Article Text


Extended report
Worsening of pain and function over 5 years in individuals with ‘early’ OA is related to structural damage: data from the Osteoarthritis Initiative and CHECK (Cohort Hip & Cohort Knee) study
  1. Janet Wesseling1,
  2. Sita M A Bierma-Zeinstra2,3,
  3. Margreet Kloppenburg4,
  4. Rik Meijer5,
  5. Johannes W J Bijlsma1
  1. 1Department of Rheumatology & Clinical Immunology, University Medical Center Utrecht, Utrecht, The Netherlands
  2. 2Department of General Practice, Erasmus MC—University Medical Center Rotterdam, Rotterdam, The Netherlands
  3. 3Department of Orthopedics, Erasmus MC-University Medical Center Rotterdam, The Netherlands
  4. 4Department of Rheumatology and Department of Clinical Epidemiology, Leiden University Medical Centre, Leiden, The Netherlands
  5. 5Department of Rheumatology, Wilhelmina Hospital Assen, Assen, The Netherlands
  1. Correspondence to J Wesseling, Department of Rheumatology & Clinical Immunology (F02.127), University Medical Center Utrecht, PO Box 85500, Utrecht 3508 GA, The Netherlands; j.wesseling{at}


Objective To analyse the relation of joint damage on evolution of pain and physical functioning in two different cohorts of early phase of osteoarthritis (OA): Osteoarthritis Initiative (OAI) and Cohort Hip & Cohort Knee study (CHECK).

Methods Longitudinal data of 4–5 years follow-up (= medium term) of CHECK study and OAI were used. The Western Ontario and McMaster Universities of Osteoarthritis Index (WOMAC) was used to measure pain and physical functioning. For comparison with CHECK a subgroup of the OAI Incidence cohort was selected, Generalised estimating equations were used and all models were adjusted for gender, Body Mass Index, age, amount of working hours, racial background and hip pain at baseline. Finally, an interaction term was added to assess a possible differential effect of radiological progression on evolution of pain and function in the two cohorts.

Results At baseline, CHECK had less radiographic OA (K&L ≥2) compared with the OAI Incidence subgroup, but at follow-up CHECK had more radiographic change (51% vs 15% ≥1 K&L point increase; p<0.001). Over 4 years of follow-up, evolution of pain and physical functioning remained fairly stable in both cohorts, however, at different levels, OAI participants had less pain and less problems with physical functioning compared to CHECK participants. Both cohorts showed the same negative effect of rapid radiological change (K&L change of 2 points) on pain and physical function.

Conclusions In participants with ‘early symptomatic OA’ rapid radiological change (change of ≥2 grades of K&L in 4 or 5 years) is related to worsening of pain and function

  • Osteoarthritis
  • Knee Osteoarthritis
  • Epidemiology
View Full Text

Statistics from


Knee pain is often the first sign of knee osteoarthritis (OA),1 and it is known that its evolution can be very different between patients. Besides pain, OA causes decline in physical function on the long term.2 Studies to analyse the natural evolution of pain and physical function in the early stage of OA are sparse. Osteoarthritis Initiative Incidence cohort (OAI) and Cohort Hip & Cohort Knee study (CHECK) focus both on the early phase of OA; we compared baseline characteristics of both cohorts and described that at baseline, radiographic knee OA was more present in OAI Incidence subgroup compared to CHECK knee pain group. This radiological observation is in discordance with the more severe clinical symptoms reported by the CHECK participants. Discrepancies between radiological findings and clinical symptoms in OA are well described, but the actual relation between radiographic damage and clinical symptoms over time has never been clear and is subject of this study in both cohorts.3 The present study aims to identify the evolution of pain and physical functioning on medium term in these different OA cohorts and to analyse the relation of structural damage with this evolution.


Design and study population of CHECK

From October 2002 till September 2005, a cohort was formed of 1002 participants with pain and/or stiffness of knee and/or hip. The study was approved by the medical ethics committees of the 12 participating centres, and all participants gave their written informed consent before entering the study. A complete description of the CHECK study has been reported elsewhere.3 Individuals were eligible if they had pain or stiffness of knee or hip, were aged 45–65 years, and had not yet consulted their physician for these symptoms, or the first consultation was within 6 months before entry.3 For the analyses in this study, only longitudinal data on medium term of CHECK participants with knee pain at baseline were included, irrespective of their annual or variable schedule to visit the study centre.

Design and study population of OAI

From March 2004 to May 2006, the OAI cohort was formed and divided into two subcohorts: progression cohort and an incidence cohort. For the incidence cohort, age-specific eligibility criteria were defined. All details of the OAI are available on the internet ( For the present study, longitudinal data on medium term was used of participants in the age range 45–65 years, which had at least frequent or infrequent knee symptoms, excluding those that just had overweight without symptoms, and excluding those that had previous knee surgery.

Outcome variables

Evolution of pain and physical functioning over medium term was assessed with the WOMAC (Western Ontario and McMaster Universities Osteoarthritis Index) questionnaire. The WOMAC4–6 is a disease-specific questionnaire with well-known and good clinimetric properties and recommended by OMERACT.7 ,8 The five-point Likert version of the WOMAC was used; item responses range from ‘none’ to ‘extreme’ and are summed to produce subscales (pain 0–20, functioning 0–68) with higher scores indicating worse health. In both cohorts, the WOMAC pain and physical function were assessed at baseline and at follow-up after 1, 2, 3 and 4 years. CHECK assessed per participant a WOMAC score for pain and physical problems; they were asked to answer the WOMAC for their knee as well as hip complaints together. OAI assessed per knee per participant a WOMAC score for pain, and physical function. In the analyses the highest WOMAC score of the OAI participant is used.

Independent variables

In both cohorts at baseline and at follow-up, posterior-anterior weight-bearing radiographs of both tibiofemoral joints were obtained to assess radiographic OA features of the knee. Follow-up time for the radiographs was in CHECK 5 years, and in OAI 4 years. All views were scored according to Kellgren & Lawrence (K&L) and original description of K&L criteria was used.9 For each knee the radiographic change from baseline to follow-up was measured, and the greatest radiographic change in one of the knees was used in the analyses. Severity of radiographic OA at baseline was defined as the worst knee, defined by the knee with higher K&L grade.1

The radiographs of CHECK were scored by trained observers independently. CHECK readers scored all consecutive radiographs at the same time knowing the sequence (baseline, 2 and 5 years). OAI readers know the true baseline visit X-ray, but were blinded to the chronological order of follow-up visits (12, 24, 36 and 48 months), with grouped images from all available time points viewed simultaneously. All readers were blinded to the clinical data of the participant.

Statistical analyses

In the present report, CHECK participants with knee problems (excluding those with exclusively hip problems) were compared with participants of the OAI incidence cohort within the same age range (45–65 years). All analyses were performed in the participants of CHECK and OAI who had data of at least two visits (CHECK n=805; OAI n=1483). The postsurgery WOMAC data of participants who underwent total knee replacement during 4 or 5 year follow-up were coded as missing. Baseline characteristics of both cohorts are presented separately using descriptive statistics: mean or percentages. Differences between cohorts are analysed using t tests or χ2 test, where appropriate.

To analyse the evolution of pain and function in the two OA cohorts, we selected participants in both cohorts with K&L grade 0 or 1, because only these two grades were present at baseline in both cohorts.

To investigate the longitudinal relationship between change in pain and function and the independent variables of two OA cohorts, generalised estimating equations (GEE) were carried out. The correction for dependency of the individual observations is performed by assuming a working correlation structure; an exchangeable structure was used. To assess differential effects between the two, cohorts’ data of CHECK and OAI were merged into one dataset. To analyse the evolution of pain and function over time, time variable is added to the GEE model. The linearity of the relationship between these outcomes and time is analysed by adding time squared to the GEE model. In a quadratic relationship, or another non-linear relationship between outcome and independent variable, the independent variable is included in the model as a categorical variable.

All GEE models were adjusted for gender, Body Mass Index (BMI), age, amount of working hours, racial background and hip pain at baseline.

Finally, possible differential effects of radiological joint damage on the evolution of pain and physical functioning in both cohorts was investigated by introducing the interaction term ‘max K&L change* cohort’.


Patient population

At baseline, 805 CHECK participants had knee pain and data of at least two visits. The subgroup of OAI Incidence cohort consisted of 1483 participants (figure 1). The characteristics of both cohorts are given in table 1. At baseline, CHECK knee subgroup had less radiographic OA (K&L ≥2) and more pain and function problems compared to OAI Incidence subgroup. During follow-up CHECK participants developed more radiographic changes (51% vs 15% K&L grade change ≥1; p<0,001), with constant mean scores of pain and function over time (figure 2 and 3). In both cohorts increase of two and three grades of K&L, were combined as K&L ≥2, as a result of low numbers of three grades increase of K&L.

Table 1

Baseline demographic and disease characteristics in CHECK knee subgroup with ≥2 visits, OAI Incidence subgroup with ≥2 visits, CHECK knee subgroup with K&L grade 0 or 1 at baseline and OAI Incidence subgroup with K&L grade 0 or 1 at baseline

Figure 1

Flow diagram from the baseline study population selected for the present study.

Figure 2

Number of participants with radiological changes (Δ0–Δ3 points) from baseline to follow-up by baseline K&L score in CHECK and OAI participants with K&L 0 or 1 at baseline.

Figure 3

Mean scores (with SDs) at baseline, after 1, 2, 3 and 4 years follow-up on WOMAC pain and physical function scale of participants with K&L 0 or 1 at baseline. Higher scores indicates more pain and physical function problems.

In CHECK knee subgroup, there were, after 5 years of follow-up, 33 joint replacements (17 hip and 16 knee replacements) performed, and in the OAI Incidence subgroup after 4 years, 30 replacements (18 hip and 12 knee replacements). In the subgroup with K&L 0 or 1, the number of replacements in CHECK was the same; in OAI there were only 12 hip replacements performed.

Effect of change in joint damage in evolution of pain and function

In table 2 univariate β coefficients of variables which are included in the multivariate longitudinal model are presented. There was an univariate association of radiological change from baseline to follow-up with pain and physical functioning. Compared to no radiological change, the pain and function score increase with more radiological change from baseline to follow-up (table 2).

Table 2

Univariate regressions in CHECK and OAI participants adjusted for age, BMI, gender, working hours, racial background and hip pain at baseline in participants with K&L score 0 or 1 at baseline

In table 3, the results of the longitudinal regression model with evolution of pain as outcome variable are presented. There was a slight decrease of pain over time for the total group (OAI and CHECK together), but at a lower level of 2.6 points in OAI participants (less pain). In the model with evolution of function as outcome variable, there was, over time, a slight decrease of physical function in the total group, but at a lower score for function of 9.7 points in OAI participants (better function). The results presented in table 3 indicate that in all participants, for example, first year of follow-up is associated with a 0.3 lower pain score compared to baseline pain score; besides that, an OAI participant has a 2.6 lower pain score compared with a CHECK participant.

Table 3

Regression model adjusted for age, BMI, gender, working hours, racial background and hip pain at baseline in participants with K&L score 0 or 1 at baseline

In the total group, a significant association of the radiological change with evolution of pain and function was noted. The pain or function score of participants with a radiological K&L change of one point compared to those without changes was not influenced by this change in radiological damage. In participants with a K&L change of at least two points compared to those with zero K&L change, a reinforcing association with evolution of pain (increase of pain) and function (decline of function) was observed.

The interaction term: ‘max K&L change * cohort’ for the outcome pain and function was not statistically significant (overall effect with p value 0.2). This indicates that no differential effect of change in joint damage from baseline to follow-up on the evolution of pain and function between CHECK and OAI participants was observed.


The objective of our study was to identify the evolution of pain and physical functioning on medium term in two different OA cohorts (CHECK and OAI), and especially to analyse the relation of structural damage with this evolution. Participants of CHECK and OAI included for the analysis were aged 45–65 years, had knee pain at baseline, and had at baseline K&L 0 or 1. It can be concluded that in the total groups the evolution of pain and physical function remains fairly stable over medium term in participants with early symptoms of symptomatic OA in The Netherlands as well as in the USA. It was noted that the level of these outcome variables are different: OAI participants had less pain and less problems with physical functioning compared with CHECK participants. In literature, others also described a stable course of limitations in activity over a period of 5 years. At individual level, a considerable variation (indicated by large SDs of the change scores) was found.10 ,11 Also, in CHECK and OAI, considerable individual variation was found.

From baseline to follow-up, half the CHECK participants showed radiological change of at least one K&L grade, and in the OAI, only one-sixth of the participants showed radiological change of at least one point. In the literature, change of K&L grade is divided into incidence and progression, both with different definitions. Incidence is defined as a change from K&L grade 0 to grade ≥1 or more from baseline to follow-up. Progress is defined as a change from K&L 1 to any higher grade.1 This definition differs from a definition by Felson, who defined incidence as new-onset K&L grade 2 and progression when K&L grade increases (≥2) or joint replacement occurs.12 Additionally, it is described that knees that are progressing are far more likely to continue to progress than knees which have been stable in terms of structural deterioration: state of inertia.13 In this study, when we used the first definition of incidence and progression, there were 16% in CHECK and 4% in OAI who changed from K&L grade 0 to grade ≥1, so have incidence knee OA. In CHECK, 34% of participants with K&L grade ≥1 progressed, compared to 6% of OAI participants. According to Felson's definition, much lower percentages were found in incidence.

Importantly, in this study, we found an association between radiological change and the evolution of pain and function, which was comparable between both cohorts. These findings suggest it was not so much the presence of joint damage at the knee that was important for the evolution of pain and function, but rather the worsening of joint damage on medium term. A limited radiographic change of one point in K&L grade did not have an association with the evolution of pain and function. This limited change could be characterised as slow progression. However, an increase of two points or more in K&L grade is associated with more pain and function problems. Apparently, there must be a larger change in structural deterioration to have effect on the evolution of pain and function. This radiological change could be characterised as rapid progression. In the literature, it has been suggested that individuals with slow progression could have more opportunity to adapt to the changes in physiology and decreases in pain.14 The reported differences in pain and function levels could not be explained by difference in BMI, age, gender, working hours, racial background or hip pain at baseline. The approaches to recruit participants differed slightly: only 6% of the CHECK participants were recruited by a healthcare provider, compared to 1% in OAI. Further, it cannot be ruled out that the social, cultural and healthcare system differences between the USA and Europe, specifically The Netherlands, account for (part of) the difference in reported WOMAC levels. There are factors which could limit the relevance of the findings of this study. First, the follow-up time for radiological change is slightly different between the cohorts. Radiological change in CHECK is measured after 5 years and in OAI after 4 years. It is unlikely that the observed radiological change between the cohorts can be caused by a difference of 1 year follow-up. The difference in radiological change is probably caused by a higher frequency of participants with K&L 1 at baseline in the CHECK group. Knees with grade K&L 1 have a higher chance to progress in comparison with K&L 0.15 ,16

Second, the reading of the X-rays in the USA is performed by other readers than the readers who performed the reading in The Netherlands. Both groups used the same original description of K&L, however, it is known that cohort studies disagree between each other on the interpretation of grading according to K&L.17 Importantly, in both cohorts, it was known if the X-rays represented baseline or follow-up, and in both cohorts they were read in a paired fashion. It has been suggested that progression might be overestimated when reading paired radiographs with sequence known, but that effect was similar in both cohorts.18

Third, there were missing data in the longitudinal data of the WOMAC. The advantage of GEE is, that subjects with missing WOMAC data are not excluded from the analyses and imputation methods are not necessary to analyse longitudinal dataset with missing data.19 The baseline characteristics of the included CHECK participants were compared with CHECK participants who were excluded from this study because they had only one visit or a missing radiograph. There was no statistical significant difference between these variables (data not shown). Also, baseline characteristics of the included OAI participants were compared with the OAI participants who were excluded for the same reason in this study. At baseline, there was a difference, the excluded participants had more pain (mean 3.5 with SD 3.4) and more functional problems (mean 10.7 with SD 10.9). Besides, this group consisted of a different racial background (28% African–American) and a differential education level (4% primary school). The other baseline variables were not different. So, it could be that the level of pain and function evolution of the excluded OAI participants is higher than the level of the included OAI participants. However, baseline values of pain and function of the excluded OAI participants is still lower than the level of the CHECK participants.

Besides these possible methodological limitations, there are other variables that are related to evolution of pain and functional problems. People vary in their reporting of pain and their pain-related behaviour; however, in this study only severity of pain and physical function are analysed. It should be interesting to evaluate different aspects of pain itself, and, for example, to evaluate intermittent and constant pain.20 ,21 However, this is not possible in this report, because the ICOAP (measure of intermittent and constant OA pain) is not measured in CHECK and OAI at baseline. It would also be interesting to analyse the use of different coping strategies between OAI and CHECK, because lower levels of pain and function could be caused by better coping strategies. Unfortunately coping was not measured in OAI. Also of interest are the comorbidities, since they may have a negative effect on pain and physical health status.22 Unfortunately, comorbidities were measured in both cohorts using different questionnaires. This underscores the importance of international, common databases which consist of the same items measured at the same time points. Hopefully, in future, answers can be given as to which factors contribute to the observed difference in evolution in pain and function between CHECK and OAI.

We conclude that on average in patients with early symptomatic OA, pain and physical function remain fairly stable over 4–5 years, and especially in participants with a slow progression (radiological change of K&L grade 1). In participants with a rapid progression (radiological change of K&L grade ≥2) pain will increase and function will decline. For the physician, these findings, which are based on analyses on groups’ level, may provide a legitimation for symptomatic treatment.


CHECK-cohort study is funded by the Dutch Arthritis Association. It is led by a steering committee comprising 16 members with expertise in different fields of OA. It is chaired by Professor J W J Bijlsma and coordinated by J Wesseling, MSc. The following institutions are involved: Erasmus Medical Center Rotterdam; Kennemer Gasthuis Haarlem; Leiden University Medical Center; Maastricht University Medical Center; Martini Hospital Groningen/Allied Health Care Center for Rheum and Rehabilitation Groningen; Medical Spectrum Twente Enschede/Ziekenhuisgroep Twente Almelo; Reade, formerly Jan van Breemen Institute/VU Medical Center Amsterdam; St Maartens-kliniek Nijmegen; University Medical Center Utrecht and Wilhelmina Hospital Assen. The OAI is a public-private partnership comprised of five contracts (N01-AR-2-2258; N01-AR-2-2259; N01-AR-2-2260; N01-AR-2-2261; N01-AR-2-2262) funded by the National Institutes of Health, a branch of the Department of Health and Human Services, and conducted by the OAI Study Investigators. Private funding partners include Merck Research Laboratories; Novartis Pharmaceuticals Corporation, GlaxoSmithKline; and Pfizer, Inc. Private sector funding for the OAI is managed by the Foundation for the National Institutes of Health. This manuscript was prepared using an OAI public use dataset and does not necessarily reflect the opinions or views of the OAI investigators, the NIH, or the private funding partners.


View Abstract

Supplementary materials

  • Lay summary

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Tore K Kvien

  • Contributors JW, SMAB-Z, MK, RM, JWJB contributed to conception and design of this study. JW contributed to the analysis of data SMAB-Z, MK, RM and JWJB contributed to the interpretation of data. Article drafts were written by JW and critically revised by all authors. The final version of the article was approved by all authors. JW takes responsibility for the integrity of the work as a whole (

  • Funding This study was financially supported by the Dutch Arthritis Association.

  • Competing interests None.

  • Ethics approval Medical Ethics Committees of the Participating Centres.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.