Double Disadvantage in the Process of Disablement: Race as a Moderator in the Association Between Chronic Conditions and Functional Limitations

Double Disadvantage in the Process of Disablement: Race as a Moderator in the Association Between... Abstract Objectives This study evaluated (a) whether the association between chronic conditions and functional limitations vary by race/ethnicity, and (b) whether socioeconomic status accounted for any observed racial variation in the association between chronic conditions and functional limitations. Method The Health and Retirement Study data were used to assess whether race/ethnicity moderated the association between chronic conditions and functional limitations, and whether education, income, and/or wealth mediated any of the observed moderation by race/ethnicity. Results Results from structural equation models of latent growth curves with random onset indicated that (a) the positive association between chronic conditions and functional limitations onset was larger for African Americans and Hispanics than it was for Whites, but (b) this difference largely persisted net of socioeconomic status. Discussion African Americans and Hispanics endure a multiplicative double disadvantage in the early stages of the disablement process where they experience (a) a more rapid onset and higher levels of functional limitations, and (b) greater risk of functional limitation onset associated with chronic conditions compared to their White counterparts. Moreover, basic economic policies are unlikely to curtail the greater risk of functional limitations onset associated with chronic conditions encountered by African Americans and Hispanics. Disability, Functional health status, Minority aging (race/ethnicity), Socioeconomic status In later life, there tend to be significant racial/ethnic disparities across a number of health outcomes. Among these, one particularly concerning outcome is functional limitations. Racial and ethnic minorities tend to be more functionally limited than White Americans. African Americans and Hispanics are generally at greater risk of becoming functionally limited (Dallo, Booza, & Nguyen, 2015; Haas & Rohlfsen, 2010) and tend to suffer from more functional limitations (Haas & Rohlfsen, 2010; Ostchega, Harris, Hirsch, Parsons, & Kington, 2000). In short, there are clear racial/ethnic disparities in functional limitation outcomes. Although racial disparities in functional limitation outcomes are relatively well established, functional limitations are situated within the broader context of the process of disablement. The disablement process model provides a conceptual understanding for the interrelations between illness/disease, limitations in functional mobility, disability, and, ultimately, mortality (Clarke & George, 2005; Lawrence & Jette, 1996; Verbrugge & Jette, 1994). However, it remains unclear whether there are racial/ethnic differences in various interrelationships within disablement processes. This omission is important because, if there are indeed racial/ethnic differences in relationships within the disablement process, this suggests interventions targeting racial differences in later stage processes (e.g., functional limitations) may be too late to reduce racial disparities in health, and targeting early stage processes (e.g., chronic disease) may be more appropriate. Additionally, racial minorities may experience a kind of double disadvantage wherein they experience elevated levels of chronic disease, and the deleterious association between chronic disease and subsequent functional limitations is stronger. However, this possibility remains heretofore untested. Furthermore, if there are indeed racial differences among the interrelationships defined within the disablement processes model, it is also important to ask why they exist. One prevailing framework for understanding health disparities is fundamental cause theory (Link & Phelan, 1995). In its original formulation, fundamental cause theory postulates that differences in flexible resources conveyed through socioeconomic status (SES) are largely responsible for disparities in health outcomes. Because racial minorities tend to have access to fewer socioeconomic resources, differences in SES may go a long way in helping explain racial disparities in health (Phelan & Link, 2015). However, it remains unclear whether SES does indeed underpin racial disparities in the interrelationships defined within the disablement processes model. To that end, we draw on the disablement process model and fundamental cause theory to address the following two research questions: (a) does the association between chronic conditions and functional limitations vary by race/ethnicity; and (b) if so, is this variation explained by differences in SES. We test these questions using the 1994 to 2012 waves of the Health and Retirement Study (HRS), using latent growth curves with random onset (LGCRO) in a structural equation modeling (SEM) framework. We focus on the association between chronic conditions and functional limitations because this is (a) after people have been diagnosed with a chronic disease, so they have begun an observable process of disablement, but (b) early enough in the disablement process that interventions may be useful. Background Racial Differences in Function Limitations On average, African Americans are both more likely to experience functional limitations, and to experience more severe levels of functional limitation than do Whites in later life (Haas & Rohlfsen, 2010; Kail & Taylor, 2014; Melvin, Hummer, Elo, & Mehta, 2014; Warner & Brown, 2011). Similar disparities are documented between Hispanics and their White counterparts as well. For instance, among U.S.-born adults age 54 to 65, 44% of Whites suffer from one or more functional limitations, compared to 60% of African American, and 57% of Hispanics (Dunlop, Song, Lyons, Manheim, & Chang, 2003). Clearly, African Americans and Hispanics are, on average, more functionally limited than non-Hispanic Whites. Although understanding racial disparities in gross measures of functional limitations (i.e., overall level) provides important information about health disparities, more information can be derived from disaggregating functional limitations into its various temporal parts (Haas & Rohlfsen, 2010; Taylor, 2008). Functional limitations involve three distinct components: (a) the risk of onset; (b) initial level at onset; and (c) subsequent levels (or trajectories) after initial level (for review see Taylor, 2008). Although African Americans and Hispanics have comparable accumulation of functional limitations compared to Whites after becoming functionally limited (Warner and Brown, 2011 do find that Black women experience larger growth in functional limitations over time relative to White men), both African Americans and Hispanics are at greater risk of experiencing functional limitation onset and experience higher initial levels of functional limitations than do Whites (Kail & Taylor, 2014; Warner & Brown, 2011). Thus, when considering racial differences in functional limitations, it is important to separately consider racial differences in risk of onset, initial level, and subsequent trajectories after onset. Disablement Process According to the disablement process model, disablement begins with illness/disease, which then leads to impairments in basic physical activities (generally referred to as functional limitations). These limitations may then accumulate over time and progress into more severe disablement which substantially interferes with daily life (i.e., activities of daily living; Jette, 2006; Lawrence & Jette, 1996; Verbrugge & Jette, 1994). Although much of the disablement research focuses on outcomes within the last stage of disablement because they are most burdensome, the disablement process model itself views the pathway from chronic conditions to functional limitations as particularly important because this is (a) after people have been diagnosed with a chronic disease, so they have begun the process of disablement, but (b) early enough in the disablement process interventions may be effective (Kail & Carr, 2017; Lawrence & Jette, 1996). The disablement process model highlights that the rate at which functional limitations accumulate over time is often conditional upon certain risk factors (Braungart Fauth, Zarit, Malmberg, & Johansson, 2007; Verbrugge & Jette, 1994). These risk factors include factors that precede the start of the disablement process (Verbrugge & Jette, 1994). Race may be one such factor along which the process as a whole is likely patterned. However, although the association between race/ethnicity and disablement outcomes has been studied extensively, the degree to which there are racial differences in interrelationships within the disablement process has not. As mentioned earlier, the latter is important because if there are indeed racial/ethnic differences in the disablement process, this suggests interventions targeting racial/ethnic differences in later stages of the process (e.g., functional limitations) may be too late to help level racial disparities in functional limitations, and targeting early stage processes (e.g., chronic disease) may be more appropriate. There are several reasons to expect interrelationships within the disablement process vary base race. First, racial minorities are less likely than Whites to receive quality care in general (Fiscella, Franks, Gold, & Clancy, 2000), and poorer quality of care than Whites among those enrolled in Medicare managed care health plans (Schneider, Zaslavsky, & Epstein, 2002). Racial minorities are also less likely to be screened for chronic diseases, as well as receive different types of prescriptions and smaller supply of medication to treat chronic diseases (Dominick, Dudley, Grambow, Oddone, & Bosworth, 2003; Goel et al., 2003). These differences persist net of access to care and SES. Second, more recently, Phelan and Link (2015) have argued racial inequalities in health may be rooted in the combination of racial differences in SES, and—more important—the unique experience of racism. Therefore, the experience of racism itself may be a fundamental cause of racial health disparities (Phelan & Link, 2015). Because of (a) the experience of various forms of racism and racial discrimination (Krieger, 2014; Williams & Mohammed, 2013), (b) the associated stress and psychological consequences of racism and discrimination (Thoits, 2010; Williams, Neighbors, & Jackson, 2003), (c) the reciprocal relationship between psychological and physical health (Gayman, Pai, Kail, & Taylor, 2013; Gayman, Turner, & Cui, 2008), and (d) healthcare providers’ racial stereotypes and biases play a large role in the racial differences in quality of care that non-Whites receive (Nelson, Stith, & Smedley, 2002), it is likely, not only will minorities experience more frequent chronic conditions, but the impact of conditions on functional limitations will be amplified. Third, there is some evidence aspects of the disablement process vary by race. For instance, prior research shows the influence of chronic conditions and functional limitations on more severe, late stage outcomes (i.e., activities of daily living limitations) is greater for African Americans and Hispanics compared to Whites (Zsembik, Peek, & Peek, 2000). This suggests the processes leading up to later stage disability vary by racial group membership. However, previous findings bypass a critical component of the disablement process where initial disablement often emerges (i.e., functional limitations). Findings from other research (Haas & Rohlfsen, 2010; Taylor, 2008) suggest focusing on advanced stages of disablement alone may be too late in the process to speak to effective interventions in slowing the progression of disability. As such, we are unaware of any research on racial differences in the ways in which chronic conditions are associated with subsequent functional limitations, and how they persist or grow over time, particularly in an onset, level, and growth model framework. This is an important omission because the pathway from chronic conditions to functional limitations is a good target for interventions as it occurs earlier in the disablement process and, thus, it makes it possible to potentially slow the development of disability before it becomes too late to effectively intervene (Lawrence & Jette, 1996). Thus, for medical and social interventions to help address any racial/ethnic disparities in functional limitations, it is likely they would be most efficacious if they target racial differences in how chronic conditions lead to subsequent functional limitations. Therefore, based on (a) theoretical insights from the disablement process model and (b) empirical research on racial/ethnic differences in functional limitations, and (c) the limited research on racial differences in various stages of the disablement process model, we develop the following hypothesis: H1: The associations between chronic conditions and (i) risk of onset of functional limitations; (ii) initial level of functional limitations, and (iii) subsequent growth of functional limitations will be greater for African Americans and Hispanics than for non-Hispanic Whites. Insights From Fundamental Cause Theory According to fundamental cause theory, differences in SES are the primary drivers of health disparities in general (Link & Phelan, 1995; Phelan & Link, 2005), and racial differences in health specifically (Williams & Jackson, 2005). According to the fundamental cause model, SES is linked to health because (a) it shapes knowledge of and access to both “health enhancing behaviors” and health care, and (b) it locates individuals in various contexts that expose them to various forms of “risk and protective factors” (e.g., work environment, residential environment) related to health (Phelan & Link, 2005). Indeed, empirical research on the relationship between race and functional limitations suggests differences in SES account for much of the disparity. In fact, according to some estimates, adjusting for education, wealth, income, and private insurance explains the majority of racial differences in functional limitations (Kail & Taylor, 2014; Schoenbaum & Waidmann, 1997). Other research finds variation by gender and age, but accounting for education and income accounts for between 53% and 100% of racial differences in level of functional limitations among people between the ages of 55 and 74 (Fuller-Thomson, Nuru-Jeter, Minkler, & Guralnik, 2009). In short, this body of research indicates adjusting for SES explains a considerable amount of the variation in functional limitations outcomes by race/ethnicity. Although clearly important for variations in functional limitations as an outcome, it remains unclear whether differences in SES will account for racial variation in interrelationships within the early stages of the disablement processes—in this case, the association between chronic conditions and subsequent functional limitations. Therefore, based on fundamental cause theory and empirical research on racial differences in functional limitations we derive the following hypothesis: H2: Racial differences in the association between chronic conditions and subsequent functional limitations will be explained by accounting for differences in education, household income, household wealth, and private insurance. Data and Method Data for this study come from the 1994 through 2012 waves of the HRS. The HRS began in 1992, and included 12,652 people between the ages of 51 and 61, plus their spouses (regardless of the spouse’s age). In 1998, 2004, and 2010, new cohorts of 51 to 56 years olds were added to the existing cohorts. In this study, the sample was limited in three primary ways. First, because functional limitations were measured differently in wave 1 than the subsequent waves, the 1992 wave was excluded. Second, because age eligibility for respondents in the HRS study was 51, people ages 50 and younger (including spouses to primary respondents) were excluded. Third, the analytic sample was limited to people who reported their race/ethnicity as being non-Hispanic White, non-Hispanic Black, or Hispanic. All other race/ethnicities (mostly coded “other”) were excluded. The resulting sample included 21,796 people, with an average of 5.65 observations per person across waves. Dependent Variable Functional limitations was a time varying variable, measured at each wave as the sum of eleven dichotomously coded items indicating self-reported difficulty with each of the following activities: walking several blocks; sitting for 2 hr; pushing or pulling large objects; reaching or extending arms up; stooping, kneeling, or crouching; getting up from a chair; climbing several flights of stairs; walking one block; lifting or carrying 10 pounds; picking a dime up off the ground; and climbing one flight of stairs (Jette, 1980, 2006; Jette & Deniston, 1978). This 11-item index was then split into two time-varying measures capturing (a) the binary process of onset over time and (b) the sum of eleven items representing level of functional limitations over time given the individual had experienced initial onset. Independent Variables There were four sets of time invariant study variables in the following analyses. First, chronic conditions were measured as the sum of seven items indicating the total number of the following conditions with which the respondent had ever been diagnosed by a doctor in the first wave the respondent was included in the HRS (Brown, O’Rand, & Adkins, 2012; Ferraro & Farmer, 1999). These conditions include: cancer; high blood pressure; diabetes; lung disease; arthritis; stroke; or heart disease. Second, three indicators of race/ethnicity were included in the following study. If the respondent reported their race as White but that they were not Hispanic, they were coded “1” on a dummy indicator for non-Hispanic White. If the respondent reported their race as Black but that they were not Hispanic, they were coded “1” on a dummy indicator for non-Hispanic Black. If the respondent reported their ethnicity as Hispanic, they were coded “1” on a dummy indicator for Hispanic, regardless of their race. Third, to test for moderation between chronic conditions and race, we include the multiplicative interactions between each of the indicators of race and the summed measure of chronic conditions. Fourth, the models also test for five time-invariant measures of SES. Education was measured as the total number of years of education the respondent reported (capped at 17). Household income was coded as the natural log of household income in the first wave the respondent was included in the HRS. Household wealth was measured as the total housing and non-housing wealth minus any debt in the first wave the respondent was included in the HRS and converted using the inverse hyperbolic sine function to account for skew and negative values. Employer insurance was coded “1” if people reported having insurance through their own employment or their spouse’s employment in the first wave the respondent was included in the HRS, and market insurance was coded “1” if people reported purchasing private insurance in the non-group marker in the first wave the respondent was included in the HRS but were without employer insurance (the reference category is no private insurance) (Kail & Taylor, 2014). Control Variables In addition to the study variables, the models also adjust for five time invariant control variables. First, gender was measured with a dichotomous indicator coded “1” for men. Second, age was measured as the age in which the respondent was in the first year they were included in the HRS. Third, marital status was coded “1” for people who were married in the first year they were included in the HRS. Fourth, to help reduce the impact of the healthy immigrant effect on ethnicity, a measure of whether or not someone was an immigrant was coded “1” for people who were not born in the United States. Finally, to account for both left and right censoring in the data, we take into account the number of waves each individual contributes to the data. Analysis Plan We estimate LGCRO to examine individual timing of onset and accumulation of functional limitations simultaneously as two separate processes. Modeling timing and accumulation as separate processes can illuminate where and how disparities unfold over time, providing clearer implications for effective interventions and policy (Kail & Taylor, 2014). Substantively, it is likely important predictors like race and chronic conditions vary in their impact on the timing and progression of functional limitations. Methodologically, failure to account for shifts in timing (e.g., delayed onset) may produce biased results in more tradition latent growth curve models (Taylor, 2008). LGCRO utilizes binary onset variables to estimate a discrete-time hazard model capturing the timing of first onset of some outcome while simultaneously estimating a latent growth curve starting at each individual’s first onset, such that some nonzero level of the continuous outcome and the change in the level over time (given onset) can be examined alongside the timing of onset. Early applications of this model include examination of tumor growth in mice (Albert & Shih, 2003). More recently, they have been applied to health processes in older populations (Haas & Rohlfsen, 2010; Taylor, 2008). The model, equations, assumptions, and relevant data coding are described in detail elsewhere (Kail & Taylor, 2014; Taylor, 2008). For further discussion of incorporating a discrete-time hazard model into the SEM framework, including simultaneous estimation of the hazard model with other latent variable techniques, see Masyn (2004). All missingness on the outcome variables was handled using a Full Information Maximum Likelihood (FIML) estimator, which relies on the assumption that observations are missing at random (MAR). Further sensitivity to mortality or other nonrandom missingness was handled using a continuous covariate measuring the number of waves each respondent contributed to the analysis, consistent with the measure used by Warner and Brown (2011). A linear model fit the limitation level trajectory best (models not shown), the factor loadings representing time were fixed at increasing integers ranging from 0 to 9. The LGCRO model has a number of strengths, most notably the ability to distinguish between earlier, more severe, and more quickly accelerating processes of health decline in establishing where disparities exist over time within one model. However, there are limitations of this combined model relative to its more widely used components (discrete-time hazard model, growth curve model). The most notable is because the model estimates so many parameters simultaneously, the examination of large numbers of covariates (e.g., time varying covariates) is often not possible. Alternate techniques should be used to examine models with large numbers of covariates. All analyses were performed using Mplus Version 6.12. Results Descriptive Results To understand how the sample experienced functional limitations over time, we created two new measures solely for descriptive presentation. First, we calculated the average risk of onset across all 10 waves of our data among those people who had not previously experienced functional limitations up until they initially experienced onset. As shown in Table 1, the within-person average of functional limitations onset (through initial onset) was 0.68, and 83% of the sample experienced functional limitations onset during the period of observation. Second, we calculated the average level of functional limitations across all 10 waves of our data among those people who had experienced functional limitation onset. On average, people experiencing functional limitation onset experienced 3.33 functional limitations. Additionally, on average, people in this sample had ever been diagnosed with 1.45 chronic conditions in the first wave they were interviewed for the HRS. The sample was 19% African American, 12% Hispanic, and 69% White. The average person in this sample had 12.43 years of education, an average household income of $61,786, and $279,124 in wealth (although as mentioned earlier, for the analyses and in Table 1, these values are converted using the natural log and inverse hyperbolic sine functions, accordingly). Finally, 64% of the sample had employer insurance and 9% had market insurance in the first wave they were interviewed for the HRS. Table 1. Descriptive Statistics Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Notes. n = 21,796. aAveraged across waves within persons—83.91% of the sample experience onset at some point during the period of observation. bAverage level after onset—this was only calculated for the 18,288 people that experienced onset. cIncome was transformed using the natural log function and the pre-transformed mean of income was $61,786. dWealth was transformed using the inverse hyperbolic sine function and the pre-transformed mean of wealth was $279,124. View Large Table 1. Descriptive Statistics Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Notes. n = 21,796. aAveraged across waves within persons—83.91% of the sample experience onset at some point during the period of observation. bAverage level after onset—this was only calculated for the 18,288 people that experienced onset. cIncome was transformed using the natural log function and the pre-transformed mean of income was $61,786. dWealth was transformed using the inverse hyperbolic sine function and the pre-transformed mean of wealth was $279,124. View Large Differences in summary statistics across the three race/ethnic groups are shown in Table 2. African Americans and Hispanics were more likely to experience functional limitation onset than Whites (70%, 70%, and 67% respectively), and have higher levels of functional limitations given onset (4.02, 3.83, and 3.05 respectively). African Americans in this sample were diagnosed with more chronic conditions than Whites, and both groups were diagnosed with more chronic conditions than Hispanics (2.45, 2.36, and 2.10 respectively). Finally, Whites had more education, income, wealth, and higher rates of insurance coverage than African Americans. Although African Americans were more educated, had larger household incomes, and greater employer provided insurance than Hispanics, Hispanics had more wealth than African Americans. Table 2. Descriptive Statistics, by Race African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 Notes. aSignificances tests indicate that the values vary across the three racial groups. bAveraged across waves within persons. cAverage limitation level is only calculated for people who experienced limitation onset, the ns for this were 3,479 among African Americans, 2,083 among Hispanics, and 12,726 among Whites. dIncome was transformed using the natural log function. eWealth was transformed using the inverse hyperbolic sine function. Differences tests were chi-square for dichotomous variables and ANOVA for continuous variables. *p < .05. **p < .01. ***p < .001. View Large Table 2. Descriptive Statistics, by Race African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 Notes. aSignificances tests indicate that the values vary across the three racial groups. bAveraged across waves within persons. cAverage limitation level is only calculated for people who experienced limitation onset, the ns for this were 3,479 among African Americans, 2,083 among Hispanics, and 12,726 among Whites. dIncome was transformed using the natural log function. eWealth was transformed using the inverse hyperbolic sine function. Differences tests were chi-square for dichotomous variables and ANOVA for continuous variables. *p < .05. **p < .01. ***p < .001. View Large Multivariable Results From Structural Equation Models Results from LGCRO estimated in a SEM framework are shown in Table 3. In model 1, African Americans experience 0.43 more function limitations at onset (and, not shown, but available in Supplementary Material, African Americans experience greater odds of becoming functionally limited prior to adjusting for chronic conditions). Similarly, Hispanics have 48% greater odds of becoming functionally limited than non-Hispanic Whites, and experience 0.60 more function limitations at onset. Additionally, consistent with expectations derived from the disablement process model, those who suffer from more chronic conditions have greater odds of becoming functionally limited (OR = 1.88, p < .001), higher initial level of functional limitations (B = 0.836, p < .001), and an increased growth of functional limitations over time (B = 0.01, p < .001). Table 3. Latent Growth Curve Models with Random Onset Predicting Functional Limitations Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Notes. n = 21,796. Coefficients for the hazards are odds ratios, numbers in parentheses are standard errors. (α) denotes intercept of the latent intercept, and (ß) denotes intercept of the latent slopes. aReferences is White. bIncome was transformed using the natural log function. cWealth was transformed using the inverse hyperbolic sine function. dReferences is no private insurance. eReferences is female. fReferences is unmarried. *p < .05. **p < .01. ***p < .001. View Large Table 3. Latent Growth Curve Models with Random Onset Predicting Functional Limitations Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Notes. n = 21,796. Coefficients for the hazards are odds ratios, numbers in parentheses are standard errors. (α) denotes intercept of the latent intercept, and (ß) denotes intercept of the latent slopes. aReferences is White. bIncome was transformed using the natural log function. cWealth was transformed using the inverse hyperbolic sine function. dReferences is no private insurance. eReferences is female. fReferences is unmarried. *p < .05. **p < .01. ***p < .001. View Large In model 2 we include the interactions between African American and chronic conditions, and Hispanic and chronic conditions. The inclusions of these interaction terms resulted in a akaike information criterion (AIC) decrease of ~19 (464491.145–464471.999), and when added parameters decrease the AIC by more than 2 it is generally interpreted as in improvement in model fit (Burnham & Anderson, 2003, pp. 87–88). According to these results, the impact of each chronic condition on the odds of becoming functionally limited is greater for both African Americans and Hispanics than it is for non-Hispanic Whites. Model estimated predicted odds ratios of functional limitation onset based on this model are shown in the first panel of Figure 1. According to this figure, the racial differences in the odds of functional limitations onset are comparable when the number of chronic conditions are low—and at zero chronic conditions, there are no significant differences between African Americans and Whites. However, as the number of chronic conditions increases, the racial differences in the risk of functional limitations onset increase. Thus, the association between chronic conditions and functional limitations onset is clearly more deleterious for African Americans and Hispanics than it is for Whites, at least prior to adjusting for SES. Figure 1. View largeDownload slide Model estimated odds of functional limitation onset by race and number of chronic conditions. Notes: 1Estimated from model 2 of Table 3, 2Estimated from model 3 of table 3. Values are (1-odds ratios), and the comparison. Figure 1. View largeDownload slide Model estimated odds of functional limitation onset by race and number of chronic conditions. Notes: 1Estimated from model 2 of Table 3, 2Estimated from model 3 of table 3. Values are (1-odds ratios), and the comparison. In Model 3, we show results of models wherein we assess the extent to which socioeconomic variables mediate the moderating effect of race on the association between chronic conditions and functional limitations. In this model, education, household income, wealth, and insurance coverage were all significantly associated with functional limitations. However, despite accounting for all these socioeconomic measures, the association between chronic conditions and functional limitations onset remained stronger for African Americans and Hispanics than for Whites. Model estimated predicted odds ratios of functional limitation onset based on this model are shown in the second panel of Figure 1. Here it becomes apparent that adjusting for SES reduces the odds of functional limitation onset across all three racial groups; however, the general pattern of racial differences in the association between chronic conditions and functional limitations remains. That is, the racial differences in the odds of functional limitations onset are fairly comparable when the number of chronic conditions are low—and in fact, African Americans experience fewer functional limitations at zero chronic conditions than do Whites. However, as the number of chronic conditions increases, the racial differences in the risk of functional limitations onset increase. Thus, after adjusting for SES, the association between chronic conditions and functional limitations onset remains more deleterious for African Americans and Hispanics than it is for Whites. Discussion Using the 1994 to 2012 waves of the HRS, we assessed (a) the extent to which the association between chronic conditions and functional limitations onset, level, and growth over time varied by race/ethnicity, and (b) the extent to which SES accounted for racial variation in the association between chronic conditions and functional limitations. These findings not only provide a more nuanced understanding of racial variation in the disablement process, but also have important implications for how and when interventions may be most efficacious in terms of reducing racial disparities in functional health. Moreover, this study makes important contributions to existing literature on the disablement process model, and fundamental cause theory more broadly. Our first hypothesis was the associations between chronic conditions and (a) risk of onset of functional limitations; (b) initial level of functional limitations; and (c) subsequent growth of functional limitations would be greater for African Americans and Hispanics than for non-Hispanic Whites. This hypothesis was partially confirmed. Before adjusting for socioeconomic variables, we found that the rate at which chronic conditions translate to functional limitation onset is faster for racial and ethnic minorities and that chronic conditions do translate to higher initials levels (but not subsequent growth) of limitations among minority groups. This has several important insights for the disablement process model. First, the disablement process model highlights that the rate at which functional limitations occur over time is often conditional upon certain risk factors, including demographic characteristics like race (Braungart Fauth et al., 2007; Verbrugge & Jette, 1994). Our research illuminates the way in which race/ethnicity may be considered a double risk factor in the process of disablement. First, when considered by itself, there is increased direct risk in terms of the progression of transitioning from healthy to disabled among African Americans and Hispanics largely due to the experience of chronic disease. Second, there is a multiplicative risk wherein each chronic condition experienced among minorities hastens the onset and initial level of functional limitations significantly more than for Whites. This suggests racial and ethnic minorities experience the disablement process even more rapidly in the presence of chronic conditions. This finding is concordant with other research on racial disparities in health that highlights the multiplicative (rather than just additive) ways in which racial inequality manifests itself in health inequities (Brown, Richardson, Hargrove, & Thomas, 2016). Second, we agree with others who suggest, because the disablement process operates differently for members of racial/ethnic minority groups, it is essential we develop targeted interventions that focus on the specific needs of these groups (Zsembik et al., 2000). Moreover, understanding the pathway from chronic conditions to functional limitations is particularly important because this is (a) after people are diagnosed with chronic disease, so they have begun the process of disablement, but (b) early enough in the disablement process that interventions may be effective. In this case, these results suggest trying to reduce functional limitations once they occur may be too late in the disablement process to help attenuate racial disparities in disablement. Rather, if practitioners and educators were interested in limiting racial disparities in functional limitations, they would be wise to consider more upstream processes because both an additive process (greater disparities) and a multiplicative process (with disease) are driving racial disparities in the onset of functional limitations. Specifically, any program that does not specifically focus—at least in part—on limiting chronic conditions and their severity is likely missing a key element in decreasing not only racial disparities in functional limitations, but likely more severe subsequent disablement processes as well. Our second hypothesis was racial differences in the association between chronic conditions and functional limitations would be explained by accounting for differences in education, household income, household wealth, and private insurance. In other words, we expected SES to mediate the extent to which race moderated the relationship between chronic conditions and functional limitations. This hypothesis was only partially supported. Although education, household income, household wealth, and insurance mediated the increased initial level of functional limitations associated with chronic conditions among Hispanics, they did not account for the significant increased odds of functional limitations onset due to chronic conditions that African Americans and Hispanics experience relative to Whites or the increased initial level of functional limitations onset due to chronic conditions among African Americans. This suggests something other than socioeconomic resources is driving the accelerated onset of functional limitations associated with increased chronic conditions among racial and ethnic minorities. According to fundamental cause theory (Phelan & Link, 2005), SES is linked to functional limitations because knowledge of and access to both “health enhancing behaviors” and health care are contingent upon SES, and SES exposes people to various forms of “risk and protective factors.” The findings from this study highlight that, although SES is fundamentally important in understanding racial differences in functional limitations, when considering the relationship between chronic conditions and functional limitations, SES fails to account fully for the exacerbated risk of functional limitation among African Americans and Hispanics because it does not account for how rapidly disease translates to limitations among these groups. Although this does not preclude that SES is operating more upstream in the disablement processes, it is also possible other barriers to care, for example, may account for the more severe or rapidly disabling nature of conditions among minorities. Additionally, the observed racial/ethnic differences are certainly not directly attributable to race or ethnicity, but rather the experience of being a racial/ethnic minority living in a racially stratified society. These findings are consistent with articulations of fundamental cause theory (Phelan & Link, 2015) positing racial differences in health would not be entirely explained by differences in socioeconomic resources. Rather, racial differences in health persist net of socioeconomic resources, in part, because racial minorities are often exposed to a wide array of racism—including, but not limited to, discrimination, segregation, or structural racism (Gee & Ford, 2011; Krieger, 2014; Williams & Mohammed, 2013). Although it is untested here because of limitations in the data, it is likely racial/ethnic minorities’ unique experiences of racism contributed to the unexplained racial/ethnic differences in the extent to which chronic conditions are associated with functional limitations. The findings presented here also emphasize the importance of developing a more nuanced understanding of risk factors and the progression of functional limitations over time, including testing the direct, mediating, and moderating effects of variables using methodological tools equipped to separate onset, initial level, and growth such as the LGCRO models used here. In the context of intervention strategies and policy, these models highlight that programs targeting economic differences may be important for ameliorating some of the disparities encountered by African Americans and Hispanics in functional limitations. However, our findings also suggest economic policies alone may do little to reduce the increased risk of functional limitations onset associated with chronic conditions among African Americans and Hispanics. For example, policies aimed at reducing racial and ethnic disparities in the preventive or early health care access and utilization necessary to diagnose and treat potentially disabling illness early need to move beyond considering financial resources alone and expand to a broader examination of factors underlying barriers to health care access and utilization among minority groups, and the deleterious ways in which exposure to racism shapes the health of racial/ethnic minorities. Limitations and Future Research Although our analysis has the benefits of modeling multiple temporal components of the disablement process, one limitation to our findings is the lack of time-varying covariate effects. Although other modeling techniques can incorporate time-varying measures of chronic conditions and SES, for example, the LGCRO which estimates multiple processes simultaneously often cannot also incorporate these effects. Since our findings suggest a robust and differential impact of chronic conditions among race and ethnic minorities that is primary in the process of accelerated onset, future studies modeling onset timing alone may incorporate time varying effects to replicate and further explore the findings shown here. Another key limitation is we were limited in the type of chronic conditions measured in the HRS and, to retain sufficient statistical power for our interactions terms, we use a simple count of conditions in estimating our interactions. Although our findings are telling, they suggest future research should examine specific conditions and broaden the range of chronic conditions studied in order to isolate the most potentially harmful conditions driving the interactive effects shown here. The focus of this research was on the early stages of the disablement process because this is an ideal time to target health-mitigating interventions. However, we hope future research delves deeper into understanding racial differences in the disablement process as a whole—something we do not do here. It would be extremely informative to understand racial variation in this process beginning with when people are healthy, and following them through to mortality—something we are unable to do here because of the high rate of survivorship in our sample. One important implication of this is that what we identify as double disadvantage here based on one interrelationship could actually be much greater or more widespread when considering racial differences in other interrelationships across the entire disablement process model. Conclusion African Americans and Hispanics experience more rapid onset and higher levels of functional limitations, but also experience greater risk of onset associated with chronic conditions compared to their White counterparts. This suggests a multiplicative double disadvantage in early stages of the disablement process for minority groups. Although traditional measures of SES attenuated disparities at similar levels of chronic conditions, the greater impact of chronic conditions for minorities remained robust for onset of functional limitations. This highlights the importance of considering how and when chronic conditions accumulate and translate to decreased functioning across disadvantaged groups, because these early stages of the disablement process may be more amenable to intervention than later stages marked by more severe disability. It also highlights the importance of moving beyond basic economic policies in decreasing racial disparities in later life health and functioning. Supplementary Material Supplementary data is available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. Conflict of Interest B. L. Kail and M. G. Taylor are both on the editorial board for the Journal of Gerontology: Social Sciences. References Albert , P. S. , & Shih , J. H . ( 2003 ). Modeling tumor growth with random onset . Biometrics , 59 , 897 – 906 . doi: 10.1111/j.0006-341X.2003.00104.x Google Scholar CrossRef Search ADS PubMed Braungart Fauth , E. , Zarit , S. H. , Malmberg , B. , & Johansson , B . ( 2007 ). Physical, cognitive, and psychosocial variables from the disablement process model predict patterns of independence and the transition into disability for the oldest-old . The Gerontologist , 47 , 613 – 624 . doi: 10.1093/geront/47.5.613 Google Scholar CrossRef Search ADS PubMed Brown , T. H. , O’Rand , A. M. , & Adkins , D. E . ( 2012 ). Race-ethnicity and health trajectories: Tests of three hypotheses across multiple groups and health outcomes . Journal of Health and Social Behavior , 53 , 359 – 377 . doi: 10.1177/0022146512455333 Google Scholar CrossRef Search ADS PubMed Brown , T. H. , Richardson , L. J. , Hargrove , T. W. , & Thomas , C. S . ( 2016 ). Using multiple-hierarchy stratification and life course approaches to understand health inequalities . Journal of Health and Social Behavior , 57 , 200 – 222 . doi: 10.1177/0022146516645165 Google Scholar CrossRef Search ADS PubMed Burnham , K. P. , & Anderson , D. R . ( 2003 ). Model selection and multimodel inference: A practical information-theoretic approach . New York : Springer Science & Business Media . Clarke , P. , & George , L. K . ( 2005 ). The role of the built environment in the disablement process . American Journal of Public Health , 95 , 1933 – 1939 . doi: 10.2105/AJPH.2004.054494 Google Scholar CrossRef Search ADS PubMed Dallo , F. J. , Booza , J. , & Nguyen , N. D . ( 2015 ). Functional limitations and nativity status among older Arab, Asian, black, Hispanic, and white Americans . Journal of Immigrant and Minority Health , 17 , 535 – 542 . doi: 10.1007/s10903-013-9943-0 Google Scholar CrossRef Search ADS PubMed Dominick , K. L. , Dudley , T. K. , Grambow , S. C. , Oddone , E. Z. , & Bosworth , H. B . ( 2003 ). Racial differences in health care utilization among patients with osteoarthritis . The Journal of Rheumatology , 30 , 2201 – 2206 . Google Scholar PubMed Dunlop , D. D. , Song , J. , Lyons , J. S. , Manheim , L. M. , & Chang , R. W . ( 2003 ). Racial/ethnic differences in rates of depression among preretirement adults . American Journal of Public Health , 93 , 1945 – 1952 . doi: 10.2105/AJPH.93.11.1945 Google Scholar CrossRef Search ADS PubMed Ferraro , K. F. , & Farmer , M. M . ( 1999 ). Utility of health data from social surveys: Is there a gold Standard for measuring morbidity ? American Sociological Review , 64 , 303 – 315 . doi: 10.2307/2657534 Google Scholar CrossRef Search ADS Fiscella , K. , Franks , P. , Gold , M. R. , & Clancy , C. M . ( 2000 ). Inequality in quality: Addressing socioeconomic, racial, and ethnic disparities in health care . JAMA , 283 , 2579 – 2584 . doi: 10.1001/jama.283.19.2579 Google Scholar CrossRef Search ADS PubMed Fuller-Thomson , E. , Nuru-Jeter , A. , Minkler , M. , & Guralnik , J. M . ( 2009 ). Black-White disparities in disability among older Americans: Further untangling the role of race and socioeconomic status . Journal of Aging and Health , 21 , 677 – 698 . doi: 10.1177/0898264309338296 Google Scholar CrossRef Search ADS PubMed Gayman , M. D. , Pai , M. , Kail , B. L. , & Taylor , M. G . ( 2013 ). Reciprocity between depressive symptoms and physical limitations: Pre- and post-retirement: Exploring racial differences . Journal of Aging and Health , 25 , 555 – 573 . doi: 10.1177/0898264313480239 Google Scholar CrossRef Search ADS PubMed Gayman , M. D. , Turner , R. J. , & Cui , M . ( 2008 ). Physical limitations and depressive symptoms: Exploring the nature of the association . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 63 , S219 – S228 . doi: 10.1093/geronb/63.4.S219 Google Scholar CrossRef Search ADS Gee , G. C. , & Ford , C. L . ( 2011 ). Structural racism and health inequities: Old issues, new directions . Du Bois Review: Social Science Research on Race , 8 , 115 – 132 . doi: 10.1017/S1742058X11000130 Google Scholar CrossRef Search ADS PubMed Goel , M. S. , Wee , C. C. , McCarthy , E. P. , Davis , R. B. , Ngo-Metzger , Q. , & Phillips , R. S . ( 2003 ). Racial and ethnic disparities in cancer screening: The importance of foreign birth as a barrier to care . Journal of General Internal Medicine , 18 , 1028 – 1035 . doi: 10.1111/j.1525-1497.2003.20807.x Google Scholar CrossRef Search ADS PubMed Haas , S. , & Rohlfsen , L . ( 2010 ). Life course determinants of racial and ethnic disparities in functional health trajectories . Social Science & Medicine , 70 , 240 – 250 . doi: 10.1016/ j.socscimed.2009.10.003 Google Scholar CrossRef Search ADS Jette , A. M . ( 1980 ). Functional Status Index: Reliability of a chronic disease evaluation instrument . Archives of Physical Medicine and Rehabilitation , 61 , 395 – 401 . Google Scholar PubMed Jette , A. M . ( 2006 ). Toward a common language for function, disability, and health . Physical Therapy , 86 , 726 – 734 . Google Scholar CrossRef Search ADS PubMed Jette , A. M. , & Deniston , O. L . ( 1978 ). Inter-observer reliability of a functional status assessment instrument . Journal of Chronic Diseases , 31 , 573 – 580 . doi: 10.1016/0021-9681(78)90017-6 Google Scholar CrossRef Search ADS PubMed Kail , B. L. , & Carr , D. C . ( 2017 ). Successful aging in the context of the disablement process: Working and volunteering as moderators on the association between chronic conditions and subsequent functional limitations . Journals of Gerontology: Social Sciences , 72 , 340 – 350 . doi: 10.1093/geronb/gbw060 Kail , B. L. , & Taylor , M. G . ( 2014 ). Cumulative inequality and racial disparities in health: Private insurance coverage and black/white differences in functional limitations . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 69 , 798 – 808 . doi: 10.1093/geronb/gbu005 Google Scholar CrossRef Search ADS Krieger , N . ( 2014 ). Discrimination and health inequities . International Journal of Health Services: Planning, Administration, Evaluation , 44 , 643 – 710 . doi: 10.2190/HS.44.4.b Google Scholar CrossRef Search ADS PubMed Lawrence , R. H. , & Jette , A. M . ( 1996 ). Disentangling the disablement process . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 51 , S173 – S182 . doi: 10.1093/geronb/51B.4.S173 Google Scholar CrossRef Search ADS Link , B. G. , & Phelan , J . ( 1995 ). Social conditions as fundamental causes of disease . Journal of Health and Social Behavior , 35 , 80 – 94 . doi: 10.2307/2626958 Google Scholar CrossRef Search ADS Masyn , K. E . ( 2004 ). Discrete-time survival analysis using latent variables . Retrieved June, 20, 2005 . Melvin , J. , Hummer , R. , Elo , I. , & Mehta , N . ( 2014 ). Volume 31-Article 17| Pages 497–510 Age patterns of racial/ethnic/nativity differences in disability and physical functioning in the United States . Age , 31 , 497 – 510 . doi: 10.4054/DemRes.2014.31.17 Nelson , A. R. , Stith , A. Y. , & Smedley , B. D . ( 2002 ). Unequal treatment: Confronting racial and ethnic disparities in health care (full printed version) . Washington, DC : National Academies Press . Ostchega , Y. , Harris , T. B. , Hirsch , R. , Parsons , V. L. , & Kington , R . ( 2000 ). The prevalence of functional limitations and disability in older persons in the US: Data from the National Health and Nutrition Examination Survey III . Journal of the American Geriatrics Society , 48 , 1132 – 1135 . doi: 10.1111/j.1532-5415.2000.tb04791.x Google Scholar CrossRef Search ADS PubMed Phelan , J. C. , & Link , B. G . ( 2005 ). Controlling disease and creating disparities: A fundamental cause perspective . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 60 ( Special Issue 2 ), S27 – S33 . doi: 10.1093/geronb/ 60.Special_Issue_2.S27 Google Scholar CrossRef Search ADS Phelan , J. C. , & Link , B. G . ( 2015 ). Is racism a fundamental cause of inequalities in health ? Annual Review of Sociology , 41 , 311 – 330 . doi: 10.1146/annurev-soc-073014-112305 Google Scholar CrossRef Search ADS Schneider , E. C. , Zaslavsky , A. M. , & Epstein , A. M . ( 2002 ). Racial disparities in the quality of care for enrollees in medicare managed care . JAMA , 287 , 1288 – 1294 . doi: 10.1001/jama.287.10.1288 Google Scholar CrossRef Search ADS PubMed Schoenbaum , M. , & Waidmann , T . ( 1997 ). Race, socioeconomic status, and health: Accounting for race differences in health . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 52 , 61 – 73 . doi: 10.1093/geronb/ 52B.Special_Issue.61 Google Scholar CrossRef Search ADS Taylor , M . ( 2008 ). Timing, Accumulation, and the black/white disability gap in later life: A test of weathering . Research on Aging , 30 , 226 . doi: 10.1177/0164027507311838 Google Scholar CrossRef Search ADS Thoits , P. A . ( 2010 ). Stress and health: Major findings and policy implications . Journal of Health and Social Behavior , 51 ( Suppl ), S41 – S53 . doi: 10.1177/0022146510383499 Google Scholar CrossRef Search ADS PubMed Verbrugge , L. M. , & Jette , A. M . ( 1994 ). The disablement process . Social Science & Medicine ,  38 ,  1 – 14 . doi: 10.1016/0277- 9536(94)90294-1 Warner , D. F. , & Brown , T. H . ( 2011 ). Understanding how race/ethnicity and gender define age-trajectories of disability: An intersectionality approach . Social Science & Medicine , 72 , 1236 – 1248 . doi: 10.1016/j.socscimed.2011.02.034 Google Scholar CrossRef Search ADS Williams , D. R. , & Jackson , P. B . ( 2005 ). Social sources of racial disparities in health . Health affairs (Project Hope) , 24 , 325 – 334 . doi: 10.1377/hlthaff.24.2.325 Google Scholar CrossRef Search ADS PubMed Williams , D. R. , & Mohammed , S. A . ( 2013 ). Racism and health I . American Behavioral Scientist , 57 , 1152 – 1173 . doi: 10.1177/0002764213487340 Google Scholar CrossRef Search ADS Williams , D. R. , Neighbors , H. W. , & Jackson , J. S . ( 2003 ). Racial/ethnic discrimination and health: Findings from community studies . American Journal of Public Health , 93 , 200 – 208 . doi: 10.2105/ajph.93.2.200 Google Scholar CrossRef Search ADS PubMed Zsembik , B. A. , Peek , M. K. , & Peek , C. W . ( 2000 ). Race and ethnic variation in the disablement process . Journal of Aging and Health , 12 , 229 – 249 . doi: 10.1177/089826430001200205 Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journals of Gerontology Series B: Psychological Sciences and Social Sciences Oxford University Press

Double Disadvantage in the Process of Disablement: Race as a Moderator in the Association Between Chronic Conditions and Functional Limitations

Loading next page...
 
/lp/ou_press/double-disadvantage-in-the-process-of-disablement-race-as-a-moderator-ayDdanJNtr
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
1079-5014
eISSN
1758-5368
D.O.I.
10.1093/geronb/gby027
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives This study evaluated (a) whether the association between chronic conditions and functional limitations vary by race/ethnicity, and (b) whether socioeconomic status accounted for any observed racial variation in the association between chronic conditions and functional limitations. Method The Health and Retirement Study data were used to assess whether race/ethnicity moderated the association between chronic conditions and functional limitations, and whether education, income, and/or wealth mediated any of the observed moderation by race/ethnicity. Results Results from structural equation models of latent growth curves with random onset indicated that (a) the positive association between chronic conditions and functional limitations onset was larger for African Americans and Hispanics than it was for Whites, but (b) this difference largely persisted net of socioeconomic status. Discussion African Americans and Hispanics endure a multiplicative double disadvantage in the early stages of the disablement process where they experience (a) a more rapid onset and higher levels of functional limitations, and (b) greater risk of functional limitation onset associated with chronic conditions compared to their White counterparts. Moreover, basic economic policies are unlikely to curtail the greater risk of functional limitations onset associated with chronic conditions encountered by African Americans and Hispanics. Disability, Functional health status, Minority aging (race/ethnicity), Socioeconomic status In later life, there tend to be significant racial/ethnic disparities across a number of health outcomes. Among these, one particularly concerning outcome is functional limitations. Racial and ethnic minorities tend to be more functionally limited than White Americans. African Americans and Hispanics are generally at greater risk of becoming functionally limited (Dallo, Booza, & Nguyen, 2015; Haas & Rohlfsen, 2010) and tend to suffer from more functional limitations (Haas & Rohlfsen, 2010; Ostchega, Harris, Hirsch, Parsons, & Kington, 2000). In short, there are clear racial/ethnic disparities in functional limitation outcomes. Although racial disparities in functional limitation outcomes are relatively well established, functional limitations are situated within the broader context of the process of disablement. The disablement process model provides a conceptual understanding for the interrelations between illness/disease, limitations in functional mobility, disability, and, ultimately, mortality (Clarke & George, 2005; Lawrence & Jette, 1996; Verbrugge & Jette, 1994). However, it remains unclear whether there are racial/ethnic differences in various interrelationships within disablement processes. This omission is important because, if there are indeed racial/ethnic differences in relationships within the disablement process, this suggests interventions targeting racial differences in later stage processes (e.g., functional limitations) may be too late to reduce racial disparities in health, and targeting early stage processes (e.g., chronic disease) may be more appropriate. Additionally, racial minorities may experience a kind of double disadvantage wherein they experience elevated levels of chronic disease, and the deleterious association between chronic disease and subsequent functional limitations is stronger. However, this possibility remains heretofore untested. Furthermore, if there are indeed racial differences among the interrelationships defined within the disablement processes model, it is also important to ask why they exist. One prevailing framework for understanding health disparities is fundamental cause theory (Link & Phelan, 1995). In its original formulation, fundamental cause theory postulates that differences in flexible resources conveyed through socioeconomic status (SES) are largely responsible for disparities in health outcomes. Because racial minorities tend to have access to fewer socioeconomic resources, differences in SES may go a long way in helping explain racial disparities in health (Phelan & Link, 2015). However, it remains unclear whether SES does indeed underpin racial disparities in the interrelationships defined within the disablement processes model. To that end, we draw on the disablement process model and fundamental cause theory to address the following two research questions: (a) does the association between chronic conditions and functional limitations vary by race/ethnicity; and (b) if so, is this variation explained by differences in SES. We test these questions using the 1994 to 2012 waves of the Health and Retirement Study (HRS), using latent growth curves with random onset (LGCRO) in a structural equation modeling (SEM) framework. We focus on the association between chronic conditions and functional limitations because this is (a) after people have been diagnosed with a chronic disease, so they have begun an observable process of disablement, but (b) early enough in the disablement process that interventions may be useful. Background Racial Differences in Function Limitations On average, African Americans are both more likely to experience functional limitations, and to experience more severe levels of functional limitation than do Whites in later life (Haas & Rohlfsen, 2010; Kail & Taylor, 2014; Melvin, Hummer, Elo, & Mehta, 2014; Warner & Brown, 2011). Similar disparities are documented between Hispanics and their White counterparts as well. For instance, among U.S.-born adults age 54 to 65, 44% of Whites suffer from one or more functional limitations, compared to 60% of African American, and 57% of Hispanics (Dunlop, Song, Lyons, Manheim, & Chang, 2003). Clearly, African Americans and Hispanics are, on average, more functionally limited than non-Hispanic Whites. Although understanding racial disparities in gross measures of functional limitations (i.e., overall level) provides important information about health disparities, more information can be derived from disaggregating functional limitations into its various temporal parts (Haas & Rohlfsen, 2010; Taylor, 2008). Functional limitations involve three distinct components: (a) the risk of onset; (b) initial level at onset; and (c) subsequent levels (or trajectories) after initial level (for review see Taylor, 2008). Although African Americans and Hispanics have comparable accumulation of functional limitations compared to Whites after becoming functionally limited (Warner and Brown, 2011 do find that Black women experience larger growth in functional limitations over time relative to White men), both African Americans and Hispanics are at greater risk of experiencing functional limitation onset and experience higher initial levels of functional limitations than do Whites (Kail & Taylor, 2014; Warner & Brown, 2011). Thus, when considering racial differences in functional limitations, it is important to separately consider racial differences in risk of onset, initial level, and subsequent trajectories after onset. Disablement Process According to the disablement process model, disablement begins with illness/disease, which then leads to impairments in basic physical activities (generally referred to as functional limitations). These limitations may then accumulate over time and progress into more severe disablement which substantially interferes with daily life (i.e., activities of daily living; Jette, 2006; Lawrence & Jette, 1996; Verbrugge & Jette, 1994). Although much of the disablement research focuses on outcomes within the last stage of disablement because they are most burdensome, the disablement process model itself views the pathway from chronic conditions to functional limitations as particularly important because this is (a) after people have been diagnosed with a chronic disease, so they have begun the process of disablement, but (b) early enough in the disablement process interventions may be effective (Kail & Carr, 2017; Lawrence & Jette, 1996). The disablement process model highlights that the rate at which functional limitations accumulate over time is often conditional upon certain risk factors (Braungart Fauth, Zarit, Malmberg, & Johansson, 2007; Verbrugge & Jette, 1994). These risk factors include factors that precede the start of the disablement process (Verbrugge & Jette, 1994). Race may be one such factor along which the process as a whole is likely patterned. However, although the association between race/ethnicity and disablement outcomes has been studied extensively, the degree to which there are racial differences in interrelationships within the disablement process has not. As mentioned earlier, the latter is important because if there are indeed racial/ethnic differences in the disablement process, this suggests interventions targeting racial/ethnic differences in later stages of the process (e.g., functional limitations) may be too late to help level racial disparities in functional limitations, and targeting early stage processes (e.g., chronic disease) may be more appropriate. There are several reasons to expect interrelationships within the disablement process vary base race. First, racial minorities are less likely than Whites to receive quality care in general (Fiscella, Franks, Gold, & Clancy, 2000), and poorer quality of care than Whites among those enrolled in Medicare managed care health plans (Schneider, Zaslavsky, & Epstein, 2002). Racial minorities are also less likely to be screened for chronic diseases, as well as receive different types of prescriptions and smaller supply of medication to treat chronic diseases (Dominick, Dudley, Grambow, Oddone, & Bosworth, 2003; Goel et al., 2003). These differences persist net of access to care and SES. Second, more recently, Phelan and Link (2015) have argued racial inequalities in health may be rooted in the combination of racial differences in SES, and—more important—the unique experience of racism. Therefore, the experience of racism itself may be a fundamental cause of racial health disparities (Phelan & Link, 2015). Because of (a) the experience of various forms of racism and racial discrimination (Krieger, 2014; Williams & Mohammed, 2013), (b) the associated stress and psychological consequences of racism and discrimination (Thoits, 2010; Williams, Neighbors, & Jackson, 2003), (c) the reciprocal relationship between psychological and physical health (Gayman, Pai, Kail, & Taylor, 2013; Gayman, Turner, & Cui, 2008), and (d) healthcare providers’ racial stereotypes and biases play a large role in the racial differences in quality of care that non-Whites receive (Nelson, Stith, & Smedley, 2002), it is likely, not only will minorities experience more frequent chronic conditions, but the impact of conditions on functional limitations will be amplified. Third, there is some evidence aspects of the disablement process vary by race. For instance, prior research shows the influence of chronic conditions and functional limitations on more severe, late stage outcomes (i.e., activities of daily living limitations) is greater for African Americans and Hispanics compared to Whites (Zsembik, Peek, & Peek, 2000). This suggests the processes leading up to later stage disability vary by racial group membership. However, previous findings bypass a critical component of the disablement process where initial disablement often emerges (i.e., functional limitations). Findings from other research (Haas & Rohlfsen, 2010; Taylor, 2008) suggest focusing on advanced stages of disablement alone may be too late in the process to speak to effective interventions in slowing the progression of disability. As such, we are unaware of any research on racial differences in the ways in which chronic conditions are associated with subsequent functional limitations, and how they persist or grow over time, particularly in an onset, level, and growth model framework. This is an important omission because the pathway from chronic conditions to functional limitations is a good target for interventions as it occurs earlier in the disablement process and, thus, it makes it possible to potentially slow the development of disability before it becomes too late to effectively intervene (Lawrence & Jette, 1996). Thus, for medical and social interventions to help address any racial/ethnic disparities in functional limitations, it is likely they would be most efficacious if they target racial differences in how chronic conditions lead to subsequent functional limitations. Therefore, based on (a) theoretical insights from the disablement process model and (b) empirical research on racial/ethnic differences in functional limitations, and (c) the limited research on racial differences in various stages of the disablement process model, we develop the following hypothesis: H1: The associations between chronic conditions and (i) risk of onset of functional limitations; (ii) initial level of functional limitations, and (iii) subsequent growth of functional limitations will be greater for African Americans and Hispanics than for non-Hispanic Whites. Insights From Fundamental Cause Theory According to fundamental cause theory, differences in SES are the primary drivers of health disparities in general (Link & Phelan, 1995; Phelan & Link, 2005), and racial differences in health specifically (Williams & Jackson, 2005). According to the fundamental cause model, SES is linked to health because (a) it shapes knowledge of and access to both “health enhancing behaviors” and health care, and (b) it locates individuals in various contexts that expose them to various forms of “risk and protective factors” (e.g., work environment, residential environment) related to health (Phelan & Link, 2005). Indeed, empirical research on the relationship between race and functional limitations suggests differences in SES account for much of the disparity. In fact, according to some estimates, adjusting for education, wealth, income, and private insurance explains the majority of racial differences in functional limitations (Kail & Taylor, 2014; Schoenbaum & Waidmann, 1997). Other research finds variation by gender and age, but accounting for education and income accounts for between 53% and 100% of racial differences in level of functional limitations among people between the ages of 55 and 74 (Fuller-Thomson, Nuru-Jeter, Minkler, & Guralnik, 2009). In short, this body of research indicates adjusting for SES explains a considerable amount of the variation in functional limitations outcomes by race/ethnicity. Although clearly important for variations in functional limitations as an outcome, it remains unclear whether differences in SES will account for racial variation in interrelationships within the early stages of the disablement processes—in this case, the association between chronic conditions and subsequent functional limitations. Therefore, based on fundamental cause theory and empirical research on racial differences in functional limitations we derive the following hypothesis: H2: Racial differences in the association between chronic conditions and subsequent functional limitations will be explained by accounting for differences in education, household income, household wealth, and private insurance. Data and Method Data for this study come from the 1994 through 2012 waves of the HRS. The HRS began in 1992, and included 12,652 people between the ages of 51 and 61, plus their spouses (regardless of the spouse’s age). In 1998, 2004, and 2010, new cohorts of 51 to 56 years olds were added to the existing cohorts. In this study, the sample was limited in three primary ways. First, because functional limitations were measured differently in wave 1 than the subsequent waves, the 1992 wave was excluded. Second, because age eligibility for respondents in the HRS study was 51, people ages 50 and younger (including spouses to primary respondents) were excluded. Third, the analytic sample was limited to people who reported their race/ethnicity as being non-Hispanic White, non-Hispanic Black, or Hispanic. All other race/ethnicities (mostly coded “other”) were excluded. The resulting sample included 21,796 people, with an average of 5.65 observations per person across waves. Dependent Variable Functional limitations was a time varying variable, measured at each wave as the sum of eleven dichotomously coded items indicating self-reported difficulty with each of the following activities: walking several blocks; sitting for 2 hr; pushing or pulling large objects; reaching or extending arms up; stooping, kneeling, or crouching; getting up from a chair; climbing several flights of stairs; walking one block; lifting or carrying 10 pounds; picking a dime up off the ground; and climbing one flight of stairs (Jette, 1980, 2006; Jette & Deniston, 1978). This 11-item index was then split into two time-varying measures capturing (a) the binary process of onset over time and (b) the sum of eleven items representing level of functional limitations over time given the individual had experienced initial onset. Independent Variables There were four sets of time invariant study variables in the following analyses. First, chronic conditions were measured as the sum of seven items indicating the total number of the following conditions with which the respondent had ever been diagnosed by a doctor in the first wave the respondent was included in the HRS (Brown, O’Rand, & Adkins, 2012; Ferraro & Farmer, 1999). These conditions include: cancer; high blood pressure; diabetes; lung disease; arthritis; stroke; or heart disease. Second, three indicators of race/ethnicity were included in the following study. If the respondent reported their race as White but that they were not Hispanic, they were coded “1” on a dummy indicator for non-Hispanic White. If the respondent reported their race as Black but that they were not Hispanic, they were coded “1” on a dummy indicator for non-Hispanic Black. If the respondent reported their ethnicity as Hispanic, they were coded “1” on a dummy indicator for Hispanic, regardless of their race. Third, to test for moderation between chronic conditions and race, we include the multiplicative interactions between each of the indicators of race and the summed measure of chronic conditions. Fourth, the models also test for five time-invariant measures of SES. Education was measured as the total number of years of education the respondent reported (capped at 17). Household income was coded as the natural log of household income in the first wave the respondent was included in the HRS. Household wealth was measured as the total housing and non-housing wealth minus any debt in the first wave the respondent was included in the HRS and converted using the inverse hyperbolic sine function to account for skew and negative values. Employer insurance was coded “1” if people reported having insurance through their own employment or their spouse’s employment in the first wave the respondent was included in the HRS, and market insurance was coded “1” if people reported purchasing private insurance in the non-group marker in the first wave the respondent was included in the HRS but were without employer insurance (the reference category is no private insurance) (Kail & Taylor, 2014). Control Variables In addition to the study variables, the models also adjust for five time invariant control variables. First, gender was measured with a dichotomous indicator coded “1” for men. Second, age was measured as the age in which the respondent was in the first year they were included in the HRS. Third, marital status was coded “1” for people who were married in the first year they were included in the HRS. Fourth, to help reduce the impact of the healthy immigrant effect on ethnicity, a measure of whether or not someone was an immigrant was coded “1” for people who were not born in the United States. Finally, to account for both left and right censoring in the data, we take into account the number of waves each individual contributes to the data. Analysis Plan We estimate LGCRO to examine individual timing of onset and accumulation of functional limitations simultaneously as two separate processes. Modeling timing and accumulation as separate processes can illuminate where and how disparities unfold over time, providing clearer implications for effective interventions and policy (Kail & Taylor, 2014). Substantively, it is likely important predictors like race and chronic conditions vary in their impact on the timing and progression of functional limitations. Methodologically, failure to account for shifts in timing (e.g., delayed onset) may produce biased results in more tradition latent growth curve models (Taylor, 2008). LGCRO utilizes binary onset variables to estimate a discrete-time hazard model capturing the timing of first onset of some outcome while simultaneously estimating a latent growth curve starting at each individual’s first onset, such that some nonzero level of the continuous outcome and the change in the level over time (given onset) can be examined alongside the timing of onset. Early applications of this model include examination of tumor growth in mice (Albert & Shih, 2003). More recently, they have been applied to health processes in older populations (Haas & Rohlfsen, 2010; Taylor, 2008). The model, equations, assumptions, and relevant data coding are described in detail elsewhere (Kail & Taylor, 2014; Taylor, 2008). For further discussion of incorporating a discrete-time hazard model into the SEM framework, including simultaneous estimation of the hazard model with other latent variable techniques, see Masyn (2004). All missingness on the outcome variables was handled using a Full Information Maximum Likelihood (FIML) estimator, which relies on the assumption that observations are missing at random (MAR). Further sensitivity to mortality or other nonrandom missingness was handled using a continuous covariate measuring the number of waves each respondent contributed to the analysis, consistent with the measure used by Warner and Brown (2011). A linear model fit the limitation level trajectory best (models not shown), the factor loadings representing time were fixed at increasing integers ranging from 0 to 9. The LGCRO model has a number of strengths, most notably the ability to distinguish between earlier, more severe, and more quickly accelerating processes of health decline in establishing where disparities exist over time within one model. However, there are limitations of this combined model relative to its more widely used components (discrete-time hazard model, growth curve model). The most notable is because the model estimates so many parameters simultaneously, the examination of large numbers of covariates (e.g., time varying covariates) is often not possible. Alternate techniques should be used to examine models with large numbers of covariates. All analyses were performed using Mplus Version 6.12. Results Descriptive Results To understand how the sample experienced functional limitations over time, we created two new measures solely for descriptive presentation. First, we calculated the average risk of onset across all 10 waves of our data among those people who had not previously experienced functional limitations up until they initially experienced onset. As shown in Table 1, the within-person average of functional limitations onset (through initial onset) was 0.68, and 83% of the sample experienced functional limitations onset during the period of observation. Second, we calculated the average level of functional limitations across all 10 waves of our data among those people who had experienced functional limitation onset. On average, people experiencing functional limitation onset experienced 3.33 functional limitations. Additionally, on average, people in this sample had ever been diagnosed with 1.45 chronic conditions in the first wave they were interviewed for the HRS. The sample was 19% African American, 12% Hispanic, and 69% White. The average person in this sample had 12.43 years of education, an average household income of $61,786, and $279,124 in wealth (although as mentioned earlier, for the analyses and in Table 1, these values are converted using the natural log and inverse hyperbolic sine functions, accordingly). Finally, 64% of the sample had employer insurance and 9% had market insurance in the first wave they were interviewed for the HRS. Table 1. Descriptive Statistics Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Notes. n = 21,796. aAveraged across waves within persons—83.91% of the sample experience onset at some point during the period of observation. bAverage level after onset—this was only calculated for the 18,288 people that experienced onset. cIncome was transformed using the natural log function and the pre-transformed mean of income was $61,786. dWealth was transformed using the inverse hyperbolic sine function and the pre-transformed mean of wealth was $279,124. View Large Table 1. Descriptive Statistics Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Mean/proportion Standard deviation Minimum Maximum Any limitation onseta 0.68 — 0.00 1.00 Limitation levelb 3.33 2.54 0.10 11.00 Chronic conditions 1.45 1.29 0.00 8.00 African American 0.19 — 0.00 1.00 Hispanic 0.12 — 0.00 1.00 White 0.69 — 0.00 1.00 Education 12.43 3.23 0.00 17.00 Incomec 10.50 1.09 1.39 15.88 Wealthd 10.30 6.11 −15.99 17.53 Employer insurance 0.64 — 0.00 1.00 Market insurance 0.09 — 0.00 1.00 Male 0.49 — 0.00 1.00 Age 58.09 6.39 51.00 100.00 Married 0.73 — 0.00 1.00 Immigrant 0.11 — 0.00 1.00 # of Waves 5.65 3.22 1.00 10.00 Notes. n = 21,796. aAveraged across waves within persons—83.91% of the sample experience onset at some point during the period of observation. bAverage level after onset—this was only calculated for the 18,288 people that experienced onset. cIncome was transformed using the natural log function and the pre-transformed mean of income was $61,786. dWealth was transformed using the inverse hyperbolic sine function and the pre-transformed mean of wealth was $279,124. View Large Differences in summary statistics across the three race/ethnic groups are shown in Table 2. African Americans and Hispanics were more likely to experience functional limitation onset than Whites (70%, 70%, and 67% respectively), and have higher levels of functional limitations given onset (4.02, 3.83, and 3.05 respectively). African Americans in this sample were diagnosed with more chronic conditions than Whites, and both groups were diagnosed with more chronic conditions than Hispanics (2.45, 2.36, and 2.10 respectively). Finally, Whites had more education, income, wealth, and higher rates of insurance coverage than African Americans. Although African Americans were more educated, had larger household incomes, and greater employer provided insurance than Hispanics, Hispanics had more wealth than African Americans. Table 2. Descriptive Statistics, by Race African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 Notes. aSignificances tests indicate that the values vary across the three racial groups. bAveraged across waves within persons. cAverage limitation level is only calculated for people who experienced limitation onset, the ns for this were 3,479 among African Americans, 2,083 among Hispanics, and 12,726 among Whites. dIncome was transformed using the natural log function. eWealth was transformed using the inverse hyperbolic sine function. Differences tests were chi-square for dichotomous variables and ANOVA for continuous variables. *p < .05. **p < .01. ***p < .001. View Large Table 2. Descriptive Statistics, by Race African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 African Americans Hispanics Whites Mean/a proportion Standard deviation Mean/ proportion Standard deviation Mean/ proportion Standard deviation Any limitation onsetb 0.70*** — 0.70 — 0.67 — Limitation levelb,c 4.02*** 2.81 3.83 2.74 3.05 2.38 Chronic conditions 1.72*** 1.37 1.43 1.33 1.38 1.25 Education 12.03*** 3.03 9.46 4.57 13.05 2.66 Incomed 10.10*** 1.16 10.02 1.13 10.70 1.00 Wealthe 7.15*** 7.69 7.78 7.37 11.61 4.76 Employer insurance 0.54*** — 0.40 — 0.70 — Market insurance 0.05*** — 0.05 — 0.12 — Male 0.43*** — 0.48 — 0.51 — Age 57.16*** 5.55 57.03 5.56 58.53 6.68 Married 0.55*** — 0.74 — 0.79 — Immigrant 0.07*** — 0.60 — 0.04 — # of Waves 4.71*** 3.16 4.59 3.10 6.09 3.17 n = 4,211 n = 2,564 n = 15,021 Notes. aSignificances tests indicate that the values vary across the three racial groups. bAveraged across waves within persons. cAverage limitation level is only calculated for people who experienced limitation onset, the ns for this were 3,479 among African Americans, 2,083 among Hispanics, and 12,726 among Whites. dIncome was transformed using the natural log function. eWealth was transformed using the inverse hyperbolic sine function. Differences tests were chi-square for dichotomous variables and ANOVA for continuous variables. *p < .05. **p < .01. ***p < .001. View Large Multivariable Results From Structural Equation Models Results from LGCRO estimated in a SEM framework are shown in Table 3. In model 1, African Americans experience 0.43 more function limitations at onset (and, not shown, but available in Supplementary Material, African Americans experience greater odds of becoming functionally limited prior to adjusting for chronic conditions). Similarly, Hispanics have 48% greater odds of becoming functionally limited than non-Hispanic Whites, and experience 0.60 more function limitations at onset. Additionally, consistent with expectations derived from the disablement process model, those who suffer from more chronic conditions have greater odds of becoming functionally limited (OR = 1.88, p < .001), higher initial level of functional limitations (B = 0.836, p < .001), and an increased growth of functional limitations over time (B = 0.01, p < .001). Table 3. Latent Growth Curve Models with Random Onset Predicting Functional Limitations Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Notes. n = 21,796. Coefficients for the hazards are odds ratios, numbers in parentheses are standard errors. (α) denotes intercept of the latent intercept, and (ß) denotes intercept of the latent slopes. aReferences is White. bIncome was transformed using the natural log function. cWealth was transformed using the inverse hyperbolic sine function. dReferences is no private insurance. eReferences is female. fReferences is unmarried. *p < .05. **p < .01. ***p < .001. View Large Table 3. Latent Growth Curve Models with Random Onset Predicting Functional Limitations Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Model 1 Model 2 Model 3 Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent intercept (α) Latent slope (ß) Hazards odds ratio Latent Intercept (α) Latent Slope (ß) African Americana 1.057 0.427*** 0.002 0.978 0.151 0.018 0.805*** −0.094 0.006 (0.030) (0.062) (0.010) (0.046) (0.107) (0.016) (0.047) (0.105) (0.016) Hispanica 1.476*** 0.603*** 0.017 1.310*** 0.406** 0.049* 0.929 −0.202 0.031 (0.043) (0.090) (0.014) (0.057) (0.133) (0.020) (0.060) (0.133) (0.020) Chronic conditions 1.878*** 0.836*** 0.008** 1.831*** 0.799*** 0.011** 1.786*** 0.716*** 0.009** (0.011) (0.018) (0.003) (0.013) (0.021) (0.003) (0.013) (0.020) (0.003) African American × Chronic Conditions — — — 1.068* 0.152** −0.009 1.066* 0.091* −0.004 (0.029) (0.047) (0.007) (0.029) (0.046) (0.007) Hispanic × Chronic conditions — — — 1.127** 0.117* −0.019* 1.127** 0.090 −0.012 (0.038) (0.058) (0.009) (0.038) (0.056) (0.009) Education — — — 0.959*** −0.052*** −0.006*** (0.004) (0.008) (0.001) Incomeb — — — — — — 0.852*** −0.301*** 0.003 (0.014) (0.028) (0.004) Wealthc — — — — — — 0.981*** −0.031*** 0.001 (0.002) (0.005) (0.001) Employer insuranced — — — — — — 0.868*** −0.582*** 0.017 (0.031) (0.061) (0.010) Market insuranced — — — — — — 0.948 −0.290*** 0.004 (0.045) (0.083) (0.014) Malee 0.634*** −0.444*** 0.000 0.636*** −0.440*** 0.000 0.647*** −0.427*** −0.001 (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) (0.022) (0.045) (0.007) Age 1.020*** −0.050*** 0.013*** 1.020*** −0.050*** 0.013*** 1.012*** −0.061*** 0.013*** (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) (0.002) (0.004) (0.001) Marriedf 0.959 -0.299*** −0.009 0.959 −0.296*** −0.009 1.168*** 0.179** −0.017 (0.027) (0.054) (0.008) (0.027) (0.054) (0.008) (0.029) (0.058) (0.009) Immigrant 0.845*** −0.063 −0.039** 0.848*** −0.059 −0.039** 0.795*** −0.176* −0.048*** (0.041) (0.089) (0.014) (0.041) (0.089) (0.014) (0.042) (0.088) (0.013) # of Waves 0.915*** −0.150*** 0.018*** 0.915*** −0.150*** 0.018*** 0.919*** −0.135*** 0.015*** (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) (0.004) (0.008) (0.001) Mean/intercept 5.743*** −0.797*** 5.768*** −0.799*** 10.714*** −0.740*** (0.245) (0.037) (0.245) (0.037) (0.365) (0.058) Variance 4.146*** 0.085*** 4.141*** 0.085*** 3.845*** 0.083*** Cov (α, β) −0.253*** −0.253*** −0.247*** Log-likelihood −232196.572 −232180.999 −231214.638 AIC 464491.145 464471.999 462569.277 Notes. n = 21,796. Coefficients for the hazards are odds ratios, numbers in parentheses are standard errors. (α) denotes intercept of the latent intercept, and (ß) denotes intercept of the latent slopes. aReferences is White. bIncome was transformed using the natural log function. cWealth was transformed using the inverse hyperbolic sine function. dReferences is no private insurance. eReferences is female. fReferences is unmarried. *p < .05. **p < .01. ***p < .001. View Large In model 2 we include the interactions between African American and chronic conditions, and Hispanic and chronic conditions. The inclusions of these interaction terms resulted in a akaike information criterion (AIC) decrease of ~19 (464491.145–464471.999), and when added parameters decrease the AIC by more than 2 it is generally interpreted as in improvement in model fit (Burnham & Anderson, 2003, pp. 87–88). According to these results, the impact of each chronic condition on the odds of becoming functionally limited is greater for both African Americans and Hispanics than it is for non-Hispanic Whites. Model estimated predicted odds ratios of functional limitation onset based on this model are shown in the first panel of Figure 1. According to this figure, the racial differences in the odds of functional limitations onset are comparable when the number of chronic conditions are low—and at zero chronic conditions, there are no significant differences between African Americans and Whites. However, as the number of chronic conditions increases, the racial differences in the risk of functional limitations onset increase. Thus, the association between chronic conditions and functional limitations onset is clearly more deleterious for African Americans and Hispanics than it is for Whites, at least prior to adjusting for SES. Figure 1. View largeDownload slide Model estimated odds of functional limitation onset by race and number of chronic conditions. Notes: 1Estimated from model 2 of Table 3, 2Estimated from model 3 of table 3. Values are (1-odds ratios), and the comparison. Figure 1. View largeDownload slide Model estimated odds of functional limitation onset by race and number of chronic conditions. Notes: 1Estimated from model 2 of Table 3, 2Estimated from model 3 of table 3. Values are (1-odds ratios), and the comparison. In Model 3, we show results of models wherein we assess the extent to which socioeconomic variables mediate the moderating effect of race on the association between chronic conditions and functional limitations. In this model, education, household income, wealth, and insurance coverage were all significantly associated with functional limitations. However, despite accounting for all these socioeconomic measures, the association between chronic conditions and functional limitations onset remained stronger for African Americans and Hispanics than for Whites. Model estimated predicted odds ratios of functional limitation onset based on this model are shown in the second panel of Figure 1. Here it becomes apparent that adjusting for SES reduces the odds of functional limitation onset across all three racial groups; however, the general pattern of racial differences in the association between chronic conditions and functional limitations remains. That is, the racial differences in the odds of functional limitations onset are fairly comparable when the number of chronic conditions are low—and in fact, African Americans experience fewer functional limitations at zero chronic conditions than do Whites. However, as the number of chronic conditions increases, the racial differences in the risk of functional limitations onset increase. Thus, after adjusting for SES, the association between chronic conditions and functional limitations onset remains more deleterious for African Americans and Hispanics than it is for Whites. Discussion Using the 1994 to 2012 waves of the HRS, we assessed (a) the extent to which the association between chronic conditions and functional limitations onset, level, and growth over time varied by race/ethnicity, and (b) the extent to which SES accounted for racial variation in the association between chronic conditions and functional limitations. These findings not only provide a more nuanced understanding of racial variation in the disablement process, but also have important implications for how and when interventions may be most efficacious in terms of reducing racial disparities in functional health. Moreover, this study makes important contributions to existing literature on the disablement process model, and fundamental cause theory more broadly. Our first hypothesis was the associations between chronic conditions and (a) risk of onset of functional limitations; (b) initial level of functional limitations; and (c) subsequent growth of functional limitations would be greater for African Americans and Hispanics than for non-Hispanic Whites. This hypothesis was partially confirmed. Before adjusting for socioeconomic variables, we found that the rate at which chronic conditions translate to functional limitation onset is faster for racial and ethnic minorities and that chronic conditions do translate to higher initials levels (but not subsequent growth) of limitations among minority groups. This has several important insights for the disablement process model. First, the disablement process model highlights that the rate at which functional limitations occur over time is often conditional upon certain risk factors, including demographic characteristics like race (Braungart Fauth et al., 2007; Verbrugge & Jette, 1994). Our research illuminates the way in which race/ethnicity may be considered a double risk factor in the process of disablement. First, when considered by itself, there is increased direct risk in terms of the progression of transitioning from healthy to disabled among African Americans and Hispanics largely due to the experience of chronic disease. Second, there is a multiplicative risk wherein each chronic condition experienced among minorities hastens the onset and initial level of functional limitations significantly more than for Whites. This suggests racial and ethnic minorities experience the disablement process even more rapidly in the presence of chronic conditions. This finding is concordant with other research on racial disparities in health that highlights the multiplicative (rather than just additive) ways in which racial inequality manifests itself in health inequities (Brown, Richardson, Hargrove, & Thomas, 2016). Second, we agree with others who suggest, because the disablement process operates differently for members of racial/ethnic minority groups, it is essential we develop targeted interventions that focus on the specific needs of these groups (Zsembik et al., 2000). Moreover, understanding the pathway from chronic conditions to functional limitations is particularly important because this is (a) after people are diagnosed with chronic disease, so they have begun the process of disablement, but (b) early enough in the disablement process that interventions may be effective. In this case, these results suggest trying to reduce functional limitations once they occur may be too late in the disablement process to help attenuate racial disparities in disablement. Rather, if practitioners and educators were interested in limiting racial disparities in functional limitations, they would be wise to consider more upstream processes because both an additive process (greater disparities) and a multiplicative process (with disease) are driving racial disparities in the onset of functional limitations. Specifically, any program that does not specifically focus—at least in part—on limiting chronic conditions and their severity is likely missing a key element in decreasing not only racial disparities in functional limitations, but likely more severe subsequent disablement processes as well. Our second hypothesis was racial differences in the association between chronic conditions and functional limitations would be explained by accounting for differences in education, household income, household wealth, and private insurance. In other words, we expected SES to mediate the extent to which race moderated the relationship between chronic conditions and functional limitations. This hypothesis was only partially supported. Although education, household income, household wealth, and insurance mediated the increased initial level of functional limitations associated with chronic conditions among Hispanics, they did not account for the significant increased odds of functional limitations onset due to chronic conditions that African Americans and Hispanics experience relative to Whites or the increased initial level of functional limitations onset due to chronic conditions among African Americans. This suggests something other than socioeconomic resources is driving the accelerated onset of functional limitations associated with increased chronic conditions among racial and ethnic minorities. According to fundamental cause theory (Phelan & Link, 2005), SES is linked to functional limitations because knowledge of and access to both “health enhancing behaviors” and health care are contingent upon SES, and SES exposes people to various forms of “risk and protective factors.” The findings from this study highlight that, although SES is fundamentally important in understanding racial differences in functional limitations, when considering the relationship between chronic conditions and functional limitations, SES fails to account fully for the exacerbated risk of functional limitation among African Americans and Hispanics because it does not account for how rapidly disease translates to limitations among these groups. Although this does not preclude that SES is operating more upstream in the disablement processes, it is also possible other barriers to care, for example, may account for the more severe or rapidly disabling nature of conditions among minorities. Additionally, the observed racial/ethnic differences are certainly not directly attributable to race or ethnicity, but rather the experience of being a racial/ethnic minority living in a racially stratified society. These findings are consistent with articulations of fundamental cause theory (Phelan & Link, 2015) positing racial differences in health would not be entirely explained by differences in socioeconomic resources. Rather, racial differences in health persist net of socioeconomic resources, in part, because racial minorities are often exposed to a wide array of racism—including, but not limited to, discrimination, segregation, or structural racism (Gee & Ford, 2011; Krieger, 2014; Williams & Mohammed, 2013). Although it is untested here because of limitations in the data, it is likely racial/ethnic minorities’ unique experiences of racism contributed to the unexplained racial/ethnic differences in the extent to which chronic conditions are associated with functional limitations. The findings presented here also emphasize the importance of developing a more nuanced understanding of risk factors and the progression of functional limitations over time, including testing the direct, mediating, and moderating effects of variables using methodological tools equipped to separate onset, initial level, and growth such as the LGCRO models used here. In the context of intervention strategies and policy, these models highlight that programs targeting economic differences may be important for ameliorating some of the disparities encountered by African Americans and Hispanics in functional limitations. However, our findings also suggest economic policies alone may do little to reduce the increased risk of functional limitations onset associated with chronic conditions among African Americans and Hispanics. For example, policies aimed at reducing racial and ethnic disparities in the preventive or early health care access and utilization necessary to diagnose and treat potentially disabling illness early need to move beyond considering financial resources alone and expand to a broader examination of factors underlying barriers to health care access and utilization among minority groups, and the deleterious ways in which exposure to racism shapes the health of racial/ethnic minorities. Limitations and Future Research Although our analysis has the benefits of modeling multiple temporal components of the disablement process, one limitation to our findings is the lack of time-varying covariate effects. Although other modeling techniques can incorporate time-varying measures of chronic conditions and SES, for example, the LGCRO which estimates multiple processes simultaneously often cannot also incorporate these effects. Since our findings suggest a robust and differential impact of chronic conditions among race and ethnic minorities that is primary in the process of accelerated onset, future studies modeling onset timing alone may incorporate time varying effects to replicate and further explore the findings shown here. Another key limitation is we were limited in the type of chronic conditions measured in the HRS and, to retain sufficient statistical power for our interactions terms, we use a simple count of conditions in estimating our interactions. Although our findings are telling, they suggest future research should examine specific conditions and broaden the range of chronic conditions studied in order to isolate the most potentially harmful conditions driving the interactive effects shown here. The focus of this research was on the early stages of the disablement process because this is an ideal time to target health-mitigating interventions. However, we hope future research delves deeper into understanding racial differences in the disablement process as a whole—something we do not do here. It would be extremely informative to understand racial variation in this process beginning with when people are healthy, and following them through to mortality—something we are unable to do here because of the high rate of survivorship in our sample. One important implication of this is that what we identify as double disadvantage here based on one interrelationship could actually be much greater or more widespread when considering racial differences in other interrelationships across the entire disablement process model. Conclusion African Americans and Hispanics experience more rapid onset and higher levels of functional limitations, but also experience greater risk of onset associated with chronic conditions compared to their White counterparts. This suggests a multiplicative double disadvantage in early stages of the disablement process for minority groups. Although traditional measures of SES attenuated disparities at similar levels of chronic conditions, the greater impact of chronic conditions for minorities remained robust for onset of functional limitations. This highlights the importance of considering how and when chronic conditions accumulate and translate to decreased functioning across disadvantaged groups, because these early stages of the disablement process may be more amenable to intervention than later stages marked by more severe disability. It also highlights the importance of moving beyond basic economic policies in decreasing racial disparities in later life health and functioning. Supplementary Material Supplementary data is available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. Conflict of Interest B. L. Kail and M. G. Taylor are both on the editorial board for the Journal of Gerontology: Social Sciences. References Albert , P. S. , & Shih , J. H . ( 2003 ). Modeling tumor growth with random onset . Biometrics , 59 , 897 – 906 . doi: 10.1111/j.0006-341X.2003.00104.x Google Scholar CrossRef Search ADS PubMed Braungart Fauth , E. , Zarit , S. H. , Malmberg , B. , & Johansson , B . ( 2007 ). Physical, cognitive, and psychosocial variables from the disablement process model predict patterns of independence and the transition into disability for the oldest-old . The Gerontologist , 47 , 613 – 624 . doi: 10.1093/geront/47.5.613 Google Scholar CrossRef Search ADS PubMed Brown , T. H. , O’Rand , A. M. , & Adkins , D. E . ( 2012 ). Race-ethnicity and health trajectories: Tests of three hypotheses across multiple groups and health outcomes . Journal of Health and Social Behavior , 53 , 359 – 377 . doi: 10.1177/0022146512455333 Google Scholar CrossRef Search ADS PubMed Brown , T. H. , Richardson , L. J. , Hargrove , T. W. , & Thomas , C. S . ( 2016 ). Using multiple-hierarchy stratification and life course approaches to understand health inequalities . Journal of Health and Social Behavior , 57 , 200 – 222 . doi: 10.1177/0022146516645165 Google Scholar CrossRef Search ADS PubMed Burnham , K. P. , & Anderson , D. R . ( 2003 ). Model selection and multimodel inference: A practical information-theoretic approach . New York : Springer Science & Business Media . Clarke , P. , & George , L. K . ( 2005 ). The role of the built environment in the disablement process . American Journal of Public Health , 95 , 1933 – 1939 . doi: 10.2105/AJPH.2004.054494 Google Scholar CrossRef Search ADS PubMed Dallo , F. J. , Booza , J. , & Nguyen , N. D . ( 2015 ). Functional limitations and nativity status among older Arab, Asian, black, Hispanic, and white Americans . Journal of Immigrant and Minority Health , 17 , 535 – 542 . doi: 10.1007/s10903-013-9943-0 Google Scholar CrossRef Search ADS PubMed Dominick , K. L. , Dudley , T. K. , Grambow , S. C. , Oddone , E. Z. , & Bosworth , H. B . ( 2003 ). Racial differences in health care utilization among patients with osteoarthritis . The Journal of Rheumatology , 30 , 2201 – 2206 . Google Scholar PubMed Dunlop , D. D. , Song , J. , Lyons , J. S. , Manheim , L. M. , & Chang , R. W . ( 2003 ). Racial/ethnic differences in rates of depression among preretirement adults . American Journal of Public Health , 93 , 1945 – 1952 . doi: 10.2105/AJPH.93.11.1945 Google Scholar CrossRef Search ADS PubMed Ferraro , K. F. , & Farmer , M. M . ( 1999 ). Utility of health data from social surveys: Is there a gold Standard for measuring morbidity ? American Sociological Review , 64 , 303 – 315 . doi: 10.2307/2657534 Google Scholar CrossRef Search ADS Fiscella , K. , Franks , P. , Gold , M. R. , & Clancy , C. M . ( 2000 ). Inequality in quality: Addressing socioeconomic, racial, and ethnic disparities in health care . JAMA , 283 , 2579 – 2584 . doi: 10.1001/jama.283.19.2579 Google Scholar CrossRef Search ADS PubMed Fuller-Thomson , E. , Nuru-Jeter , A. , Minkler , M. , & Guralnik , J. M . ( 2009 ). Black-White disparities in disability among older Americans: Further untangling the role of race and socioeconomic status . Journal of Aging and Health , 21 , 677 – 698 . doi: 10.1177/0898264309338296 Google Scholar CrossRef Search ADS PubMed Gayman , M. D. , Pai , M. , Kail , B. L. , & Taylor , M. G . ( 2013 ). Reciprocity between depressive symptoms and physical limitations: Pre- and post-retirement: Exploring racial differences . Journal of Aging and Health , 25 , 555 – 573 . doi: 10.1177/0898264313480239 Google Scholar CrossRef Search ADS PubMed Gayman , M. D. , Turner , R. J. , & Cui , M . ( 2008 ). Physical limitations and depressive symptoms: Exploring the nature of the association . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 63 , S219 – S228 . doi: 10.1093/geronb/63.4.S219 Google Scholar CrossRef Search ADS Gee , G. C. , & Ford , C. L . ( 2011 ). Structural racism and health inequities: Old issues, new directions . Du Bois Review: Social Science Research on Race , 8 , 115 – 132 . doi: 10.1017/S1742058X11000130 Google Scholar CrossRef Search ADS PubMed Goel , M. S. , Wee , C. C. , McCarthy , E. P. , Davis , R. B. , Ngo-Metzger , Q. , & Phillips , R. S . ( 2003 ). Racial and ethnic disparities in cancer screening: The importance of foreign birth as a barrier to care . Journal of General Internal Medicine , 18 , 1028 – 1035 . doi: 10.1111/j.1525-1497.2003.20807.x Google Scholar CrossRef Search ADS PubMed Haas , S. , & Rohlfsen , L . ( 2010 ). Life course determinants of racial and ethnic disparities in functional health trajectories . Social Science & Medicine , 70 , 240 – 250 . doi: 10.1016/ j.socscimed.2009.10.003 Google Scholar CrossRef Search ADS Jette , A. M . ( 1980 ). Functional Status Index: Reliability of a chronic disease evaluation instrument . Archives of Physical Medicine and Rehabilitation , 61 , 395 – 401 . Google Scholar PubMed Jette , A. M . ( 2006 ). Toward a common language for function, disability, and health . Physical Therapy , 86 , 726 – 734 . Google Scholar CrossRef Search ADS PubMed Jette , A. M. , & Deniston , O. L . ( 1978 ). Inter-observer reliability of a functional status assessment instrument . Journal of Chronic Diseases , 31 , 573 – 580 . doi: 10.1016/0021-9681(78)90017-6 Google Scholar CrossRef Search ADS PubMed Kail , B. L. , & Carr , D. C . ( 2017 ). Successful aging in the context of the disablement process: Working and volunteering as moderators on the association between chronic conditions and subsequent functional limitations . Journals of Gerontology: Social Sciences , 72 , 340 – 350 . doi: 10.1093/geronb/gbw060 Kail , B. L. , & Taylor , M. G . ( 2014 ). Cumulative inequality and racial disparities in health: Private insurance coverage and black/white differences in functional limitations . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 69 , 798 – 808 . doi: 10.1093/geronb/gbu005 Google Scholar CrossRef Search ADS Krieger , N . ( 2014 ). Discrimination and health inequities . International Journal of Health Services: Planning, Administration, Evaluation , 44 , 643 – 710 . doi: 10.2190/HS.44.4.b Google Scholar CrossRef Search ADS PubMed Lawrence , R. H. , & Jette , A. M . ( 1996 ). Disentangling the disablement process . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 51 , S173 – S182 . doi: 10.1093/geronb/51B.4.S173 Google Scholar CrossRef Search ADS Link , B. G. , & Phelan , J . ( 1995 ). Social conditions as fundamental causes of disease . Journal of Health and Social Behavior , 35 , 80 – 94 . doi: 10.2307/2626958 Google Scholar CrossRef Search ADS Masyn , K. E . ( 2004 ). Discrete-time survival analysis using latent variables . Retrieved June, 20, 2005 . Melvin , J. , Hummer , R. , Elo , I. , & Mehta , N . ( 2014 ). Volume 31-Article 17| Pages 497–510 Age patterns of racial/ethnic/nativity differences in disability and physical functioning in the United States . Age , 31 , 497 – 510 . doi: 10.4054/DemRes.2014.31.17 Nelson , A. R. , Stith , A. Y. , & Smedley , B. D . ( 2002 ). Unequal treatment: Confronting racial and ethnic disparities in health care (full printed version) . Washington, DC : National Academies Press . Ostchega , Y. , Harris , T. B. , Hirsch , R. , Parsons , V. L. , & Kington , R . ( 2000 ). The prevalence of functional limitations and disability in older persons in the US: Data from the National Health and Nutrition Examination Survey III . Journal of the American Geriatrics Society , 48 , 1132 – 1135 . doi: 10.1111/j.1532-5415.2000.tb04791.x Google Scholar CrossRef Search ADS PubMed Phelan , J. C. , & Link , B. G . ( 2005 ). Controlling disease and creating disparities: A fundamental cause perspective . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 60 ( Special Issue 2 ), S27 – S33 . doi: 10.1093/geronb/ 60.Special_Issue_2.S27 Google Scholar CrossRef Search ADS Phelan , J. C. , & Link , B. G . ( 2015 ). Is racism a fundamental cause of inequalities in health ? Annual Review of Sociology , 41 , 311 – 330 . doi: 10.1146/annurev-soc-073014-112305 Google Scholar CrossRef Search ADS Schneider , E. C. , Zaslavsky , A. M. , & Epstein , A. M . ( 2002 ). Racial disparities in the quality of care for enrollees in medicare managed care . JAMA , 287 , 1288 – 1294 . doi: 10.1001/jama.287.10.1288 Google Scholar CrossRef Search ADS PubMed Schoenbaum , M. , & Waidmann , T . ( 1997 ). Race, socioeconomic status, and health: Accounting for race differences in health . The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences , 52 , 61 – 73 . doi: 10.1093/geronb/ 52B.Special_Issue.61 Google Scholar CrossRef Search ADS Taylor , M . ( 2008 ). Timing, Accumulation, and the black/white disability gap in later life: A test of weathering . Research on Aging , 30 , 226 . doi: 10.1177/0164027507311838 Google Scholar CrossRef Search ADS Thoits , P. A . ( 2010 ). Stress and health: Major findings and policy implications . Journal of Health and Social Behavior , 51 ( Suppl ), S41 – S53 . doi: 10.1177/0022146510383499 Google Scholar CrossRef Search ADS PubMed Verbrugge , L. M. , & Jette , A. M . ( 1994 ). The disablement process . Social Science & Medicine ,  38 ,  1 – 14 . doi: 10.1016/0277- 9536(94)90294-1 Warner , D. F. , & Brown , T. H . ( 2011 ). Understanding how race/ethnicity and gender define age-trajectories of disability: An intersectionality approach . Social Science & Medicine , 72 , 1236 – 1248 . doi: 10.1016/j.socscimed.2011.02.034 Google Scholar CrossRef Search ADS Williams , D. R. , & Jackson , P. B . ( 2005 ). Social sources of racial disparities in health . Health affairs (Project Hope) , 24 , 325 – 334 . doi: 10.1377/hlthaff.24.2.325 Google Scholar CrossRef Search ADS PubMed Williams , D. R. , & Mohammed , S. A . ( 2013 ). Racism and health I . American Behavioral Scientist , 57 , 1152 – 1173 . doi: 10.1177/0002764213487340 Google Scholar CrossRef Search ADS Williams , D. R. , Neighbors , H. W. , & Jackson , J. S . ( 2003 ). Racial/ethnic discrimination and health: Findings from community studies . American Journal of Public Health , 93 , 200 – 208 . doi: 10.2105/ajph.93.2.200 Google Scholar CrossRef Search ADS PubMed Zsembik , B. A. , Peek , M. K. , & Peek , C. W . ( 2000 ). Race and ethnic variation in the disablement process . Journal of Aging and Health , 12 , 229 – 249 . doi: 10.1177/089826430001200205 Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

The Journals of Gerontology Series B: Psychological Sciences and Social SciencesOxford University Press

Published: Apr 16, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off