Determinants of disability in osteoarthritis of the

258 Annals of the Rheumati Diseases 1993; 52: 258-262 Rheumatology Unit, Bristol Royal nfirmary, Bristol BS2 8HW, United Kingdom T E MAlindon C Cooper J R Kirwan P A Dieppe Correspondene to: Dr T E MAlindon, Rheumatology Department, St Thomas's Hospital, London SEl 7EH, United Kingdom. Aepted for publiation 3 November 1992 Determinants of disability in osteoarthritis of the knee T E MAlindon, C Cooper, J R Kirwan, P A Dieppe Abstrat Objetives To evaluate the influenes of radiographi severity, quadrieps strength, knee pain, age, and gender on funtional ability in patients with osteoarthritis of the knee. Methods Equal numbers of knee pain positive and negative respondents to a survey of registrants aged more than 55 years at a general pratie were invited to attend for knee radiographs and quadrieps femoris isometri strength estimations. Disability was measured using the Stanford Health Assessment Questionnaire. Results Complete data were available on 7 men (mean age 72-7 years) and 89 women (mean age 68-1 years); 44% reported knee pain, 48% had radiographi features of osteoarthritis, and 32% reported some degree of disability. Signifiant orrelations were observed between disability and radiographi sore, quadrieps strength, and knee pain. Logisti regression analysis, however, showed signifiant independent ontributions from quadrieps strength (odds ratio -84 kgf), knee pain (odds ratio 1.67), and age (odds ratio 1-6 per year) only; the radiographi sore had no influene on the model. These results were not influened by onfining the analysis to the group with radiographi features of osteoarthritis. Conlusions Quadrieps strength, knee pain, and age are more important determinants of funtional impairment in elderly subjets than the severity of knee osteoarthritis as assessed radiographially. Strategies designed to optimise musle strength may have the potential to redue a vast burden of disability, dependeny, and ost. (Ann Rheum Dis 1993; 52: 258-262) Physial disability is now a major publi health problem in elderly subjets. Problems with loomotion are the most ommon disorders, affeting more than four million adults in Great Britain. Osteoarthritis is blamed for most of this2 and the knee joint is a partiularly ommon site to be affeted one in four people aged over 55 years have knee pain3 and by the age of 65 years 3% of men and 4% of women have radiographi hanges of knee osteoarthritis.4 Knee osteoarthritis is therefore likely to provide a major ontribution to disability in the general population. There are problems, however, with this assumption. Firstly, although radiographi hanges are preditive of the later development of disability,5 the relation of knee osteoarthritis with urrent loomotor impairment has never been speifially examined. Seondly, there are reasons to suspet that the orrelation between radiographi severity and level of disability may be poor. Traditionally the severity of osteoarthritis has been assessed using a system whih sores radiographi features believed to be harateristi of this disorder.6 Conordane of radiographi osteoarthritis with pain and linial signs is poor, however, partiularly at the knee.7 8 Problems with the soring system (for example, not inluding patellofemoral views9) may aount for some of these diffiulties, but this does not appear to be the whole story as some work suggests that many diverse fators may be important orrelates of pain in osteoarthritis.' Although pain annot be assumed to be synonymous with disability, the disordane between symptoms and radiographi appearanes suggests that there may be determinants of funtional impairment in this disorder other than radiographi severity. One important ontender is quadrieps femoris weakness: the assoiation of thigh musle wasting with knee osteoarthritis has been reognised sine its earliest desriptions" and treatment traditionally inludes exerises intended speifially to inrease quadrieps strength. 12 Researh has eluidated mehanisms for an interation between knee damage and quadrieps dysfuntion'3 whih may be partiularly adverse in the ontext of the profound redutions in musle mass and strength whih already our with inreasing age.'4 To larify some of these issues we have therefore investigated the relations of a number of putative determinants of disability inluding radiographi sore and quadrieps strength with loomotor impairment. Subjets and methods SUBJECTS Reruitment was based on respondents to a postal questionnaire asking about knee pain and disability whih was sent to all 211 men and women aged over 55 registered at a general pratie in Bristol. t reeived a response rate of 81% and has been desribed elsewhere.3

Determinants of disability in osteoarthritis of the knee 259 Respondents with knee pain and an equal number of ontrols were ontated in random order by telephone and invited to partiipate in the study. Cases and ontrols were mathed by gender and were born within 12 months of eah other. KNEE PAN The question about knee pain was taken verbatim from the Health and Nutrition Examination Survey'5 whih asks: (a) have you ever had pain in or around a knee on most days for at least a month? (b) if so, have you had any knee pain during the last year? An affirmative response to the two parts of the question was required for a subjet to be designated 'knee pain positive'. For the purposes of analysis a sore was derived from this in whih O=no pain, =pain in one knee, and 2=pain in two knees. DSABLTY The measure of disability used in this analysis was the total sore provided by the Stanford Health Assessment Questionnaire modified for use in this ountry.'6 Three parts of the questionnaire (setions 2, 4, and 8) are related to lower limb funtion and were taken together as a 'lower limb sore' (range -9) for the analysis. ASSESSMENT OF QUADRCEPS FUNCTON Quadrieps femoris isometri strength was measured using a system based on the established method of Edwards et al,7 whih has been shown to be reliable and reproduible.'8 '9 n our study we used a onverted ouh rather than a hair. The subjet lay supine with the knee flexed at 9 over the end of the ouh. A foot harness was attahed via an inextensible wire to an eletroni strain gauge (UF2 45 kg strain gauge transduer, Pioden Controls Ltd, Canterbury, United Kingdom) whih relayed a signal to a BBC omputer. This voluntary effort was displayed as a fore v time graph and a ursor provided a measurement of maximum voluntary ontration, defined as the maximum fore sustained for at least one seond. The best of three reordings was taken, and if there was a large disrepany between the last two readings, a fourth measurement was obtained. The mean of the right and left maximum voluntary ontrations was used for analysis. Table 1 Clinial data for 159 subjets responding to questionnaire Men RADOGRAPHY Weightbearing anteroposterior knee radiographs and reumbent laterals at 3 of flexion were performed using a standardised tehnique. To reflet the variability in ompartmental prediletion and the bilaterality of knee osteoarthritis, and to optimise reproduibility, a soring system based on the entral three features of the Kellgren and Lawrene sale was used.6 Joint spae narrowing (-2), osteophytosis (-2), and slerosis (-1) were sored in the medial, lateral, and patellofemoral ompartments of eah knee. This provided a maximum sore of 5 for eah ompartment, and the sores from eah of the six ompartments were summated to provide an overall value for eah subjet ranging from to 3. To assess the repeatability of this system we seleted 1 knee radiographs representing the spetrum of disease severity from normal to severe. These were then sored twie in a blinded fashion by a single trained observer (TEM). Repeatability was assessed using the K statisti. STATSTCAL METHODS The distributions of age, knee pain, disability, and radiographi knee osteoarthritis in the study subsamples were analysed using a x2 test. The relations between variables were investigated first using simple linear regression for ontinuous variables (age and quadrieps strength) and Spearman rank orrelation for disontinuous measures. The relations of quadrieps strength and radiographi sore were also investigated using a proportionate analysis based on a x2 test, or analysis of variane as desribed by Gardner and Altman.2 The determinants of disability were then explored using a logisti regression model. Results SUBJECTS Complete data were obtained on the first 159 from a total of 513 subjets who attended (table 1). These 159 subjets were younger (mean age 7-2 years v 72-4; p<5) and less disabled (X2=l 1-2; one degree of freedom; p<5) than the remainder. The response rate to the telephone invitation was about 8%, though this remains neessarily an estimate beause a number were unobtainable and lear Women Knee pain negative Knee pain positive Knee pain negative Knee pain positive (n=36) (n=34) (n=53) (n=36) Mean (SD) age (years) 72 6(7 5) 71-8(4-7) 7-8(9-1) 65 4(7-5) Mean (SD) quadrieps maximum 27 8(9-3) 25-8(11-5) 16-7(5-7) 15-(5-8) voluntary ontration (kgf) Subjets with osteoarthritis (/) 13-9 82-4 33 69-4 Median radiographi sore 4 5 Disabled subjets* (%) 38-9 44-1 15-1 38-9 *Lower limb Health Assessment Questionnaire sore greater than.

26 MAlindon, Cooper, Kinvan, Dieppe Table 2 Pearson's orrelation oeffiients for age, quadrieps strength (maximum voluntary ontration), radiographi osteoarthritis severity sore, knee pain sore, and disability sore (lower limb setions of the Stanford Health Assessment Questionnaire) Quadrieps Radiographi Knee pain Disability strength sore Age -{5 (women) 21 NS NS -26 (men) Quadrieps strength -21 NS -35 Radiographi sore 43-24 Knee pain -32 X- Th 1 ', 8.2 4W E 6 'A u). 4. 1t T reords of the large number of alls made were diffiult to keep. The data suggest, however, that there was a bias in favour of younger and fitter subjets attending. CORRELATONS Table 2 gives the orrelation matrix between the variables in the subjets tested. For all variables higher orrelations were provided by linear regression analysis than by rank orrelation. Quadrieps strength was more losely orrelated with age in women than men but other orrelations did not relate to gender. QUADRCEPS STRENGTH AND DSABLTY A lear inverse relation between disability and quadrieps strength remained after adjustment for age using the oeffiient derived from linear nen regression of mean quadrieps maximum Men Women voluntary ontration against age in subjets with normal knees (fig 1). Subjets with a CU ZD - measured quadrieps strength of less than 1 1-4 >4 1-4 > 4 kgf were more likely to report disability, but Disability sore differenes between higher inrements of Figure 1 Mean quadrieps isometri strength orretedfor a, ge at three levels ofdisability quadrieps strength were not signifiant, (lines indiate 95% onfidene limits). suggesting a threshold effet (fig 2). 2 Cr S A a ZU. VU) Os. m C a J 1 8-6- 4 2 t 1 - = 8 n 6, 4 Ca ) 2" 2- - ' -1 1-2 2-3 > Mean quadrieps strength (kgf) Figure 2 Perentage of reported disability (lower limb Health Assessment Questionnaire sore greater than ) in subjets with inreasing inrements ofquadrieps strength (lines indiate 95% onfidene limits). u-, 1-7 8-14 15-21 > 22 x Ray sore Figure 3 Perentage ofsubjets reporting disability (lower limb Health Assessment Questionnaire sore greater than ) with inreasing severity of radiographi knee osteoarthritis (lines indiate 95% onfidene intervals). RADOGRAPHC SCORE AND DSABLTY The K statisti for radiographi repeatability derived from the sore ompressed into a three point sale was 57. Figure 3 shows radiographi hanges of knee osteoarthritis in relation to levels of disability. The frequeny of reported disability (lower limb Health Assessment Questionnaire sore greater than ) among subjets with radiographi sores of - 15 was signifiantly higher than normal subjets (p<5), but differenes were not apparent below this sore, again suggesting a threshold effet. REGRESSON MODELS Logisti regression was undertaken beause of the non-normal distribution of the Health Assessment Questionnaire sore, and beause of a possibility that linear regression might favour a ontinuous dependent variable suh as quadrieps strength in favour of a disontinuous measure suh as radiographi sore. Despite this, when added to the model as four ategories, radiographi sore did not show any signifiant effet (p>o- 1). Presenting quadrieps strength as a ontinuous variable or as two ategories made little differene to these results, nor did onfining the analysis to those with radiographi hanges (sore greater than ). Table 3 gives the odds ratios (odds ratio=probability/1-probability) and their 95%/o onfidene limits. Disussion The entral onlusion of this study is that, of the variables studied, three exert important independent effets on fumtional ability in the age group investigated: quadrieps weakness,

Determinants of disability in osteoarthritis of the knee 261 Table 3 Logisti regression analysis of quadrieps strength (as ontinuous variable), knee pain, sex, age, and radiographi sore as explainants of variation in lower limb Health Assessment Questionnaire sore in the whole study group and onfined to those with radiographi osteoarthritis of the knee Whole group (n=159) Subgroup with osteoarthritis (n=98) Odds ratio 95% onfidene limit Odds ratio 95% onfidene limit Quadrieps strength -84 (per kgf) -77 to -91-84 77 to -92 Knee pain Oneknee 1-67 (vnopain) 1-8to 15-4 3-41 -84to 13-71 Two knees 1-48 (v no pain) 1-5 to 12-4 1-72 -47 to 6-48 Age 1-6(peryear) 1-to1-12 1-4 97to1 11 Gender -38 (men v women) - 13 to 1-12 -43-11 to 1-61 the presene of knee pain, and inreasing age. Of these, quadrieps weakness appears to be the most important, whereas radiographi sore itself makes no signifiant independent ontribution to disability. The odds ratios give some idea of the magnitude of the impat of these variables in terms of reported disability. For example, the risk of disability dereases by -88 with eah 1 kgf inrement in quadrieps strength. Therefore a subjet with a mean quadrieps strength of 2 kgf ompared with another with 3 kgf will therefore have an inreased risk of disability of (1/-84)=5 7. Knee pain onfers a risk of approximately 1-5 ompared with no pain, and an inrement of 2 years of age (within the older than 55 years age group) onfers a risk of disability of about 1-8. Further insight into the interplay between quadrieps funtion and loomotor impairment is provided by the threefold inrease in frequeny of reported disability in subjets with quadrieps strength less than 1 kgf. This finding suggests some form of threshold effet, and is in keeping with the onept of a ritial strength below whih ertain ativities of daily life beome impossible, as suggested by Young 13 nteration may also exist between musle strength as measured by a voluntary ontration and psyhologial fators. Summers et al showed an assoiation between psyhometri variables and pereived levels of pain and disability.1 t is possible that similar fators might influene maximum effort during strength testing whih ould enhane a orrelation between weakness and self reported disability. The absene of any detetable independent effet of radiographi osteoarthritis on funtional ability supports the original hypothesis, but is, perhaps, surprising. One possible explanation relates to the lak of validity of the radiographi soring system. Early researh using the Kellgren and Lawrene radiographi osteoarthritis sore showed relatively poor reproduibility,21 but in later studies its performane has been improved by assigning strit verbal riteria to eah grade based on the original radiographi features.22 For this reason, and to reflet the heterogeneity of knee osteoarthritis in terms of ompartmental involvement and bilaterality, eah of the six ompartments were sored independently for joint spae narrowing, osteophytosis, and slerosis. These sores were then summated to provide an overall grade for eah subjet whih proved adequately repeatable in this study. t is possible that this methodology provided a non-representative refletion of severity, though this seems unlikely in the ontext of urrent onepts of osteoarthritis. Alternatively, the problem may relate to the sored radiographi features as indiators of the osteoarthriti proess itself. Osteoarthritis is a disease defined by a onstellation of histologial features, but in epidemiologial surveys the radiographi sore has provided the diagnosti gold standard.4 1 22 The relation between radiographi hanges of knee osteoarthritis and symptoms is known to be poor' 2 and ertain features whih ontribute to the radiographi diagnosis suh as osteophytosis may relate more to ageing than any disease.23 Some reent work has also shown that the preditive value of joint spae narrowing for artilage loss is suboptimal in individuals, though the degree of orrelation is at a level aeptable for epidemiologial study.24 n our analysis, similar results were generated by the use of a joint spae narrowing, osteophyte, or slerosis sore, suggesting little differene in validity between them. The most plausible explanation for the findings of the regression model is that pain and musle weakness are the mehanisms through whih osteoarthritis exerts an influene on funtional ability, and the orrelations of radiographi sore with knee pain and quadrieps strength agree with this. Alternative hypotheses are that radiographi features reflet outome and tell us little about the disease proess, or, more provoatively, that the radiopathologial hanges believed to onstitute osteoarthritis do not represent a disease entity and are not a ause of disability. Finally, disability is a onept, measurement of whih has philosophial was well as pratial problems. nterest in this field over the last deade has generated validated funtional assessment instruments suh as the Health Assessment Questionnaire, without whih a study suh as this would not have been possible. The Health Assessment Questionnaire itself has been extensively validated in subjets with rheumati diseases and orrelations between reported and observed performane range from 47 to -88 for individual questions with higher values obtained for aggregated sores.25 26 Work using this instrument showing greater levels of disability in women has raised a question about liability to female bias27 and it is therefore interesting that, though the ommunity survey

262 MAlindon, Cooper, Kirwan, Dieppe also showed an exess of disability in women, the regression analysis found no signifiant independent effet of gender. This lends further weight to the suggestion that women are more likely to experiene funtional limitation than men. The general appliability of the findings of this study may be redued by biases in patient seletion. As radiographi hanges of osteoarthritis are well reognised to our in the absene of symptomatology it was neessary to examine a knee pain negative group as well as symptomati subjets. The seletion method used represented a pratial ompromise between seeing a random population sample (whih would have resulted in a low yield of ases) and using hospital outpatients (whih would have resulted in unaeptable seletion bias). The study sample, though ontated in random order, annot be assumed to be representative of the study population and it is likely that biases in favour of healthier and fitter subjets attending for assessment were operative. This effet would be partiularly important if either the prevalene of radiographi hange was onsiderably less in the tested group, or if the relation between musle strength and disability was dereased in the fae of more advaned osteoarthritis. Neither of these, however, appears to be so; subjets with disability and with radiographi knee osteoarthritis were well represented in the study group (32 and 48% respetively), and the logisti regression analysis onfined to the subgroup with knee osteoarthritis provided idential results. The publi health impliations of this study are therefore not onfined to those with osteoarthritis alone. Deline in musle funtion after about 55 years of age is an apparently almost universal phenomenon 18 19 28 and signifiant negative orrelations between age and quadrieps maximum voluntary ontration were found in this sample. The auses of this derease in musle funtion are not fully understood but it has been suggested that a large proportion is likely to be due to inativity.29 Muh of this may be reversible: improvements in musle funtion have been doumented in elderly volunteers partiipating in exerise programmes and the benefits appear to be substantial.29 Applied to the general population, strategies designed to optimise musle strength may have the potential to redue a vast burden of disability, dependeny, and ost. We thank Shelagh Snow, researh sister, Dr Rosie Jones, musle physiologist, Dr Alan Preee, hief medial physiist, Bristol Royal nfirmary, Mr Nik Taube, statistiian, St Thomas's Hospital, and Dr Tim Paine and patients of the Whiteladies Road Health Centre, Bristol. The projet was funded by the Arthritis and Rheumatism Counil. 1 Martin J, Meltzer H, Eliot D. The prevalene of disability among adults. London: Offie of Population Censuses and Surveys, Soial Survey Division, HMSO, 1988. 2 Badley E M, Thompson R P, Wood P H N. The prevalene and severity of major disabling onditions. nty Epidemiol 1978;7: 145-51. 3 MAlindon T E, Cooper C, Kirwan J R, Dieppe P A. Knee pain and disability in the ommunity. Br Y Rheumatol 1992; 31: 189-92. 4 Van Saase J L C M, Van Romunde L K J, Cats A, Vandenbrouke J P, Valkenberg H A. Epidemiology of osteoarthritis: Zoetermeer survey. Comparison of radiologial osteoarthritis in a Duth population with that in 1 other populations. Ann Rheum Dis 1989; 48: 271-8. 5 Hohberg M C, Lawrene R C, Everett D F, Comoni- Huntley J. Epidemiologi assoiations of pain in osteoarthritis of the knee: data from the National Health and Nutrition Examination Survey and from the National Health and Nutrition Examination- epidemiologi follow-up survey. Semin Arthritis Rheum 1989; 4 (suppl 2): 4-9. 6 Kellgren J H, Lawrene J S. The epidemiology of hroni rheumatism: atlas of standard radiographs of arthritis. Oxford: Blakwell Sientifi, 1963. 7 Lawrene J S, Bremner J M, Bier F. Osteoarthrosis: prevalene in the population and relationship between symptoms and x-ray hanges. Ann Rheum Dis 1966; 25: 1-23. 8 Claessens A A M, Shouten J S A G, Van den Ouweland F A, Valkenburg H A. Do linial findings assoiate with radiographi osteoarthritis of the knee? Ann Rheum Dis 199; 49:771-4. 9 MAlindon T E, Snow S, Cooper C, Dieppe P A. Radiographi patterns of osteoarthritis of the knee joint in the ommunity: the importane of the patellofemoral joint. Ann Rheum Dis 1992; 51: 844-9. 1 Summers M N, Haley W E, Reveille J D, Alaron G S. Radiographi assessment and psyhologial variables as preditors of pain and funtional impairment in osteoarthritis of the knee or hip. Arthritis Rheum 1988; 31: 24-9. 11 Adams R. A treatise on rheumati gout or hroni rheumati arthritis of all the joints. London: Churhill, 1857. 12 Rudd E. Physiatri management of osteoarthritis. Clin Rheum Dis 1985; 2: 433-5. 13 Young A. Exerise physiology in geriatri pratie. Ata Med Sand Suppl 1986; 711: 227-32. 14 Aniansson A, Sperling L, Rundgren A, Lehnberg E. Musle funtion in 75 year old men and women: a longitudinal study. Sandj Rehabil Med Suppl 1983; 9: 92-12. 15 Maurer K. Basi data on arthritis of the hip, knee and saroilia joints in adults ages 25-74, United States 1971-1975 vital and health statistis, series 11, no. 213. Hyatsville, MD: National Center for Health Statistis; 1979. DHEW Publiation No.: (PUS) 79-1661. 16 Kirwan J R, Reebak J S. Stanford Health Assessment Questionnaire modified to assess disability in British patients with rheumatoid arthritis. Br J Rheumatol 1986; 25: 26-9. 17 Edwards RH T, Young A, Hosking G P, Jones D A. Human skeletal musle funtion: desription of tests and normal values. Clin Si Mol Med 1977; 52: 283-9. 18 Young A, Stokes M, Crowe M. Size and strength of the quadrieps musles of old and young women. Eur J Clin nvest 1984; 14: 282-7. 19 Young A, Stokes M, Crowe M. Size and strength of the quadrieps musles of old and young men. Clin Physiol 1985; 5:145-54. 2 Gardner M J, Altman D G. Calulating onfidene intervals for proportions and their differenes. n: Gardner M J, Altman D G, eds. Statistis with onfidene. London: BMJ Publiations, 1989: 28-33. 21 Kellgren J H, Lawrene J S. Radiologial assessment of osteo-arthrosis. Ann Rheum Dis 1957; 16: 494-51. 22 Felson D T, Naimark A, Anderson J, et al. The prevalene of knee osteoarthritis in the elderly: the Framingham osteoarthritis study. Arthritis Rheum 1987; 3: 914-8. 23 Hernborg J, Nilsson B E. Age and sex inidene of osteophytes in the knee joint. Ata Orthop Sand 1973; 44: 66-8. 24 Fife R S, Brandt K D, Braunstein E M, et al. Relationship between arthrosopi evidene of artilage damage and radiographi evidene of joint spae narrowing in early osteoarthritis of the knee. Arthritis Rheum 1991; 34: 377-82. 25 Fries J F, Spitz P, Kraines G, Holman H R. Measurement of patient outome in arthritis. Arthritis Rheum 198; 23: 137-45. 26 Fries J, Spitz P, Young D. The dimensions of health outomes: the Health Assessment Questionnaire, disability and pain sales. 7 Rheumatol 1982; 9: 789-93. 27 Thompson P W, Pegley F S. A omparison of disability measured by the Stanford Health Assessment Questionnaire disability sales in male and female rheumatoid patients. BrJ Rheumatol 1991; 3: 298-3. 28 Asmussen E, Fruensgaard K, Norgaard S. A follow-up survey of seleted physiologi funtions in former physial eduation students after forty years. J Am Geriatr So 1975; 23: 442-5. 29 Bassey E J. Benefits of exerise in the elderly. n: saas B, ed. Reent advanes in geriatri mediine, No 3. London: Churhill Livingstone, 1985.