Journal of Clinical Epidemiology 53 (2000) 315–322
Physical activity and self-reported, physician-diagnosed osteoarthritis: is physical activity a risk factor? Yiling Chenga, Caroline A. Macerab,*, Dorothy R. Davisa, Barbara E. Ainswortha, Philip J. Tropeda, Steven N. Blairc a School of Public Health, University of South Carolina, Columbia, SC 29208, USA National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, 4770 Buford Highway (K46), Atlanta, GA 303 41, USA c Divison of Epidemiology, Cooper Institute for Aerobics Research, Dallas, TX 75230, USA Received 13 May 1997; received in revised form 16 August 1999; accepted 23 August 1999
b
Abstract This prospective study evaluated regular physical activity and self-reported physician-diagnosed osteoarthritis of the knee and/or hip joints among 16,961 people, ages 20–87, examined at the Cooper Clinic between 1970 and 1995. Among those aged 50 years and older, osteoarthritis incidence was higher among women (7.0 per 1000 person-years) than among men (4.9 per 1000 person-years, P ⫽ 0.001), while among those under 50 years of age, osteoarthritis incidence was similar between men (2.6) and women (2.7). High levels of physical activity (running 20 or more miles per week) were associated with osteoarthritis among men under age 50 after controlling for body mass index, smoking, and use of alcohol or caffeine (hazard ratio ⫽ 2.4, 95% CI: 1.5, 3.9), while no relationship was suggested among women or older men. These findings support the conclusion that high levels of physical activity may be a risk factor for symptomatic osteoarthritis among men under age 50. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Alcohol; Body mass index; Caffeine, cohort study; Etiology; Exercise; Obesity; Proportional hazards; Smoking
1. Introduction Osteoarthritis of the hip and knee represent two of the most important causes of pain and physical disability in community-dwelling adults [1,2]. This condition can be classified into two types: primary osteoarthritis (OA) is generally related to aging and heredity [3] while the specific etiology is not clear; secondary OA is caused by other diseases or conditions such as obesity, joint trauma, or repetitive joint use [3–8]. In addition, Hart et al. [9] found that hypertension, hypercholesterolemia, and high blood glucose are associated with both unilateral and bilateral knee OA independent of obesity and suggested that OA had an important systemic and metabolic component in its etiology. While many types of physical activity involve repetitive joint use that may cause cartilage attrition, physical activity should be helpful in preventing OA in several ways. First, physical activity strengthens the muscular support around joints and thereby reduces the risk of joint injury. Second, physical activity prevents the joints from ‘freezing up’ and
* Corresponding author. Tel.: (770) 488-5018; fax: (770) 488-5473. E-mail address:
[email protected] (C.A. Macera)
improves and maintains joint mobility. Third, physical activity helps to avoid obesity, a risk factor for some forms of OA. Finally, because cartilage has no blood vessels or nerves, mature cartilage cells receive nourishment only from the diffusion of substances through the cartilage matrix from joint fluid, and physical activity enhances this process [10]. However, persons who participate in competitive sports and certain types of occupational activity have been shown to be at higher risk for OA [11–13]. Regular physical activity is increasingly recognized as a major protective factor for coronary heart disease as well as a variety of related health benefits [14]. In the United States and other developed countries public health efforts are under way to encourage regular physical activity. Because leisure-time physical activity is widely promoted as a way to improve and maintain health, it is important to understand its potential effect on OA. There are a number of studies on the role of physical activity in the development of OA [5– 9,13,15–18], but the results are inconclusive. The purpose of this prospective study is to evaluate the association between physical activity and the incidence of self-reported physician-diagnosed OA, while controlling for other potential risk factors such as age, gender, body weight, current and past smoking, and consumption of alcohol and caffeine.
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(99)00 1 6 8 - 7
316
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
2. Methods 2.1. Study population The study population consists of individuals aged 20 years or older examined at the Cooper Clinic, in Dallas, TX between 1970 and 1995. At each visit the individuals received a medical evaluation and clinical examination, and completed a detailed questionnaire that included items on health status, past medical history, and lifestyle habits. Individuals come to the clinic from all 50 states, are mostly white, well-educated, and in the middle to upper socioeconomic strata. Follow-up mail-back surveys completed in 1990 and 1995 provided detailed self-reported information on physician-diagnosed illnesses and conditions. The surveys were sent to all individuals ever examined at the Cooper Clinic. 2.2. Case ascertainment The definition of OA in this study included knee and hip joints only. Knee and hip osteoarthritis were combined for analysis because the surveys did not collect the information separately. To be included as a case, subjects had to report the presence of physician-diagnosed OA of either the knee or hip joints in the 1990 or 1995 follow-up survey. All of these individuals were well-educated, and other studies of this cohort have found good correspondence between selfreported physician-diagnosed events (such as hypertension) and clinical records [19]. To further support our reliance on self-report data, we conducted a small validity study. We randomly selected 36 subjects who reported hip or knee OA and 38 who did not and conducted a chart review to determine the presence of OA of the hip or knee in the medical chart. Of the 36 subjects who reported OA on the survey, 32 (89%) had OA documented in their medical records; of the 38 subjects who did not report OA on the survey, 8 (21%) had OA documented in their medical records. The kappa test was used to statistically assess agreement between selfreported data and the person’s actual medical record. Our kappa coefficient was 0.68, which indicates substantial agreement [20,21]. A recent validation study comparing self-reported physician-diagnosed osteoarthritis (with and without pain or swelling) and a standard clinical assessment obtained agreement similar to our small validation study [22]. They suggest that self-reported physician-diagnosed OA may actually be an underestimate of the burden of disease in the community, which is also similar to our findings. 2.3. Definition of variables The major exposure variable in this analysis, physical activity, was assessed at baseline by self-reported regular exercise patterns. Those who reported no regular exercise (sedentary) were the reference group. Among those who reported any regular exercise, four levels of activity were defined: high (those who walked or jogged more than 20 miles per week); moderate (those who walked or jogged between
10 and 20 miles per week); low (those who walked or jogged up to 10 miles per week); and other (those who participate in some other regular physical activity but not walking or jogging). Walking and jogging were chosen as the basis for the physical activity index used here because it was the most common activity for this population. In addition to physical activity, six other variables (gender, age, BMI, smoking, alcohol use, and caffeine consumption) also were considered. These variables were obtained as part of the clinic visit that included a medical examination and a detailed questionnaire on lifestyle habits. Smoking and alcohol use were included as markers for unmeasured lifestyle factors that may play a role in OA. Caffeine consumption is related to bone density, which may play a role in OA. Race, education, and other sociodemographic variables were not included because of the homogeneity of this study population. All analyses were gender specific. When used as a dichotomous variable, age at baseline was categorized as: 20 to 49 years (reference level), and 50 years and older. Weight and height were measured at the baseline clinic visit. Body mass index (BMI) was calculated as: weight (in kilograms) divided by height2 (in meters). BMI was treated as a categorical variable in this study using quartiles to identify four levels with gender-specific cutpoints. Based on current smoking behavior, subjects were categorized into three groups: those who currently smoked cigarettes; those who previously smoked cigarettes; and those who never smoked cigarettes. To measure alcohol use, the ethanol content of various beverages was estimated as 1.1 g for 1 oz of beer, 2.7 g for 1 oz of wine, and 15.1 g for 1 oz of liquor [23]. Four groups were created using group cutpoints as follows: none (reference level), 0.01–50 g/per week, 50.01–150 g/per week, and 150.01 or more g/per week. Caffeine intake was summarized by equating one cup of coffee to one unit of caffeine, and one cup of tea or cola to 0.5 unit of caffeine [24]. Units of caffeine intake were grouped as follows: none (reference level); 1–24 units (or cups) per week; and 25 or more units (or cups) per week. 2.4. Statistical analysis Considering the differences between men and women in physiology, lifestyle, and other unmeasured factors, the analyses were done separately for men and women. Cox proportional hazard regression procedures were used so that the length of follow-up could be included in the model [25]. The hazard ratio (HR), or instantaneous relative risk, was used as the measure of association between exposure (physical activity) and outcome (OA). Ninety-five percent confidence intervals (95% CI) were calculated for each HR. Interaction terms for physical activity and each of the other potential risk factors were created and tested. For testing the interaction terms, other variables were treated as continuous measures, while physical activity remained a categorical variable. The likelihood ratio test was used to select vari-
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
ables and interaction terms in the model. For testing trends, the variables were included in the model as continuous variables. For describing the sample, continuous variables were summarized using means and standard deviations. Cox proportional hazard regression models were also developed for each individual variable. After the interaction between physical activity and age was found, age was divided into two groups (20–49, 50⫹) for ease in interpreting results. The models were built separately by gender and age groups. 3. Results 3.1. Descriptive analyses The study population consisted of individuals aged 20– 87 years old. The median age was 44 years for men and 43 years for women. After excluding subjects who had a diagnosis of OA prior to the baseline visit and subjects who had missing information on the outcome variable, 16,961 individuals (12,888 (76%) men, 4,073 (24%) women) were included in this study. The average follow-up time was 10.9 years (SD ⫽ 6.13) for men, and 9.9 years (SD ⫽ 5.73) for women. Overall, there were 439 incident cases of OA for men, and 162 cases for women occurring during the followup period. In this study, OA was more common among women (7.0 per 1000 person-years) than among men (4.9 per 1000 person-years) age 50 years and older (P ⫽ 0.001), while OA incidence was similar between men (2.3 per 1000 person-years) and women (2.7 per 1000 person-years) under age 50 (P ⫽ 0.116). 3.2. Univariate analysis Among men, only the highest level physical activity (20.01+ miles per week) was associated with OA but the test for trend was not statistically significant. Additionally, the highest 2 quartiles of BMI, past cigarette smoking, alcohol use (more than 14 drinks per week), and older age (50⫹) were associated with OA and the trend tests for these variables were statistically significant (Table 1). Among women there was no relationship between OA and regular physical activity, current smoking, or alcohol use. The point estimates for the three highest quartiles of BMI, drinking 25 or more caffeine units (equivalent to one cup of coffee) per week, and older age (50⫹) were associated with OA and the trend tests for these variables were statistically significant (Table 2). 3.3. Multivariate analysis In the multivariate Cox proportional hazard model, the highest level physical activity was associated with OA among men (HR ⫽ 1.9; 95% CI 1.2–2.6). After introducing the interaction terms between physical activity and other potential risk factors, only the physical activity and age interaction was statistically significant. Therefore, we con-
317
structed models for two age groups (age 20–49 and age 50 and older) for men and women separately to minimize the interaction effect of physical activity and age (Table 3). The major finding of these analyses was that a high level of physical activity (running more than 20 miles per week) was significantly associated with OA among younger men (HR ⫽ 2.4; 95% CI 1.5–3.9), but not among older men (HR ⫽ 1.2; 95% CI 0.6–2.3). Among women there were no statistically significant associations between physical activity and OA. BMI was an important OA risk factor for all groups except older men. Most of the lifestyle factors, such as alcohol use (for men) and caffeine consumption (for women), were associated with OA only among the younger subjects. No relationship between physical activity and osteoarthritis was found among older subjects even though the number of cases among subjects age 50 and older (n ⫽ 212) was close to the number of cases among younger subjects (n ⫽ 227). The mean miles per week for high mileage men under and over age 50 were 34.7 miles per week and 34.3 miles per week, respectively. The mileage in the moderate and low groups were also similar for men and women under and over age 50. There were more than twice as many men in the younger age group (n ⫽ 8,820) compared to the men aged 50 and older (n ⫽ 4,068). Because the inability to detect relationships may have been due to a relatively small sample size among older men, analyses were developed combining men and women and the two highest physical activity categories. Assuming a relative risk worth detecting of 1.5, the power increased from 42% to 85%. After controlling for BMI, there was still no evidence that OA was associated with any level of physical activity among older adults (HR ⫽ 1.3 95% CI 0.7–2.2). To ensure that the results were not due to prevalent but undiagnosed cases of osteoarthritis, all cases that developed within 3 years of the clinic visit were excluded. The results from this analysis did not vary from what was reported in Table 3. 4. Discussion This prospective study on a large cohort of men and women, aged 20–87, provides additional knowledge about the relation of physical activity to knee and hip OA. We found a positive association between high levels of physical activity and OA among young men (age 20–49) after controlling for BMI, age, smoking, alcohol, and caffeine use. However, this association was not found for women or older men. The most probable explanation for these findings is that other unmeasured factors (such as the intensity of chosen activities) may vary by the age and gender of the subjects. Many other studies have investigated the role of physical activity, particularly sports activities, on the development of OA. In a case-control study of 233 men under 50 years of age, Vingard et al. [13] found that long-term exposure to
318
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
Table 1 Cox proportional hazard univariate model for men: the hazard ratios (HR) of the potential risk factors for self-reported osteoarthritis: the CIAR, Dallas, TX, 1970–1995 Variables
No. of cases (N ⫽ 439)
No. of subjects (N ⫽ 12,888)
Person-years (N ⫽ 143,871)
Regular physical activity (milesⲐper week) High (⬎20) Moderate (10–20) Low (⬍10) Other Sedentary
34 47 85 111 162
1003 1760 3006 2846 4273
9046 16292 29505 33319 55709 Trend test : P ⫽ 0.1669
1.6 (1.1, 2.3) 1.1 (0.9, 1.2) 1.1 (0.9, 1.5) 1.2 (1.0, 1.6) 1.0
Age (years) 50⫹ 20–49
212 227
4068 8820
2.2 (1.8, 2.6) 1.0
BMI (quartiles & kgⲐm2) Q4 (27.2⫹) Q3 (25.3–27.2) Q2 (23.7–25.3) Q1 (⬍23.7)
43737 100134 Trend test : P ⬍ 0.001
118 120 110 97
3222 3184 3261 3221
32262 35547 37866 38196 Trend test : P ⬍ 0.001
1.6 (1.2, 2.1) 1.4 (1.1, 1.9) 1.2 (0.9, 1.6) 1.0
Smoking Current smoker Past smoker Never smoked
64 162 213
1711 4074 7103
1.1 (0.9, 1.2) 1.4 (1.2, 1.8) 1.0
Absolute ethanol (gⲐper week) 150.01⫹ 50.01–150 0.01–50 None
20642 43637 79592 Trend test : P ⫽ 0.041
163 103 45 128
3657 3343 2066 3822
42155 36634 19348 45734 Trend test : P ⬍ 0.001
1.4 (1.1, 1.8) 1.1 (0.8, 1.4) 0.9 (0.6, 1.3) 1.0
Caffeine equivalent (cups of coffee/per week) 25⫹ 1–24 None
113 237 89
3079 7684 2125
35318 78693 29860 Trend test : P ⫽ 0.6916
1.2 (0.9, 1.6) 1.2 (0.9, 1.5) 1.0
medium and high sports activities increased the risk of developing OA of the hip, while another study found weightbearing sports activity in women to be associated with a two- to three-fold increased risk of radiologic OA of the knees and hips [26]. However, Roos et al. [15], after excluding subjects with known knee injuries, found no difference between nonelite players and controls, but a higher rate of gonarthrosis was found among the elite players. These studies suggest that competitive sports and associated injuries may be involved in the development of OA. Younger men may be more likely to compete and to participate in vigorous physical activity and, therefore, more likely to be injured. In our study, we did not have access to specific information on injuries and were not able to control for this potential confounder. Two Framingham cohort studies of 1,415 participants with a mean age of 73 years found that habitual physical activity did not increase the risk of knee OA (confirmed by radiography) for either men or women but significantly elevated rates of asymptomatic osteophytes (radiographic knee OA) [27,28]. A 6-year prospective longitudinal study of 410
HR (95% CI)
runners’ club members and 289 community controls, age 53–75 years at study initiation, found that vigorous running activity over many years was not associated with an increase in musculoskeletal pain [29]. After comparing the prevalence of degenerative joint disease among 17 male runners (mean age, 56 years) with 18 male nonrunners (mean age, 60 years), Panush et al. [16] did not find an association between OA and years of running. Of course, in these studies, men who were injured or developed OA before age 50 may not have been available to be selected. These studies do suggest that, for many people, long-term vigorous activity does not necessarily lead to joint problems and pain. In 1990, Cooper et al. [30] followed up 583 men and women aged 55 years and older. Later, 370 of these subjects were traced and underwent repeat radiography and interview. They found the only significant determinant of progressive radiographic knee OA was obesity, and that knee pain was a predictor of incident radiographic change. They suggested that occupational or leisure physical activity influenced incidence rather than progression of radiographic knee OA. Overall, these studies are consistent with
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
319
Table 2 Cox proportional hazard univariate model for women: the hazard ratios (HR) of the potential risk factors for self-reported osteoarthritis: the CIAR, Dallas, TX, 1970–1995 Variables
No. of cases (N ⫽ 162)
No. of subjects (N ⫽ 4,073)
Person-years (N ⫽ 41,321)
Regular physical activity (milesⲐper week) High (⬎20) Moderate (10–20) Low (⬍10) Other Sedentary
6 20 23 40 73
211 495 1029 1042 1296
1721 4061 9334 10937 15268 Trend test : P ⫽ 0.896
1.0 (0.4, 2.30) 1.1 (0.9, 1.3) 0.7 (0.4, 1.1) 0.9 (0.6, 1.3) 1.0
Age (years) 50⫹ 20–49
85 77
1210 2863
2.7 (2.0, 3.7) 1.0
BMI (quartiles & kgⲐm2) Q4 (23.3⫹) Q3 (21.7–23.3) Q2 (20.4–21.7) Q1 (⬍20.4)
12001 29320 Trend test : P ⬍ 0.001
55 58 36 13
1011 1011 1028 1023
9233 10515 10727 10846 Trend test : P ⬍ 0.001
5.2 (2.8, 9.4) 4.5 (2.5, 8.3) 2.8 (1.5, 5.2) 1.0
Smoking Current smoker Past smoker Never smoked
21 44 97
369 1141 2563
1.2 (0.9, 1.5) 1.0 (0.7, 1.5) 1.0
Ethanol (gⲐweek) 150.01⫹ 50.01–150 0.01–50 None
4090 11333 25898 Trend test : P ⫽ 0.277
25 37 28 72
568 944 1003 1558
6241 9670 9092 16318 Trend test : P ⫽ 0.149
1.0 (0.6, 1.5) 0.9 (0.6, 1.4) 0.8 (0.5, 1.3) 1.0
Caffeine equivalent (cups of coffee/per week) 25⫹ 1–24 None
40 84 38
565 2764 744
6177 26330 8814 Trend test : P ⫽ 0.008
1.7 (1.1, 2.7) 0.9 (0.6, 1.3) 1.0
our findings that physical activity may be associated with OA among younger persons, but not among older persons, possibly because of the intensity of the activity or injuries experienced by younger adults. The wide age range of our study population (20–87 years) allowed us to stratify by age group. The age/physical activity interaction found in our study may explain some of the inconsistencies among previous studies and demonstrates the necessity of analyzing by age groups for exploring the relationship between OA and physical activity. There were a few differences between men and women. The incidence of OA was similar for men and women before age 50, but after age 50, the incidence for women was higher than that of men. Our findings are similar to the results of Lawrence et al. [31] and Oliveria et al. [32] in which OA was more common among men under age 45 and more common among women over age 45. These findings are consistent with the suggestion that OA among men may be associated with injuries and some specific types or intensity of physical activity, while among women, OA may be associated with systemic and metabolic components such as BMI, caffeine use, or smoking.
HR (95% CI)
Similar to others [5,7,33], we found high BMI to be directly associated with OA, supporting the hypothesis that obesity causes OA. After separate analysis by age group, BMI was not associated with OA among older men, similar to the findings of Tepper and Hochberg [6] that obesity and fat distribution were not associated with hip OA among subjects aged 55–74 years. However, our finding of a strong association of BMI and OA persisted for men and women under age 50 and for women over age 50. While some studies have found a weak negative association between cigarette smoking and large joint OA [34,35], this finding has not been consistently demonstrated in other studies [36]. Our findings do not support a protective role for past or current cigarette smoking for men or women in either age group, even though, in the unadjusted analyses, we found that smoking cigarettes was a risk factor for OA among men. Because the prevalence of current smoking is low in this population these findings are inclusive. High bone density is suggested as a risk factor for OA, while caffeine consumption is associated with low bone density, calcium loss, and hip fracture [35,37,38]. Therefore, it is possible that caffeine consumption could have a
320
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
Table 3 Cox proportional hazard model by gender and age, controlling for all other potential risk factors: the CIAR, Dallas, TX, 1970–1995 Men Cases/subjects Age group: ⬍50 (years) Regular physical activity (milesⲐper week) High (⬎20) Moderate (10–20) Low (⬍10) Other Sedentary BMI (quartiles) Q4 Q3 Q2 Q1 Absolute ethanol (gⲐper week) 150.01⫹ 50.01–150 0.01–50 None Smoking Current smoker Past smoker Never smoked Caffeine equivalent (cups of coffeeⲐper week) 25⫹ 1–24 None Age group: 50⫹ (years) Regular physical activity (milesⲐper week) High (⬎20) Moderate (10–20) Low (⬍10) Other Sedentary BMI (quartiles) Q4 Q3 Q2 Q1 Ethanol (gⲐper week) 150.01⫹ 50.01–150 0.01–50 None Smoking Current smoker Past smoker Never smoked Caffeine equivalent (cups of coffeeⲐper week) 25⫹ 1–24 None
Women HR (95% CI)
Cases/subjects
HR (95% CI)
23Ⲑ733 28Ⲑ1207 34Ⲑ2013 59Ⲑ1910 83Ⲑ2957
2.4 (1.5, 3.9) 1.2 (1.0, 1.4) 1.0 (0.6, 1.5) 1.4 (0.9, 2.0) 1.0
3Ⲑ166 9Ⲑ338 11Ⲑ718 21Ⲑ733 33Ⲑ908
1.5 (0.4, 5.1) 1.2 (0.9, 1.5) 0.8 (0.4, 1.6) 1.1 (0.6, 2.0) 1.0
63Ⲑ2123 52Ⲑ1891 71Ⲑ2465 41Ⲑ2341
2.3 (1.6, 3.5) 1.8 (1.2, 2.8) 1.8 (1.2, 2.6) 1.0
21Ⲑ607 22Ⲑ599 26Ⲑ840 8Ⲑ817
4.2 (1.8, 9.6) 3.7 (1.7, 8.4) 3.0 (1.4, 1.5) 1.0
81Ⲑ2391 55Ⲑ2415 30Ⲑ1504 61Ⲑ2510
1.5 (1.1, 2.2) 1.1 (0.7, 1.7) 1.2 (0.8, 2.0) 1.0
9Ⲑ371 15Ⲑ692 17Ⲑ778 36Ⲑ1022
0.6 (0.3, 1.3) 0.6 (0.3, 1.2) 0.8 (0.4, 1.5) 1.0
31Ⲑ1213 73Ⲑ2512 123Ⲑ5095
0.9 (0.7–1.1) 1.1 (0.8–1.5) 1.0
10Ⲑ276 21Ⲑ793 46Ⲑ1794
1.2 (0.8, 1.6) 1.0 (0.6, 1.8) 1.0
62Ⲑ2132 117Ⲑ5260 48Ⲑ1428
0.9 (0.5, 1.3) 0.9 (0.6, 1.2) 1.0
19Ⲑ402 41Ⲑ1969 17Ⲑ492
2.2 (1.0, 4.6) 1.2 (0.6, 2.3) 1.0
11Ⲑ270 19Ⲑ553 51Ⲑ993 52Ⲑ936 79Ⲑ1316
1.2 (0.6, 2.3) 1.0 (0.8, 1.2) 1.3 (0.9, 1.8) 1.1 (0.7, 1.5) 1.0
3Ⲑ45 11Ⲑ157 12Ⲑ311 19Ⲑ309 40Ⲑ388
1.4 (0.4, 4.6) 1.2 (0.9, 1.5) 0.6 (0.3, 1.2) 0.7 (0.4, 1.3) 1.0
55Ⲑ1099 68Ⲑ1293 39Ⲑ796 50Ⲑ880
1.1 (0.8, 1.7) 1.0 (0.7, 1.4) 0.9 (0.6, 1.3) 1.0
34ᐊ404 36Ⲑ412 10Ⲑ188 5Ⲑ206
3.9 (1.5, 10.1) 3.4 (1.3, 8.7) 2.3 (0.8, 6.7) 1.0
82Ⲑ1266 48Ⲑ928 15Ⲑ562 67Ⲑ1312
1.2 (0.8, 1.7) 1.0 (0.7, 1.5) 0.6 (0.3, 1.1) 1.0
16Ⲑ197 22Ⲑ252 11Ⲑ225 36Ⲑ536
1.1 (0.5, 2.0) 1.2 (0.7, 2.2) 1.0 (0.5, 2.1) 1.0
33Ⲑ498 89Ⲑ1562 90Ⲑ2008
1.1 (0.9–1.4) 1.4 (0.9–1.9) 1.0
11Ⲑ93 23Ⲑ348 51Ⲑ769
1.1 (0.8, 1.6) 1.1 (0.7, 1.9) 1.0
51Ⲑ947 120Ⲑ2424 41Ⲑ697
1.0 (0.6, 1.6) 1.1 (0.7, 1.7) 1.0
21Ⲑ163 43Ⲑ795 21Ⲑ252
2.0 (1.0, 4.0) 1.0 (0.5, 1.7) 1.0
protective effect for OA. However, in our study, weekly consumption of 25 or more units of caffeine (a unit ⫽ one cup of coffee or two cups of tea or cola) was positively associated with OA among young women. This finding suggests that a more complex mechanism other than through
decreased bone mineral density may be involved in untangling role of caffeine consumption and OA. While not directly related to OA, alcohol consumption is associated with high bone density, particularly for postmenopausal women taking estrogen [39], thus suggesting a possible role in the
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322
development of OA. However, alcohol consumption (more than 2 drinks per day) was a risk factor for OA only among men under age 50. This finding may imply that alcohol consumption may be a marker for an unmeasured lifestyle risk factor for OA. The major strengths of this prospective study include a large sample of men and women (over 16,000 persons) with a wide age range (20–87 years), assessment of physical activity levels before OA diagnosis, and a relatively long follow-up (average of 10 years). However, there are some limitations that also should be considered. Our study population was demographically similar (well-educated, high socioeconomic status, and white), thus limiting our ability to make generalizations about other population groups. While our study had a good measure of leisure-time physical activity, we were unable to identify occupational physical activity, though persons at this socioeconomic level do not usually have occupational activity that includes lifting or carrying heavy loads, kneeling, squatting, or similar physical tasks. Related to the outcome measure of hip or knee OA, these sites were collected as one item and could not be separated for analysis. One of the major limitations of our study is the reliance on self-report data, which could be influenced by poor recall. Although our subjects were highly educated and aware of their health status, there remains considerable controversy as to whether people can accurately report physician-diagnosed OA. While radiological tests are an important component for diagnosis, it is not practical to do radiological screening in large community studies. This issue is further complicated because results from radiological tests are not always consistent with clinical symptoms [22]. Because there is no definitive diagnostic test for OA, the diagnosis involves both clinical symptoms and radiological findings. While symptoms have been shown to be closely associated with the presence of osteophytes for knee OA, in general, only about half of the patients with radiographic OA have symptoms [40]. Given the difficulty of diagnosing OA, self-report information on physician-diagnosed OA may be an acceptable though not ideal method to use for case ascertainment in large epidemiologic studies. In summary, our study found that high levels of physical activity (running 20 miles or more per week) were associated with OA among young men and no evidence that similar levels of physical activity were associated with OA of the hip or knee among women or older men. Among young men, BMI and alcohol consumption were also associated with an increased risk of OA. For young women, BMI and caffeine consumption were associated with risk of OA. For those age 50 and older, no risk factor other than age is suggested for men, while BMI is still associated with OA among women. Overall, these results suggest that participation in moderate-intensity physical activity at levels recommended by recent public health guidelines [14] is not likely to increase risk of hip or knee OA.
321
Acknowledgments We thank the Cooper Clinic physicians and technicians for baseline data collection, Carolyn E. Barlow for data and survey management, and H. W. Kohl III for the design and administration of the survey. Supported in part by U.S. Public Health Service research grant AG06945 from the National Institute on Aging, Bethesda, MD. References [1] Murray CJL, Lopez AD. Disability: the invisible burden. In: A Summary of the Global Burden of Disease Cambridge: Harvard University Press; 1996. p. 21. [2] Felson DT, Zhang Y. An update on the epidemiology of knee and hip osteoarthritis with a view to prevention. Arthritis Rheum 1998;41: 1343–55. [3] Felson DT, Couropmitree NN, Chaisson CE, Hannan M, Zhang Y, McAlindon TE, LaValley M, Levy D, Myers RH. Evidence for a mendelian gene in a segregation analysis of generalized radiographic osteoarthritis. Arthritis Rheum 1998;41:1064–71. [4] Hochberg MC. Epidemiology and genetics of osteoarthritis. Curr Opin Rheumatol 1991;3:662–85. [5] Davis MA, Ettinger WH, Neuhaus JM. Obesity and osteoarthritis of the knee: evidence from the National Health and Nutrition Examination Survey (NHANES I). Semin Arthritis Rheum 1990;20(suppl. 1):34–41. [6] Tepper S, Hochberg MC. Factors associated with hip osteoarthritis: data from the National Health and Nutrition Examination Survey (NHANES-I). Am J Epidemiol 1993;137:1081–8. [7] Anderson JJ, Felson DT. Factors associated with osteoarthritis of the knee in the first National Health and Nutrition Examination Survey (NHANES-I). Am J Epidemiol 1988;128:179–89. [8] Felson DT. The epidemiology of knee osteoarthritis: results from the Framingham Osteoarthritis Study. Semin Arthritis Rheum 1990; 20(suppl. 1):42–50. [9] Hart DJ, Doyle DV, Spector TD. Association between metabolic factors and knee osteoarthritis in women: the Chingford Study. J Rheumatol 1995;22:1118–23. [10] Hall AC, Urban JPG, Gehl KA. The effects of hydrostatic pressure on matrix synthesis in articular cartilage. J Orthop Res 1991;9:1–10. [11] Cooper C, McAlindon T, Coggon D, Egger P, Dieppe P. Occupational activity and osteoarthritis of the knee. Ann Rheum Dis 1994; 53:90–3. [12] Cooper C, Inskip H, Croft P, Campbell L, Smith G, McLauren M, Coggon D. Individual risk factors for hip osteoarthritis: obesity, hip injury, and physical activity. Am J Epidemiol 1998;147:516–22. [13] Vingard E, Alfredsson L, Goldie I, Hogstedt C. Sports and osteoarthritis of the hip: an epidemiologic study. Am J Sports Med 1993;21: 195–200. [14] U.S. Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996. [15] Roos H, Lindberg H, Gardsell P, Lohmander LS, Wingstrand H. The prevalence of gonarthrosis and its relation to meniscectomy in former soccer players. Am J Sports Med 1994;22:219–22. [16] Panush RS, Schmidt C, Caldwell JR, Edwards NL, Longley S, Yonker R, Webster E, Nauman J, Stork J, Pettersson H. Is running associated with degenerative joint disease? JAMA 1986;255:1152–4. [17] Lane NE, Bloch DA, Jones HH, Marshall WH, Wood PD, Fries JF. Long-distance running, bone density, and osteoarthritis. JAMA 1986; 255:1147–51. [18] Imeokparia RL, Barrett JP, Arrieta MI, Leaverton PE, Wilson AA,
322
[19]
[20]
[21] [22] [23]
[24] [25] [26]
[27]
[28]
[29]
Y. Cheng et al. / Journal of Clinical Epidemiology 53 (2000) 315–322 Hall BJ, Marlowe SM. Physical activity as a risk factor for osteoarthritis of the knee. Ann Epidemiol 1994;4:221–30. Blair SN, Goodyear NN, Gibbons LW, Cooper KN. Physical fitness and incidence of hypertension in health normotension men and women. JAMA 1984;252:487–90. Harlow SD, Linet MS. Agreement between questionnaire data and medical records: The evidence for accuracy of recall. Am J Epidemiol 1989;129:233–48. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biotmetrics 1977;33:159–74. March LM, Schwarz JM, Carfrae BH, Bagge E. Clinical validation of self-reported osteoarthritis. Osteoarthritis Cartilage 1998;6:87–93. Provisional Table on the Nutrient Content of Beverages. Washington, DC: Human nutrition information service, U.S. Department of Agriculture, 1982. Bunker ML, McWilliams M. Caffeine content of common beverages. J Am Diet Assoc 1979;74:28–32. Cox DR. Regression models and life-tables. J R Stat Soc B 1972;34: 187–220. Spector TD, Harris PA, Hart DJ, Cicuttini FM, Nandra D, Etherington J, Wolman R, Doyle DV. Risk of osteoarthritis associated with long-term weight-bearing sports: a radiologic survey of the hips and knees in female ex-athletes and population controls. Arthritis Rheum 1996;39:988–95. Hannan MT, Felson DT, Anderson JJ, Naimark A. Habitual physical activity is not associated with knee osteoarthritis: the Framingham Study. J Rheumatol 1993;20:704–9. Felson DT, Zhang Y, Hannan M, Naimark A, Weissman B, Aliabadi P, Levy D. Risk factors for incident radiographic knee osteoarthritis in the elderly. Arthritis Rheum 1997;40:728–33. Fries JF, Singh G, Morfeld D, O’Driscoll P, Hubert H. Relationship of running to musculoskeletal pain with age. Arthritis Rheum 1996; 39:64–72.
[30] Cooper C, Dieppe PA, Snow S, Kellingray S, Coggon D, Stuart B. Determinants of incidence and progression of radiographic knee osteoarthritis. Arthritis Rheum 1997;40(suppl.):S331. [31] Lawrence RC, Hochberg MC, Kelsey JL, McDuffie FC, Medsger TA Jr, Felts WR, Shulman LE. Estimates of the prevalence of selected arthritic and musculoskeletal diseases in the United States. J Rheumatol 1989;16:427–41. [32] Oliveria SA, Felson DT, Reed JI, Cirillo PA, Walker AM. Incidence of symptomatic hand, hip, and knee osteoarthritis among patients in a health maintenance organization. Arthritis Rheum 1995;38:1134–41. [33] Felson DT, Zhang Y, Anthony JM, Naimark A, Anderson JJ. Weight loss reduces the risk for symptomatic knee osteoarthritis in women. The Framingham study. Ann Intern Med 1992;116:535–9. [34] Samanta A, Jones A, Regan M, Wilson S, Doherty M. Is osteoarthritis in women affected by hormonal changes or smoking? Br J Rheumatol 1993;32:366–70. [35] Slemenda CW. The epidemiology of osteoarthritis of the knee. Curr Opin Rheumatol 1992;4:546–51. [36] Hart DJ, Spector TD. Cigarette smoking and risk of osteoarthritis in women in the general population: the Chingford study. Ann Rheum Dis 1993;52:93–6. [37] Massey LK, Bergman EA, Wise KJ, Sherrard DJ. Interactions between dietary caffeine and calcium on calcium and bone metabolism in older women. J Am Coll Nutr 1994;13:592–6. [38] Kiel DP, Felson DT, Hannan MT, Anderson JJ, Wilson PW. Caffeine and the risk of hip fracture: the Framingham Study. Am J Epidemiol 1990;132:675–84. [39] Felson DT, Zhang Y, Hannan MT, Kannel WB, Kiel DP. Alcohol intake and bone mineral density in elderly men and women. The Framingham study. Am J Epidemiol 1995;142:485–92. [40] Altman R. The syndrome of osteoarthritis. J Rheumatol 1997;24: 776–7.