HEALTH has traditionally been defined as the absence of illness in the United States. Recently, however, the focus has shifted to consideration of other aspects of health, such as functional status and global self-perceptions. This shift is supported by research findings that self-assessed health status is a strong predictor of morbidity and mortality, even when risk factors, the number of comorbidities, or health behaviors are controlled in the analysis.1,2
This evolution in our understanding of what health means has profound implications for people with disabilities because a definition of health that encompasses factors other than simply the absence of illness or disability raises the possibility that anyone, regardless of his or her disabling conditions, can be healthy. The purpose of this article is to explore how health has been defined and measured for people with disabilities. The article discusses the limitations of traditional measurements and concludes with recommendations to improve assessment of health status among people with disabilities. But first, consider how the definition of health has evolved.
THE MULTIDIMENSIONAL NATURE OF HEALTH
Sociologist Lerner3 asserted that chronic disease became a "recognizable entity in its own right" as mortality rates declined in the early 20th century. Medical advances have prolonged life so that a person can live with a chronic condition for many years.
In 1974, the World Health Organization defined health as a multidimensional construct, encompassing physical, mental, and social well-being, not just the absence of disease or infirmity.4 Lerner3 argued that health status may contain the same components for everyone but different weights may be assigned to these components based on the individual's situation. While measuring mortality is straightforward, because it reflects a clearly defined discrete event, measuring morbidity is more challenging. He argued that one's definition of health depends on the purpose of defining it, which might be to measure the outcome of health interventions, to describe the health of a community, to determine resource allocation, or to compare variables associated with health and health behaviors. The purpose will drive both the content and structure of the resulting measure. As McDowell and Newell explicate in the following quotation, our concept of health and its measurement are inexorably linked: "Just as language molds the way we think, our health measurements influence (and are influenced by) the way we define and think about health."5(p14)
THE EVOLVING CONCEPT OF HEALTH FOR PEOPLE WITH DISABILITIES
While past research on people with disabilities focused heavily on healthcare utilization and secondary medical conditions, recent research has begun addressing well-being and health promotion.6 In 2001, the World Health Organization General Assembly adopted the International Classification of Functioning, Disability, and Health.7 This classification separates disability from health status and examines the relationship between activity limitations and barriers encountered in the environment. Thus a person with a disabling condition can be considered "healthy." This system also recognizes participation and inclusion in society as a critical part of health. The increased emphasis on the environment's contribution to disability challenges the traditional "functional limitations paradigm," which treats disability as restriction of "normal" human activity and ignores other sources of disablement.8
Rimmer9 argued that health is multifaceted and dynamic, shifting back and forth throughout our lives. The only difference for people with disabilities is that they may begin at the lower end of the health continuum, because of secondary conditions that result from their initial disabling condition. Consequently, they have as much, if not more, need for health promotion as does the general population, because of their susceptibility to secondary conditions.
DISABILITY AND HEALTH FROM THE PUBLIC HEALTH PERSPECTIVE
In Healthy People 2010, the nation's health agenda specifically addressed health for people with disabilities for the first time.10 In a 2000 symposium held to propose strategies to achieve the Healthy People 2010 Disability and Secondary Condition objectives, a number of recommendations were made that have relevance for the measurement of health and health promotion among people with disabilities. First, disability should be treated as a demographic descriptor rather than as a health outcome. As a descriptor, it can be used to monitor progress toward eliminating disparities in health outcomes. Second, Disability-Adjusted Life-Years, which assumes that disability reflects negative health status, should be eliminated. Third, we must recognize that disability is not a static phenomenon but varies by developmental phase, point in time, environmental context, and type of disability. In other words, individuals with disabilities are a heterogeneous group, and their health self-ratings will reflect this heterogeneity.
People with disabilities generally report poorer health than do people without disabilities.9,11,12 Drum13 used data from the 1998 Behavioral Risk Factor Surveillance System Survey aggregated across 14 US states to explore differences in health perceptions between people with disabilities and those without disabilities. The telephone survey is conducted yearly to a random sample of households in the United States. Sixteen percent of the sample (n = 8,455) was considered disabled on the basis of their indication that an impairment or health problem limited their activities. These respondents reported lower general health status and fewer healthy days in the past month than did people without disabilities. Furthermore, within each descriptive health category, those with disabilities reported fewer healthy days than those without disabilities. Among those respondents who rated their health "excellent," for example, people with disabilities reported in the last month on average 5 days of poor physical health, 3 days of poor mental health, and 6 days when poor health prevented usual activities. During the same period of time, nondisabled respondents had 1 poor physical health day, 1 poor mental health day, and 1 day when health problems prevented usual activities. At the other end of the continuum, persons with disabilities who described their health as poor reported in the last month an average of 18 days when health problems prevented usual activities compared with 10 days for nondisabled respondents who rated their overall health as poor. Drum concluded that compared with nondisabled persons, individuals with disabilities and chronic health conditions may consider additional factors in evaluating their health status. Therefore, concepts of health and disability may need to encompass factors other than poor physical and mental health, as measured in the Behavioral Risk Factor Surveillance System, to capture people's alternative definitions of health. Krahn14 has suggested that a broader conceptualization may include psychological and social constructs such as self-esteem, spirituality, and participation in one's environment.
UNCOVERING THE MEANING OF HEALTH
McDowell and Newell5 have stated that subjective measures of health can be grouped into 3 major categories: global ratings of well-being, measures of symptoms of illness, and functional status measures. Researchers have used cognitive interviewing to explore the meaning of global health ratings. Krause and Jay15 instructed 158 individuals to rate their health as excellent, good, fair, or poor, and then asked them to explain their rating. Nine conceptually meaningful categories were identified from their responses. While the presence or absence of health problems was the most frequent response, many interviewees also cited physical functioning, physical condition, and health behaviors as the rationale for their global health rating. The specific referents used varied by respondents' age, education, and race/ethnicity, but not by their gender. Consistent with previous findings, those who compare their health with that of others when making a global health rating tend to make downward social comparisons (ie, seeing others' health as worse off than theirs). As might be expected, the majority of those who mentioned health problems rated their health as "fair" or "poor." However, 47% of those who used negative health behaviors when explaining their health rating had rated their health status as "good." Krause and Jay concluded that the variation in health referents people use when determining health ratings suggest that multiple factors explain how health constructs are translated into summary health ratings.
Idler and Benjamini1 have speculated about why global health self-ratings are such a strong predictor of survival, and their hypotheses have implications for the meaning of self-reported health in people with disabilities. They suggested that health self-ratings may interact with health behaviors, such that those who consider their health poor may subsequently engage in fewer health-promoting behaviors, thus making health rating a self-fulfilling prophesy. Idler and Benjamini also speculated that health self-ratings may reflect the presence or absence of resources that can improve or diminish health. When making a global self-rating, the individual is able to consider information not easily elicited in measures that focus just on risk factors or comorbidities when evaluating health status.
Researchers have begun to interview people with disabilities about the meaning of health to them. On the basis of 19 focus groups with people with disabilities, Powers16 identified 4 major themes about health and wellness: the ability to function and the opportunity to do what you want, independence and self-determination, an interrelated physical and emotional state of well-being, and being unencumbered by pain. When asked how health differs for people with and without disabling conditions, these participants stressed the importance of resilience and experiencing health as distinct from their disability. They also spoke about the additional challenges people with disabilities often face around preventing secondary conditions, such as pain or infection. They perceived that the larger societal environment plays a major part in their health; the stronger the net of society and health system links that support their health promotion, the higher their perceived health and wellness.
Functional status was also a major theme in interviews with 44 individuals receiving services from Independent Living Centers.17 Approximately half of these interviewed described being healthy in multiple ways, but most responses could be categorized in functional terms. For example, one woman stated that being healthy meant being "able to take care of myself, take care of my family, keep my home, and live as I did before I lost my sight." While approximately one third mentioned the absence of pain or illness, few respondents referred specifically to their disability. Interesting differences in responses to close-ended health questions were also noted among the larger sample of 135 individuals from the Independent Living Centers. While 73% of the sample rated their current health as good or excellent, 74% indicated that their health status interfered to at least some extent with doing what they wanted.
Nosek et al18 interviewed 18 middle-aged women with physical disabilities. Their analysis revealed that functional capacity was a key component of women's health conception; the criterion for "being sick" was the inability to conduct their normal life activities. Their illness definitions emphasized acute, temporary conditions, including a remission/recurrence of their primary disabling condition, or manifestation of secondary or coexisting conditions. By contrast, being healthy meant feeling in control and able to fulfill social roles and was strongly influenced by mental status. "For individuals whose disability is stable, it becomes a context within which they function, and illness is defined in terms of a departure from that normal level of functioning."18(p17) Individuals whose disabling condition was relapsing or degenerative tended to incorporate weakness, pain, or fatigue when defining their health. Therefore, the nature of the disabling condition influences the individual's definition of health. While these women acknowledged a "ceiling" for their health and wellness imposed by their physical limitations, they clearly did not equate disability and health.
I tested Nosek's contention that the nature of the disabling condition influences one's definition of health by contrasting the health self-ratings of 1,730 individuals who have post-polio syndrome (PPS) with 423 individuals who were polio survivors, but not experiencing post-polio syndrome (non-PPS). As previously reported, the average age of both largely female groups was 62 years old.19 Those with PPS had significantly lower total scores on the Philadelphia Geriatric Center Multi-Level Assessment Instrument 4-item summated health self-rating20 than those without the syndrome. On each of the items, the PPS group rated its health status more negatively. The 2 items with the greatest difference between groups concerned health problems standing in the way and comparison of health with others. Half of the PPS group, compared with 30% of the non-PPS group, indicated that its health problems stood a great deal in the way. Sixty-five percent of the PPS group, compared with 38% of the non-PPS group, rated its health as not as good as most people their age. This finding illustrates that simply examining health status by general diagnosis may obscure differences among individuals who had the same initial disabling condition, but a different symptom history.
QUANTITATIVE MEASURES OF HEALTH STATUS
These studies confirm that people with disabilities consider functional status when assessing their own health. However, their definitions of functional status may differ from how it is operationalized in many existing health measures. Measures of daily living activities are often used to assess functional status. These scales typically include items that focus on skills such as feeding, dressing, housework, and getting around in the community. Individuals are asked whether they can perform these activities with or without assistance. These items may not tap the activities of most concern to many people with disabilities. For example, a person with cerebral palsy who needs personal assistance to carry out basic daily activities may be more concerned about the ability to participate in preferred community activities than the ability to dress independently when self-assessing health status. Furthermore, scores on these scales may be confounded with the way that the presence of disability itself is measured. For example, the National Health Interview Survey asks the following questions:
1. Are you limited in any way in any activities because of physical, mental, or emotional problems?
2. Do you have any health problem that requires you to use special equipment, such as a cane, a wheelchair, a special bed, or a special telephone?
Because many functional assessment measures also ask about limitations in activities of daily living or needed assistance to perform these activities, these 2 sets of questions are getting at the same thing. Allen et al observed a similar confound with conventional gender roles when measuring functional status.21
To investigate how health status changes across time for people with disabilities, a secondary analysis was conducted with data from Stuifbergen's 8-year longitudinal study of 421 people with multiple sclerosis (MS). Recruited from the National MS Society chapters, the sample was 84% female and 94% Caucasian, and had an average age of 48 years. Despite the expected decline in health for people with MS, health status remained remarkably similar for the group as a whole. Approximately 10% of the sample rated their health as "poor" at the beginning of the 8-year study, and 9% rated their health "poor" at the end. Nine percent rated their health as "excellent" at the first data-collection period, and 12% rated their health as "excellent" 8 years later. However, many individuals did shift from 1 health category to another. The majority of those who initially rated their health as "poor" rated their health as "fair," "good," or "excellent" on at least one of the subsequent data-collection periods. Approximately half the respondents who rated their health as "fair" or "excellent" also changed their health rating from baseline. The results suggest that among individuals with a chronic health condition, self-ratings of health status may be quite variable across time.
CRITIQUE OF EXISTING HEALTH MEASURES FOR PEOPLE WITH DISABILITIES
Froehlich-Grobe22 asserted that the SF-36, a commonly used measure of health-related quality of life, may not assess the construct it is intended to measure in people with disabilities or be sensitive to differences in their experiences. In particular, the physical functioning subscale has not allowed people with mobility impairments to consider their adaptive devices when answering health questions. Meyers and Adresen23 developed the SF-36E (ie, enabled) by altering 6 of the 10 items. "Go" was substituted for "walk" and "climb." The examples of vigorous physical activity substituted "wheeling" for "running." The stem for the scale was expanded to instruct the respondent to consider "using your normal assistive devices" when considering how health is limited in the 10 listed activities. Froehlich-Grobe22 tested the SF-36E with 158 individuals with physical disabilities, including 70 wheelchair users. The total group reported between 9% and 23% better performance on the 6 Physical Function items, which had been altered, and wheelchair users demonstrated the greatest change in scores. Internal consistency was slightly lower on the SF-36E, particularly for wheelchair users, than on the SF-36 Physical Function Scale. On most of the other scales, where items had not been altered, the internal consistency reliabilities were similar for both versions of the scale. However, on the Mental Health and Vitality Scales, the internal consistencies were somewhat higher on the SF-36E. A different factor structure was found in this sample for both the SF-36 and SF-36E than the 8-factor structure previously reported for the SF-36 in the general population. The item clusters were also different for the wheelchair and nonwheelchair users. These findings tend to support the contention that health-related quality of life may have a meaning for people with physical disabilities different from that for other SF-36 users. Froehlich-Grobe and her colleagues also found that less than 6% of the General Health subscale items for their total sample (9% for wheelchair users) revealed a floor or ceiling effect on either the SF-36 or SF-36E, well below the proposed 15% cutoff used to indicate inadequate measurement. Internal consistency reliability was above 0.80. These findings support the use of the general health subscale among people with disabilities, although their factor analysis results suggest a different factor structure for people with disabilities than for the SF-36 norm group.
Looking more generally at health outcomes research with people with disabilities, Andresen and Meyers24 cited numerous ways in which they have been disadvantaged in research studies. First, standard sampling methods-particularly the typical random digit dialing survey methodology used in many population-based studies-could miss people with disabilities, who may have difficulty answering the telephone in time, or use Telecommunication Services for the Deaf. Population-based studies may also overlook people who are institutionalized, such as people living in state hospitals or state schools. Some people may have difficulty completing written surveys, yet alternative administration modes are rarely offered. Finally, the concepts and wording of some health surveys are alien to the experiences of people with disabilities, and in some cases may be offensive to them. For example, Meyers and Andresen related that a number of people with mobility impairments became angry and hung up on them when they tried to administer the SF-36 by phone because they could not "walk" or "climb," as stated in some of the Physical Functioning items. Questions that solicit ratings of health within the past year or past month may have little meaning to people with relapsing or remitting health conditions, such as MS, or to individuals whose short-term memory is impaired.
As a result of such disadvantages, Meyers and Andresen23 urged researchers to include people with disabilities in the design of future research and to modify standard research practices to make them accessible and acceptable to people with disabilities. They acknowledged that such efforts will entail substantial work, as we seek to determine the psychometric properties of revised instruments and convince funders that the extra expense required to include people with disabilities in "mainstream" research is worth the price. To begin this process, Andresen25 proposed the following criteria for evaluating health outcomes measures for people with disabilities:
* Conceptual: Reflects ICIDH-2* interactive model of impairment & environment
* Norms: Includes standard comparative data from people with disabilities
* Measurement model: Appropriate model; does not evidence severe ceiling/floor effects
* Item/instrument bias: Data support lack of bias for people with disabilities
* Burden: Brief, easy to administer, acceptable to people with disabilities
* Reliability: Consistency of response, particularly when proxies are used
* Validity: Adequate convergent/discriminant validity for people with disabilities
* Responsiveness: Sensitive to meaningful change over time for people with disabilities
* Accessible forms: Alternative administration ways show no mode effects for people with disabilities
* Cultural adaptations: Adequate testing in subgroups; validated alternative language formats available
DISCUSSION AND RECOMMENDATIONS
People with disabilities generally rate their health lower than those without disabilities on a variety of different measures; it is important to go beyond this simple finding to understand why this might be so. Rimmer9 has suggested that people with disabilities start at the lower end of the health continuum, which is a shifting dynamic process for everyone. Obviously, functional status measures that do not give people "credit" for the adaptations they make to achieve key functional outcomes will lower their scores on certain measures. Idler and Benjamin1 have speculated that health self-ratings may reflect the presence or absence of resources that can improve or diminish health. If people with disabilities lack the environmental supports to take care of their health, these impediments may contribute to their lower self-ratings of health.
Developers of health measures are encouraged to consider whether their measures may be expanded to consider environmental supports for staying healthy and avoiding secondary conditions. At the same time, it will be important to determine whether any modifications made enhance rather than detract from the reliability, validity, and sensitivity of these measures.
Because the answers we get about health status clearly depend on the questions we ask, researchers are encouraged to ensure that the measures they use clearly reflect what they want to know. If we want to understand how individuals' functional limitations or health problems affect their health, then measures that focus respondents' attention on these issues are appropriate. If, however, we want to understand individuals' self-perceived health status, asking a global question enables them to weigh the various components of health in ways that make sense to them.
Interview studies of people with disabilities have provided important contextual information about the meaning of health for them. The findings suggest that they may consider a broader range of factors when defining their health than do nondisabled individuals. The cognitive interviewing process described by Krause and Jay15 should be used to determine the pattern of relationships between global health self-ratings and explanatory themes among people with disabilities.
As a number of researchers have asserted, much work remains to be done to determine whether health measures developed for the general population are appropriate for people with disabilities. Research with people with disabilities is often plagued with small sample sizes. It is hoped that the increased attention to the health status of people with disabilities, particularly in the public health community, will lead to larger population-based studies. The heterogeneity of this population on many variables contributes to increased variance in the denominators of many statistical analyses. Only with adequate sample sizes will we have the statistical power to appropriately determine the psychometric properties of our measuring instruments.
CONCLUSIONS
We must recognize that individuals with disabilities and chronic health conditions are a heterogeneous group. People with the same diagnosis may have varying levels of disability, often depending on their environments. In addition, health is a dynamic construct, which can change because of exacerbations of the underlying condition, new health challenges, and their interplay with the environment. Therefore, what people focus on when asked to assess their health will vary from person to person and within the same person over time.
People with disabilities have historically not been at the table when key decisions were made about how disability and health were to be conceptualized and measured, often to their disadvantage. While their exclusion appears to be diminishing, we must work to ensure that how we measure health for people with disabilities authentically reflects their experiences.
REFERENCES