Nuffield Orthopaedic Centre, Headington, Oxford OX3 7LD, UK and
1 University of California, Los Angeles, CA, USA
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. A total of 402 Caucasians consecutively undergoing total hip replacement (THR) or total knee replacement (TKR) for idiopathic OA at a major centre was surveyed.
Results. Previous joint injury was more common in the TKR group (P < 0.0001). However, both groups manifested a mixed occupational background, body mass indices similar to the general population and a predominance of females (F:M = 1.31.4:1). The TKR group had a significantly younger age of symptom onset (56 yr) than the THR group (61 yr) but both groups had a tendency to bilateral arthroplasty (33%), nodal involvement (5459%), a significant excess of right-sided replacements (1.8:1, THR; 2.2:1, TKR) and similar levels of pre-operative pain and disability. Up to 40% of hips manifested acetabular dysplasia and 10% possible previous slipped upper femoral epiphyses. Eighty-five per cent with end-stage coxarthrosis or gonarthrosis had an identical pattern of radiographic disease contralaterally.
Conclusions. Our data suggest the importance of a constitutional tendency to idiopathic, end-stage OA, a disorder traditionally associated with environmental factors leading to wear and tear.
KEY WORDS: Osteoarthritis, Idiopathic, End-stage, Symptomatic, Hip, Knee, Aetiology, Clinical, Radiographic.
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Insights into the aetiology of idiopathic coxarthrosis and gonarthrosis have been gained by epidemiological surveys on asymptomatic radiographic disease [3]. These investigations have implicated age, gender, race, obesity, occupation, injury, heredity and developmental deformity as likely risk factors for OA. There are differences in the importance of these factors to hip and knee OA, with injury and obesity, for example, associated with gonarthrosis whilst occupation and dysplasia are associated with coxarthrosis. Studies exploring the importance of such factors in symptomatic OA have been limited by size [4], the inclusion of cases with secondary OA [5] and the exclusion of female patients [6]. Despite the prevalence of coxarthrosis and gonarthrosis, no one study has compared possible aetiological factors for these conditions. An absence of long-term, prospective studies prevents prediction of the proportion of those with asymptomatic radiographic disease who progress to OA requiring therapeutic intervention. Indeed, risk factors for asymptomatic OA might be different to those for symptomatic disease.
Despite the prevalence of idiopathic OA warranting THR and TKR, we lack basic information on disease presentation. The proportion with uni- or bilateral disease and localized or generalized OA is unknown and prevents us from classifying end-stage OA by accepted criteria [7]. Rates of disease progression from symptom onset to arthroplasty remain unclear. The nature and duration of joint pain and associated disability are factors determined in everyday clinical practice but are hardly discussed in the literature.
Although radiographic surveys in the community have described the appearances of asymptomatic OA of the hip and knee joints [8, 9] and some studies have described radiographic patterns in end-stage coxarthrosis, different scoring systems have been used to localize disease [1012]. Radiographic studies implicating acetabular and proximal femoral dysplasia in the background of idiopathic coxarthrosis are widely known [13, 14] but have not been validated. There is little published literature on the intrajoint localization of gonarthrosis requiring arthroplasty.
We conducted a study of patients with idiopathic OA undergoing THR or TKR in order to describe and compare the aetiological associations, clinical features and radiological appearances of end-stage coxarthrosis and gonarthrosis.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In the days following arthroplasty, 402 patients were invited to participate in an interview designed to ascertain disease associations and clinical patterns. Interviews were conducted by a single researcher. Nine patients who could not be interviewed prior to departure from hospital were telephoned and taken through the same questionnaire as those reviewed in hospital.
The information obtained from each patient is given in Appendix 1.
The hands of those interviewed in person were examined for Heberden's nodes.
Patients were divided by lifetime history of THR or TKR for OA, excluding those who had undergone both procedures.
We assessed the association of site of replacement with potential covariates [for example, age at symptom onset; gender; body mass index (BMI) at 20 yr, 40 yr or current; deformity; occupational risk; occupational sum score and previous joint injury] using classification and regression tree analysis (CART) [16] and logistic regression. CART was used to identify covariates whose values improved the ability to discriminate between THR and TKR. Because CART is less subject to selection error due to multicollinearity among the covariates, and since its non-parametric tree-building algorithm does not assume a normally distributed error variance, CART was favoured over other methodologies in the initial selection of predictive variables. Although CART ranks variables by importance, it does not assign statistical significance. Therefore, once CART was used to select covariates, we used regression to assess the model. Linear and logistic regression analyses were performed using STATA (statistical analysis package, version 5.0, Stata Corp., Texas, USA). CART analyses were performed using SYSTAT [17].
Available standard hip and knee radiographs taken immediately prior to arthroplasty were obtained from hospital records. Anteroposterior views of the pelvis and upper femora, weight-bearing anteroposterior and lateral views of both knees were examined and scored for the severity and localization of OA by two researchers working together. The severity of OA for the hip and tibiofemoral joints was scored using the Kellgren and Lawrence scheme [18], whilst the scheme of Burnett et al. [19] was used to score patellofemoral OA in association with a radiographic atlas. Measurements of the centre-edge angle (CEA) [13] and femoral head ratio (FHR) [14] were made on available radiographs by a single researcher.
A subsample of radiographs was re-examined to assess the reproducibility of CEA and FHR measurements.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Gender.
A slight female excess was evident for patients undergoing THR or TKR [5, 21, 22]. High perimenopausal oestrogen levels might predispose women to OA [23]. Oestrogen receptors present in articular cartilage could potentiate cytokines involved in cartilage metabolism [24] and gender differences exist in the prevalence of these receptors in articular cartilage [25].
BMI.
A steady rise in BMI with increasing age, very similar to that obtained in a UK population study [26] was evident for those undergoing TKR or THR. Only at 60 yr of age were patients clearly overweight for height (BMI 2530). However, mean BMI remained below the obese category (BMI > 30).
Occupation.
Farming and construction work have been implicated in the onset of end-stage coxarthrosis [27, 28] but our results indicate that patients undergoing THR or TKR had a similar background of work intensity. Although the most common occupations presenting for THR and TKR included farming, a large proportion were administrators and teachers, suggesting that those in lighter occupations are also vulnerable to OA.
Injury and menisectomy.
A history of a previous injury to the joint replaced was more than five times more common in the TKR than in the THR group. Given that nearly a third of men undergoing TKR could recollect a significant knee injury, these results suggest that soft-tissue injury might be an important factor in idiopathic gonarthrosis. Further to other studies implicating menisectomy as a risk factor for gonarthrosis [5, 29], nearly one in five men and one in 10 men or women undergoing TKR had previously undergone ipsilateral menisectomy. Furthermore, the proportion with a history of menisectomy in the TKR group was significantly greater than that in the THR group, suggesting direct association between menisectomy and gonarthrosis.
Clinical patterns
Disease distribution.
Further to previous work showing that bilateral radiographic changes are common in those with symptomatic coxarthrosis and gonarthrosis [30, 31] our results indicate that between a quarter and a third of patients with THR or TKR had bilateral replacements at survey. Moreover, nearly one fifth of those with unilateral THR admitted to chronic contralateral hip pain whilst two fifths of those with unilateral TKR had chronic pain in the opposite knee. These results suggest that many of those with unilateral, end-stage OA progress to warrant contralateral arthroplasty for OA of the hip or knee.
Whether coxarthrosis and gonarthrosis occur together or independently has been debated [32]. In our survey, a quarter of those with THR manifested chronic knee pain whilst an eighth of those with TKR had chronic hip pain. Knee and hand OA have been previously associated and are implicated more often in the presentation of generalized disease than hip OA [22, 31]. In the present study, the percentage of women with nodal OA undergoing THR was not significantly different from the percentage of women with nodal OA undergoing TKR. As distal interphalangeal nodal arthropathy suggests polyarticular disease, our results imply that more than a half of those undergoing either TKR or THR manifested generalized OA [7].
A significant tendency towards right-sided arthroplasty was noted [5, 32]. It is possible that idiopathic coxarthrosis and gonarthrosis are more prevalent on the right because of an excess of right-handedness and right-footedness subjecting joints to increased repetitive impulse loading [33].
Pain from arthritis, patients' perceived limitations and analgesic use.
Patients with end-stage coxarthrosis and gonarthrosis admitted to similar levels of joint pain and functional limitations from OA. Women were more likely to admit rest pain or night pain. Patients with end-stage coxarthrosis and gonarthrosis felt similarly limited by OA.
Over 80% of patients admitted to long-term analgesic use prior to THR or TKR. Despite recognized complications, non-steroidal anti-inflammatory drugs (NSAIDs) were required by more than half the sample [34]. Women were more likely to have used analgesics than men.
Radiological features
Coxarthrosis.
Our study confirms the predominance of superior joint localization noted by previous workers [22, 35]. Concentric disease was present in over half of all cases, a far greater proportion than in previous studies on less severe coxarthrosis [12] implying that localized OA progresses to involve the entire joint. Women had a tendency to superolateral OA and men to superomedial OA [11, 12, 35]. Nearly 70% of radiographs manifested OA in the contralateral hip and in 85% of these cases a symmetrical pattern of localization was present. These findings suggest an intrinsic tendency to bilateral disease in patients with end-stage coxarthrosis. If extrinsic influences such as injury were predominant then differences in the localization of OA might have been expected in adjacent hips.
CEA.
From his studies, Wiberg [13] concluded that angles below 25° could be regarded as abnormal. Population surveys have established mean CEA values between 36 and 38° [13, 3638], whilst hospital-based series have associated low angles with advanced coxarthrosis [13, 14, 36, 39]. However, the selection of cases in hospital-based surveys was not described and included cases of secondary OA. To date, mean values of CEA have not been determined in idiopathic, end-stage OA. Our results suggest that between 20 and 40% of those with idiopathic coxarthrosis manifested shallow acetabulae as defined by a CEA of less than 20 and 25°. Women averaged lower angles than men and were twice as likely to have angles less than 20° (Table 3). The fact that subjective assessment failed to detect dysplasia, except for coxa valga deformity in one hip, suggests that acetabular dysplasia can be overlooked unless the CEA is measured. Our results suggest that the CEA is a repeatable parameter, even in patients with advanced OA.
FHR.
Murray [14] suggested that ratios greater than 1.35 indicated a slipped upper femoral epiphysis (SUFE) and felt that 39% of those with idiopathic, end-stage coxarthrosis had sustained SUFE in adolescence. Our patients had a mean FHR of 1. The overall proportion of our series (10%) with possible underlying SUFE, although less than Murray's estimate, is identical to that obtained in a study of Caucasians with end-stage coxarthrosis [35]. Our results suggest that the FHR is less repeatable than the CEA and might therefore be less useful in clinical practice.
Gonarthrosis.
The radiological patterns of end-stage gonarthrosis have not previously been comprehensively described [4042]. In our study (Table 4) the medial tibiofemoral joint was predominantly involved by OA in two thirds of cases whilst in one fifth of cases it was not possible to specify which site was most affected. The lateral tibiofemoral joint was most involved in about one tenth of cases whilst predominant patellofemoral joint OA was present in only 1%. The same proportion of men and women were affected by medial tibiofemoral OA but women were three times more at risk of predominant patellofemoral OA. Bilateral radiographic OA was present in nearly two thirds of cases, with symmetrical patterns of intrajoint localization nearly universal. The tendency to symmetrical disease is suggestive of an inherent predisposition to knee OA.
Limitations of the study and sources of possible error
It might be argued that the patients in the study do not accurately represent all those with end-stage OA. Concurrent medical conditions, as well as health service variations in the practice and provision of arthroplasty might have influenced the composition of our patient sample. Although personal experience suggests that patient selection criteria for THR and TKR are broadly similar across the UK, further studies conducted in other countries will be necessary to validate our results.
Reliance on the recall of middle-aged and elderly respondents regarding events occurring much earlier in life might be criticized. However, the capacity of patients to remember past events was generally good. Intra-observer errors were ascertained for sample questions, using some of the respondents in this study, as part of a larger study. Correlation coefficients in excess of 0.8 were obtained for all questions (J. Chitnavis, unpublished data). The use of in-person interviews, rather than self-administered questionnaires, is likely to have encouraged responses.
Errors could also have resulted from the overweight underestimating their weight and in the misclassification of occupational strength demands. Furthermore, the obese could have been denied surgery. However, the reliability of previous studies on bodyweight and occupation based on patient-volunteered information has been acceptable [43, 44]. Moreover, at our centre, few are denied arthroplasty because of obesity. Assessments of pain and disability would have been enhanced by the use of better measurement, including visual analogue scales, but many existing instruments for determining disability are limited by complexity or lack of specificity for end-stage OA [45].
![]() |
Conclusions |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The contralateral joint was affected by chronic pain or had been previously replaced in more than 40% of those undergoing hip or knee replacements for OA. However, of those undergoing unilateral THR or TKR, a significant excess of right-sided replacements was seen. Although most patients from both groups denied chronic pain in other joints, more than 50% of patients had clinical evidence of Heberden's nodes indicating a tendency to generalized OA in the majority of patients. Similar degrees of pain and disability were experienced by the two groups with women admitting more joint pain than men.
Although superior joint localization prevailed in those with coxarthrosis, more than half manifested a concentric pattern of OA. Although predominant involvement of the medial tibiofemoral joint was typical of the majority with end-stage gonarthrosis, tricompartmental radiographic disease was universal. The severity of OA was less marked in the patellofemoral and lateral tibiofemoral compartments. About 40% of those with idiopathic OA of the hip manifested acetabular dysplasia, whilst 10% showed signs of SUFE. Our findings support suggestions that idiopathic coxarthrosis is often associated with subtle and bilateral deformities of the hip joint.
Obesity and heavy occupation, factors traditionally considered relevant to the development of wear and tear in hips and knees, are not mandatory to the development of severe, symptomatic OA. In contrast, the prevalence of bilateral clinical and symmetrical radiographic disease, dysplasia and nodal change highlights the importance of a constitutional predisposition to idiopathic, end-stage coxarthrosis and gonarthrosis.
![]() |
Appendix 1 |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Age.
The date of birth, year of onset of symptoms of arthritis in the joint most recently treated by primary arthroplasty and the year of primary arthroplasty for that joint. In cases of bilateral replacement, these details were obtained for the joint which had developed symptoms earliest.
Gender
Longstanding BMI.
Patients were asked to estimate their:
Using these estimates, Quetelet's index (BMI = mass in kg/height in m2) was determined for specified times during the patients' lives.
Occupational history.
The nature and duration of the occupation(s) undertaken by patients in their lifetimes were recorded. Periods of service less than 6 months were ignored.
We modified the Nordic Occupational Classification used by Vingaard [6] to estimate occupational risks for hip and knee OA by including white collar workers. Blue collar workers regarded by them as low risk were classed as moderate risk and white collar workers as low risk. Blue collar workers scored by their scheme as high risk were unaffected by the modification. A points scale awarding three points for high risk blue collar workers, two points for the remaining blue collar workers and one point for white collar workers was used. An occupational sum score [46] was derived from the product of occupational risk and years in employment.
Joint injury and associated menisectomy.
Patients with an injury to either hip or knee which prevented them from walking normally for a period of at least 1 week were asked to identify the joint injured, the date of the injury and relate any surgical intervention to the joint following the injury.
Clinical patterns of end-stage coxarthrosis and gonarthrosis
Factors aggravating hip and knee pain, night pain and analgesic use were determined. The site of chronic joint pain, as defined by pain for at least 6 months within the previous year, was recorded.
Patients chose one of the following terms to describe the limitations they experienced as a result of end-stage OA: none, limited handicap, handicapped but independent, partially dependent, completely disabled.
![]() |
Appendix 2 |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Using the age of symptom onset as the outcome for best subset linear regression, those undergoing TKR developed symptoms in their joints 4 yr in advance of those undergoing THR when controlling for previous joint injury, BMI at 40 yr and deformity. The best predictive linear regression equation was
Age at symptom onset = 74.37 - 6.64 injury - 0.514 BMI 40 - 21.38 deformity - 4.26 TKR, r2 = 0.1224.
![]() |
Acknowledgments |
---|
![]() |
Notes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|