THE ASSOCIATION OF ABNORMALITIES ON PHYSICAL EXAMINATION OF THE HIP AND KNEE WITH LOCOMOTOR DISABILITY IN THE ROTTERDAM STUDY

British Journal of Rheumatology 1996;35:884-890 THE ASSOCIATION OF ABNORMALITIES ON PHYSICAL EXAMINATION OF THE HIP AND KNEE WITH LOCOMOTOR DISABILITY IN THE ROTTERDAM STUDY E. ODDING, H. A. VALKENBURG, D. ALGRA, F. A. VANDENOUWELAND,* D. E. GROBBEE and A. HOFMAN Department of Epidemiology < Biostatistics, Erasmus University Medical School, Rotterdam, The Netherlands SUMMARY To assess the influence of abnormalities found on physical examination of the hips and knees on disability 1156 men and 1739 women aged ^ 55 yr (the Rotterdam Study) were asked about locomotor disability (LD) using six questions of the Health Assessment Questionnaire (HAQ). The prevalence of LD was 20.2 for men and 31.9 for women. ly restricted range of motion of the hips and knees occurred in 34.5 of the men and 38.6 of the women. The prevalence of instability of the knees was for men and 16.8 for women. Varus deformity in men (10.1) and valgus deformity in women (15.0) were the most common deformities of the knees. Restrictedflexionof the hips was the strongest determinant of LD. Age-adjusted odds ratios for restricted hipflexionof LD were 4.7 (95 CI: 3.2-6.8) for men and 3.5 (2.7-4.5) for women. Valgus deformity, knee instability and obliquity wereriskfactors only in women. Adjustment of these odds ratios for morning stiffness and joint pain did not alter the estimates. KEY WORDS: Disability, Physical examination, Hips, Knees, General population, Elderly. LOCOMOTOR disability (LD), as defined by the difficulties people experience when carrying out basic activities of daily living depending on lower limb function, can be caused by several conditions. The overall aim of the present study was to determine which signs and symptoms affecting the major weight-bearing joints of the lower limbs were associated with LD. In a recent paper, we demonstrated the association between LD and joint pain and morning stiffness [1]. Subsequently, we studied the contribution of radiological osteoarthritis (ROA) to the occurrence of disability [2]. While symptomatic osteoarthritis, defined as ROA and pain in the corresponding joint, was strongly associated with LD, ROA without joint symptoms had only a minimal independent effect on the occurrence of disability. Abnormalities on physical examination, such as restricted range of motion of the hips and knees, instability of the knee joints and valgus and varus deformity, may explain (part) of the disability occurring with ROA. On the other hand, these abnormalities may also independently contribute to the occurrence of LD, diminishing one's ability to carry out such common functions as rising from a chair or climbing stairs. The present study analysed in 2895 people from the Rotterdam Study cohort the association between LD and restricted range of motion of the hips and knees, valgus and varus deformity, instability of the knee joint and pelvic obliquity. Submitted 13 June 1995; revised version accepted 19 March 1996. Present address: 'Procter & Gamble Pharmaceuticals, Middlesex. Correspondence to: E. Odding, Department of Rehabilitation, University Hospital Rotterdam, Dijkzigt, Dr Molewaterplein 40, 3015 GD Rotterdam, The Netherlands. POPULATION AND METHODS Study subjects The Rotterdam Study is a prospective follow-up study of the incidence and risk factors of chronic disease and disability in the general population [3]. The source population comprises all residents aged ^ 55 yr on 1 January 1989 living in the Ommoord district of Rotterdam. Baseline data on all 10 275 eligible subjects were gathered from April 1990 to July 1993. The present study is concerned with those participants who took part in the study between April 1990 and July 1992. At this stage of the study, 2247 men and 3433 women living independently were invited to participate in a home interview on, amongst other items, disability. Complete interview data were available for 1819 men (81) and 2817 women (82). In the second phase of the study, 1690 men and 2577 women visited the research centre for further examination. This corresponds with a response rate of 93 and 92, respectively, of the interviewed subjects. The average time between the interview and the centre examination was 2 weeks. Because of logistic reasons during the first year, not all measurements were carried out at the centre. This meant that complete data on locomotor signs and symptoms were available for 1156 men and 1739 women, i.e. 64 of the men and 62 of the women with complete interview data. Measurements Locomotor disability was defined as proposed by the International Classification of Impairments Disabilities and Handicaps (ICIDH) and comprised the relevant items from the ambulation subcategory, i.e. walking, climbing stairs, getting in and out of bed and a car, bending, and rising from a chair [4], and was assessed with the Stanford Health Assessment Questionnaire (HAQ) [5-9]. The HAQ measures disability in eight 884 1996 British Society for Rheumatology

ODDING ET AL.: PHYSICAL ABNORMALITIES OF HIP AND KNEE AND DISABILITY 885 components (dressing and grooming, rising, reach, hygiene, eating, walking, grip and activity), each of which consists of 2-4 questions starting with: 'Are you able to...'. Each question is answered by one of four possible answers with score 0 = without difficulty, 1 = with difficulty, 2 = with much difficulty and 3 = unable to do. A comprehensive description of the way the HAQ was assessed during the home interview carried out by one of our nine extensively trained interviewers, who were standardized on a regular basis, has been presented earlier [1]. The mean of the scores on the six questions related to lower limb functions constituted the LD index (range 0.00-3.00). Locomotor disability was considered to be present when the participants indicated at least some difficulty with three or more out of six functions; this is in concurrence with a score of >0.50 on the LD index. Pain of the hips and knees was denned as joint pain occurring during the past month at the left and/or right side. The duration of morning stiffness was assessed at three levels (<ih, f-1 h, >lh), and subsequently dichotomized to no morning stiffness or ~&\\v [1]. At the research centre, which was located in the health centre of the study district, one of 10 physicians examined both hips and knees. In the supine position, internal and external rotation of the hips,flexionof the hips and knees, and the ligamental apparatus of the knees were tested. Restriction in range of motion (ROM) was categorized as follows: 1 = no; 2 = mild; 3 = moderate restriction; 4 = severe restriction. Categories 2-4 were denned as 2: 10-20 restriction of the normal ROM, 3: 20-60 restriction, 4: >60 restriction. Normal ROM is maximal ROM for the healthy adult (i.e. 120 hip flexion, 40 internal and external rotation, and 130 knee flexion). This means that the categories for hip flexion were: 2 = ROM 97-108, 3 = ROM 49-96, 4 = ROM < 48 ; for internal and external rotation of the hip: 2 = ROM 33-36, 3 => ROM 17-32, 4 = ROM < 16 ; for knee flexion: 2 = ROM 105-117, 3 = ROM 53-104, 4 = ROM «S 52. Much effort was put into the instruction and training of the physicians who, for the purposes of physical examination, were all initially coached and standardized by an experienced orthopaedic surgeon. They were trained to assess ROM in the four categories mentioned above, without using a goniometer. The cruciate and collateral ligaments of the knees were tested for instability. In the case of a positive anterior drawer test, special attention was given to left and right comparison. The same procedure was followed with a positive lateral or medial stress test. With the participant standing upright, barefoot and without trousers or dress, valgus and varus deformity was assessed. The physicians were instructed to draw an imaginary line down from midway through the groin, through the centre of the patella towards the floor; if the medial malleolus was lateral of this line, the participant was classified as having a valgus deformity (knock-knee); if the lateral malleolus was medial of the line, the knee was considered to be in a varus position (bow-leg). Pelvic obliquity, regardless of its cause, was tested by placing the thumbs on the spinae iliacae anteriores superiores and deciding whether the line between the thumbs was in a horizontal plane (the protocol of the physical examination is available on request). Data analysis All analyses were carried out for men and women separately. Wefirstestimated the prevalence of LD and abnormalities on physical examination. Two composite measures were defined: ROM-hip = restricted flexion and/or internal rotation and/or external rotation of the hip; ROM-hip+knee = restricted ROM-hip and/or restricted flexion of the knee. All figures represent abnormalities in the left and/or right joint. To assess the association between the physical examination variables and LD, age-adjusted univariate odds ratios for LD of all separate physical examination variables were estimated using a logistic regression model. In Age range (yr) Mean age (±S.D.) (yr) Locomotor disability Hip pain Knee pain Morning stiffness Age range (yr) Mean age (±S.D.) (yr) Locomotor disability Hip pain Knee pain Morning stiffness TABLE I Some characteristics of the subjects of the present study compared with all participants of the Rotterdam Study 20.2 8.3 12.6 31.9 16.6 22.3 8.6 Present study 1156 55.0-93.2 68.6 (±7.5) (95 CI) (17.9-22.5) (6.7-9.9) (10.7-14.5) (3.4-5.8) 1739 55.0-94.0 69.4 (±8.1) (95 CI) (29.7-34.1) (14.9-18.3) (20.3-24.3) (7.3-9.9) 19.9 8.3 12.2 32.5 16.3 22.9 9.2 Research centre 1690 55.0-94.3 6 (±7.7) (95 CI) (18.0-21.8) (7.0-9.6) (10.6-13.8) (3.6-5.6) 2577 55.0-95.6 69.2 (±8.3) (95 a) (30.7-34.3) (14.9-17.7) (21.3-24.5) (8.1-10.3) 21.9 34.8 16.4 22.6 9.2 Interview 1819 55.0-94.3 68.9 (±8.0) (95 CI) (20.0-23.8) (7.2-9.8) (11.0-14.0) (3.6-5.6) 2817 55.0-95.6 69.7 (±8.8) (95 O) (33.0-36.6) (15.0-17.8) (21.1-24.1) (8.1-10.3)

886 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 9 TABLE n Prevalence () of locomotor disability, restriction in range of motion of the hips and knees, valgus and vanis deformity, and obliquity in men and women by age Age group (yr) Locomotor disability Hip flexion Knee flexion Anterior cruciate Medial collateral Lateral collateral Knee deformity Valgus Varus Locoraotor disability Hip flexion Knee flexion Anterior cruciate Medial collateral Lateral collateral Knee deformity Valgus Varus 55-64 404 10.7 8.0 1.7 16.9 15.4 3.7 17.4 13.2 2.7 9.0 3.0 0.2 4.2 5.4 3.5 7.2 10.1 589 15.8 8.9 2.7 16.8 4.3 15.2 13.5 3.6 5.3 1.2 7.3 8.8 10.9 2.0 12.9 65-74 501 19.8 15.0 11.4 1.6 18.6 19.8 9.0 18.0 17.0 8.0 14.4 6.4 1.2 8.4 5.2 6.2 3.0 11.4 10.2 684 30.4 18.8 15.1 4.8 21.0 7.3 19.4 17.7 5.9 1 10.4 2.6 8.6 9.9 9.8 15.6 3.7 75-84 234 34.2 15.9 19.0 4.7 20.7 14.7 1 20.7 10.3 15.5 11.2 3.9 7.3 5.1 3.8 10.7 416 51.0 22.5 22.7 26.8 12.1 21.5 24.4 10.9 16.9 5.3 7.9 6.7 18.3 6.0 16.6 85 + 17 58.8 0.0 5.9 5.9 0.0 0.0 50 82.0 30.0 26.0 18.0 24.0 28.0 24.0 20.0 32.0 14.0 6.0 28.0 30.0 Total 1156 20.2 13.7 12.0 2.4 18.4 18.6 17.8 16.4 7.0 6.5 1.4 8.2 4.8 5.5 4.2 10.1 10.1 1739 31.9 17.9 15.1 5.4 19.5 18.4 8.0 1 18.1 6.7 17.7 10.6 2.9 7.8 15.0 3.9 13.6 (95 CI) (17.9-22.5) (11.7-15.7) (10.1-13.9) (1.5-3.3) (16.2-20.6) (16.4-20.8) (7.1-10.3) (15.6-20.0) (14.3-1) (5.5-) (-14.9) (5.1-7.9) (0.7-2.1) (6.6-9.8) (3.6-6.0) (4.5-6.5) (3.0-5.4) (8.4-) (8.4-) (29.7-34.1) (16.1-19.7) (13.4-16.8) (4.3-6.5) (-21.4) (16.6-20.2) (6.7-9.3) (16.7-20.3) (16.3-19.9) (5.5-7.9) (15.9-19.5) (9.2-12.0) (2.1-3.7) (6.5-9.1) (7.2-9.8) (7.4-) (13.3-16.7) (3.0-4.8) (12.0-15.2) view of the results of the logistic regression discussed in Table III, the cut-off for the prevalence of restricted ROM was set at moderate severity (i.e. >20). To assess the influence of joint pain and morning stiffness, all odds ratios for LD were subsequently adjusted for joint pain in the relevant joint and for morning stiffness. Finally, age, joint pain, morning stiffness and the physical examination variables were entered jointly in a multiple logistic regression model of LD to estimate adjusted odds ratios and aetiologic fractions for all independent variables. The aetiologic fraction (EF) is defined as the proportion of disabled persons which is attributable to the determinant of interest [10].

ODDING ET AL.: PHYSICAL ABNORMALITIES OF HIP AND KNEE AND DISABILITY 887 The EF was calculated using the formula: EF = p(aor - l)/{p(aor - 1) + 1} where p is the prevalence of the determinant in the population and aor is the odds ratio adjusted for all variables in the model. In this analysis, all variables were dichotomized using no or only mild restriction of ROM as the reference category for all ROM variables. The variable 'knee deformity' was categorized for men as 0 = no deformity, 1 = varus deformity and for women as 0 = no deformity, 1 = valgus deformity. RESULTS In Table I, the prevalence of LD, pain in hips and knees, and morning stiffness of the participants of the present study is compared with that of all interviewed subjects and with that of all participants who visited the research centre. Restriction of the originally interviewed cohort by non-response and missing data on physical examination had no effect on the age distribution of the study group. People visiting the research centre were somewhat less disabled than those interviewed, but the differences were very small. The occurrence of pain and morning stiffness did not differ between the three groups, taking the 95 confidence intervals into account. Table II gives the prevalence of LD and abnormalities on physical examination of the hips and knees in men and women by age. Internal and external rotation were the most prevalent restricted ROMs. and severe restriction of ROM rose with age in both men and women, as did lateral instability. In TABLE III Age-adjusted univariate odds ratios (OR) and 95 confidence intervals (in parentheses) of abnormalities on physical examination for locomotor disability Hip Flexion Knee Flexion Instability Anterior Medial Lateral Deformity Valgus Varus 2.3 (1.5-3.5) 1.3 (0.9-1.7) 4.0 (2.7-6.1) 2.8 (2.1-3.7) 9.7 (4.2-22.3) ( 5.7(4.0-11.2) 1.2 (0.8-1.8) 1.5 (1.0-2.2) 4.2 (2.6-6.8) 1.2 (0.8-1.8) 2.0 (1.3-2.9) 3.2 (1.9-5.3) 1.4 (0.9-2.2) 1.2 (0.9-1.6) 3.9 (2.4-6.5) : 2.9 (2.1-4.1) 15.3 (4.2-55.6) ' >.6 (2.4-8.9) 0.9 (0.5-1.5) 1.4 (0.7-2.7) 1.1 (0.6-2.1) 1.5 (0.8-2.8) 0.5 (0.3-0.9) 1.4 (0.9-2.2) 1.0 (0.8-1.4) 1.9 (1.4-2.5) 3.7 (2.5-5.6) 1.1 (0.8-1.5) 1.7 (1.2-2.2) U (2.7-6.4) 1.7 (1.1-2.4) 1.6 (1.1-2.3) 1.8 (1.2-2.5) 1.8 (1.3-2.4) 1.5 (0.9-2.5).9 (1.4-2.5) men, there was an increase with age of varus deformity and in women of deformities of the knee as well as obliquity. restriction of motion occurred in 1.4- of the men and in 2.9-8.0 of the women. Instability of the knee joint, as expressed by laxity of the separate ligaments of the knees, was observed in < 10 of the men and women. Valgus deformity in men and varus deformity in women were present in ~4, but varus deformity in men (bow-legs) and valgus deformity in women (knock-knees) were 2-3 times as common. Pelvic obliquity while standing was somewhat more common in women than in men. Table III presents the age-adjusted univariate odds ratios of the abnormalities on physical examination for LD in men and women in detail. Considerable differences between men and women were observed for the odds ratios for knee deformities, instability and obliquity. In men, varus deformity (bow-legs) was negatively associated with LD. Valgus deformity (knock-knees), laxity of the knee ligaments and pelvic obliquity were significantly associated with LD in women, but not in men. restriction of ROM was either not or only weakly associated with LD. For this reason, we presented prevalence figures of at least moderate restriction (Table IV) and decided to estimate odds ratios at this cut-off point (Table V). Table IV presents the prevalence rates of the composite measures. Restriction of internal and external rotation of the hips was observed in about a quarter of the men and women, but the figures for restricted flexion of the hips and knees are 2-3 times lower. The composite measures ROM-hip and ROM-hip+knee amounted to more than a third of the men and women. Most prevalence figures for the composite measures increased steeply with age, in women more so than in men. Instability of the knees was present in of the men and 16.8 of the women, and these prevalences did not change with age. A fifth of the men and women had some abnormality on physical examination of the knees and half on examination of both hips and knees. Table V shows the effect on measures of association when parts of a composite measure have a low prevalence. While the odds ratios of restricted flexion of the hips and knees with a relatively low prevalence are, especially in men, higher than those of internal and external rotation, the odds ratios of ROM-hip and ROM-hip+knee are closer to the odds ratios of internal and external rotation, which have a higher prevalence. To assess the effect of joint pain and morning stiffness, all univariate odds ratios of the composite measures were subsequently adjusted for these variables. Morning stiffness did not alter the odds for disability of abnormalities on physical examination. Pain in the relevant joint had only a minor effect, which was not significant. For instance, the age-adjusted odds ratio of restricted flexion of the hip in men decreased from 4.7 (95 CI: 3.2-6.8) to 4.1 (2.8-6.0) when adjusted for hip pain. (Data available on request.) Table VI presents the results of the multiple regression model. The odds ratios given are adjusted

888 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 9 TABLE IV Prevalence () of at least moderate restricted range of motion ( ) and composite measures of abnormalities on physical examination of the hips and knees in men and women by age Flexion hip hip hip Flexion knee ROM-hip ROM-hip+knee Abnormalities of knee Abnormalities total Flexion hip hip hip Flexion knee ROM-hip ROM-hip + knee Abnormalities of knee Abnormalities total 55-64 404 9.7 19.1 15.9 3.2 25.2 25.5 12.9 41.6 589 11.6 17.3 17.1 6.5 25.6 26.8 16.6 17.5 43.5 65-74 501 28.8 25.0 7.6 35.5 35.9 12.8 21.8 52.7 684 19.9 24.9 23.6 35.8 3 18.3 21.5 52.0 75-84 234 23.7 36.7 31.0 15.1 44.4 45.7 18.8 57.3 416 31.4 38.9 21.5 50.5 52.2 14.4 19.2 65.9 Age group (yr) 85 + 17 41.2 47.1 52.9 52.9 52.9 76.5 50 44.0 52.0 44.0 36.0 58.0 62.0 18.0 80.0 Total 1156 14.4 27.3 23.4 7.9 34.0 34.5 19.9 50.1 1739 20.5 26.4 24.8 13.5 36.5 38.6 16.8 19.6 53.2 95 Q (12.4-16.4) (24.7-29.9) (21.0-25.8) (6.3-9.5) (31.3-36.7) (31.8-37.2) (10.6-14.4) (-22.2) (47.2-53.0) (18.6-22.4) (24.3-2) (22.8-26.8) (11.9-15.1) (34.2-38.8) (36.3-40.9) (15.0-18.6) (17.7-21.5) (50.9-55.5) ROM-hip - restricted flexion and/or internal rotation and/or external rotation; ROM-hip + knee - restricted ROM-hip and/or flexion knee; abnormalities of knee = restricted flexion knee and/or valgus/varus, and/or instability of knee ligaments; abnormalities total abnormalities of knee and/or restricted ROM-hip and/or obliquity. for age and all variables. Both in men and women, morning stiffness, joint pain and restricted flexion of the hips were independently and significantly associated with LD. Additional associated variables in men were restricted flexion of the knees and varus deformity, be it that varus deformity was negatively associated. In women, restricted internal rotation, valgus deformity, instability of the knees and obliquity were also independently associated with LD. As compared to Table V, the multivariate odds ratios of the ROM variables were lower. In view of the fact that the univariate odds ratios did not change very much when adjusted for pain and morning stiffness, we concluded that the occurring decrease in odds ratios resulted from the inter-relationship between the ROM variables. A multivariate regression with all physical examination variables, but leaving out pain and morning stiffness, confirmed this conclusion: the odds ratios were of the same magnitude as those presented in Table VI. (Data available on request.) The last column for each gender shows the proportion of LD in the total population that was attributable to the determinants of interest. For example, among women aged ^55 yr, it can be estimated that 30 of all LD in the general population is attributable to hip pain. DISCUSSION In the present study, we analysed the association between abnormalities on physical examination of the hips and knees and LD. LD occurred in a fifth of the men and in nearly a third of the women. The most prevalent abnormalities on physical examination were moderate and severe restriction of the internal and external rotation of the hips (ROM ^ 32 ) in men as well as in women (~25). and severe restricted flexion of the hips (ROM «S 96 ) and knees (ROM ^ 104 ) was 2-3 times less common. The univariate odds ratios for LD of restricted flexion of the hips and knees were, however, higher than those of restricted internal and external rotation of the hips, albeit the latter occurred more often. The prevalences of other abnormalities on physical examination were TABLE V Age-adjusted univariate odds ratios (OR) and 95 confidence intervals (in parentheses) of abnormalities on physical examination for locomotor disability * Flexion hip 4.7 (3.2-6.8) 3.5 (2.7-4.5) hip 2.2(1.6-3.0) 2.4(1.8-3.0) hip 2.3 (1.7-3.3) 2.1 (1.6-2.7) Flexion knee 4.8 (3.0-7.7) 3.2 (2.4-4.4) ROM-hip 2.4 (1.7-3.3) 2.7 (2.1-3.4) ROM-hip + knee 2.6 (1.9-3.7) 2.9 (2.3-3.8) Instability of knee ligamentst 0.9 (0.6-1.5) 1.5 (1.1-2.0) Abnormalities of knee 1.0 (0.7-1.5) 2.0 (1.5-2.6) Abnormalities total 1.9 (1.3-2.7) 2.7 (2.0-3.5) ROM dichotomized: 0 = no restriction; 1 = at least moderate restriction (>20 restriction of ROM), fligaments dichotomized: 0 no laxity; 1 = any laxity.

ODDING ET AL:. PHYSICAL ABNORMALITIES OF HIP AND KNEE AND DISABILITY 889 TABLE VI Adjusted odds ratios and aetiologic fractions of joint complaints and abnormalities on physical examination for locomotor disability Hip pain Knee pain Morning stiffness Flexion hip Flexion knee Knee stability Valgus/varus aor 2.3 2.7 5.2 2.2 2.0 1.1 1.0 0.9 0.5 1.4 95 a (1.4-3.8) (1.7-4.1) (2.7-9.7) (1.4-3.6) (1.1-3.5) (0.7-1.8) (0.6-1.8) (0.6-1.6) (0.3-0.9) (0.8-2.3) EF 9.5 17.3 16.0 14.7 7.0 (3.4)t (0-9)t _ _ (3.4)t aor 3.6 2.1 4.8 1.9.4 1.5 ().9 1.3 1.4.4 95 CI (2.6-4.9) (1.6-2.8) (3.1-7.3) (1.3-2.7) (0.9-2.0) (1.0-2.2) (0.6-1.3) (1.0-1.8) (1.0-2.0) (1.0-1.9) EF 30.3 19.4 24.4 15.3 (4.7)t _ 5.4 5.8 4.5 aor, odds ratio adjusted for age and all variables in the model. 95 CI, 95 confidence interval of aor. EF, aetiologic fraction - p(aor - l)/{p(aor - 1) + 1}. aor < 1. for not significantly higher than 1. much lower, ranging from 3.9 for varus deformity to 15.0 for valgus deformity in women. Of these findings, varus deformity (bow-legs) occurred more often in men with a LD score <0.5. In women, valgus deformity, instability of the knee and obliquity were weakly associated with LD. In the multivariate analyses, the estimation of the aetiologic fraction showed that, when joint complaints and findings on physical examination are analysed together, joint pain, morning stiffness and restricted flexion of the hips and knees contribute most to the occurrence of LD in men. In women, the same variables, except for knee flexion, as well as restricted internal rotation, instability of the knees, valgus deformity and obliquity contribute to this occurrence. Apart from this, there is a relatively strong association between the four ROMs, resulting in lower independent associations with disability when adjusting for this association (Tables V and VI). Or to put it differently: the high univariate odds ratios, of for example flexion of the hip in men, can in part be explained by the co-occurrence of restriction in other joint motions. Selection bias is a possible source of bias in our study. Not all participants of the interview were part of the present study, either due to non-response (~8) or to missing data (~ 30). The prevalence of LD in the reduced study group is somewhat lower than the figure for all interviewed participants. Selection of participants as the result of the 8 non-response to the invitation to visit the research centre reduced the prevalence estimates of self-reported LD, but did not significantly influence the prevalence estimates of joint pain. The further reduction of the study group because of missing data occurred more or less randomly as it did not materially change the prevalence estimates or the odds ratios of joint pain and morning stiffness for LD in comparison to the results of the analyses on all interviewed participants. In follow-up studies of individual patients, it is common to measure range of joint motion with a goniometer. In epidemiological surveys, where more examiners are involved, intra- and interobserver variation is of great concern. When each observer has only a restricted amount of time to carry out all necessary assessments, the choice of an appropriate measurement instrument and subsequently standardization of measurements are crucial. Physical examination of joints is known to be subject to considerable interobserver variation. While in clinical follow-up studies of individual patients by a single physician a goniometer might be able to detect small differences in joint motion over time, this instrumental assessment, although precise, is difficult to standardize and hence impractical for survey purposes. Therefore, ROM of the joints was assessed in broad categories. Particularly moderate (grade 3) restriction, which was used as the cut-off point in the analyses, encompasses a wide range of 20-60 reduction of normality. This wide range increases reproducibility between examiners and reduces interobserver variation, but clearly also precision. Correspondingly, the evaluation and categorization of stability and deformity of the knee, and obliquity, should be considered as an approximation of the presence of problems with stature rather than a precise measurement. In an earlier study, it was shown that by grading ROM in rather broad categories interobserver variation could be minimized and reproducibility enhanced [11]. Evaluating the results of the physical examination for each individual observer revealed differences in the average prevalence rates for the various joint motions. The relative distribution over the categories of restricted motion, however, remained the same for each physician. As both participants and observers were randomly allocated to the days of the survey week, introducing more physicians mainly increases the variance, but probably leaves the mean of all examiners as a fair estimate of the prevalence rate intact. This assumption is supported by the observation that the odds ratios of restricted ROM in Table III regularly go up with increasing severity of restricted ROM. The results of this epidemiological study cannot directly be transposed to the clinical situation. They merely indicate what limitations of ROM relate (independently) to LD and where possible intervention strategies might be successful. There are not many population-based studies to compare our results with. A study among people aged > 70 yr in Massachussettes presented a prevalence of reduced flexion of the hips and knees of 16 in men and 29 in women [12]. A study among people aged 79 yr in Goteborg reported that 67 of the participants had 'fairly good' flexion of the hips and 80 of the knees [13]. Of the participants with joint complaints, 84 had a restricted ROM of the hips and 19 a restricted ROM of the knees. This study also reported a strong correlation between restricted ROM of the hips and climbing stairs, and between restricted ROM of the knees andrisingfrom a chair and climbing stairs [14]. The major conclusion of the present study is that, on physical examination, restricted flexion of the hips and

890 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 9 knees is the most important independent correlate of disability of the locomotor functions. Other joint functions, such as internal and external rotation of the hips and laxity of the ligaments of the knee, are more difficult to examine and are only associated with LD in women. We therefore propose to restrict the assessment of limitation of joint motion in epidemiological population surveys to flexion of the hips and knees. ACKNOWLEDGEMENTS Supported by the NESTOR programme for geriatric research (Ministry of Health and Ministry of Education), the Netherlands Organization for Scientific Research (NWO) and the Municipality of Rotterdam. REFERENCES 1. Odding E, Valkenburg HA, Algra D, Vandenouweland FA, Grobbec DE, Hofman A. Association of locomotor complaints and disability in the Rotterdam study. Ann Rheum Dis 1995;54:721-5. 2. Odding E. Locomotor disability in the elderly. An epidemiologjcal study of its occurrence and determinants in a general population of 55 years and over. Thesis, Erasmus University Medical School, Rotterdam, The Netherlands, 1994:103-15. 3. Hofman A, Grobbee DE, Dejong PTVM, Vandenouweland FA. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur J Epidemiol 1991;7:403-22. 4. WHO. International classification of impairments, disabilities, and handicap. A manual of classification relating to the consequences of disease. Geneva: World Health Organization, 1980; 1989 reprint. 5. Fries JF, Spitz PW, Kraines RG, Holman HR. Measurement of patient outcome in arthritis. Arthritis Rheum 1980^3:137-45. 6. Fries JF, Spitz PW, Young DY. The dimensions of health outcomes: the Health Assessment Questionnaire, Disability and Pain Scales. / Rheumatol 1982;9:789-93. 7. Siegert CEH, Vleming LJ, Vandenbroucke JP, Cats A. Measurement of disability in Dutch rheumatoid arthritis patients. Clin Rheumatol 1984;3:305-9. 8. Thompson PW. Functional outcome in rheumatoid arthritis. Br J Rheumatol 1988;27(suppl I.):37-43. 9. Sullivan FM, Eagers RC, Lynch K, Berber JH. Assessment of disability caused by rheumatic diseases in general practice. Ann Rheum Dis 1987;46:598-600. 10. Kleinbaum DG, Kupper LL, Morgenstera H (eds) Epidemiologic research. Principles and quantitative methods. New York: Van Nostrand Reinhold, 1982: 160-4. 11. Valkenburg HA, Haanen HCM. The epidemiology of low back pain. In: White III AA, Gordon SL, eds. Idiopathic low back pain. Mosby Co., 1982:9-22. 12. Jette AM, Branch LG. Musculoskeletal impairment among tile non-institutionalized aged. Int Rehabil Med 1984;6:157-61. 13. Bergstrom G, Bjelle A, Sorensen LB, Sundh V, Svanborg A. Prevalence of symptoms and signs of joint impairment at age 79. Scand J Rehabil Med 1985;17:173-82. 14. Bergstrdm G, Aniansson A, Bjelle A, Grimby G, Lundgren-Lindquist B, Svanborg A. Functional consequences of joint impairment at age 79. Scand J Rehabil Med 1985;17:183-90.