Need for and receipt of hip and knee replacement--a national population survey.

OBJECTIVES
Hip and knee joint replacements are effective, and yet little is known about how closely the need for joint replacement matches supply in different population groups. Our objective was to compare the prevalence of existing joint replacements with that of need in population groups in England.


METHODS
A total of 7101 people aged 60 yrs or older, representative of the population of England, were interviewed. Participants were asked about both receipt and need for joint replacement, socio-economic status and co-morbidity. 'Need' classification was based on hip or knee pain and difficulty walking, with adjustment for potential surgical contraindications. Associations between participants' characteristics and both need and receipt were estimated.


RESULTS
The prevalence of existing joint replacement (receipt) was 6% [95% confidence intervals (CI) 5, 6], and this was lower in the North than the South [adjusted odds ratio (OR) 0.72, CI 0.53, 0.96]. In contrast, the prevalence of estimated need was higher in the North (OR 1.27, CI 1.03, 1.58). Need was greater in women than men (OR 1.30, CI 1.09, 1.53), and showed an increasing gradient from the wealthiest to poorest quintile (ORs 1.00, 1.52, 2.18, 2.49, 3.23). In contrast, receipt did not differ significantly by sex or socio-economic group.


CONCLUSIONS
People living in the North of England, women and the less wealthy experience relatively high levels of need, yet do not receive relatively more hip and knee joint replacements.


Introduction
Arthritis is the commonest cause of disability, is a frequent reason for medical consultations, and causes considerable costs to the health service and to society [1][2][3]. Pain and disability in the knee or hip joint caused by osteoarthritis can be effectively treated by surgical replacement of the joint [4][5][6][7][8].
Higher need for hip and knee joint replacement has been found in women and in those with lower socio-economic status in the Somerset and Avon Survey of Health and other regional studies [9][10][11][12][13][14][15]. Similar variations in need have been reported from Sweden, Canada and the USA [16][17][18][19][20]. These variations in need follow the familiar pattern of worse general health and disability in England being associated with lower socio-economic position, female gender and living in the North [21,22]. The inverse care law [23] appears to apply to joint replacement, and those with lower socio-economic status experience lower rates of joint replacement in both England [24,25] and the USA [26][27][28][29].
There has been no previous estimate of both need and receipt of hip or knee joint replacement on a national sample in England. We present data on the national distribution of symptomatic indications for hip or knee joint replacement surgery, and of existing joint replacements, with detailed estimation of comorbidities and socio-economic status, including direct measures of wealth. We include replacements carried out in the independent sector, which accounts for approximately one-quarter of hip or knee joint replacements [30].
We aimed to estimate the prevalence of need and receipt of hip and knee joint replacements, and to estimate their associations with respondent characteristics. We wanted to determine which population groups in England most needed a hip or knee joint replacement, and whether they were also the most likely to receive one.

Data collection
The English Longitudinal Study of Ageing (ELSA) is an interview survey of a sample of the population of England aged 50 yrs or older. During 2002, interviewers collected data using face-to-face interviews in participants' homes. The sample was drawn from the Health Survey for England sample, which was designed to be representative of the English population living in private households, and was stratified by health authority and deprivation. The questions on joint replacement were administered only to those aged 60 yrs or over, the age group of those who receive over 85% of hip and knee replacements in England [31]. All analyses in this study refer only to this age group. The survey methodology is described in more detail elsewhere [32].
Participants were asked about receipt of joint replacement, pain and functional limitation (Table 1) as well as diagnosed illness, smoking and socio-economic status. Obesity had been previously measured in the Health Survey for England. Questions about socio-economic status included direct measures of total non-pension wealth for the first time in a survey of this nature. For older individuals, wealth is a better measure than income of an individual's past circumstances and lifetime living standards.

Estimation of need
Classification of need for hip or knee joint replacement required respondents to be troubled often by pain, with pain in the knee or hip when walking on the flat rated as 5 or more on a scale of 0-10, plus some or much difficulty or inability to walk a quarter of a mile. These criteria are consistent with the conclusions of the National Institutes of Health Consensus Development Panel on total hip replacement, and consensus statement on total knee replacement [33,34]. The criteria are simpler than the three most widely used scoring systems for hip and knee disability (New Zealand, WOMAC and Lequesne), but measure similar domains of self-rated pain and functional activity [35][36][37].
We excluded all those with a previous hip or knee replacement from the classification of need. We estimated need for hip and knee replacement together rather than separately because arthritis of the hip can present with knee pain, and less than half of those with knee pain have pain affecting one hip or knee joint only [9]. We excluded hip replacements due to fracture, as they are often emergency operations, and have different aetiology from elective hip replacement.

Statistical analysis
The data were weighted to adjust the respondent age and sex distribution to the Census 2001 non-institutionalized population distribution. Binary logistic regression modelling was used to evaluate the independent determinants of having received one or more replacement knee or hip joints, and of needing a hip or knee joint replacement. The independent variables included in the model were: age, sex, smoking status, education, occupational class, obesity, total wealth and geographic region. They were chosen on the basis that they were likely to influence joint replacement rates. The statistical software used was STATA SE 8.
We conducted a sensitivity analysis to estimate the effect of contraindications to surgery. We excluded from the classification of need those with the following possible contraindications to surgery: reported diagnosis of a heart attack, stroke, chronic lung disease or cancer, and difficulty walking a quarter of a mile caused by either chest pain or shortness of breath. Those with intermittent claudication on the Edinburgh questionnaire [38] were also excluded, because claudication is a possible cause of leg pain and difficulty walking.

Results
Interviews were achieved with 3885 women and 3216 men, giving a total of 7101 eligible sample members, out of 10 599 invited to participate (response rate 67%). Data were present for all the variables in the regression analysis for 'receipt' on 6028 people, for all variables for 'need' on 5641 people, and for all variables for 'need with no major contraindication' for 5524 people. The application of the weights as described before made a very slight difference to the results, and weighted data were used in the following analyses.

Prevalence of existing joint replacement
The prevalence of a lifetime receipt of hip or knee joint replacement, excluding hip replacements due to fracture, in the weighted population aged 60 yrs or over was 6% [95% confidence interval (CI) 5, 7] in women, 5% (CI 5, 6) in men and 6% (95% CI 5, 6) overall (Table 2).
Receipt was lower in the North than the South [adjusted odds ratio (OR) 0.72, 95% CI 0.53, 0.96]. It did not differ significantly by sex, wealth (population quintiles), education or occupational class. As expected, lifetime prevalence in those aged 75 and over was greater than in those aged 60-74 (OR 1.98, 95% CI 1.57, 2.50). Receipt was lower for current smokers (OR 0.6, 95% CI 0.39, 0.94) ( Table 3).

Sensitivity analysis
Among the respondents classified as in 'need', 412 people (44%) reported either a contraindication to surgery or intermittent claudication, and were excluded from the classification of 'need with no major contraindication to surgery'. After exclusion of those with potential contraindications, the overall prevalence of need was 6% (95% CI 5, 6). Repeating the regression analysis using this measure of need showed that the excess of need previously seen in poorer quintiles, and in smokers, either disappeared or reduced, and the gradient of increasing need from rich to poor disappeared (Table 3).

Discussion
In this national population-based study, we found a mismatch between need and receipt for hip and knee joint replacements. Need in the North of England is 30% higher than in the South, yet existing replacements are 30% less prevalent. Women have 30% higher need than men, yet receive replacements no more frequently. The poorest fifth of the population experience nearly three times as much need as the wealthiest, and replacements are How would you rate your pain if you were walking on a flat surface? Please rate your pain from 0 to 10 for each of the following where 0 is no pain and 10 is severe or excruciating pain, as bad as you can imagine: In your hips? In your knees? By yourself and without using any special equipment, how much difficulty do you have walking for a quarter of a mile? Do you have: no difficulty; some difficulty; much difficulty; or, are you unable to do this? no commoner, and this inequality is at least partly explained by higher comorbidity in the less-wealthy quintiles. The fewer replacements in smokers are also explained by higher comorbidity levels. These findings are unlikely to be due to inaccurate estimation of wealth or exclusion of replacements provided by the private sector. This study provides the first national estimate of both need and receipt of hip or knee joint replacement. Strengths of the study are that it includes detailed estimates of socio-economic status rather than indirect measures, it allows for differential rates of comorbidities, and it includes joint replacements from both the National Health Service and independent sector.
Our estimate of potential need for hip or knee replacement surgery of 6% is slightly higher than previous estimates, which is not surprising given an older population and broad definition of need. Previous estimates of need vary from 2.4% in people aged over 35 yrs in Avon and Somerset [13,14], to 3.4% in North Yorkshire, Wiltshire and Sheffield and Tameside [10][11][12]39], and 4.5% among women and 2.1% among men in Canada [19].

Limitations
The estimate of need for joint replacement relies on self-reported pain and disability, and these preliminary findings require confirmation using radiological and skilled clinical examination. Our method was designed to produce a reasonably accurate epidemiological estimate of need, given a low correlation between radiography and disease, and unsatisfactory evidence to define criteria for knee or hip replacement [7,14,40].
There is a possibility that some of our subjects were suffering from referred back pain, rather than pain originating in the knee or hip, although respondents were asked separately about pain in the hip, knee and back. Any misclassification for this reason is likely to have affected all groups similarly, making bias unlikely. No account was taken of respondents' willingness to consider hip and knee arthroplasty, but this is unlikely to introduce a bias as willingness to consider arthroplasty has not been shown to be associated with sex, education or income, nor to differ between hips and knees [18,19,41].

Implications
Our findings suggest that access to hip or knee replacement is reduced by barriers which exist particularly for people living in the North, for women, and for the less wealthy; however, they need to be confirmed by other studies. The barrier for the less-wealthy groups appears, at least partly, to be high levels of co-morbid illness. Painful, disabling osteoarthritis is significantly commoner In 'need' of hip or knee replacement* (n ¼ 5641) In 'need' of hip or knee replacement, with no major contraindication** (n ¼ 5  *Need for hip or knee replacement (severe joint disease) ¼ often troubled by pain, with pain in the knee or hip on walking on the flat rated as 5 or more on a scale of 0-10, and reported difficulty ('some' or 'much') or inability to walk a quarter of a mile. **Exclusions were diagnosed: heart attack (including myocardial infarction or coronary thrombosis), stroke (cerebral vascular disease), chronic lung disease such as chronic bronchitis or emphysema, or cancer or a malignant tumour (excluding minor skin cancers), difficulty with walking a quarter of a mile caused by either chest pain or shortness of breath and grade 2 intermittent claudication (Edinburgh claudication questionnaire [38]). ***NS-SEC, National Statistics Socio Economic Classification.
in less-wealthy groups, and so are several other diseases, yet it is these comorbid illnesses that may be preventing access to needed hip and knee replacement. For those in the North, there appear to be fewer opportunities to have a joint replacement than in the South. For women, the data suggest no simple explanation, and further research is needed to demonstrate where in the pathway from pain and disability to joint replacement the barriers to treatment occur. One possible barrier may be the views of referring practitioners on the patient characteristics associated with a favourable outcome from hip replacement [42]. There may be a case for targeted funding to reduce avoidable mobility disabilities in lower-access regions and groups. Bold font is used where the 95% CIs exclude 1. *Need for hip or knee replacement (severe joint disease) ¼ often troubled by pain, with pain in the knee or hip on walking on the flat rated as 5 or more on a scale of 0-10, and reported difficulty ('some' or 'much') or inability to walk a quarter of a mile. **Exclusions were diagnosed: heart attack (including myocardial infarction or coronary thrombosis), stroke (cerebral vascular disease), chronic lung disease such as chronic bronchitis or emphysema, or cancer or a malignant tumour (excluding minor skin cancers), difficulty with walking a quarter of a mile caused by either chest pain or shortness of breath, and grade 2 intermittent claudication (Edinburgh claudication questionnaire [38]). ***NS-SEC, National Statistics Socio Economic Classification.

Key messages
Joint replacements are 30% rarer in the North than the South of England, despite higher need. Women and less-wealthy people experience higher need for joint replacement than men and wealthier groups, and relatively more of this need goes unmet.