BODY FAT DISTRIBUTION AND OSTEOARTHRITIS

Transcription

1 AMERICAN JOURNAL OF EPIDEMIOLOGY Vol. 132, No. 4 Copyright 1990 by The Johns Hopkins University School of Hygiene and Public Health Printed in U.S.A. All rights reserved BODY FAT DISTRIBUTION AND OSTEOARTHRITIS MARADEE A. DAVIS, 1 JOHN M. NEUHAUS, 1 WALTER H. ETTINGER, 2 AND WILLIAM H. MUELLER 3 Davis, M. A. (Dept. of Epidemiology and Biostatistics, School of Medicine, U. of California, San Francisco, CA ), J. M. Neuhaus, W. H. Ettinger, and W. H. Mueller. Body fat distribution and osteoarthritis. Am J Epidemiol 1990;132: The association of body fat distribution with single and combined site osteoarthritis was investigated using data from the US Health Examination Survey I, (HES I) and the first National Health and Nutrition Examination Survey 1, (NHANES I). The study included 1,636 adults aged years from HES I with hands and feet radiographs and four anthropometric fat distribution measures subscapular and triceps skinfolds, waist girth, and seat breadth and 3,885 adults aged from NHANES I with knee radiographs and subscapular and triceps skinfold measures. Sex-specific data, adjusted for age, race, and body mass index, were analyzed using polychotomous logistic regression. There was a positive association of body mass index with knee osteoarthritis and with combined hands and feet osteoarthritis. A peripheral body girth pattern was associated with combined site osteoarthritis of the hands and feet; however, there was no consistent pattern of association of body fat distribution with knee osteoarthritis nor with osteoarthritis of the hands or feet only. These findings suggest that the central body fat pattern observed in previous studies to be associated with cardiovascular and gallbladder disease, and with diabetes, is not associated with osteoarthritis of the hands, feet, or knees. body composition; obesity; osteoarthritis; skinfold thickness Osteoarthritis affects approximately 15.8 the hips (4-6), and inconsistently of the million Americans (1). Although the patho- feet (4, 7-9). However, an association of genesis and risk factors for osteoarthritis obesity with osteoarthritis of the hands, a are not well understood, increasing evi- nonweightbearing joint, has been shown dence shows that obesity is associated with (7-10). osteoarthritis in several joints (2, 3). Stud- Obesity may be a risk factor for osteoaries have found an association of obesity thritis of weightbearing joints such as the with osteoarthritis of the knee but not of knee because of increased mechanical Received for publication July 24, 1989, and in final Health Science Center at Houston, Houston, TX. form April 20, Reprint requests to Dr. Maradee A. Davis, Abbreviations: HES I, Health Examination Survey Department of Epidemiology and Biostatistics, Uni- I; NHANES I, first National Health and Nutrition versity of California, San Francisco, CA Examination Survey; OR, odds ratio. This work was supported in part by grants from 1 Department of Epidemiology and Biostatistics, the National Institute on Aging (AG07802 and School of Medicine, University of California, San AG00421), the National Institute of Arthritis and Francisco, CA. Musculoskeletal and Skin Diseases (AM-21393), and 2 Department of Medicine, Bowman Gray School the Northern California Arthritis Foundation. of Medicine, Wake Forest University, Winston-Salem, The authors thank Kenneth Mallon for assistance NC. with data analysis and Christine Choy for assistance 3 School of Public Health, The University of Texas with preparation of the manuscript. 701

2 702 DAVIS ET AL. stress across the joint. However, the association of obesity with nonweightbearing joints such as the small joints of the hands suggests other mechanisms, such as the metabolic consequences of obesity that may adversely affect joint components (2). Additionally, not only the degree of obesity but the distribution of adipose tissue may be independent risk factors. Central body fat distribution has been associated with metabolic abnormalities such as glucose intolerance, hyperlipidemia, and the development of diabetes and cardiovascular and gallbladder disease (11-16). We explored the relation of obesity and body fat distribution with osteoarthritis of the knees, feet, and hands. Specifically, we examined whether a centralized body fat distribution is associated with osteoarthritis at single or multiple joint sites. MATERIALS AND METHODS Study population The first Health Examination Survey, (HES I) (17) includes radiographs of the hands and feet for persons aged years. The first National Health and Nutrition Examination Survey (NHANES I) (18) includes bilateral knee radiographs for persons aged years. The surveys also contain extensive anthropometric data, including body fat distribution measures. Variables Classification of osteoarthritis used in HES I and NHANES I was based on standard radiographic criteria (19): 0, no osteoarthritis; 1, questionable osteoarthritis; 2, minimal osteoarthritis; 3, moderate osteoarthritis; and 4, severe osteoarthritis. In HES I, standard radiographs of hands and feet were read independently by three rheumatologists and a standard protocol was followed to adjudicate discordant gradings (20). Interobserver correlations ranged from 0.75 to 0.77 for hand films and 0.59 to 0.61 for foot films (20). In NHANES I, nonweightbearing, anterior-posterior, bilateral radiographs of the knees were read independently by two rheumatologists. Radiographs graded 1-4 were additionally read by a third rheumatologist (21). Radiographic films of every 20th examinee read as normal were reread (21); only three were reclassified (22). All radiographs were also graded for presence of osteophytes on the lateral and medial femur, lateral and medial tibia, and joint space narrowing in medial and lateral compartments. From HES I, we categorized persons aged years by radiograph status as follows: 1) normal radiographs (grade 0) for hands and feet (n = 500), 2) grades 2-4 for feet and grade 0 for hands (n = 100), 3) grades 2-4 for hands and grade 0 for feet (n = 343), and 4) grades 2-4 for both hands and feet (n = 693). Persons with missing radiographs (n = 66) were excluded. Additionally, persons with grade 1 at either joint site (n = 2,604) were excluded in order to reduce the potential effect of osteoarthritis misclassification. From NHANES I, we categorized persons aged years by radiograph status as follows: 1) normal radiographs (grade 0) for both knees (n = 3,584), 2) unilateral osteoarthritis with grades 2-4 and osteophytes or joint space narrowing in only one knee (n = 75), 3) bilateral knee osteoarthritis with grades 2-4 and osteophytes or joint space narrowing in both knees (n = 226). Excluded were persons with overall grade of 1 (n = 151), missing radiographs (n = 78), grade 0 (normal) and specific radiograph abnormalities (n = 19), and grades 2-4 and no specific radiograph abnormalities {n = 1). Height and weight were measured by standard protocol in both surveys (23)., a common indicator of obesity, was calculated as weight (kg)/ height (m) 2 (24). Central body fat distribution was measured at subscapular skinfolds and waist girth, and peripheral fat distribution at tri-

3 BODY FAT DISTRIBUTION AND OSTEOARTHRITIS 703 ceps skinfolds and seat breadth. Subscapular and triceps skinfolds were measured similarly in both surveys by trained technicians using calipers (9, 25). Waist girth and seat breadth were measured only in HES I. Waist girth was measured with a steel tape measure at the indentation of the waist line. Seat breadth was measured with an anthropometer across the greatest lateral protrusion of each buttock without compression of soft tissue; the examinee was seated with knees together (9). Reliability of anthropometric measures from HES and NHANES have been reported to be high (26, 27). Circumferential and breadth measures appear to be valid indicators of body fat distribution (28, 29). Because there is scarce previous analysis of body fat distribution and osteoarthritis, the association of fat distribution variables was examined without assuming a ratio relationship of central to peripheral measures. Race was dichotomized as white and others (black, Asian, Eskimo/Indian, other). Statistical analyses Because the upper ages differed between the two surveys and age distribution of osteoarthritis of hands and feet differed from that of the knees, different age groups were used (35-79 years, HES I; years, NHANES I). The two surveys were analyzed separately. We conducted sex-specific analyses of HES I and NHANES I, and we incorporated sample weights allowing generalization to the US population (30). Standard errors were estimated by a Taylor series approximation that takes into consideration HES I and NHANES I complex sample designs (31-33). We used polychotomous logistic regression (34) to assess the effects of body fat distribution on hands, feet, and knee' osteoarthritis, while controlling for age, race, and body mass index to assess for interaction of body fat distribution and body mass index. The CATMOD procedure of SAS Institute (35) was used to estimate parameters of the polychotomous model. Goodness-of-fit of the model was assessed by the Hosmer-Lemeshow chi-square goodness-of-fit test (36) applied to individual regressions using the approach of Begg-Gray (37). RESULTS We found no significant interactions of body fat variables with body mass index. After controlling for age, race, and each of the body girth measures in HES I, neither body mass index, waist circumference, nor seat breadth were associated with hands or feet osteoarthritis only, except for women, where waist circumference was negatively associated with hand osteoarthritis (table 1). For men and women, body mass index was positively associated with combined hands and feet osteoarthritis after controlling for the two body girth measures. For men and women, waist circumference was negatively associated with combined hands and feet osteoarthritis, and seat breadth was positively associated with hands and feet osteoarthritis, but the 95 percent confidence interval for women included one. In HES I (table 2), for men, body mass index was positively associated with hand osteoarthritis and combined hands and feet osteoarthritis; for women, it was positively associated with feet osteoarthritis. For men, neither skinfold measure was significantly associated with osteoarthritis, but there was a suggestion of a negative association of subscapular skinfolds with osteoarthritis. For women, there was no association of subscapular skinfold with osteoarthritis, but triceps skinfold was positively associated with combined hands and feet osteoarthritis (table 2). In NHANES I, for men, body mass index was positively associated with both types of knee osteoarthritis (OR = 1.36 for unilateral, 1.14 for bilateral); for women, an association was observed only for bilateral knee osteoarthritis (table 3). For men, triceps skinfold was negatively associated with unilateral knee osteoarthritis. The logistic models reported in tables 1-3 fit the data reasonably by Hosmer-

4 704 DAVIS ET AL. TABLE 1 Odds ratios and 95 percent confidence intervals (CIs), by body girth and body mass index for hands and feet osteoarthritis in persons aged years data from the US Health Examination Survey I, lands Feet Hands and feet Men Waist girth Seat breadth ( ) ( ) ( ) ( ) ( ) ( ) 1.22* 0.83* 1.45* ( ) ( ) ( ) Women Waist girth Seat breadth * 1.13 ( ) ( ) ( ) ( ) ( ) ( ) *p<0.05. t Odds ratios are adjusted for age, race, and each body fat measure in this table. TABLE * 0.83* 1.20 ( ) ( ) ( ) Odds ratios and 95 percent confidence intervals (CIs), by skinfold and body mass index, for hands and feet osteoarthritis in persons aged years data from the US Health Examination Survey I, Men Subscapular skinfold Triceps skinfold Women Subscapular skinfold Triceps skinfold 1.15* Hands ( ) ( ) ( ) ( ) ( ) ( ) * Feet ( ) ( ) ( ) ( ) ( ) ( ) *p<0.05. t Odds ratios are adjusted for age, race, and each body fat measure in this table. TABLE 3 Odds ratiol 1.16* * Hands and feet ( ) ( ) ( ) ( ) ( ) ( ) Odds ratios and 95 percent confidence intervals (CIs), by skinfold and body mass index, for unilateral and bilateral knee osteoarthritis in persons aged years data from the first National Health and Nutrition Examination Survey, Odds ratiot Unilateral knee osteoarthritis Odds ratiot Bilateral knee osteoarthritis Men 1.36* ( ) 1.14* ( ) Subscapular skinfold (cm) 0.92 ( ) 1.03 ( ) Triceps skinfold (cm) 0.89 ( ) 0.98 ( ) Women 1.07 ( ) 1.24* ( ) Subscapular skinfold (cm) 1.00 ( ) 0.99 ( ) Triceps skinfold (cm) 1.03 ( ) 1.00 ( ) * Different from normal group at p < t Odds ratios are adjusted for age, race, and each body fat measure in this table.

5 BODY FAT DISTRIBUTION AND OSTEOARTHRITIS 705 Lemeshow chi-square goodness-of-fit statistics. The significance probabilities ranged from 0.1 to 0.9. To compare the findings in NHANES I, ages 45-74, with those from HES I, ages 35-79, we additionally limited the analysis of HES I data to ages The results were substantively the same as for persons aged years in HES I (table 2). The odds ratios measuring the association of body mass index, subscapular skinfold, and triceps skinfold with combined hands and feet osteoarthritis were 1.14, 0.75, and 1.12, respectively, for men and 0.99, 1.45, and 1.58, respectively, for women. DISCUSSION We found a positive association of body mass index with knee osteoarthritis and with combined hands and feet osteoarthritis, and an inconsistent association with osteoarthritis in the hands only or feet only. In contrast, we found no association of a centralized or upper body fat pattern with osteoarthritis of the hands, feet, or knees in contrast to the reported association in studies of fat patterning with cardiovascular and gallbladder disease, and diabetes (11-16). In fact, we found some evidence of association for a peripheral body fat pattern and combined hands and feet osteoarthritis. Taken together, these data suggest that obesity increases the risk of osteoarthritis through mechanical factors or yet unknown factors, but not through metabolic consequences of a central fat pattern. It is generally hypothesized that obesity increases the risk of osteoarthritis through mechanical stress resulting in excessive wear and tear on weightbearing joints. Our data are consistent with this hypothesis with regard to osteoarthritis of the knees; however, there was no clear association of obesity with osteoarthritis of the feet, also weightbearing joints. These findings support the increasing consensus that osteoarthritis is not a single disorder but a heterogeneous group of disorders and that a complex interplay of several factors results in a final common pathway of joint damage (38). Each joint or joint group may have its own risk factors that account for the differences in association of osteoarthritis with obesity and other risk factors. Interestingly, combined hands and feet osteoarthritis was positively associated with body mass index. If combined hands and feet osteoarthritis is a form of generalized osteoarthritis, some unknown systemic effect of obesity may act adversely on diarthrodial joints. We found some evidence for a peripheral body fat pattern associated with combined hands and feet osteoarthritis. Hartz et al. (39) found a peripheral fat pattern associated with arthritis of unspecified type; they suggest that the association was the result of mechanical factors such as added weight or the effect of fat tissue in certain regions on posture and ambulatory motion. It is, however, difficult to understand how mechanical factors would account for a positive association of combined hands and feet osteoarthritis when no association with a peripheral fat pattern was observed for solo hand or foot osteoarthritis. There are several limitations to our study. The overall osteoarthritis readings in HES I were based on the most affected joint, and there may be an association of obesity or fat distribution with specific joints in the hands or feet such as the distal and proximal interphalangeal or carpalmetacarpal joints that we could not assess with these data. Further, our measures of fat distribution may not have been sensitive enough to demonstrate a weak association with osteoarthritis. Thus, an association of metabolic factors and osteoarthritis may exist that we were not able to identify with our measures of fat distribution. We have, however, previously reported that controlling for serum cholesterol, serum uric acid, diabetes, and blood pressure did not reduce the association of obesity with knee osteoarthritis (40). Further, our data are cross-sectional and do not prove a causal relation or lack of relation of body fat distribution with osteoarthritis. In conclusion, our data are consistent

6 706 DAVIS ET AL. with the hypothesis that obesity acts through a mechanical effect, at least on the knees, and that the association of central body fat distribution with cardiovascular disease is not found with osteoarthritis of the knees, hands, or feet. Further studies need to be done to explore the relation of a peripheral body fat distribution with osteoarthritis and with unknown factors by which obesity could cause cartilage degradation and joint damage. REFERENCES 1. Lawrence RC, Hochberg MC, Kelsey JL, et al. Estimates of the prevalence of selected arthritic and musculoskeletal diseases in the United States. J Rheumatol 1989;16: Davis MA. Epidemiology of osteoarthritis. Clin Geriatr Med 1988;4: Peyron JG. The epidemiology of osteoarthritis. In: Moskowitz RW, Howell DS, Goldberg VM, et al, eds. Osteoarthritis: diagnosis and management. Philadelphia: WB Saunders, 1984: Peyron JG. Epidemiologic and etiologic approach of osteoarthritis. Semin Arthritis Rheum 1979; 8: Acheson RM. Heberden Oration 1981: epidemiology and the arthritides. Ann Rheum Dis 1982; 41: Felson DT. Epidemiology of hip and knee osteoarthritis. Epidemiol Rev 1988;10:l Kellgren JH, Lawrence JS. Osteo-arthrosis and disk degeneration in an urban population. Ann Rheum Dis 1958;17: Van Saase JLCM, Vandenbroucke JP, Van Romunde LKJ, et al. Osteoarthritis and obesity in the general population: a relationship calling for explanation. J Rheumatol 1988;15: Engel A. Osteoarthritis and body measurements: the relationship of osteoarthritis to body measurements as shown in data from the Health Examination Survey, Washington, DC: National Center for Health Statistics, (Vital and health statistics, series 11, no. 29). 10. Acheson RM, Collart AB. New Haven Survey of Joint Diseases XVII: relationship between some systematic characteristics and osteoarthrosis in a general population. Ann Rheum Dis 1975;34: Bjorntorp P. Regional patterns of fat distribution. Ann Intern Med 1985; 103: Vague J. The degree of masculine differentiation of obesities: a factor determining predisposition to diabetes, atherosclerosis, gout, and uric calculous disease. Am J Clin Nutr 1956;4: Kalkhoff RK, Hartz AH, Rupley D, et al. Relationship of body fat distribution to blood pressure, carbohydrate tolerance, and plasma lipids in healthy obese women. J Lab Clin Med 1983; 102: Joos SK, Mueller WH, Hanis CL, et al. Diabetes Alert Study: weight history and upper body obesity in diabetic and non-diabetic Mexican American adults. Ann Hum Biol 1984;11: Reichley KB, Mueller WH, Hanis CL, et al. Centralized obesity and cardiovascular disease risk in Mexican Americans. Am J Epidemiol 1987;125: Haffner SM, Diehl AK, Stern MP, et al. Central adiposity and gallbladder disease in Mexican Americans. Am J Epidemiol 1989;129: National Center for Health Statistics. Plan and initial program of the Health Examination Survey: a description of the first-cycle program of the survey, including the statistical design, the pattern of examination, and the field procedures. Rockville, MD: National Center for Health Statistics, (Vital and health statistics, series 1, no. 4) (DHEW publication no. (HRA) ). 18. Miller HW. Plan and operation of the Health and Nutrition Examination Survey, United States, Hyattsville, MD: National Center for Health Statistics, (Vital and health statistics, series 1: Data from the National Health Survey, no. 10a) (DHEW publication no. (PHS) ). 19. Department of Rheumatology and Medical Illustration, University of Manchester and Manchester Royal Infirmary and the Empire Rheumatism Council's Field Unit. The epidemiology of chronic rheumatism. In: Atlas of standard radiographs of arthritis. Vol 2. Oxford: Blackwell Scientific Publications, Roberts J, Burch TA. Prevalence of osteoarthritis in adults by age, sex, and geographic area, United States, Washington, DC: National Center for Health Statistics, (Vital and health statistics, series 11: no. 15) (PHS publication no. (PHS) 1000). 21. Maurer K. Basic data on arthritis knee, hip, and sacroiliac joints in adults ages years, United States, Hyattsville, MD: National Center for Health Statistics, (Vital and health statistics, series 11: Data from the National Health Survey no. 213) (DHEW publication no. (PHS) ). 22. National Center for Health Statistics. Public use data tape documentation: arthritis, ages Hyattsville, MD: National Center for Health Statistics, (National Health and Nutrition Examination Survey, , tape no. 4121). 23. Harlan WR, Landis JR, Flegal KM, et al. Secular trends in body mass in the United States, Am J Epidemiol 1988;128: Bray GA. Definition, measurement, and classification of the syndromes of obesity. Int J Obes 1978;2: US Department of Health, Education and Welfare. Examination staff procedures manual for the health examination survey, Rockville, MD: National Center for Health Statistics, (Instruction manual, data collection, part 15C). 26. Mueller WH, Malina RM. Relative reliability of circumferences and skinfolds as measures of body fat distribution. Am J Phys Anthropol 1987; 72: Marks GC, Habicht J-P, Mueller WH. Reliability, dependability, and precision of anthropometric

7 BODY FAT DISTRIBUTION AND OSTEOARTHRITIS 707 measurements: the secopd National Health and Nutrition Examination Survey Am J Epidemiol 1989;130: Ashwell M, Chinn S, Stalley S, et al. Female fat distribution a photographic and celjularity study. Int J Obes 1978;2: Ashwell M, Chinn S, Stalley S, et al. Female fat distribution a simple classification based on two circumference measurements. Int J Obes 1982; 6: Landis J, Lepkowski S, Ek und S, et al. A statistical methodology for analyzing data from a complex survey, the First National Health and Nutrition Examination Survey. Washington, DC: National Center for Health Statistics, (Vital and health statistics, series 2, no. 92) IDHHS publication no. (PHS) ). 3\. Hidiroglou MA, Fuller WA, Hickman RD. SUPER CARP, 6th ed. Ames, IA: Survey Section, Statistical Laboratory, Iowa State University, Binder D. On the variances of asymptotically normal estimators from complex surveys. Int Stat Rev 1983;51: Shah BV, Folsom RE, Harrell FE, et a). RTILOGIT: procedure for logistic regression on survey data. Research Triangle Park, NC: Research Triangle Institute, Dubin N, Pasternack BS. Risk assessment for case-control subgroups by polychotomous logistic regression. Am J Epidemiol 1986;123: SAS Institute, Inc: SAS user's guide: statistics, version 5. Cary, NC: SAS Institute, Inc., Lemeshow S, Hosmer D Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982; 115: Begg CB, Gray R. Calculation of polychotomous logistic regression parameters using individualized regressions. Biometrika 1984;71: Dieppe PA. Osteoarthritis: aje we asking the wrong questions? (Editorial). Br J Rheumatol 1984;23: Hartz AJ, Rupley DC, Rimm AA. The association of girth measurements with disease in 32,856 women. Am J Epidemiol 1984;119: Davis MA, Ettinigr WH, Neuhaus JM. The role of metabolic factors, and blood pressure in the association of Qbgsjty with osteoarthritis of the knee. J Rheumatpl 1988;15: