CHRONIC conditions in older adults contribute to loss

Transcription

1 Journal of Gerontology: MEDICAL SCIENCES 2007, Vol. 62A, No. 9, Copyright 2007 by The Gerontological Society of America Chronic Pain Increases the Risk of Decreasing Physical Performance in Older Adults: The San Luis Valley Health and Aging Study Lucinda L. Bryant, 1 Jim Grigsby, 2 Carolyn Swenson, 3 Sharon Scarbro, 1 and Judith Baxter 1 Departments of 1 Preventive Medicine and Biometrics and 2 Medicine, Division of Health Care Policy and Research, University of Colorado at Denver and Health Sciences Center. 3 Colorado Clinical Guidelines Collaborative, Denver. Background. Pain often accompanies chronic disease in older adults and may exacerbate physical limitations, which the Disablement Model suggests may increase disability and decrease independence. This study tests the hypothesis that chronic pain and change in levels of pain over time have associations with worsening physical performance independent of disease conditions. Methods. We studied the effects of initial and changing levels of pain on observed physical performance over approximately 22 months in 925 community-dwelling Hispanic and non-hispanic white participants in the San Luis Valley Health and Aging Study. Logistic regression models controlled for demographic variables, baseline performance, and comorbidities. Results. We found that chronic pain has an independent association with worsening physical performance, regardless of ethnicity. The intensity of the pain appears to have no independent effect. Although the presence of multiple comorbidities (or vascular disease or diabetes singly) also increases the risk of a worsened physical performance outcome, an independent effect of chronic pain remains after adjusting for these disease conditions. Furthermore, ongoing chronic pain increases the risk of worsening performance; obversely, recovery from chronic pain has a significant and substantial protective effect. Conclusions. Pain in and of itself appears to increase physical impairment. These results strongly suggest that controlling chronic pain may interrupt the negative disease impairment disability trajectory by significantly reducing impaired physical performance, no matter the disease conditions that may underlie the pain. CHRONIC conditions in older adults contribute to loss of physical abilities, which in turn often leads to decreased function and increasing dependence. Pain often accompanies these conditions. Many reports have documented the relationships between chronic conditions (e.g., arthritis, diabetes, high blood pressure, heart disease) and functional disability (1 5) or physical performance, which the Disablement Model (6) suggests lie in the path from disease to disability (7 11). Incident as well as prevalent conditions predicted 3-year decline in observed physical performance in the highfunctioning cohort of the MacArthur Studies of Successful Aging (12). A review by Stuck and colleagues of 78 studies of functional decline found that depression and multiple comorbidities had the greatest effects (13). In a substudy of the Women s Health and Aging Study, Guralnik and colleagues (14) did not find an association between number of acute or chronic conditions and change in performance test results, but their assessment lasted only 6 months. Less research has examined the effect of chronic condition related pain on physical and functional outcomes, but significant relationships exist. A study of Mexican- American elders reported that pain on weight bearing significantly increased 2-year incident limitations in lowerbody activities of daily living and adversely affected observed lower body physical performance (15). A recent study of 659 adults 70 years old and older found chronic, restricting back pain associated with worsening lower extremity physical function over 18 months of follow-up (16). Researchers in another recent study, a cross-sectional assessment of approximately 1500 adults years old in the Health, Aging and Body Composition study, similarly found that participants with more severe lower back pain had significantly lower physical performance (lower body) scores (17). The Study of Assets and Health Dynamics Among the Oldest Old (AHEAD) (11) and the Established Populations for Epidemiologic Studies of the Elderly (EPESE) (10) found, respectively, that frequent pain and exertional leg pain predicted impairment, but the studies did not address independent effects of pain. A small Finnish study found pain and pain behavior the most important determinants of self-reported disability, with an additional association between pain and observed impaired performance (18). Many studies have linked socioeconomic status (SES) and poorer physical and functional outcomes; Hispanic ethnicity has not generally had an additional significant effect. Exceptions exist. Hispanic elders in the San Luis Valley Health and Aging Study (SLVHAS) reported greater prevalent dependence in activities of daily living (19,20) but no significant difference in incidence (21). In another community, researchers also found a greater prevalence 989

2 990 BRYANT ET AL. of disability among Hispanic elders, and attributed the disparity to less education and greater disease prevalence (22). A study of Mexican-American elders in five Southwestern states (23) reported that medical conditions primarily stroke and hip fracture increased the risk of self-reported lower extremity dysfunction. Multiple conditions may interact significantly: An Hispanic EPESE study of depression (24) found a significant interactive relationship with diabetes and arthritis. Data from the biethnic SLVHAS make it possible to examine the issues of chronic conditions, pain and change in pain over time, physical performance, and Hispanic or non-hispanic white ethnicity. In the study reported here we hypothesize that pain has an independent association with worsening physical performance beyond the effects of the conditions themselves, regardless of ethnicity. METHODS Population and Sampling The community-based SLVHAS examined health and disability among older Hispanic and non-hispanic white residents of rural Alamosa and Conejos counties in southern Colorado. The study design has been described in detail elsewhere (19); it is briefly summarized here. All occupied households were enumerated (97.2% response rate). Study eligibility required the participant to be 60 years old or older, to reside in either county, and to be Hispanic or non- Hispanic white based on the 1980 U.S. Census question, Are you of Spanish or Hispanic origin or descent? (25). Differential sampling ensured appropriate numbers of participants for ethnic contrasts, yielding 1757 eligible persons. Bilingual interviewers collected responses twice from 1358 community-dwelling participants between 1993 and 1995 (as well as from 75 nursing home residents, whom this study does not include because they were not offered all questions about pain; 81.6% response rate); Spanishlanguage forms were available. Refusers were more likely to be non-hispanic white (response rate 78.2%; Hispanic response rate 84.3%), less likely to report any functional disability, and slightly older. They did not differ significantly by sex, education, self-rated health, or hospitalization in the past year (19). This study excludes 188 individuals who required proxy informants because they were unable or unwilling to respond. Inability generally meant cognitive impairment based on Folstein s Mini-Mental Status Examination (MMSE) scores, 18 (26). Bilingual interviewers revisited the respondents an average of 22 months after their first visits and readministered the assessment. By that time, 67 of the participants eligible for this analysis had died, 71 refused the second visit, and 107 required proxy assistance, leaving 925 individuals available for this study. Participants who refused at visit 2 had significantly lower visit 1 MMSE scores than the respondents had (p,.01), but otherwise they did not differ significantly. Outcome Measure We measured worsened physical performance, the primary outcome, as a decrease (vs increase or no change), over the 22 months between study visits, in scores that combine Tinetti-type observations of lower extremity balance, gait, and coordination into a summary scale (27) with a maximum score of 17 (no difficulty with any item). Meaningful change was defined as change of at least 1 standard deviation in the difference between visit scores. The tests included standing up from and sitting down in a chair; bending over and picking up an object; maintaining sitting and standing balance; initiating gait, walking smoothly along a 10-foot course, and staying on a path while walking; and turning a full circle. Cronbach s alpha for the summary scale was Explanatory Variables Visit 1 observed performance. Because of the skewed distribution at visit 1 (two thirds of respondents had no difficulty with any item), we recoded scores from the continuous scale 0 17 to a dichotomous measure of no versus any difficulty with any item. Pain. Measures of pain included intensity (or severity) and chronicity. They were drawn from questions used in the San Antonio Longitudinal Study of Aging (SALSA) (28) and the EPESE (29). The question [If you are ever troubled by pain], when it is at its worst would you describe it as mild, moderate, severe or unbearable? assessed intensity, generating a dichotomous variable that compared mild with the other three responses combined. Responses to the question During the past week, how much of the time have you been troubled with pain? Would you say: all of the time, most of the time, some of the time, rarely or never? generated a dichotomous chronicity variable that compared all, most, and some of the time with rarely or never. We also evaluated new (incident) chronic or intense pain and recovery from pain at visit 2. Chronic disease. Unlike the situation in many multiethnic and rural communities, a Valley-wide network of clinics has provided reliable access to health care for both Hispanic and non-hispanic white residents of the San Luis Valley for many years. SLVHAS reports have consistently used a comorbidity score that sums (a) positive responses to a series of questions asking if a doctor had ever told respondents that they had cancer, heart attack, mild (transient ischemic attack) or severe stroke, angina, Parkinson s disease, heart failure, pulmonary disease, cirrhosis, kidney failure, osteoporosis, seizure, migraine, angioplasty or vascular disease requiring surgery, diabetes, or depressive symptoms (that is, current symptoms if ever diagnosed); (b) severe arthritis, a conservative estimate that combines selfreport of having arthritis or rheumatism accompanied by swelling or stiffness with self-report of a doctor s specific diagnosis of osteoarthritis, degenerative arthritis, or rheumatoid arthritis; and (c) high blood pressure, ascertained as self-report only in the absence of information from auscultation and medication history (n ¼ 26 of the 925 study participants). For the analyses here, we have substituted responses to the Center for Epidemiologic Studies-Depression (CES-D) inventory of depressive symptoms (30) for the

3 CHRONIC PAIN AND PHYSICAL PERFORMANCE 991 self-reported depressive symptoms score; a score of 16 indicated a high level of depressive symptoms. The comorbidity score summed the prevalence of these 18 conditions; one categorical variable distinguished among 0, 1, 2, and 3 conditions, and an incidence variable registered any visit 2 report of new chronic conditions. We also assessed cumulative incidence of chronic conditions at visit 2. Other explanatory variables. Analyses included age; gender; ethnicity (Hispanic or non-hispanic white); education (0 8 years, 9 12 years, or more than high school); general cognitive function (MMSE. 24); and body mass index (BMI) of, 22 (undernourished), (normal), or 30 (overweight/obese) kg/m 2 (11). Analysis Analysis began with descriptions of characteristics of the study sample. Pearson chi-square tests and Student t tests provided assessments of the degree to which characteristics differed by ethnicity. We examined the relationships between chronic conditions and pain to determine which conditions had the strongest associations with chronic pain. We used logistic regression to model the contributions of visit 1 pain, chronic conditions, and other explanatory variables to worsened versus improved or unchanged physical performance at visit 2. Preliminary analyses modeled the age-adjusted relationship between the outcome and each explanatory variable, including both individual chronic conditions and the summed comorbidity score. We then constructed multivariable models with factors found significant in the age-adjusted bivariate models, and tested for interactions between pain and chronic conditions. For all analyses, we used the Statistical Analysis System (SAS) version 9.1 for Windows (SAS Institute, Cary, NC). RESULTS Table 1 presents visit 1 characteristics of respondents; Table 2 adds information about incident characteristics. The mean age of the study sample at visit 1 was 72.2 years (range years), nearly 59% were female, 53% reported Hispanic ethnicity, and 30% were obese (BMI 30 kg/m 2 ). Education and MMSE scores differed significantly by ethnicity; 41% of non-hispanic white compared to 12% of Hispanic persons had more than a high school education ( p,.0001), and 4% of non-hispanic white compared to 26% of Hispanic persons scored, 24 on the MMSE ( p,.0001). A significantly greater proportion of non-hispanic white than Hispanic respondents reported a loss of any physical performance ability. Approximately 47% of respondents reported chronic pain. Chronicity did not differ by ethnicity, but significantly more non-hispanic white than Hispanic persons reported intense pain. Intensity and chronicity were moderately correlated (Spearman rho 0.59, p,.0001). Forty percent of respondents reported no pain, the same percentage reported both chronic and intense pain, and the remainder reported one or the other. Non-Hispanic white respondents were more likely to report any chronic conditions, but differences in reported total numbers of conditions were not significant. Hypertension (51%), arthritis (31%), and diabetes (24%) were the most commonly reported conditions. Angina (12% vs 8%, p ¼.05), arthritis (34% vs 28%, p ¼.04), vascular surgery (10% vs 6%, p ¼.04), any cancer (22% vs 9%, p,.0001), emphysema (12% vs 7%, p ¼.03), heart failure (8% vs 4%, p ¼.01), and osteoporosis (12% vs 4%, p,.0001) were significantly more prevalent among non-hispanic white respondents. Conversely, Hispanic persons reported significantly greater prevalences of diabetes (30% vs 16%, p,.0001) and CES-D depressive symptoms (13% vs 8%, p ¼.03). Patterns of cumulative incidence paralleled prevalence, with highest rates for hypertension and arthritis. Table 3 describes physical performance. Data include summary scores at visits 1 and 2 and meaningful differences between the two scores. The standard deviation of the difference score was 2.53 (around the mean of 0.72, which represents worsened performance); meaningful difference, therefore, represents visit 2 scores at least 3 points lower (or higher) than visit 1 scores. The mean visit 1 score was 15.9 on the 17-point scale; 16% of participants experienced worsened performance at visit 2. Applying the same definition to better versus unchanged or worse performance (not shown), 50 persons (5.5%) improved their performance. Hispanic elders had slightly but significantly higher visit 1 performance scores, and a smaller percentage of them had worsened performance at visit 2. Persons who died during the interval between study visits had lower visit 1 performance scores. Figure 1 shows the burden of chronic pain associated with the most prevalent chronic conditions. At least half of participants with at least one of these conditions high blood pressure, arthritis, diabetes, cancer, depressive symptoms, osteoporosis, and vascular disease reported chronic pain; the percentages reached 70% for persons with arthritis or depressive symptoms, singly or in combination with other conditions. Results of a logistic regression model of chronic pain on these conditions, adjusted for demographic variables (results not shown), showed that arthritis had the strongest independent association with chronic pain (odds ratio [OR] 4.23; 95% confidence interval [CI], ), followed by depressive symptoms (OR 2.09; 95% CI, ) and diabetes (OR 1.81; 95% CI, ). Similar modeling identified the chronic conditions that had statistically significant independent associations with the worsening physical performance outcome vascular disease requiring surgery (OR 1.94; 95% CI, ) and diabetes (OR 1.89; 95% CI, ). When adjusted for these other conditions, depressive symptoms did not retain a statistically significant association (OR 1.43; 95% CI, ). There were no significant interactions between pain and chronic conditions. Table 4 presents the results of logistic regression models of worsened physical performance on visit 1 pain and chronic conditions, adjusted for other explanatory factors and the presence of any visit 1 performance limitation. The first column presents the effects of baseline pain, adjusted for the basic explanatory variables. The second column shows the differential effect of changes in pain over time (i.e., recovery from pain, pain when it did not exist before). The

4 992 BRYANT ET AL. Table 1. Baseline (Visit 1) Characteristics of the Study Sample (N ¼ 925) by Ethnicity, San Luis Valley Health and Aging Study, Hispanic (N ¼ 492) Non-Hispanic White (N ¼ 433) Characteristic N % N % p Value of Ethnic Difference* Observed performance score.03 No loss (score ¼ 17) % % Any loss (score 16) % % Age,.0001, 70 years % % years % % 80 years % % Female % %.96 Education, years % % 9 12 years % % More than high school % % MMSE, % %,.0001 Body mass index % % % %, % % Pain Any chronic pain % %.12 Any intense pain % %.01 Combinations,.0001 Chronic and intense % % Chronic only % % Intense only % % Neither % % Chronic conditions Sum of chronic conditions (range 0 9) % % % % % % % % Total No. of reported chronic conditions y Individual conditions z Hypertension % %.74 Severe arthritis % %.04 Diabetes % %,.0001 Cancer % %,.0001 Heart attack % %.27 Depressive symptoms (CES-D) % %.33 Angina % %,.05 Emphysema/COPD % %.03 Migraine % %.48 Vascular disease requiring surgery % %.04 Osteoporosis % %,.0001 Notes: *Significance by chi-square test unless otherwise noted. y Significance by Student t test. z Conditions not listed had numbers too small for comparisons. MMSE¼ Mini-Mental State Examination; COPD ¼ chronic obstructive pulmonary disease; CES-D ¼ Center for Epidemiologic Studies Depression Scale. third column provides information about the degree to which the presence of chronic conditions contributes to the relationship between pain and physical performance modeled first as a sum of all comorbidities and then, in separate models, individually for the two chronic conditions that had significant relationships with the outcome. BMI did not contribute significantly to any model; it has been excluded from the results. Neither did general cognitive status (MMSE) contribute significantly, but we retained it in the model to confirm the absence of confounding with education, a common proxy for SES. (Neither variable contributed significantly to performance models, either in the presence or in the absence of the other.) Gender also had no significant effect but was retained as a common demographic adjustment. OR values for explanatory covariates were nearly identical across models for summed and separate

5 CHRONIC PAIN AND PHYSICAL PERFORMANCE 993 chronic conditions; column three of Table 4 shows results from the model that includes the sum of comorbidities. Other results are available from the corresponding author. Presence of chronic pain more than doubled the likelihood of worsened performance, but the intensity of pain did not add additional risk. Recovery from chronic pain had a significant and substantial protective effect, reducing the risk of worsened performance by about one third (OR 0.35; 95% CI, ). No other pain change variable had a significant effect. The presence of multiple comorbidities (3) also doubled the risk of worsened physical performance. Vascular disease and diabetes each separately contributed to worsened performance. Although depressive symptoms assessed by the CES-D had no significant association with the performance outcome, a model (not shown) that assessed self-reported depressive symptoms did find a statistically significant effect (OR 1.96; 95% CI, ). Neither education (a proxy measure of SES) nor ethnicity had a significant effect. DISCUSSION As hypothesized, we found that chronic pain has an independent association with worsening physical performance, regardless of ethnicity or specific condition. The intensity of pain, however, appears not to have any additional effect. Although the presence of multiple comorbidities also increases the risk of worsened physical performance, the effect of chronic pain appears substantially independent of disease condition. That is, the effect of pain on worsened physical performance changed only negligibly with the addition of disease conditions to the model, either in the aggregate or as individual conditions. Furthermore, chronic pain has a larger effect on physical performance when it persists over an extended period of time. That is, when we adjusted the presence of chronic pain at visit 1 for recovery from that pain over the period of the study, we found the risk of worsened physical performance for participants who continued to have pain to be about one third greater than estimated by the unadjusted model. Obversely, recovery from chronic pain appears to reduce the risk of worsening physical performance by two thirds. As in earlier studies reviewed by Stuck and colleagues (13), we found that multiple comorbidities and, individually, diabetes and vascular disease had the greatest significant associations with poor outcomes. Although the conservatively estimated prevalence of severe arthritis had an independent association with pain, it had no effect on the performance outcome. Because we had no information about the location of arthritis (10), we could not assess the degree to which lower extremity arthritis impairment might affect the types of physical performance measured here. The nearly 2:1 ratio of Hispanic to non-hispanic white selfreported prevalence of diabetes, higher than generally reported among older adults, is consistent with the 2- to 4- fold excess of confirmed Type II diabetes in this population (31). The estimate of the prevalence of depression (12.7% among Hispanic respondents, 8.4% among non-hispanic white respondents) falls in the middle of the range of 5% Table 2. Incident Characteristics of the Study Sample (N ¼ 925), San Luis Valley Health and Aging Study, Incidence (Recovery) at Visit 2* Characteristic N % Pain New chronic pain, from none % Recovery, from any chronic pain % New intense pain, from none % Recovery, from any intense pain % Chronic conditions Total new conditions across population (range 0 5 per individual) % y New conditions individually Hypertension % Arthritis % Diabetes % Cancer % Heart attack % Depressive symptoms (CES-D) % Angina % Emphysema/COPD % Migraine 9 1.1% Vascular disease requiring surgery % Osteoporosis % Notes: *Cumulative incidence unless otherwise noted. Cumulative incidence ¼ number of new events over the period between study visits among individuals without condition at visit 1 / number of individuals without condition at visit 1. y % of study population, not cumulative incidence. COPD ¼ chronic obstructive pulmonary disease; CES-D ¼ Center for Epidemiologic Studies Depression Scale. 30% found by Swenson and colleagues (32) in communitydwelling elderly persons, from various ethnic and racial backgrounds. It is interesting that models (not shown) that included self-reported depressive symptoms rather than CES-D assessments found a significant relationship with poor outcomes. We have much to learn about the complexities and wisdom embedded in self-reports. We did not find that incident conditions predict worsened physical performance, as found in the MacArthur study (12). It may be that new illness in a high-functioning cohort such as that in the MacArthur study has a more noticeable impact on mean levels of changes in the outcome. Although Hispanic elders in this population bear greater socioeconomic risks, have greater prevalence of diabetes, and report more functional disability than their non- Hispanic white contemporaries (19,20), on average they scored higher on tests of physical performance and reported less intense pain. We have observed this as yet unexplained Hispanic paradox in this population in other realms as well. For example, 16.8% of community-dwelling Hispanic men compared to 29.7% of non-hispanic white men in the SLVHAS cohort reported heart-related disease. Reverse migration, the tendency to return to the country of origin when ill, has been suggested as a contributing factor, but that phenomenon is uncommon in this well-established, non-immigrant old Hispanic population. Disentangling the relationships will require further study. As with all studies, this one has certain limitations. A greater proportion of persons who were eliminated from

6 994 BRYANT ET AL. Table 3. Observed Performance Scores at Visits 1 and 2 and Change in Scores, San Luis Valley Health and Aging Study, All Respondents Hispanic Non-Hispanic White Visit status N Mean SD N Mean SD N Mean SD p Visit 1 Performance Score* All visit y Visit 1 by visit 2 status,.0001 z Death Refusal Completed visit Visit 2 Performance Score* All visit 2 k y Change in Physical Performance Status Between Visit 1 and Visit 2 Change in status N % of Total k N % of Total k N % of Total k p Worse { Better/No change z Notes: Missing performance data: visit 1 ¼ 6, visit 2 ¼ 12 (excluding deaths and refusals). *Higher score indicates better performance. y Significance of ethnic difference, by Student t test. z Significance of ethnic difference, by chi-square test. Significance of differences among visit 2 outcomes, by analysis of variance (ANOVA) F test. k Total ¼ completed visit 2 with no missing relevant data. { Change defined as 1 SD (i.e., decrease of 3 points in change score). SD ¼ standard deviation. analysis because they required proxy assistance had at least one physical limitation (83.3% vs 33.1%). The absolute number of lost responses, however, is not large (n ¼ 18). The relatively short time to follow-up (22 months on average between visits) may mean that we did not observe the full impact of chronic disease and pain over time. It is interesting, however, that even in this relatively short time, recovery from pain provided dramatic protection against Figure 1. Pain associated with chronic conditions.

7 CHRONIC PAIN AND PHYSICAL PERFORMANCE 995 Table 4. Predictors of Worsened Physical Performance, San Luis Valley Health and Aging Study, Pain Pain and Change in Pain Pain, Change in Pain, and Chronic Conditions Characteristic OR 95% CI OR 95% CI OR 95% CI Baseline performance loss Age (5 years) Female gender Hispanic ethnicity Education 9 12 vs 0 8 years vs 0 8 years Poor cognition (MMSE, 24) Baseline chronic pain Recovery vs no change Incidence vs no change Baseline severe pain Recovery vs no change Incidence vs no change Comorbid conditions 1vs vs vs Selected conditions (separate models)* Vascular disease CES-D assessed depressive symptoms Diabetes Notes: *The contribution of basic explanatory variables varied little between the model of summed comorbidities and chronic conditions models that included only single items (i.e., vascular disease, depressive symptoms, diabetes); the table therefore provides estimates and confidence intervals for these basic variables only for the model that includes the sum of all comorbidities. OR¼ odds ratio; CI ¼ confidence interval; MMSE ¼ Mini-Mental Status Examination; CES-D ¼ Center for Epidemiologic Studies-Depression Scale. worsened physical performance. The definitions of intensity and chronicity of pain may have affected our results. The question about intensity asked if the respondent was ever troubled by pain, whereas the chronicity question asked about frequency of pain during the previous week. Perhaps the greatest concern has to do with self-report. Chronic conditions are among the main predictors included in the multivariate models, and this study relied primarily on self-report, which previous studies have shown may lead to misclassification. The SLVHAS assessment of chronic disease was not intended as a diagnostic test for every condition that individuals might have but as a means of estimating the chronic disease load in the population. Correlations between measured and self-reported presence of high blood pressure (0.61, p,.0001) and depressive symptomatology (0.28, p,.0001), the two conditions for which we have both types of measures, show the range of self-report validity. The parallel patterning between selfreported diabetes here and clinically ascertained Type II diabetes in the overlapping SLV Diabetes Study increases our trust in these data. These findings have important clinical and social implications. It appears that controlling chronic pain may interrupt the negative trajectory from disease to functional disability by significantly reducing impaired physical performance, no matter the disease condition or conditions that may underlie or accompany the pain. The Disablement Model has shown us that functional disability may in turn lead to loss of independence and the need for costly interventions and care. We may now have evidence to support interventions to delay or prevent those undesirable consequences while simultaneously decreasing discomfort and distress due to pain. ACKNOWLEDGMENTS The research reported here was supported by National Institute on Aging Grant R01 AG We wish to acknowledge our appreciation to the residents, most especially the older residents, of Alamosa and Conejos counties for their cooperation and participation in this study. We also thank the many investigators and staff associated with the project for their general contributions to study design, data collection, and analysis. CORRESPONDENCE Address correspondence to Lucinda L. Bryant, PhD, MSHA, University of Colorado at Denver and HSC, Department of Preventive Medicine and Biometrics, 4200 East Ninth Ave., Box C245, Office Annex 2C17, Denver, CO lucinda.bryant@uchsc.edu REFERENCES 1. Lammi UK, Kivela SL, Nissinen A, Punsar S, Puska P, Karvonen M. Predictors of disability in elderly Finnish men a longitudinal study. J Clin Epidemiol. 1989;42: Stewart AL, Greenfield S, Hays RD, et al. Functional status and wellbeing of patients with chronic conditions. Results from the Medical Outcomes Study. JAMA. 1989;262: Boult C, Kane RL, Louis TA, Boult L, McCaffrey D. Chronic conditions that lead to functional limitation in the elderly. J Gerontol. 1994;49:M28 M36.

8 996 BRYANT ET AL. 4. Guralnik JM, Kaplan GA. Predictors of healthy aging: prospective evidence from the Alameda County study. Am J Public Health. 1989;79: Strawbridge WJ, Cohen RD, Shema SJ, Kaplan GA. Successful aging: predictors and associated activities. Am J Epidemiol. 1996;144: Verbrugge LM, Jette AM. The disablement process. Soc Sci Med. 1994;38: Harris T, Kovar MG, Suzman R, Kleinman JC, Feldman JJ. Longitudinal study of physical ability in the oldest-old. Am J Public Health. 1989;79: Mor V, Murphy J, Masterson-Allen S, et al. Risk of functional decline among well elders. J Clin Epidemiol. 1989;42: Idler EL, Kasl SV. Self-ratings of health: do they also predict change in functional ability? J Gerontol Soc Sci. 1995;50B:S344 S Guralnik JM, LaCroix AZ, Abbott RD, et al. Maintaining mobility in late life. I. Demographic characteristics and chronic conditions. Am J Epidemiol. 1993;137: Clark DO, Stump TE, Hui SL, Wolinsky FD. Predictors of mobility and basic ADL difficulty among adults aged 70 years and older. J Aging Health. 1998;10: Seeman TE, Charpentier PA, Berkman LF, et al. Predicting changes in physical performance in a high-functioning elderly cohort: MacArthur Studies of Successful Aging. J Gerontol. 1994;49:M97 M Stuck AE, Walthert JM, Nikolaus T, Bula CJ, Hohmann C, Beck JC. Risk factors for functional status decline in community-living elderly people: a systematic literature review. Soc Sci Med. 1999;48: Guralnik JM, Ferrucci L, Penninx BW, et al. New and worsening conditions and change in physical and cognitive performance during weekly evaluations over 6 months: the Women s Health and Aging Study. J Gerontol Med Sci. 1999;54A:M410 M Al Snih S, Markides KS, Ray L, Goodwin JS. Impact of pain on disability among older Mexican Americans. J Gerontol Med Sci. 2001;56A:M400 M Reid MC, Williams CS, Gill TM. Back pain and decline in lower extremity physical function among community-dwelling older persons. J Gerontol A Biol Sci Med Sci. 2005;60: Hicks GE, Simonsick EM, Harris TB, et al. Cross-sectional associations between trunk muscle composition, back pain, and physical function in the health, aging and body composition study. J Gerontol A Biol Sci Med Sci. 2005;60: Koho P, Aho S, Watson P, Hurri H. Assessment of chronic pain behaviour: reliability of the method and its relationship with perceived disability, physical impairment and function. J Rehabil Med. 2001;33: Hamman RF, Mulgrew CL, Baxter J, Shetterly SM, Swenson C, Morgenstern NE. Methods and prevalence of ADL limitations in Hispanic and non-hispanic white subjects in rural Colorado: The San Luis Valley Health and Aging Study. Ann Epidemiol. 1999;9: Shetterly SM, Baxter J, Morgenstern NE, Grigsby J, Hamman RF. Higher instrumental activities of daily living disability in Hispanics compared with non-hispanic whites in rural Colorado. The San Luis Valley Health and Aging Study. Am J Epidemiol. 1998;147: Bryant LL, Shetterly SM, Baxter J, Hamman RF. Changing functional status in a biethnic rural population: The San Luis Valley Health and Aging Study. Am J Epidemiol. 2002;155: Haan MN, Weldon M. The influence of diabetes, hypertension, and stroke on ethnic differences in physical and cognitive functioning in an ethnically diverse older population. Ann Epidemiol. 1996;6: Ma J, Markides KS, Perkowski LP, Stroup-Benham CA, Lichtenstein M, Goodwin JS. Impact of selected medical conditions on self-reported lower-extremity function in Mexican-American elderly. Ethn Dis. 1998;8: Black SA, Goodwin JS, Markides KS. The association between chronic diseases and depressive symptomatology in older Mexican Americans. J Gerontol Med Sci. 1998;53A:M188 M U. S. Department of Commerce Bureau of the Census Census of Population and Housing, Summary Characteristics for Governmental Units and Standard Metropolitan Statistical Areas, Colorado. PHC Washington, DC: Government Printing Office; Folstein MF, Folstein SE, McHugh PR. Mini-mental state : a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12: Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc. 1986;34: Lichtenstein MJ, Dhanda R, Cornell JE, et al. Disaggregating pain and its effect on physical functional limitations. J Gerontol Med Sci. 1998;53A:M361 M Established Populations for Epidemiologic Studies of the Elderly. Resource Data Book. Washington, DC: National Institutes of Health; Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. App Psychol Meas. 1977;1: Hamman RF, Marshall JA, Baxter J, et al. Methods and prevalence of non-insulin-dependent diabetes mellitus in a biethnic Colorado population: The San Luis Valley Diabetes Study. Am J Epidemiol. 1989;129: Swenson CJ, Baxter J, Shetterly SM, Scarbro SL, Hamman RF. Depressive symptoms in Hispanic and non-hispanic white rural elderly: The San Luis Valley Health and Aging Study. Am J Epidemiol. 2000;152: Received June 6, 2006 Accepted December 1, 2006 Decision Editor: Luigi Ferrucci, MD, PhD