What is new?
- •
This large study of community-dwelling elderly examined 1-year change in both self-rated health and objective comorbidity, as measured by the Charlson comorbidity score. Fractional polynomials were used to model nonlinearity in the association between comorbidity change and self-rated health worsening.
- •
Self-rated health was responsive to changes in Charlson comorbidity. However, the relationship was nonlinear and was moderated by interactions involving age and baseline comorbidity level.
- •
Younger, healthier individuals were most likely to lower their health perceptions in response to new or worsening chronic conditions. Older individuals and those with greater preexisting comorbidity were less likely to further reduce their health perceptions following additional comorbidity change.
- •
Examination of specific Charlson diagnosis categories indicated that self-rated health was most likely to decline after new diagnoses for metastatic tumors, paralysis, and dementia.
An impressive array of work spanning several decades has demonstrated that a single question about self-rated health has a remarkable ability to predict mortality. This relationship has been borne out in numerous populations with follow-up times ranging from as short as 30 days to as long as 28 years [1], [2], [3]. In many studies, the ability of self-rated health to predict mortality has persisted even after adjustment for more objective health measures [4], [5], [6]. Self-rated health has also been shown to predict health-care utilization, functional ability, and recovery from illness [7], [8], [9], [10].
Despite these well-documented relationships, many questions remain about what aspects of health are captured in self-ratings and how people's health perceptions affect their responses to future illness [11]. One area in which data are still relatively sparse relates to how older adults modify their health perceptions in response to changes in objective health status. Understanding the factors involved in such change is important, as recent work suggests that short-term changes in self-rated health predict mortality better than a single baseline assessment [12], [13], [14].
Although many cross-sectional studies have compared objective and subjective health measures, relatively few studies have examined longitudinal relationships between these domains, and their methodologies and findings have been mixed. Several studies with follow-up times of 1–3 years have reported significant associations between self-reported new or worsening medical conditions and declines in self-rated health [15], [16]. A number of studies of longer duration have also found significant associations among various objective health measures and changing health perceptions [17], [18], [19], and several studies have also reported associations between hospitalization or other acute medical events and self-rated health [20], [21]. However, results of other studies suggest that the degree to which individuals alter their self-ratings of health after new illnesses or medical events may be influenced more by longstanding health perceptions than by the direct impact of the new illness [22], [23].
The interpretation of the collective findings of prior research is complicated by variability in the measurement of objective health. Although some studies have used physician reports or other clinical measures, objective health has frequently been assessed through self-report using condition checklists. A potentially useful alternative approach is offered by the combined use of administrative health services data and comorbidity indices, which are used extensively in epidemiological studies. The most frequently used comorbidity measure, the Charlson comorbidity index, is a summation of chronic conditions, which are weighted for severity based on their previously observed association with 1-year mortality [24]. Although the original Charlson index and administrative record-based modifications [25], [26] were first developed and validated with hospital inpatient data, more recent studies have demonstrated that incorporating outpatient claims as an additional data source improves comorbidity measurement, and the Charlson is increasingly used in outpatient research settings [27], [28], [29].
Despite its widespread use in epidemiological and health services research, only two previously published studies have explored the association of the Charlson index with self-rated health. One cross-sectional study of women with breast cancer reported a significant correlation between Charlson comorbidity and self-rated health [30]. A second cross-sectional study found that hypertensive veterans who rated their health optimistically compared with their Charlson scores had better health behaviors and greater perceived control of their hypertension [31]. To our knowledge, no prior studies have examined changes in self-rated health within the context of Charlson comorbidity change. Because the Charlson index was validated to maximally predict mortality from clinical data, understanding the impact of Charlson comorbidity change on health perceptions may provide new insights into the factors that mediate the relationship between self-rated health and mortality.
The goal of the present study was to examine longitudinal associations between self-rated health and Charlson comorbidity in a large sample of U.S. elderly by linking survey data with Medicare health services claims. Specifically, this study examined how 1-year change in Charlson comorbidity affected the probability of decline in self-rated health. These associations were explored in two ways: (1) by examining the association of change in total Charlson score, as an overall measure of individuals' disease burden, with worsened self-rated health and (2) by examining changes in the occurrence of specific diagnoses that constitute the Charlson index, to determine which conditions have the greatest effect on health perceptions.