self-rated health. Like perceptions of disability, this is a largely subjective indicator. However, research has long shown that these perceptions are as good as or better indicators of mortality than more “objective” indicators including physician’s ratings of their patient’s health (Giltay, Vollaard, & Kromhout, 2012; Mohan et al., 2011). Recent research has noted that comorbidities and disability translate into poorer self-rated health among older adults (Hardy, Acciai, & Reyes, 2014). In the oncology literature, research has shown that among breast cancer survivors both poorer functioning and the presence of comorbidities are associated with poorer self-rated health (Nápoles, Ortíz, O’Brien, Sereno, & Kaplan, 2011) as are surgical side-effects (Schootman, Homan, Weaver, Jeffe, & Yun, 2013).

Research Questions and Conceptual Model

The research cited above suggests that older adults who have had cancer face the dual health vulnerabilities conferred by both advancing age and related illnesses, and those that result more directly from the lingering effects of cancer or its treatment. However, two largely unanswered questions remain:

Research Question 1: What is the relative impact of age-related factors such as greater comorbidity and related illness symptoms when compared with health issues resulting from cancer or its treatment on the functioning of older survivors?

Research Question 2: How do these age-related and cancer-related factors along with functional difficulties translate into perceived disability and self-rated health?

The complexity of these relationships can be more clearly understood looking at our conceptual model (see Figure 1). It portrays the potential linkages among survivors’ personal characteristics, cancer-related illness factors, other health conditions, and symptoms as they may affect functioning and how all of these in turn affect perceptions of disability and self-rated health. The model portrays the direct effects that survivors’ personal characteristics and both cancer and noncancer illness factors may have on the two global subjective health outcomes. However, an important aspect of this research is determining how these effects are mediated by the functional difficulties they may produce.

Figure 1.

Conceptual model.

Method

Data Source

The data presented in this article were derived from a larger NCI-funded study of older adult cancer survivors conducted between 1998 and 2010 at the Cancer Survivors’ Research Program (CSRP) at Case Western Reserve University. This research was conducted in collaboration with the Seidman Cancer Center of University Hospitals of Cleveland. The sample was randomly selected from more than 6,000 individuals in the tumor registry of the Seidman Cancer Center (at that time the Ireland Cancer Center) at University Hospitals Health System (UHHS) of Cleveland who met the study’s age (60+), duration of survivorship (5+ years post diagnosis), and cancer type (breast, colorectal, and prostate cancers). Mindful of the racial disparities in cancer outcomes, the study oversampled African Americans who represented 41% of the final sample. Unfortunately, there were too few potential respondents of other racial or ethnic groups in the registry that met the other study selection criteria. Women comprised 50% of the final sample.

The analysis presented here is from the first wave of in-person interviews with 321 survivors. That specific wave of data was selected because it included all of the variables identified as important in the proposed model. Respondents were on average 72 years of age at the time of interview and on average had survived cancer 10 years since diagnosis. Table 1 provides further detail on the sample and key measures included in the model.

Table 1.

Description of Sample (N = 321).

	M/%	SD
Age	72.3	7.51
Female	50%	—
African American	41%	—
Years education	13.9	3.54
Years since diagnosis	10.3	5.48
Cancer stage at diagnosis
Local/in situ	68.1	—
Regional	29.9	—
Distal	2.0	—
Number of treatment types	1.6	0.78
Breast cancer	41%	—
Prostate cancer	29%	—
Colorectal cancer	30%	—
Number of noncancer symptoms	2.6	2.45
Number of current cancer symptoms	0.8	1.51
Number of health conditions	3.7	2.40
Functional difficulty index	5.2	5.35
Perceived disability burden	0.3	0.74
Self-rated health score	11.6	2.06

Model variables

Outcome variables

This study assesses two health quality of life outcomes. The first, Perceived Disability, is a single item indicator in which respondents indicated their response to the question “Do you consider yourself disabled” on a 5-point Likert-type scale ranging from not at all to very much. Because the distribution of this variable demonstrated considerable skew and relatively high Kurtosis (2.7 and 7.2), we “Winsorized” the variable into a four category item ranging from 0 to 3 (Hastings, Mosteller, Tukey, & Winsor, 1947; see Dixon, 1960). With this transformation, the indicator demonstrated acceptable skew and Kurtosis (2.2, 3.9), making this variable appropriate for use in the correlation analysis and as an outcome measure in the ordinary least squares (OLS) regression analyses. The arithmetic mean of the transformed variable was 0.33; with a standard deviation (SD) of 0.83.

The second outcome, Self-Rated Health, was measured using a three-item indicator. These items include questions asking “In general, do you consider yourself to be very health, healthy, fairly health, sick or very sick”; “In general considering your health over the past year would you say your health is excellent, good, fair poor or very poor?” and “Compared to others your age would you say that your health in the past year is much better, better, about the same, worse or much worse.” The final measure used is a simple sum of the scores on the three specific items that had a potential range of five to 15. The mean of the three-item scale in our analysis was 11.64; with a SD = 2.06.

Predictors

The health/illness measures that constitute the focal predictors in our model are described below. Health Conditions/comorbidities is measured by a sum of the number of diagnosed health conditions (other than cancer) reported by respondents from a list of 27 possible conditions from the Older Americans’ Resources and Services (OARS) assessment (George & Fillenbaum, 1985). The actual range was 0 to 11 (M = 3.70; SD = 2.40).

Two measures of current illness symptoms were constructed from respondent’s reports of the presence of specific symptoms from a list of 22 that could be the result of cancer or another illness. The same list of symptoms was used for both cancer and noncancer symptom indices. These include symptoms such as weakness, pain, swelling, imbalance, incontinence, nausea, and so on which are relevant for all older adults. The first indicator represents the total number of symptoms reported by the respondent as not cancer-related (Current Noncancer Symptoms, M = 2.55; SD = 2.45). The second indicator represents the total number of symptoms reported by the respondent to be cancer-related (Current Symptoms Attributed to Cancer, M = 0.80; SD = 1.51).

The last health-related predictor used in the analysis was an index of Functional Difficulties developed by Nagi (1976). This widely used index provides respondent’s self-report of the ability to perform 11 specific motions or movements such as standing, reaching/stretching, lifting, and ability to stand for an extended period of time. The possible range of scores on this index is 0 to 44, with an actual range of 0 to 25 (M = 5.24; SD = 5.35).

Covariates

A number of respondent’s personal characteristics that could potentially affect health and health-related quality of life were included in the model and subsequent analyses as predictors/covariates. Age was measured by chronological age as document in the tumor registry data. Gender is included as a binary variable in the correlation and regression analyses, scored 1 = female and 0 = male. Race was operationalized as the respondents self-report racial identity (coded 1= African American/Black and 0 = Caucasian/White). Respondents from other racial/ethnic groups were not included in the study due to the relatively small numbers in the tumor registry who met the study’s inclusion criteria. Education was measured by years of formal education as reported by the respondent.

Analysis plan

The first step in the analysis was to examine the correlation among all the variables portrayed on our model (Figure 1). Pearson product–moment correlation coefficients are provided in Table 2 for all interval and categorical binary measures. Specifically, gender, race, and the three cancer types are treated as binary variables. These binary measures are appropriate and interpretable for both correlation and as predictors in regression given the distribution of scores on these items (approximately a 40-60 split on the frequency distribution for the two categories). The advantage of using these as binary variables in the correlation and regression is that it allows for direct comparison across all study measures. However, coefficient eta was also calculated. It provides an alternative measure of association appropriate for use where one measure is binary/categorical and the other is interval/interval-like. Like coefficient r, it can be squared to indicate variance explained. Eta is reported in the text along with r for comparison for specific variables pairs. In nearly all cases, eta did not differ from r to any degree either substantively or in terms of statistical significance.

Table 2.

Correlations (N = 321).

	Number of health conditions	Functional difficulties	Perceived disability	Self-rated health
Age	.07	.19**	.08	.00
Female	.20**	.33**	.13*	−.10
African American	.08	.23**	.11	−.24**
Years of education	−.07	−.15**	−.14*	.17**
Number of health conditions	1.00	.44**	.33**	−.02**
Functional difficulties index	.44**	1.00	.62**	−.51**
Number of noncancer symptoms	.42**	.48**	.37**	−.31**
Years since diagnosis	.05	.11*	−.01	.00
Cancer stage at diagnosis	−.07	.10	.06	−.09
Number of treatment types	−.03	.04	.04	−.02
Breast cancer	.16**	.19**	.06	−.06
Prostate cancer	−.14*	−.26**	−.08*	.03
Colorectal cancer	−.04	.05	.02	.04
Number of current cancer symptoms	.14*	.25**	.24**	−.28**
Perceived disability	.33**	.62**	1.00	−.47**

p < .05. **p < .01. ***p < .001.

Using the bivariate analysis as a basis, three OLS regression equations were then estimated based on the model as portrayed in Figure 1. The first equation is used to identify the strongest predictors of functional difficulties, and the latter to identify the net effect of all predictors on the two health quality of life outcomes, perceived disability, and self-rated health. OLS regression is appropriate where the outcome measures are interval or interval-like and are not excessively skewed (as indicated in above in the methods section). As in the correlation, gender, race, and cancer type are included as binary measures scored as 0 or 1.

Results

Correlation Analysis

A review of the bivariate correlation coefficients (Pearson product–moment coefficient and Eta) is an instructive preliminary step in identifying the factors associated with comorbid health conditions, functional difficulties, and the two health quality of life outcomes, perceived disability and self-rated health (see Table 2). First, in terms of personal characteristics, age is moderately (r = .19) but significantly associated with functional difficulties, but none of the other model outcomes. Gender, on the contrary, is a key correlate of both comorbidities (r = .20, eta = .15) and functional difficulties (r = .33, eta = .33), although only weakly related to perceived disabilities (r = .13 and eta = .08). These associations replicate the well-known patterns of more health problems and poorer functioning among older women compared with men.

With regard to race, being African American is significantly associated with higher levels of functional difficulty (r = .23, eta = .23) and poorer self-rated health (r = −.24, eta = .24) while higher levels of education is also modestly associated with fewer functional difficulties (–.15), less perceived disability (–.14), and better self-rated health (r = .17). The regression that follows allows us to identify the net impact of race and education on these health outcomes directly.

Turning next to general health factors, not surprisingly the number of comorbid health conditions is a relatively strong correlate of functional difficulties (r = .44) and also perceptions of disability (r = .33). Functional difficulties, in turn, are relatively strong correlates of perceived disability (r = .62) and self-rated health (r = .51). Similarly, symptoms reported as not attributed to cancer were relatively strong correlates of comorbid health conditions (r = .42) and functional difficulties (r = .48). The number of noncancer symptoms was also a moderate correlated of two focal outcomes, perceived disability (r = .37) and self-rated health (r = −.31). Although these represent considerable correlation among the predictors and outcomes, most do not explain over a third of the variation in any other measure in the model, leaving considerable variance to be accounted for in the multivariate analyses of the model.

Focusing on the role that cancer-related factors play in functional difficulties and health quality of life perceptions, the duration of time since diagnosis is only weakly related to any of the health outcomes. Surprisingly, neither of the measures that might serve as indicators of disease severity, stage at diagnosis or number of types of treatment received, was a statistically significant correlate of the health outcomes. It is clear that breast cancer survivors are most negatively affected (r = .16, eta = .16) in terms of the number of comorbid health conditions as well as functional difficulties (r = .19, eta = .19). However, this type of cancer is not related to greater perceptions of disability or poorer self-rated health. Having had prostate cancer was associated with less functional difficulty (r = −.26, eta = −.26). Given what we know about gender differences in health and functioning, it is likely that these cancer type differences may be in part a reflection of gender, a premise that will be tested in the regression analyses.

Looking at other cancer-related factors, the number of current cancer-related symptoms is a significant correlate of all four health indicators; comorbid health conditions (r = .14), functional difficulties (r = .25), perceived disability (r = .24), and self-rated health (r = .28). Finally, the number of comorbid health conditions (r = .33) and functional difficulty (r = .62) are all relatively strong correlates of perceived disability as is self-rated health (r = .47).

An important correlational finding is that the personal, general health, and cancer-related variables show different patterns of association with functional difficulties compared with the two health quality perception outcomes. For example, age, gender, and race along with two of the cancer types and noncancer symptoms show stronger correlations with functional difficulties than with either or both of the two health quality of life perceptions.

The correlation data also suggest that functional difficulty, perceived disability, and self-rated health may be treated as different factors in the model given that none explains as much as 40% of the variance in the other, or alternatively that they have at least 60% unique variance. This is an empirical reflection of the different substantive meaning and measurement of the items.

Multivariate Analyses

With the above correlations as a basis, the analysis turns to the regression findings as organized by our model and are shown in Table 3. The OLS regression analyses document the net or independent effect of each predictor on the respective outcome measures. Two coefficients are provided in the table, B and beta. The unstandardized coefficient (B) provides an estimate of these effects size in terms of the change in each unit of the outcome variable score that corresponds to a change in each unit of the predictor variable. The standardized regression coefficient (beta) provides an indication of the relative importance of each predictor on each of the three model outcomes and will be the focus of the data reported below. As noted above in the measurement section and analysis plan, all the outcomes examined in the OLS regressions are interval or interval-like measure and have acceptable skew and Kurtosis. The binary predictors (i.e., gender, race, and the cancer types) also have acceptable distributions. Note that colorectal cancer is treated as the reference category among the three cancer types as required by regression.

Table 3.

Regression Models (N = 321).

	Functional difficulties				Perceived disability				Self-rated health
	B	SE	Beta	p	B	SE	Beta	p	B	SE	Beta	p
Constant	−7.60	2.72	—	.01	0.11	0.45	—	.58	11.18	1.13	—	.00
Age	0.10	0.03	.14	.00	−0.00	0.01	−.03	.53	0.02	0.01	.08	.11
Female	2.17	0.87	.20	.01	−0.06	0.13	−.04	.65	−0.01	0.36	.00	.98
African American	1.50	0.51	.14	.00	0.07	0.08	.05	.33	−0.47	0.21	−.11	.03
Years of education	−0.02	0.07	−.01	.83	−0.01	0.01	−.03	.54	0.04	0.03	.07	.15
Health conditions	0.50	0.11	.22	.00	0.00	0.02	.00	.97	−0.05	0.05	−.06	.24
Functional difficulties	—	—	—	—	0.08	0.01	.59	.00	−0.15	0.02	−.40	.00
Noncancer symptoms	0.70	0.11	.32	.00	0.01	0.02	.02	.72	−0.09	0.05	−.11	.06
Breast cancer	−0.98	0.66	−.09	.14	−0.02	0.01	−.01	.88	0.16	0.27	.04	.57
Prostate cancer	−0.22	0.79	−.02	.78	0.15	0.12	.09	.19	−0.47	0.32	−.11	.15
Current cancer symptoms	0.83	0.16	.24	.00	0.05	0.02	.09	.05	−0.21	0.07	−.16	.00
Adjusted R²	.41				.35				.31
F test	25.39				18.43				15.38
p value	<.001				<.001				<.001

Note. The third cancer type, colorectal cancer, is treated as the “reference group” variable as required in regression analyses and as such is not included as a predictor. Coefficients in bold are statistically significant at the .05 level or greater.

Looking first at the predictors of functional difficulties, the key mediating variable in the model, age, gender, and race were modest but statistically significant predictors (beta = .14, .20, and .14, respectively) indicating that advancing age, being female, and Black each explains important variation in the levels of functional difficulties survivors report. Three of the stronger predictors of the level of functional difficulties experiences are the number of comorbid health conditions (beta = .22) and current symptoms not attributed to cancer (beta = .32). Importantly, current cancer symptoms also play an important, albeit weaker role (beta = .24). The cancer types were not found to be important predictors. The model as estimated is statistically significant and predicts more than 40% (R² = .41) of the overall variance in the functional difficulties experienced by these older adult, long-term survivors.

Turning attention to the predictors of perceived disability, it is clear that the levels of functional difficulties reported by survivors is the key predictor of perceived disability (beta = .59), far stronger than the other variables in the model. Importantly, symptoms attributed to cancer (beta = . 09), a substantially weaker predictors, is also statistically significant indicating it has an independent effect even with the effects of functional difficulties statistically controlled. This equation explains over a third of the variance (R² = .35) in survivors’ perceptions of disability.

Finally, turning our attention to survivors’ reported self-rated health, functional difficulty is again the most powerful single predictor (beta = −40). However, in this equation, the number of symptoms attributed to cancer (beta = −.16) is the only other significant predictor, while noncancer symptoms (beta = .11) have a somewhat weaker effect that approaches statistical significance (p = .06). This indicates that these two symptom factors have an impact on this health perception beyond their effects through functional difficulties. Race also is a significant predictor (beta = −.11), with African Americans reporting poorer self-rated health. This overall model is statistically significant and all of the predictors taken together explain well over a third of the variation in the outcome (R² = .37).

Summary and Discussion

Functional Difficulties Resulting From Both Cancer and Noncancer Factors

Examining the above correlation and regression data in the context of the model tells a relatively clear story. Functional difficulties among older adult cancer survivors are the key factor in explaining both their perceptions of disability and self-rated health. Equally important is that the data documents that both cancer-related and noncancer symptoms have statistically significant effects on functional difficulties. In addition, although the number of reported current noncancer symptoms is a relatively stronger predictor, continuing cancer symptoms is the second strongest predictor in that equation. Importantly, these effects persist even after controlling for the survivors’ personal characteristics, which are correlates of functioning. The findings on the impact of cancer-related factors are consistent with past studies on cancer survivors (Hewitt et al., 2003; Rowland & Yancik, 2006). The significance of the findings reported here is that the impact of cancer-related factors persists even after controlling for other health problems and noncancer symptoms.

The role that the number of comorbid health conditions plays in the health quality of life of older survivors is also substantively important. Other studies have documented that cancer survivors experience higher levels of comorbidities than the general population (Rowland & Yancik, 2006). Piccirillo (2000) revealed that head and neck, lung, breast, and prostate cancer patients had significant comorbidity problems even after controlling for age and treatment stages. In the data presented here, the number of diagnosed noncancer health conditions reported by survivors was the strongest correlate of the functional difficulties they experienced and, in the regression, it was the third most powerful predictor. However, the effects of comorbidities were shown to be almost entirely through functional difficulties as comorbidities did not have an independent direct effect on either perceived disability or self-rated health.

Survivors’ Personal Characteristics

The results also point to two important personal characteristics to be considered in evaluating the impact of cancer and noncancer factors on the health quality of life of older survivors: gender and race. Being a female is associated with greater functional difficulties among the survivors in our research. Previous studies on women with breast cancer reported increased functional difficulties linked to having this form of cancer and also reported at least one functional difficulty related to this disease (Braithwaite et al., 2010; Yancik, 2001). The regression analyses presented here documents that being female predicts greater functional impairment even when the effects of cancer type is controlled. Importantly, the regression analysis documents that the effect of gender on perceived disability are primarily through functional difficulties. Gender does not have a substantively meaningful or statistically significant direct impact on either of the health quality of life outcomes after the effects of functioning are controlled.

In terms of race, being African American (compared with being White in this sample) was associated with greater functional difficulty in the bivariate analysis and this relationship persisted in the multivariate analysis even when the effects of other noncancer health conditions (comorbidities) and education were controlled. This may well be linked to the often-reported finding that Blacks often are diagnosed later with more advanced cancer (Tammemagi, Nerenz, Neslund-Dudas, Feldkamp, & Nathanson, 2005). It is also possible that this difference may result in more invasive treatment and hence more posttreatment functional difficulty. These same authors noted that a critical factor in the racial differences in mortality is the comorbidities they experience. However, being African American has a direct negative effect on self-rated health in addition to its effects through functional difficulties.

Factors That Affect Perceptions of Disability and Self-Rated Health

The regression analysis is clear that both perceptions of disability and self-rated health are most clearly the product of functional difficulties reported by the respondents, which are predicted by both continuing cancer symptoms and noncancer symptoms. This suggests that one important way in which symptoms affect perceptions of disability and overall health is through their impact on functioning. However, beyond this mediating relationship, the regression findings also document that both continuing cancer and noncancer symptoms have significant, albeit weaker, direct effects on perceived disability. For self-rated health, continuing cancer-related symptoms has a statistically significant and somewhat stronger effect.

Implications and Conclusions

The key finding in this article is that both the noncancer and cancer-related factors play important roles in explaining health quality of life among older, long-term cancer survivors in terms of both perceptions of disability and self-rated health. However, the largest portion of these effects are indirect through the functional difficulties that older survivors report experiencing. Although the cancer-specific factors such as illness severity (as measured in this research by stage at diagnosis, number of treatment types) or the type of cancer type may not directly affect health quality of life, continuing symptoms and functional difficulties associated with cancer continue to be important even decades after diagnosis and treatment. Moreover, the research presented here suggests that cancer factors, especially continuing symptom, are important over and above the effects other illness survivors may have and the related functional difficulties they may produce.

With regard to the powerful role that functioning plays in predicting the more subjective health quality of life outcomes, perceived disability and self-rated health, it is important to note that these represent very different ways of viewing the impact that cancer has on older adults. The measurement of functioning is based on reports of very specific functional difficulties compared with the broader, subjective personal response to the illness. Treating them in this manner in the model allowed us to document the mediating role of functioning in understanding health quality of life that is an important part of understanding how specific cancer and noncancer health issues operate both directly and indirectly to affect these outcomes. For those geriatricians, geriatric nurses, and clinical gerontologists who work with cancer survivors, these results identify the need to be aware of the ways in which both cancer and noncancer illness factors work together in producing threats to health quality of life through the extent and nature of functional impairments.

The findings point to the importance of recognizing how the racial and gender linked health disparities found in the general population may be exacerbated by cancer-related factors. Especially important are differences in physical functioning that may affect perceptions of health quality of life. In the analysis presented here, the greater number of comorbidities and poorer functioning reported by African Americans and women have important implications for those who treat older minority cancer survivors. Our findings suggest that being an older, minority, cancer survivor in fact confers a “triple jeopardy.” These “at risk” groups may need special attention from health care and human service providers to ensure health quality of life after cancer.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This article was funded by The National Cancer Institute (NCI) Grant R01-CA-78975.

References

American Cancer Society. (2002). Cancer facts & figures 2002. Retrieved from http://www.uhmsi.com/docs/CancerFacts&Figures2002.pdf

Bowman

K. F.

Rose

J. H.

Deimling

G. T.

Kypriotakis

O’Toole

E. E.

(2010). Primary care physicians’ involvement in the cancer care of older long-term survivors. Journal of Aging and Health, 22, 673-686. doi:10.1177/0898264310373501

Braithwaite

Satariano

W. A.

Sternfeld

Hiatt

R. A.

Ganz

P. A.

Kerlikowske

. . . Caan

B. J.

(2010). Long-term prognostic role of functional limitations among women with breast cancer. JNCI: Journal of the National Cancer Institute, 102, 1468-1477. doi:10.1093/jnci/djq344

Catalá-López

Suárez-Pinilla

Valderas

J. M.

Gómez-Beneyto

Martinez

. . . Tabarés-Seisdedos

(2014). Inverse and direct cancer comorbidity in people with central nervous system disorders: A meta-analysis of cancer incidence in 577,013 participants of 50 observational studies. Psychotherapy and Psychosomatics, 83, 89-105. doi:10.1159/000356498

Darling

R. B.

Heckert

D. A.

(2010). Orientations toward disability: Differences over the lifecourse. International Journal of Disability, Development and Education, 57, 131-143. doi:10.1080/10349121003750489

Deimling

G. T.

Kahana

Bowman

K. F.

(2012). Adaptation and adjustment to cancer in later life: A conceptual model. In Bellizzi

K. M.

Gosney

(Eds.), Cancer and aging: Research and practice (pp. 365-384). Hoboken, NJ: Wiley-Blackwell.

Deimling

G. T.

Kahana

Bowman

K. F.

Schaefer

M. L.

(2002). Cancer survivorship and psychological distress in later life. Psycho-Oncology, 11, 479-494. doi:10.1002/pon.614

Dixon

(1960). Simplified estimation from censored normal samples. The Annals of Mathematical Statistics, 31, 385-391.

Ershler

(2006). Capecitabine use in geriatric oncology: An analysis of current safety, efficacy, and quality of life data. Critical Reviews in Oncology/Hematology, 58, 68-78. doi:10.1016/j.critrevonc.2005.08.006

10.

Ganz

P. A.

(2003). Why and how to study the fate of cancer survivors. European Journal of Cancer, 39, 2136-2141. doi:10.1016/S0959-8049(03)00489-1

11.

George

L. K.

Fillenbaum

G. G.

(1985). OARS methodology: A decade of experience in geriatric assessment. Journal of American Geriatric Society, 33, 607-615.

12.

Giltay

E. J.

Vollaard

A. M.

Kromhout

(2012). Self-rated health and physician-rated health as independent predictors of mortality in elderly men. Age and Ageing, 41, 165-171. doi:10.1093/ageing/afr161

13.

Greenwald

H. P.

Dirks

S. J.

Borgatta

E. F.

McCorkle

Nevitt

M. C.

Yelin

E. H.

(1989). Work disability among cancer patients. Social Science & Medicine, 29, 1253-1259. doi:10.1016/0277-9536(89)90065-8

14.

Hardy

M. A.

Acciai

Reyes

A. M.

(2014). How health conditions translate into self-ratings: A comparative study of older adults across Europe. Journal of Health and Social Behavior, 55, 320-341. doi:10.1177/0022146514541446

15.

Hastings

Mosteller

Tukey

Winsor

(1947). Low moments for small samples: A comparative study of order statistics. The Annals of Mathematical Statistics, 18, 413-426.

16.

Havlik

R. J.

Yancik

Long

Ries

Edwards

(1994). The National Institute on Aging and the National Cancer Institute SEER collaborative study on comorbidity and early diagnosis of cancer in the elderly. Cancer, 74(Suppl. 7), 2101-2106.

17.

Hewitt

Rowland

J. H.

Yancik

(2003). Cancer survivors in the United States: Age, health, and disability. The Journals of Gerontology, Series A: Biological Sciences & Medical Sciences, 58(1), M82-M91. doi:10.1093/gerona/58.1.M82

18.

Iezzoni

L. I.

McCarthy

E. P.

Davis

R. B.

Siebens

(2000). Mobility problems and perceptions of disability by self-respondents and proxy respondents. Medical Care, 38, 1051-1057.

19.

Kelley-Moore

J. A.

Schumacher

J. G.

Kahana

(2006). When do older adults become “disabled”? Social and health antecedents of perceived disability in a panel study of the oldest old. Journal of Health and Social Behavior, 47, 126-141. doi:10.1177/002214650604700203

20.

Krok-Schoen

J. L.

Palmer-Wackerly

A. L.

Dailey

P. M.

Wojno

J. C.

Krieger

J. L.

(2017). Age differences in cancer treatment decision making and social support. Journal of Aging and Health, 29, 187-205.

21.

Land

K. C.

Yang

(2011). Morbidity, disability, and mortality. In Binstock

R. H.

George

L. K.

Cutler

S. J.

Hendricks

Schulz

J. H.

(Eds.), Handbook of aging and the social sciences (pp. 41-58). New York, NY: Academic Press.

22.

Langlois

J. A.

Maggi

Harris

Simonsick

E. M.

Ferrucci

Pavan

. . . Enzi

(1996). Self-report of difficulty in performing functional activities identifies a broad range of disability in old age. Journal of the American Geriatrics Society, 44, 1421-1428. doi:10.1111/j.1532-5415.1996.tb04065.x

23.

Litwin

(2003). The association of disability, sociodemographic background, and social network type in later life. Journal of Aging and Health, 15, 391-408.

24.

Mohan

Beydoun

H. A.

Beydoun

M. A.

Barnes-Eley

Davis

Lance

Schellhammer

(2011). Self-rated health as a tool for estimating health-adjusted life expectancy among patients newly diagnosed with localized prostate cancer: A preliminary study. Quality of Life Research, 20, 713-721. doi:10.1007/s11136-010-9805-3

25.

Nagi

(1976). An epidemiology of disability among adults in the United States. Milbank Memorial Fund Quarterly, 92, 101-103.

26.

Nápoles

A. M.

Ortíz

O’Brien

Sereno

A. B.

Kaplan

C. P.

(2011). Coping resources and self-rated health among Latina breast cancer survivors. Oncology Nursing Forum, 38, 523-531. doi:10.1188/11.ONF.523-531

27.

Ogle

K. S.

Swanson

G. M.

Woods

Azzouz

(2000). Cancer and comorbidity: Redefining chronic diseases. Cancer, 88, 653-663.

28.

Piccirillo

J. F.

(2000). Importance of comorbidity in head and neck cancer. The Laryngoscope, 110, 593-602. doi:10.1097/00005537-200004000-00011

29.

Porter

K. E.

(2013). “Chemo brain”—Is cancer survivorship related to later-life cognition? Findings from the health and retirement study. Journal of Aging and Health, 25, 960-981. doi:10.1177/0898264313498417

30.

Ramsey

S. D.

Andersen

M. R.

Etzioni

Moinpour

Peacock

Potosky

Urban

(2000). Quality of life in survivors of colorectal carcinoma. Cancer, 88, 1294-1303. doi:10.1002/(SICI)1097-0142(20000315)88:6<1294::AID-CNCR4>3.0.CO;2-M

31.

Romoren

T. I.

Blekeseaune

(2003). Trajectories of disability among the oldest old. Journal of Aging and Health, 15, 548-566. doi:10.1177/0898264303253633

32.

Rowland

J. H.

Yancik

(2006). Cancer survivorship: The interface of aging, comorbidity, and quality care. Journal of the National Cancer Institute, 98, 504-505. doi:10.1093/jnci/djj154

33.

Schootman

Homan

Weaver

K. E.

Jeffe

D. B.

Yun

(2013). The health and welfare of rural and urban cancer survivors in Missouri. Preventing Chronic Disease, 10, E152. doi:10.5888/pcd10.130052

34.

Sergeant

J. F.

Ekerdt

D. J.

Chapin

R. K.

(2010). Older adults’ expectations to move: Do they predict actual community-based or nursing facility moves within 2 years? Journal of Aging and Health, 22, 1029-1053. doi:10.1177/0898264310368296

35.

Short

P. F.

Vasey

J. J.

BeLue

(2008). Work disability associated with cancer survivorship and other chronic conditions. Psycho-Oncology, 17, 91-97. doi:10.1002/pon.1194

36.

Tammemagi

C. M.

Nerenz

Neslund-Dudas

Feldkamp

Nathanson

(2005). Comorbidity and survival disparities among black and white patients with breast cancer. The Journal of the American Medical Association, 294, 1765-1772. doi:10.1001/jama.294.14.1765

37.

Yancik

(2001). Effect of age and comorbidity in postmenopausal breast cancer patients aged 55 years and older. The Journal of the American Medical Association, 285, 885. doi:10.1001/jama.285.7.885

38.

Zola

I. K.

(1993). Self, identity and the naming question: Reflections on the language of disability. Social Science & Medicine, 36, 167-173. doi:10.1016/0277-9536(93)90208-L