OCCUPATIONAL PHYSICAL ACTIVITIES AND OSTEOARTHRITIS OF THE KNEE

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 43, No. 7, July 2000, pp , American College of Rheumatology 1443 OCCUPATIONAL PHYSICAL ACTIVITIES AND OSTEOARTHRITIS OF THE KNEE DAVID COGGON, PETER CROFT, SAMANTHA KELLINGRAY, DAVID BARRETT, MAGNUS MCLAREN, and CYRUS COOPER Supported by a grant from the Arthritis Research Campaign. David Coggon, DM, Samantha Kellingray, MSc, Cyrus Cooper, DM: University of Southampton, Southampton, UK; Peter Croft, MD: University of Keele, Keele, UK; David Barrett, FRCS: Southampton General Hospital, Southampton, UK; Magnus McLaren, MS: Queen Alexandra Hospital, Portsmouth, UK. Address reprint requests to David Coggon, DM, MRC Environmental Epidemiology Unit, Southampton General Hospital, Southampton S016 6YD, UK. Submitted for publication October 22, 1999; accepted in revised form February 3, Objective. To assess the risk of knee osteoarthritis (OA) associated with kneeling, squatting, and other occupational activities. Methods. We compared 518 patients who were listed for surgical treatment of knee OA and an equal number of control subjects from the same communities who were matched for sex and age. Histories of knee injury and occupational activities were ascertained at interview, height and weight were measured, and the hands were examined for Heberden s nodes. Data were analyzed by conditional logistic regression. Results. After adjustment for body mass index (BMI), history of knee injury, and the presence of Heberden s nodes, risk was elevated in subjects who reported prolonged kneeling or squatting (odds ratio [] 1.9; 95% confidence interval [95% CI] ), walking >2 miles/day ( 1.9; 95% CI ), and regularly lifting weights of at least 25 kg ( 1.7; 95% CI ) in the course of their work. The risks associated with kneeling and squatting were higher in subjects who also reported occupational lifting, and appeared to interact multiplicatively with the risk conferred by obesity. People with a BMI of >30 kg/m 2 whose work had entailed prolonged kneeling or squatting had an of 14.7 (95% CI ), compared with subjects with a BMI <25 kg/m 2 who were not exposed to occupational kneeling or squatting. Conclusion. There is now strong evidence for an occupational hazard of knee OA resulting from prolonged kneeling and squatting. One approach to reducing this risk may lie in the avoidance of obesity in people who perform this sort of work. Osteoarthritis (OA) of the knee is a common cause of pain and disability in older people, and is the main reason for the 9,000 knee arthroplasties performed annually in England and Wales. Established risk factors include older age, female sex, evidence of OA in other joints, obesity, and previous injury or surgery of the knee (1). In addition, evidence is accumulating that the disease is more common in people who have performed heavy physical work (2 7), and particularly in those whose jobs have involved kneeling or squatting (8 12). Further information is needed about the occupational determinants of knee OA, both as a basis for preventive strategies and because compensation may be appropriate if individual can be attributed with reasonable confidence to employment. In Britain, this last issue was examined by the Department of Social Security s Industrial Injuries Advisory Council in 1995, but at that time, the evidence linking the disorder with work was not considered sufficiently robust to justify its classification as an occupational disease for the purposes of social security compensation (13). A limitation of most epidemiologic studies that have thus far addressed the issue has been the quality of the information collected about occupational exposure. al histories have not always been complete, and exposure has often only been inferred indirectly from the subject s job title. To provide further data on the relationship of knee OA to occupational activities, we conducted a case control study of patients requiring surgery for OA, and collected detailed lifetime histories of exposure to kneeling and other suspected risk factors in the workplace. PATIENTS AND METHODS The study population comprised residents of three English health districts, Portsmouth, Southampton (eastern

2 1444 COGGON ET AL Table 1. Association of knee osteoarthritis with body mass index, Heberden s nodes, and previous knee injury* BMI (kg/m 2 ) ( ) ( ) ( ) ( ) ( ) ( ) Heberden s nodes None Possible ( ) ( ) ( ) Definite ( ) ( ) ( ) Previous knee injury No Yes ( ) ( ) ( ) * All risk estimates are mutually adjusted. odds ratio; 95% CI 95% confidence interval; BMI body mass index. On the side that was listed for surgery (see text). part only), and Stoke-on-Trent. With help from the local orthopedic surgeons and their staffs, we attempted to identify all members of the study population who were placed on a waiting list for knee surgery (total knee arthroplasty, osteotomy, or patellar replacement) because of OA during a 2-year period. The clinical records and radiographs of each patient were reviewed, and the radiographic severity of OA in the tibiofemoral compartment of the knee listed for surgery was graded according to the Kellgren/Lawrence scale (14). Patients who had injured their knee within the previous year or fulfilled diagnostic criteria for rheumatoid arthritis or ankylosing spondylitis were excluded from further study. The remainder were eligible for inclusion as, and those who agreed were visited at home by 1 of 2 trained research nurses (one in Portsmouth and Southampton and one in Stoke-on-Trent) who interviewed them using a structured questionnaire. Among other things, the questionnaire asked about earlier injury to either knee (bad enough to impair weight bearing for 1 week) and about all jobs held since leaving school. For each job reported, we asked about the weekly frequency of different levels of lifting and whether an average working day involved any of 8 specified occupational activities, including kneeling and squatting. After the interview, the nurse measured the patient s height (using a portable stadiometer) and weight (using electronic scales), and examined his or her hands for the presence of Heberden s nodes (as a marker for OA at another joint site). For each case who was interviewed, we sought a control of the same sex and (as closely as possible) date of birth who was registered with the same general practitioner and who had not undergone previous knee surgery for OA. With permission from their general practitioners, we wrote to the potential control subjects inviting them to participate in the study. Those who agreed were visited and interviewed in the same way as the. Where possible, who could not be interviewed were replaced. Associations of knee OA with occupational activities and other risk factors were analyzed by conditional logistic regression. Since it was possible that some patients had changed their work because of early symptoms of their arthritis, when looking at occupational activities, we censored histories 10 years before entry to the study. For a few subjects, there was uncertainty about the duration of exposure to certain activities because the date of changing from one job to another was unknown. In these circumstances, we interpolated the unknown date with the assumption that each job had been held for the same length of time. Ethical approval for the study was provided by the local Research Ethics Committees in Southampton, Portsmouth, and Stoke-on-Trent. RESULTS A total of 729 OA were identified over the course of the study, of whom 675 (93%) agreed to be interviewed. The response rate among the 1,171 whom we tried to recruit was 57%, the losses occurring either because the general practitioner did not wish the person to be approached (152 subjects, or 13%) or because the control could not be contacted or declined to participate (354 subjects, or 30%). This left 665 matched pairs in which both the case and control had been interviewed; 8 of these pairs were excluded because the control had previously undergone knee surgery for OA. Among the remaining 1,314 subjects, 53 had missing information on body mass index (BMI), 77 on previous knee injury, 32 on occupational exposures, and 2 on Heberden s nodes. The analysis presented in this report is restricted to the 518 pairs of subjects (205 pairs of males and 313 pairs of females) with usable data on all of the exposure variables. The ages of the 518 ranged from 47 to 93 years (median 71.5 years), and the ages of all but 5 were matched to within 2 years. The other 5

3 KNEE OA AND OCCUPATIONAL PHYSICAL ACTIVITIES 1445 Table 2. Proportion of working life during which subjects were engaged in an occupation* Proportion of working life engaged in an occupation* 0 49% % % * Working life was defined as beginning at age 15 years and continuing until age 65/60 years (in men/women) or until 10 years before entry into the study, whichever was earlier. differed in age from their matched by up to 3.2 years. Seventy-eight percent of the had a Kellgren/Lawrence OA grade of 3 or 4 in the knee listed for surgery, and only 5% had experienced knee pain for 10 years. Table 1 summarizes the relationship of knee OA to the BMI, the presence of Heberden s nodes, and previous knee injury. In this and all further analyses, knee injuries in a case and his or her matched control were counted only if they were on the same side as the knee of the case that was listed for surgery. Each of the risk factors in Table 1 was strongly and independently associated with knee OA, and all subsequent analyses were adjusted to take into account their possible confounding effects. Table 2 shows the proportions of their lifetime for which subjects reported that they were engaged in paid work. The pattern was similar in the and the. Most men had been employed for at least 90% of their working lives, whereas many women had held jobs for less than half this time. In the analysis of occupational activities, it was necessary to interpolate the dates of job changes for 16 and 18. The association of knee OA with occupational activities is shown in Table 3. In both sexes, the disorder was more common among people who reported prolonged kneeling or squatting at work, with a summary odds ratio () of 1.9 (95% confidence interval [95% CI] ) in those who had performed either of these activities for 1 hour/day over 1 year; risk was similarly elevated in subjects who reported frequently getting up from kneeling or squatting. Significant associations were also observed with occupational lifting (both sexes), walking for 2 miles/day (especially in women), and climbing a ladder or flight of stairs 30 times/day (only in men). Table 3. Association of knee osteoarthritis with occupational activities performed for 1 year* Men Women Both Sexes Activity Lifting 10 kg 10 times/week ( ) ( ) ( ) Lifting 25 kg 10 times/week ( ) ( ) ( ) Lifting 50 kg 10 times/week ( ) ( ) ( ) Sitting 2 hours/day in total ( ) ( ) ( ) Standing or walking 2 hours/day ( ) ( ) ( ) in total Kneeling 1 hour/day in total ( ) ( ) ( ) Squatting 1 hour/day in total ( ) ( ) ( ) Kneeling 1 hour/day in total or ( ) ( ) ( ) squatting for 1 hour/day in total Getting up from kneeling or ( ) ( ) ( ) squatting 30 times/day Driving for 4 hours/day in total ( ) ( ) ( ) Walking 2 miles/day in total ( ) ( ) ( ) Climbing a ladder or flight of stairs 30 times/day ( ) ( ) ( ) * Each activity was analyzed in a separate regression model with adjustment for body mass index, Heberden s nodes, and previous knee injury classified as in Table 1. odds ratio; 95% CI 95% confidence interval.

4 1446 COGGON ET AL Table 4. Association of knee osteoarthritis with duration of exposure to selected occupational activities* Activity, duration of exposure Lifting 25 kg 10 times/week 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) Standing or walking 2 hours/day in total 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) Kneeling 1 hour/day in total or squatting 1 hour/day in total 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) Getting up from kneeling or squatting 30 times/day 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) Walking 2 miles/day in total 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) Climbing a ladder or flight of stairs 30 times/day 1.0 year years ( ) ( ) ( ) years ( ) ( ) ( ) 20.0 years ( ) ( ) ( ) * Each activity was analyzed in a separate regression model with adjustment for body mass index, Heberden s nodes, and previous knee injury classified as in Table 1. odds ratio; 95% CI 95% confidence interval. With additional adjustment for social class (based on the last occupation and classified into 6 strata), the risk estimates for lifting and walking were a little lower, although still statistically significant (data not shown). The risk estimates for kneeling, squatting, and climbing ladders and stairs were virtually unaltered; for example, in both sexes combined, the for kneeling for 1 hour/day or squatting for 1 hour/day was 2.0 (95% CI ). Table 4 shows how risk varied according to the duration of exposure to activities that were associated with knee OA. No clear exposure-response relationship was apparent for any of the activities. To explore the interaction of occupational lifting and kneeling or squatting as risk factors for knee OA, we examined their influence on risk when they occurred singly or in combination (Table 5). Risk was particularly high in subjects who reported having performed both activities in the course of their work ( 3.0; 95% CI ). The interaction of occupational kneeling and squatting with obesity was approximately multiplicative (Table 6), so that those who were exposed to the activity and had a BMI of 30 kg/m 2 had an of 14.7 (95% CI ) in comparison with unexposed subjects whose BMI was 25 kg/m 2. Table 7 shows the occupations in which kneeling or squatting was most often reported. Some of these jobs (e.g., face-trained coal miners, plumbers, and moulders and coremakers) were reported to involve kneeling or squatting by almost all subjects who had done them. For others, there was a substantial proportion of and

5 KNEE OA AND OCCUPATIONAL PHYSICAL ACTIVITIES 1447 Table 5. Association of knee osteoarthritis with combinations of occupational lifting, kneeling, and squatting* Exposure to occupational activities No kneeling/squatting or heavy lifting Kneeling/squatting but no heavy lifting ( ) ( ) ( ) Heavy lifting but no kneeling/squatting ( ) ( ) ( ) Both kneeling/squatting and heavy lifting ( ) ( ) ( ) * Kneeling/squatting was defined as kneeling for 1 hour/day or squatting for 1 hour/day for 1 year. Heavy lifting was defined as lifting 25 kg 10 times/week for 1 year. odds ratio; 95% CI 95% confidence interval. who had held the occupation but said that it did not entail kneeling or squatting. In each such job, however, the number of unexposed was similar to or greater than the number of unexposed. Repeat analyses excluding the 16 and 18 (33 matched pairs) for whom the dates of job changes had been interpolated produced similar results, as did analyses restricted to occupational activities performed before age 30 years. DISCUSSION Our findings give further support to the hypothesis that knee OA can be caused by work that involves kneeling, squatting, climbing stairs and ladders, or heavy lifting. Risk estimates remained elevated after adjustment for BMI, previous knee injury, and the presence of Heberden s nodes. One potential source of bias in this study was the possibility that patients with physically demanding jobs are more likely to be impeded by knee OA, and therefore obtain treatment for the disease at an earlier stage than those who are not engaged in strenuous work. We cannot exclude the possibility that selective presentation for treatment in this way exaggerated associations with physical activities. However, since 85% of were beyond normal retirement age at the time they were listed for surgery, any such bias was probably small. The response rate for (57%) was lower than for. It is possible that people from a poorer social background were less willing to participate as. However, the risk estimates for kneeling, squatting, and other occupational activities were little altered by adjusting for social class, suggesting that if this did occur, it was not a major source of bias. The questions used to ascertain occupational activities have previously been found to have acceptable validity in relation to subjects current work (15). However, differences in the motivation of and could have led to differential reporting of past occupational activities, with recalling exposures that they linked with their illness more completely than. To explore this possibility, we included in the interview dummy questions about activities such as sitting and driving that were unlikely to cause knee OA, and it is reassuring that no positive association was found with these variables. In addition, we looked for evidence of differential reporting of kneeling and squatting in relevant occupations (Table 7). Had the in these occupations remembered kneeling and squatting more completely than, one might have expected fewer than to report working in the jobs without kneeling and squatting. This did not occur. A few subjects could not remember the dates when they had changed from one job to another, and where this led to uncertainty about the duration of exposure to occupational activities, we were obliged to estimate the date of change. However, repeat analyses excluding these subjects gave similar results, suggesting that the missing information was not an important source of bias. Table 6. Associations of knee osteoarthritis with occupational kneeling and squatting, according to BMI (both sexes combined)* al kneeling/squatting* No Yes BMI (kg/m 2 ) ( ) ( ) ( ) ( ) ( ) * Kneeling/squatting was defined as kneeling for 1 hour/day or squatting for 1 hour/day for 1 year. BMI body mass index; odds ratio; 95% CI 95% confidence interval. Risk estimates were adjusted for Heberden s nodes and previous knee injury classified as in Table 1.

6 1448 COGGON ET AL Table 7. s in which kneeling or squatting was most frequently reported involved never involved involved kneeling or squatting kneeling or squatting kneeling or squatting Men Metal working, production, and maintenance fitters Face-trained coal miners Electricians, electrical maintenance fitters, and nonprofessional electrical engineers never involved kneeling or squatting Painters and decorators Motor mechanics Plumbers, heating and ventilating engineers Carpenters and joiners Metal plate workers, shipwrights, and boilermakers Moulders and coremakers (metal) Women Cleaners and domestics Shop assistants Agricultural workers Biased risk estimates could also have occurred if some had changed their jobs because of early symptoms of OA. If anything, such bias would be expected to obscure rather than exaggerate the impact of hazardous exposures. To minimize any such effect, we restricted our main analysis to occupations held at least 10 years before entry to the study. This was before 95% of had developed pain in the knee that required surgery. Moreover, a subsidiary analysis limited to jobs held before age 30 years produced similar results. That reported an excess of kneeling and squatting in a wide range of jobs makes it less likely that the association with knee OA is explicable by a confounding effect of other occupational activities, nor was it explained by known risk factors, such as obesity, knee injury, and the presence of Heberden s nodes (1). We did not adjust risk estimates for participation in sports, but apart from the effects of knee injury, particularly in soccer players, sporting activities have not been shown to have a major influence on the disease except in elite athletes (16). Various indices of occupational kneeling and squatting were analyzed, all of which were associated with an approximate doubling of risk, findings that are consistent with those of most previous studies and with the conclusions of 2 systematic reviews (11,12). For example, in a cross-sectional analysis of data from the first National Health and Nutrition Examination Survey, Anderson and Felson (8) found s of 2.45 in men and 3.49 in women for radiographic knee OA in occupations classified as involving knee bending at ages years. In the Framingham study, Felson and colleagues (9) reported an of 2.22 for the occurrence of radiographic knee OA in men whose jobs required knee bending and at least medium levels of physical activity. Moreover, in a case control study in Bristol, England, kneeling for at least 30 minutes per day in the main lifetime occupation carried an of 3.4 for radiographic OA associated with knee pain (10). In the current study, there was no indication that risk depended more on frequent rising from kneeling and squatting than on prolonged periods in such postures, but because the exposures were closely correlated, statistical power to discriminate between them was limited. Also, surprisingly, risk did not increase with duration of employment in work that involved kneeling or squatting. This aspect of the exposure-response relationship has not been studied previously, and it may be that longer-term exposures carry little extra risk compared with exposures of only a year or two. Alternatively, a steeper gradient in risk may have been missed through random sampling error. As in an earlier study (10), knee OA was also associated with frequent climbing of stairs and ladders at work, although only in men. The absence of an association in women could be because their pattern of expo-

7 KNEE OA AND OCCUPATIONAL PHYSICAL ACTIVITIES 1449 sure is different (e.g., relatively less use of ladders than that reported by men), or it may have occurred by chance. The association between knee OA and occupational lifting, although statistically significant, was weaker than the association with kneeling and squatting. The finding is consistent with previous reports of an increased risk of OA with heavy work (2 7), as is the apparent interaction with kneeling and squatting (9,10). People whose work had involved both heavy lifting and kneeling or squatting had a 3-fold increase in risk. In contrast, an association with extensive walking at work has not been found previously, and it may have been a chance finding. Given the evidence that has now accumulated for a hazard of knee OA from occupational kneeling and squatting, consideration should be given to possible preventive measures, especially in jobs that also involve heavy lifting. It would seem sensible to avoid prolonged work in such postures. Where avoidance of kneeling is not practical, the use of knee pads is already recommended to protect against prepatellar bursitis, but we do not know whether it reduces the risk of OA. Another option might lie in the control of nonoccupational risk factors that interact with those in the work place. For example, our findings suggest that there may be particular benefits from avoiding obesity in people who kneel or squat in their work (Table 6). In addition, our data provide further support for compensation of knee OA as an occupational disease in people who kneel or squat for long periods in their work. ACKNOWLEDGMENTS We thank Syd Anstee, Trish Byng, Lesley Campbell, Gillian Latham, and Gill Smith, who carried out the fieldwork; Vanessa Cox and Graham Wield, who supported the data handling and analysis; and Sue McIntosh, who prepared the manuscript. REFERENCES 1. Felson DT. Epidemiology of hip and knee osteoarthritis. Epidemiol Rev 1988;10: Kellgren JH, Lawrence JS. Rheumatism in miners. Part II. X-ray study. Br J Ind Med 1952;9: Partridge REH, Duthie JJR. Rheumatism in dockers and civil servants: a comparison of heavy manual and sedentary workers. Ann Rheum Dis 1968;27: Lindberg H, Montgomery F. Heavy labor and the occurrence of gonarthrosis. Clin Orthop 1987;214: Kohatsu ND, Schurman DJ. Risk factors for the development of osteoarthrosis of the knee. Clin Orthop 1990;261: Vingård E, Alfredsson L, Goldie I, Hogstedt C. and osteoarthrosis of the hip and knee: a register-based cohort study. Int J Epidemiol 1991;20: Vingård E, Alfredsson L, Fellenius E, Hogstedt C. Disability pensions due to musculo-skeletal disorders among men in heavy occupations. Scand J Soc Med 1992;20: Anderson JJ, Felson DT. Factors associated with osteoarthritis of the knee in the first National Health and Nutrition Examination Survey. Am J Epidemiol 1988;128: Felson DT, Hannan MT, Naimark A. al physical demands, knee bending and knee osteoarthritis: results from the Framingham study. J Rheumatol 1991;18: Cooper C, McAlindon T, Coggon D, Egger P, Dieppe P. al activity and osteoarthritis of the knee. Ann Rheum Dis 1994;53: Jensen LK, Eenberg W. as a risk factor for knee disorders. Scand J Work Environ Health 1996;22: Maetzel A, Mäkelä M, Hawker G, Bombardier C. Osteoarthritis of the hip and knee and mechanical occupational exposure a systematic overview of the evidence. J Rheumatol 1997;24: Department of Social Security. Disorders of the knee (Cm 2842). London: HMSO; Kellgren JH, Lawrence JS. Atlas of standard radiographs: the epidemiology of chronic rheumatism. Vol. 2. Oxford: Blackwell; Campbell L, Pannett B, Egger P, Cooper C, Coggon D. Validity of a questionnaire for assessing occupational activities. Am J Ind Med 1997;31: Lequesne MG, Dang N, Lane NE. Sport practice and osteoarthritis of the limbs. Osteoarthritis Cartilage 1997;5:75 86.