Rheumatology Advance Access originally published online on February 17, 2007
Rheumatology 2007 46(5):877-881; doi:10.1093/rheumatology/kem013
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Osteoarthritis as a public health problem: the impact of developing knee pain on physical function in adults living in the community: (KNEST 3)
Primary Care Musculoskeletal Research Centre, Keele University, Keele, Staffordshire ST5 5BG, UK.
Correspondence to: Dr Clare Jinks, Lecturer in Health Services Research, Primary Care Musculoskeletal Research Centre, Keele University, Keele, Staffordshire ST5 5BG, UK. E-mail: c.jinks{at}cphc.keele.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Objectives. To determine the effect of newly developed knee pain on general health and physical function of people
50 yrs living in the community. Methods. Prospective cohort study of 3907 people aged 50+ registered with three general practices in North Staffordshire, in the UK. The main outcome measures were self-reported knee pain; general health and physical function as measured by the Short Form 36 (SF-36).
Results. Of those with no knee pain at baseline, 24% (n = 501) reported it at follow-up. There was a steep decline in physical function in this group (mean fall in SF-36 score at follow-up 10.3 points) compared with the 1558 persons who had no knee pain at baseline or follow-up (mean fall 3.3). Those with knee pain at baseline whose pain had resolved at 3-yr follow-up (n = 409) showed only minor improvements in physical functioning scores at follow-up (mean improvement –1.3).
Conclusions. The onset of knee pain is associated with a substantial and persistent reduction of physical function in older adults living in the community. Since knee pain is common, and reduced physical function in mid to older ages is a strong predictor of future disability and dependency, effective prevention or early treatment of knee pain at these ages is likely to have a major influence on healthy ageing in the population.
KEY WORDS: Public health, Knee pain, Osteoarthritis, SF-36, Disability, Survey, Knee Pain Screening Tool (KNEST)
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Musculoskeletal conditions are an important public health problem [1]. The population burden from arthritis is considerable with an estimated 48 million people in the US, 8 million in the UK and 108 million across Europe suffering from some form of this condition [2, 3]. The implications of this burden are poor quality of life, restrictions in daily activities and disability [4, 5]. Population projections from the US estimate that by the year 2030, the prevalence of self-reported doctor-diagnosed arthritis will increase from 47.8 million in 2005 to nearly 67 million, with 25 million reporting arthritis-attributable activity limitations [3]. The biggest contributor to this is osteoarthritis, and its most frequent manifestation—disabling joint pain in older people—will be an increasingly important public health concern in older people according to these trends.
The knee is the site most affected by joint pain in older adults [5–7], where it is usually attributed to osteoarthritis in this age group. Nearly half of adults aged 50 and over report knee pain in a 1-yr period [8]. The high prevalence of this condition, its impact in terms of disability, and the existence of modifiable risk factors, mean that identifying approaches to prevention are a public health priority [9].
Recent studies have focused on how to prevent progression of pain and disability in persons with OA [10], but there has been less attention paid to prevention strategies at the general population level. One reason for this may be due to the challenges of screening for this condition [11] and only a small number of studies have addressed screening for knee pain and symptomatic knee OA at the population level [8, 11–13]. Effective screening and targeted prevention can only occur if the natural history of the disease is understood [14] and there are few population-based studies of the course of knee pain and knee osteoarthritis over time [15–18].
Given the major contribution of musculoskeletal disorders to disability, the first onset of joint pain might provide a specific target for prevention of future disability. However, little is known about the initial impact of joint pain on disability in the older general population, nor about whether such impact is reversible if the pain gets better. The main objective of this study was, therefore, to determine the incidence of knee pain over a 3-yr period and to assess its impact on general health status, and physical function specifically, in people aged 50 and over living in the community. The second objective was to investigate the rate of recovery and persistence of knee pain over 3 yrs and associated changes in health status.
|
Patients and methods |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
The baseline questionnaire was administered in March 2000 to all patients aged 50 and over registered at three general practices in North Staffordshire, UK. The follow-up survey was conducted 3 yrs later in all baseline responders who were still registered with the practices.
The questionnaires included the Knee Pain Screening Tool (KNEST) [8] which asks whether the respondent has had pain in or around the knee in the last year. The Short Form-36 (SF-36 version 1) profile tool was used to measure the general health of the responders [19]. This measure has 36 items which can be grouped and transformed into eight dimensions (for example, physical functioning, body pain, general health) with scores ranging from 0 (worst health) to 100 (best health). Self-reported height and weight at baseline were used to determine body mass index (BMI). A BMI 25 was defined as overweight and BMI 30 was defined as obese [20]. Psychological distress (Hospital Anxiety and Depression Scale [21]), social deprivation (Townsend score [22]) and other body pain (derived from pain drawing [23]) were also measured.
Ethics approval was gained from North Staffordshire Research Ethics Committee (LREC Project 862 for the baseline survey, LREC ref02/61 for the follow-up survey).
Statistical analysis
Knee pain was defined at both baseline and follow-up by the KNEST screening question. This asks about pain in or around the knee in the last year. Four groups of responders were defined according to change in knee pain over the 3 yrs: (i) no knee pain at baseline or follow-up (‘no knee pain’ group), (ii) no knee pain at baseline but knee pain at follow-up (‘new knee pain’ group), (iii) knee pain at baseline but not at follow-up (‘resolved knee pain’ group), (iv) knee pain at both baseline and follow-up (‘continuing knee pain’ group). According to the definition used (i.e. pain in the last 12 months), those whose pain had resolved must have had improvements in the first 2 yrs.
The four groups were compared according to scores on the SF-36. Mean baseline scores for new, resolved and continuing groups were each compared to the no knee pain group unadjusted, and then adjusted for age, gender, deprivation status, depression, overweight or obese, and pain elsewhere at baseline. A similar analysis was then performed based on mean 3-yr change in SF-36 scale scores, adjusting for the same baseline characteristics. Finally, the SF-36 change analysis was repeated adjusting for 3-yr changes in depression status (four categories: became depressed, became non-depressed, stayed depressed, stayed non-depressed), BMI (one category gain in BMI, one category loss, remained overweight or obese, remained normal weight) and pain elsewhere (gained pain elsewhere, lost all pain elsewhere, still had pain elsewhere, still had no pain elsewhere), as well as age, gender and social deprivation. All adjusted analyses were carried out using multiple linear regression with all independent variables entered simultaneously and checks performed on the residuals for departure from normality, homogeneity of variance and multicollinearity between the independent variables.
|
Results |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
Response
At baseline, 8995 subjects were sent a questionnaire to which 6792 responded (adjusted response 77%). 5784 of these 6792 subjects were alive and still registered at the practices at follow-up and were sent a follow-up questionnaire to which 4317 responded (adjusted response 75%). 3907 (91%) responders to the follow-up questionnaire answered the KNEST knee pain question at both baseline and follow-up. In total, 43% of the original population sent the baseline questionnaire (53% adjusted for deaths and departures from the GP list) were followed up at 3 yrs.
Those who answered the KNEST knee pain question at both baseline and follow-up were slightly younger (mean difference 1.8 yrs; 95% CI 1.3, 2.4) and had slightly better SF-36 scores (mean differences of between 3 and 6 points higher) than those who did not respond at follow-up or failed to complete the KNEST knee pain question at baseline or follow-up. However, there was no difference by gender or by response to the baseline knee pain question.
Change in knee pain status
Prevalence of ‘recent’ knee pain (i.e. in the 12 months prior to the surveys) was similar at baseline (47%) and follow-up (50%). Of those reporting no recent knee pain at baseline, 24% reported it at follow-up. Of those who did report recent knee pain at baseline, 22% were free of it at follow-up (Table 1).
|
The distribution of the four groups based on change in knee pain in this population was 40% (no knee pain), 37% (continuing), 13% (new) and 10% (resolved). The baseline characteristics of these groups are outlined in Table 2. The groups were similar in terms of age at baseline with mean ages ranging from 63.4 (S.D. 9.4: new knee pain group) to 64.6 (S.D. 9.2: continuing knee pain group). However, a larger proportion of the continuing group were female (60% compared with 52% in the no knee pain group).
|
SF-36 scores at baseline
Table 3 shows the mean SF-36 scores at baseline. The no knee pain group had the highest (better health) mean SF-36 scale scores at baseline and those subsequently classified to the continuing knee pain group had the worst. The largest differences in mean scores between these two groups were for physical functioning (mean difference 25; 95% CI 23, 27), Body Pain (mean difference 27; 95% CI 25, 29) and Role Limitations due to physical problems (mean difference 28; 95% CI 25, 31); the smallest difference was for Mental Health (mean difference 6; 95% CI 5, 8). Those who developed knee pain at follow-up (the new knee pain group) had scores 5–8 points lower at baseline on Physical Functioning, Body Pain and Role Limitations due to physical problems compared with the no knee pain group.
|
Differences in baseline SF-36 scores between groups persisted but were reduced after adjusting for other baseline characteristics (Table 4); with adjusted differences between the new knee pain group and the no knee pain group of <5 points on all the scales. For the physical functioning scale, the difference in mean scores between continuing knee pain and the no knee pain group fell from 25 to 16 points (95% CI 14, 18) after adjustment for other factors. Adjustment also led to similar-sized falls on the Body Pain and Role Limitations due to physical problems scales.
|
SF-36 scores—3-yr change
The no knee pain, resolved and continuing groups remained relatively stable in SF-36 scores at follow up, with generally only small falls in the no knee pain and continuing groups (Table 3). In contrast, the new knee pain group (i.e. those who developed knee pain between baseline and follow-up) showed a mean fall of more than 10 points for Physical Functioning, Body Pain and Role Limitations due to Physical problems, and had scores at follow-up similar or only slightly lower than those in the resolved group. The resolved group still had mean scores at follow up between 4 and 14 points lower than the no knee pain group, including 9 points lower on Physical Functioning (95% CI 6, 12), and only showed slight improvements in scores at follow-up relative to the no knee pain group on the Physical Functioning and Body Pain scales (Table 5). Adjustment for baseline differences made little difference to changes in scores for each group relative to the no knee pain group (not shown).
|
After adjustment for any changes in the other variables measured, the new knee pain group showed falls relative to the no knee pain group of 6 points (95% CI 4, 8) for Physical Functioning, 7 points (95% CI 5, 10) for Body Pain and 10 points (95% CI 6, 15) on the Role Limitations due to Physical problems scale (Table 5). There were small falls in this new pain group on all other SF-36 scales except Mental Health. The continuing group had adjusted reductions in SF-36 scores no greater than those observed in the no knee pain group. The resolved knee pain group generally showed a slight improvement in adjusted scores relative to the no knee pain group but by no more than 4 points on any of the scales.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
The major change in self-reported health status between baseline and follow-up was the decline in physical function among persons who developed knee pain. Although there was evidence that this group already had slightly poorer physical health status at baseline (i.e. prior to the onset of their knee pain) than the group who remained pain-free, their subsequent decline in physical functioning scores was substantial compared with the no pain group, even after adjustment for differences in SF-36 scores and other characteristics at baseline and for changes in these other characteristics during follow-up. This finding highlights the early limitation in physical activities that is linked with onset of knee pain. Furthermore, persons whose pain resolved during follow-up (i.e. have been pain-free for at least 1 yr at follow-up) did not return to full function, suggesting that the initial decline in SF-36 scores associated with onset of knee pain does not simply reverse if the pain gets better. The phenomenon of resolving pain not leading to an inevitable return to normal functioning has been shown in relation to joint replacement surgery [24]. In our study, persisting limitations in those whose knee pain resolves could not be attributed to changes in other health-related factors.
From a public health perspective, identifying people at an early stage in this age group who develop knee pain may be important as a strategy for secondary prevention of disability, but our findings also highlight the potential for primary prevention of knee pain in older people as a means to avoid a significant decline in physical function. This is particularly relevant in light of previous research in the general population which has shown that locomotor disability is an early marker for disability progression [25].
One in four people develop knee pain and nearly a quarter of those with knee pain at baseline experience resolution by follow-up. Previous studies have not investigated the incidence of knee pain, but have reported similar degrees of recovery. Dawson and colleagues [16] followed up a community sample of adults aged 65 with knee or hip pain and found that after 1 yr, 29% of people who had baseline knee pain (without hip pain) were symptom-free. Peters et al. [17] reported that 29% of people in the general population aged 35 and over with knee arthropathies improved over a 7-yr period.
There has been one previous study of new onset joint pain in older people which has investigated its association with functional decline [26]. Although this study did not distinguish between the different joint sites, it reported the same pattern as we have observed for the knee, of modestly higher baseline disability scores in those who subsequently developed any joint pain in a 12-month period compared with those who did not, and a strong association between acquiring joint pain and functional decline during the 12 months of the study.
We have added to the findings of Dawson et al. [16] by showing that differences between knee pain groups continue over 3 yrs, and that the biggest decline over time is in persons who develop knee pain. Furthermore, the size of the physical function decline in persons with continuing pain is no greater than the decline in those do not develop knee pain and whose change can be considered as caused by other factors related to ageing but unrelated to pain.
This was a large longitudinal study of the general population aged 50 and over. Based on the smallest group (the resolved pain group), there was a power of >90% to detect a mean difference in change scores of 5 points compared with the no knee pain group for all the SF-36 scales except the role limitation scales (7 points).
There is no established definition of what is a clinically important difference on the SF-36. However, recommended clinically meaningful differences suggest that a difference of at least 5–7 points (and probably twice that on the two role limitation scales) is needed before it can be considered a meaningful change [27]. A change of 5 points on the Physical Functioning scale, for example, would mean one more item of limitation (which may not be severe), and of 10 points would mean one extra item of severe limitation. Our results suggest that, whilst the new knee pain group scored significantly worse than the no knee pain group at baseline on all but the General Health scale, the estimated size of the baseline differences between the two groups were unlikely to be clinically meaningful. Most of these differences also lost their statistical significance after adjustment for other related factors. In contrast, the size of the subsequent decline in physical functioning, physical role limitation, body pain and general health in the new pain group appears clinically and statistically significant even after adjustment for changes in other factors.
Our results are likely to be generalizable to the wider UK Caucasian population. Overall mean SF-36 scores in our study are similar to the UK norm-based SF-36 data [28]. SF-36 scores for survey responders with knee pain are also similar to those reported in another population-based cross-sectional study [29]. Whilst the response rate was high at both baseline and follow-up (75%), responders at follow-up tended to report slightly less restriction on the baseline SF-36 than those who did not respond. This, however, is unlikely to affect the reported differences between the four responder groups reported in this article.
The Role Limitations due to Physical problems scale on the SF-36 showed large differences between groups but this scale suffers from floor and ceiling effects [27]. Analysis of the normality of the residuals from the multiple linear regression analyses suggested problems only with the two role limitations scales and these scales also had the largest number of people for whom a scale score could not be calculated. The two scales have been amended in version 2 of the SF-36 and need to be treated with caution although have been shown here for completeness. Apart from the Role Limitations scales, the Physical Functioning scale had the largest percentage of subjects who had at least one of the baseline or follow-up scores missing (9%). Implementation of the SF-36 Missing Data Estimation facility reduced this to 5% and led to no differences in the results reported here.
Our definition of having knee pain was pain at any point during the previous 12 months. This meant that a respondent may not have had pain in the 4 weeks before the survey which is the time frame of the SF-36. The effect of this is likely to be that the differences reported here between the new, resolved and continuing groups and the no knee pain group will be smaller than if an identical time frame had been used. However, it does mean that those reporting no knee pain have had no knee pain for at least a year and therefore can be considered free of knee pain rather than going through a pain-free episodic phase.
Our study is also based upon a wide spectrum of knee pain and will include short episodes of pain as well as chronic pain, along with non-severe and severe pain. The differences in SF-36 scores between groups are likely to be greater if the definition of knee pain was based on severe knee pain.
There may be other factors associated with general health and physical function decline that this study has not measured. These factors may include, for example, comorbid conditions (like cancer, stroke and diabetes), and psychological factors like illness perceptions, health beliefs and optimism and pessimism [30]. Decreased physical functioning may also be a risk factor for knee pain. A specific issue is that the decrease in physical function may occur before the onset of knee pain. There is some evidence for this from our study in the lower baseline physical function scores in the new pain group. This may also partly explain the reason that the resolved group do not return to full functioning. However, the substantial decline in functioning observed in the new knee pain group highlights the close association between pain symptoms and loss of function.
|
Conclusion |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
A quarter of adults aged over 50 who are free of knee pain will develop it during the following 3 yrs and this is the group that experiences the biggest change in their self-reported physical functioning in this time frame. This finding underlines the importance of knee pain as a potential trigger of physical decline in older people. Nearly a quarter of adults over 50 yrs who have knee pain will experience resolution of pain at 3 yrs, but their general health status and physical functioning will not have returned to the levels experienced prior to onset of knee pain. There is now an urgent need for future studies to identify strategies of prevention which might be effective at different stages of development of disabling knee pain in the adult population.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
We would like to thank the doctors and patients of the three health centres involved and also the administration team in Primary Care Sciences who helped with the surveys. The baseline survey was funded by The West Midlands New Blood Research Fellowship Committee (former West Midlands Regional Health Authority) and the Haywood Rheumatism Research and Development Foundation (HRRDF), North Staffordshire. The follow-up survey was funded by the North Staffordshire Primary Care Research Consortium. All authors are independent of the funders.
The authors have no conflicts of interest.
|
References |
|---|
|
TopAbstractIntroductionPatients and methodsResultsDiscussionConclusionAcknowledgementsReferences |
|---|
- World Health Organisation Scientific Group. (2003) The Burden of Musculoskeletal Conditions at the Start of the New Millennium(World Health Organization, Geneva).
- Dunlop DD, Manheim LM, Yelin EH, Song J, Chang RW. (2003) The costs of arthritis. Arthritis Rheum 49:101–13.[CrossRef][ISI][Medline]
- Hootman JM and Helmick CG. (2006) Projections of US prevalence of arthritis and associated activity limitations. Arthritis Rheum 54:226–9.[CrossRef][ISI][Medline]
- Sprangers MA, de Regt EB, Andries F, et al. (2000) Which chronic conditions are associated with better or poorer quality of life? J Clin Epidemiol 53:895–907.[CrossRef][ISI][Medline]
- van der Waal JM, Terwee CB, van der Windt DA. (2005) Health-related and overall quality of life of patients with chronic hip and knee complaints in general practice. Qual Life Res 14:795–803.[CrossRef][ISI][Medline]
- Thomas E, Peat G, Harris L, Wilkie R, Croft PR. (2004) The prevalence of pain and pain interference in a general population of older adults: cross-sectional findings from the North Staffordshire Osteoarthritis Project (NorStOP). Pain 110:361–8.[CrossRef][ISI][Medline]
- Urwin M, Symmons D, Allison T, et al. (1998) Estimating the burden of musculoskeletal disorders in the community: the comparative prevalence of symptoms at different anatomical sites, and the relation to social deprivation. Ann Rheum Dis 57:649–55.
[Abstract/Free Full Text] - Jinks C, Jordan K, Ong BN, Croft P. (2004) A brief screening tool for knee pain in primary care (KNEST). 2. Results from a survey in the general population aged 50 and over. Rheumatology 3:55–61.
- Hunter DJ, March L, Sambrook PN. (2002) Knee osteoarthritis: the influence of environmental factors. Clin Exp Rheumatol 20:93–100.[ISI][Medline]
- Felson DT and Zhang Y. (1998) An update on the epidemiology of knee and hip osteoarthritis with a view to prevention. Arthritis Rheum 41:1343–55.[CrossRef][ISI][Medline]
- Jinks C, Lewis M, Ong BN, Croft P. (2001) A brief screening tool for knee pain in primary care. 1. Validity and reliability. Rheumatology 40:528–36.
[Abstract/Free Full Text] - LaValley M, McAlindon TE, Evans S, Chaisson CE, Felson DT. (2001) Problems in the development and validation of questionnaire-based screening instruments for ascertaining cases with symptomatic knee osteoarthritis: the Framingham Study. Arthritis Rheum 44:1105–13.[CrossRef][ISI][Medline]
- O’Reilly SC, Muir KR, Doherty M. (1996) Screening for pain in knee osteoarthritis: which question? Ann Rheum Dis 55:931–33.
[Abstract/Free Full Text] - Wilson J and Junger J. (1968) Principles and practice of screening for disease(World Health Organisation, Geneva).
- Cooper C, Snow S, McAlindon TE, Kellingray S, Stuart B, Coggon D, Dieppe PA. (2000) Risk factors for the incidence and progression of radiographic knee osteoarthritis. Arthritis Rheum 43:995–1000.[CrossRef][ISI][Medline]
- Dawson J, Linsell L, Zondervan K. (2005) Impact of persistent hip or knee pain on overall health status in elderly people: a longitudinal population study. Arthritis Rheum 53:368–74.[CrossRef][ISI][Medline]
- Peters TJ, Sanders C, Dieppe P, Donovan J. (2005) Factors associated with change in pain and disability over time: a community-based prospective observational study of hip and knee osteoarthritis. Br J Gen Pract 55:205–11.[ISI][Medline]
- Hart DJ, Doyle DV, Spector TD. (1999) Incidence and risk factors for radiographic knee osteoarthritis in middle-aged women: the Chingford Study. Arthritis Rheum 42:17–24.[CrossRef][ISI][Medline]
- Ware JE, Snow K, Kosinski M, Gandek B. (1993) SF36 Health Survey. Manual & Interpretation Guide(Health Institute, New England Medical Centre, Boston).
- World Health Organization. Preventing and Managing the Global Epidemic of Obesity1997GenevaReport of the World Health Organization Consultation on Obesity.
- Zigmond AS and Snaith RP. (1983) The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand 67:361–70.[ISI][Medline]
- Townsend P, Phillimore P, Beattie A. (1988) Health and deprivation: inequality and the North(Croom Helm, London).
- Lacey RJ, Lewis M, Jordan K, Jinks C, Sim J. (2005) Inter-rater reliability of scoring of pain drawings in self-reported health surveys. Spine 30:E455–8.[CrossRef][ISI][Medline]
- Boutron I, Polraudeau S, Ravaud J, et al. (2003) Disability in adults with hip and knee arthroplasty. A French national community based survey. Ann Rheu Dis 62:748–54.
[Abstract/Free Full Text] - Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. (1995) Lower extremity function in persons over the age of 70 years as a predictor of subsequent disability. New Engl J Med 332:556–61.
[Abstract/Free Full Text] - Donald IP and Foy C. (2004) A longitudinal study of joint pain in older people. Rheumatology 43:1256–60.
[Abstract/Free Full Text] - Wyrwich KW, Tierney WM, Wolinsky FD. (1999) Further evidence supporting an SEM-based criterion for identifying meaningful intra-individual changes in health-related quality of life. J Clin Epidemiol 52:861–73.[CrossRef][ISI][Medline]
- Bowling A, Bond M, Jenkinson C, Lamping DL. (1999) Short form 36 (SF-36) Health Survey questionnaire: which normative data should be used? Comparisons between the norms provided by the Omnibus Survey in Britain, Health Survey for England and the Oxford Life Survey. J Public Health Med 21:255–70.
[Abstract/Free Full Text] - Walters SJ, Munro JF, Brazier JE. (2001) Using the SF-36 with older adults: a cross-sectional community-based survey. Age Ageing 30:337–43.
[Abstract/Free Full Text] - Brenes GA, Rapp SR, Rejeski WJ, Miller ME. (2002) Do optimism and pessimism predict physical functioning? J Behav Med 25:219–31.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Jinks, K. P. Jordan, M. Blagojevic, and P. Croft Predictors of onset and progression of knee pain in adults living in the community. A prospective study Rheumatology, March 1, 2008; 47(3): 368 - 374. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||