Article Text

Extended report
‘Insights in the relationship of joint space narrowing versus erosive joint damage and physical functioning of patients with RA’
  1. R Koevoets1,
  2. L Dirven1,
  3. N B Klarenbeek1,
  4. M V van Krugten2,
  5. H K Ronday3,
  6. D M F M van der Heijde1,
  7. T W J Huizinga1,
  8. P S J M Kerstens4,
  9. W F Lems4,5,
  10. C F Allaart1
  1. 1Department of Rheumatology, Leiden University Medical Centre, Leiden, The Netherlands
  2. 2Department of Rheumatology, Admiraal de Ruyter ziekenhuis, Vlissingen, The Netherlands
  3. 3Department of Rheumatology, HagaZiekenhuis, Den Haag, The Netherlands
  4. 4Department of Rheumatology, Jan van Breemen Research Institute│Reade, Amsterdam, The Netherlands
  5. 5Department of Rheumatology, VU Medical Centre, Amsterdam, The Netherlands
  1. Correspondence to Rosanne Koevoets, Stafcentrum reumatologie, C1-45, Postbus 9600, Leiden 2300 RC, The Netherlands; r.koevoets{at}


Objective To evaluate the contribution of joint space narrowing (JSN) and erosions in general and in four different joint groups in relation to physical disability in rheumatoid arthritis (RA).

Methods 5-year follow-up data from the Behandel Strategieën (BeSt) trial were used, where 508 patients with recent onset RA were treated aiming at a disease activity score ≤2.4. Joint damage was assessed annually and scored according to the Sharp-van der Heijde method. Physical disability was measured 3-monthly with the Health Assessment Questionnaire (HAQ). Generalised Estimating Equations analyses were performed to assess the relationship between the HAQ and JSN scores and erosions scores, separately and in joint groups.

Results Overall, damage scores were low, and neither total JSN nor erosions showed a significant effect on HAQ (β=0.001 95% CI −0.003 to 0.004 and β=0.002 95% CI −0.001 to 0.006, respectively). Of the total damage scores per joint group, damage in the wrist shows a trend for association with physical disability displaying the largest effect size (β=0.005 95% CI 0.000 to 0.011). Also in the analysis with erosions per joint group, the wrist was most strongly related with physical functioning (β=0.016 95% CI 0.003 to 0.029); in the analysis with JSN per joint group no joint group was significantly related to the HAQ. Analysis of all erosion and narrowing scores per joint group in one model reveals only erosions in the wrist to be independently associated with impaired physical functioning (β=0.017 95% CI 0.003 to 0.030).

Conclusions Joint damage in the wrist, erosions more than JSN, is associated with impaired physical functioning even in patients with early RA with limited overall damage after 5 years tightly controlled treatment.

View Full Text

Statistics from


Joint damage is an early and potentially progressive feature of rheumatoid arthritis (RA) and is related to functional disability together with other factors, such as disease activity and co-morbidity.1–5 Previous research has shown that in early disease, physical functioning is mostly determined by the disease activity score (DAS), whereas in late disease, it is mostly determined by the extent of joint damage.6

RA-related joint damage on radiographs involves damage to the bone (bone mineral density loss and erosions), damage to cartilage (scored as joint space narrowing (JSN)) and damage to ligaments causing malalignment, which is sometimes also scored as part of the JSN scoring, depending on the scoring method used.7–9 Recently, it has been suggested in a cross-sectional study that JSN rather than erosiveness is associated with physical disability.10 It is not clear whether the relation between JSN or erosions and physical functioning is influenced by the location of damage in particular joints or joint groups. Further insight in this could lead to specific site evaluations to assess treatment efficacy, and possibly to specific therapy targets. Therefore, we evaluated the contribution of JSN and erosions in general, and in four different joint groups, in relation to physical disability over time, in a cohort of patients with recent onset of RA, dynamically treated during 5 years in a tight control setting aimed at low disease activity.



Data from the Behandel Strategieën (BeSt) trial were used, where 508 patients with recent onset of active RA were dynamically treated according to a protocol directed by 3-monthly assessments of disease activity, aiming at a DAS ≤2.4. Patients were randomised into four different treatment strategies: (1) sequential monotherapy (n=126); (2) step-up combination therapy (n=121); (3) initial combination therapy with prednisone (n=133) and (4) initial combination therapy with infliximab (n=128). Clinical assessment of disease activity was performed 3-monthly, and included a 68/66 graded joint count for tenderness and swelling, erythrocyte sedimentation rate measurements and patient's assessment of global disease activity. This study was approved by the ethical committees of participating centres, and all patients provided informed consent. More details about the BeSt study have been described elsewhere.11 The current analysis was performed on 5-year follow-up data.

Radiological assessments

Annual radiographs of hands, wrist and feet from baseline until year 5 were scored in random order by two observers blinded for patient identity and treatment allocation. The modification of the Sharp method by van der Heijde (SHS)8 was used for the scoring of JSN and erosiveness on the radiographs. JSN is scored on a scale of 0–4 with a maximum total of 120 in the hands and 48 in the feet. Erosions are scored on a scale of 0 (no erosion)–3 (large erosion), and are scored cumulatively with a maximum of five points per joint in the hands, and with a maximum of 10 points in the feet. The maximum total erosion score in the hands is 160, and in the feet 120 points. Average scores of the two readers were used for the analysis.

Outcome assessment

Physical functioning was assessed using the Dutch version of the Health Assessment Questionnaire (HAQ).12 The HAQ consists of 24 questions in eight different categories which are answered on a 0–3 severity scale. To calculate the total HAQ score, which ranges between 0 (no disability) and 3 (severe disability), all highest scores per category are summed up and divided by eight.

Statistical methods

Statistical analyses were performed with the software program SPSS V.17.0 (SPSS, Chicago, Illinois, USA). Generalised Estimating Equations (GEE) regression models were used to investigate the relationship between joint damage (in general and per joint group) and physical functioning during 5 years follow-up, while correcting for within-patient correlation. The total HAQ score at the yearly visits was used as outcome for all analyses. GEE modelling was chosen as it is relatively robust against violations of normality, which is frequently the case with SHS, erosion and JSN scores.

Separate models were specified to be able to assess the effect size of the independent variable in relation to the dependent variable. Combined models were specified to evaluate the independent contribution of one dependent variable in the presence of other dependent variables. Adjustments were made for possible confounding variables time, DAS, gender, treatment group and body mass index (BMI) that were associated with physical functioning in the univariate analysis. p Values of ≤0.05 were considered statistically significant. For all analyses, the unstructured covariance matrix was used, which does not assume a specific covariance structure, and estimates every covariance individually for consecutive measurements. Since high disease activity was a requirement for inclusion, baseline measurements were excluded from the analysis to avoid a large contribution of high disease activity on disability.

The analyses were performed by entering variables in the model in the following three steps: (1) total score of JSN and total erosions score (in all joints); (2) total damage score (erosions and JSN together) per joint group and (3) JSN and erosion scores separately per joint group. For the second and third steps in the analysis, joint damage was evaluated in four joint groups: MCPs, PIPs, wrists and feet. For steps 1 and 3, first erosions and JSN were analysed separately and then in one model together. Finally, separate variables for JSN and erosions in all different joint groups were entered in one model to evaluate the independent contribution of all factors. As the maximum attributable points per joint group or per feature can differ, all analyses were repeated using the percentage of the maximum total damage.


At baseline, patients reported severe disability with a mean (SD) HAQ score of 1.4 (0.7), which decreased to 0.6 (0.6) at t=1 year and remained stable over the years (HAQ 0.6 (0.6) at t=5 years). Joint damage at baseline, although present in the majority of this population with severe RA, was limited, with a median SHS score (IQR) of 3.0 (0.5–9.5), and 62% of patients showing joint damage <5 points SHS. At baseline, 37% of patients had no JSN and 28% had no erosions. After 5 years, these percentages had decreased to 30% and 18%, respectively. The median SHS score (IQR) at t=5 years was 7.0 (1.5–20.3).

JSN was mostly seen in the wrists with a mean (SD) JSN score of 2.9 (6.7), and to a lesser extent in the feet with a mean (SD) JSN score of 2.3 (5.2) at year 5. Erosions were most frequently seen in the feet with a mean (SD) score 4.4 (7.4), and thereafter in the wrists with a score of 1.2 (3.6). Mean percentage JSN ranged from 2.0% of the maximum score (in the PIPs) to 6.1% (in the wrists), and mean percentage erosions ranged from 1.8% (in the PIPs) to 3.7% (in the feet). More details are described in the supplementary online table S1.

The relationship of JSN and of erosions with functional disability

In the univariate analysis (table 1), the effect on disability of erosion score and JSN score is similar. JSN and physical functioning show a statistically significant relationship with a beta estimate (β) (increase in HAQ per point increase in JSN score) of 0.004 (95% CI 0.001 to 0.008). Erosions are related to physical functioning in the same order of magnitude, however, this relationship is not statistically significant (β=0.003 95% CI −0.001 to 0.006). The main clinically relevant predictor for disability, showing the largest effect size, is the DAS (β=0.250; 95% CI 0.220 to 0.280). Other univariate predictors are (female) gender, time, BMI, baseline DAS and treatment group. When applying a multivariable model with correction for disease activity (concurrent and baseline), BMI, time, gender and treatment group, effect sizes for the erosions and JSN scores are smaller (β=0.001 and β=0.002, respectively), and the relationship is no longer statistically significant, neither separately nor combined in a model (see supplementary online table S2).

Table 1

Univariate predictors of physical functioning as measured by the Health Assessment Questionnaire

Joint damage in joint groups in relation to disability

The relationship between the total damage scores per group and the HAQ is shown in table 2. Total damage in the wrist shows a trend for association with the HAQ and is demonstrating the largest effect size in both the separate analysis and in the analysis combining with the other joint groups (β=0.005 95% CI 0.000 to 0.010 and β=0.005 95% CI 0.000 to 0.011, respectively).

Table 2

The total joint damage score per group in relation to physical functioning measured by the Health Assessment Questionnaire

In analyses including erosions and narrowing separately per joint group, only erosions in the wrist (β=0.014 95% CI 0.003 to 0.026) show a statistically significant relation with physical functioning (table 3).

Table 3

Narrowing and erosion scores per joint group in relation to physical functioning measured by the Health Assessment Questionnaire in eight different models

When applying a model including JSN in all groups there is no statistically significant relationship of any joint group with physical functioning, but in the model for erosions in all joint groups, erosions in the wrist are found as an independent explanatory variable with a large effect size (β=0.016 95% CI 0.003 to 0.029) (see supplementary online table S3).

Finally, in the extended model including all joint groups, both, as JSN score and as erosion score while correcting for BMI, DAS (baseline and concurrent), time, gender and treatment group, we find erosions in the wrist as the only independent predictor in the presence of all other associated factors (β=0.017 95% CI 0.003 to 0.030) (table 4).

Table 4

Independent contribution of narrowing and erosion scores within each joint group in relation to physical functioning measured by the Health Assessment Questionnaire.

We repeated the analyses using the percentage of the maximum score instead of absolute scores (as the maximum attributable point can differ per feature or group), and found comparable results (data not shown).


In a cohort of patients with recent onset of RA, over 5 years of tight controlled treatment, both JSN and erosions showed only a small non-significant effect on physical functioning defined by the HAQ. Of all sites evaluated with the SHS score, the wrist was the most important determinant, and erosions in the wrist were the only independent predictor of functional disability.

Previous research in patients from various early or advanced RA trials has shown that JSN, rather than erosiveness, was associated with physical functioning.10 It has been suggested that this is due to different pathophysiological mechanisms of both types of damage.13–15 In our cohort, we found the effect size on outcome HAQ of JSN or erosion scores was roughly the same and very small. This is probably because in this cohort of patients with early RA, treated with DAS-steered tight control strategies, there was very limited radiological damage progression in general. The high agreement between the analyses using the absolute scores and those using the percentages of the maximum scores also indicates that severe joint damage (at the highest end of the scale) was rare. In these patients, disability was largely determined by disease activity, even when the disease activity at baseline was discarded.

When analysing damage per joint or joint group, we found that damage in the wrist was the main determinant of disability. As JSN frequently occurs in the wrist, this finding would seem to corroborate the previously reported dominance of JSN in relation to functional disability.16 However, our analyses showed that erosive damage, not narrowing, in the wrist was an independent predictor for functional disability. A possible explanation for the effect of wrist joint damage on functional ability may be that activities requiring wrist movement constitute a large proportion of the daily activities asked after in the questions of the HAQ. Only questions on walking and climbing stairs, and possibly (but not likely) the questions on rising from an armless chair and getting out of bed, address activities where one would not need to use one's wrists. This predominance of wrist-related activities in the HAQ of course reflects the daily activities of a species that evolved to walk on his hind legs to free up the use of hands and wrists.

This study has several limitations. Within the SHS method, not only JSN, as cartilage damage, but also (sub)luxation from soft tissue damage is scored.8 Based on our scores, we do not know the separate influence of soft tissue damage on physical functioning. This is of course also a limitation in other studies that use the SHS score. However, for our conclusions on wrist damage, this appears less important, as in the wrist, the JSN is almost exclusively determined by loss of joint space and not (sub)luxation. Also in the PIPs, (sub)luxation is rarely scored, but subluxation does influence the JSN score in MTPs and MCPs.

It has been suggested that plain radiographs are less sensitive for detecting erosive damage than alternative imaging techniques such as MRI, CT or ultrasound. We have not used these techniques, but if this has led to an underdetection of erosions in the wrist, this would mean that we have underestimated and not overestimated the effect of wrist erosions on functional ability.

JSN is also frequently seen on hand radiographs of patients with osteoarthritis, although it occurs mostly in the DIPs (not in the SHS score) and PIPs, and less frequently in the wrists.17 ,18 We have not scored osteoarthritic damage separately, but as osteoarthritis frequently affects the middle aged and elderly, it is bound to be present in our population with a mean age of 54 at the onset of the study. Still, since the SHS method does not make a distinction between JSN due to osteoarthritis or RA, the combined occurrence may have enlarged the effect of JSN and thus masked the effect of rheumatoid erosions. Again, this would mean we could have underestimated the effect of erosive damage on functional disability, and it could explain why in the total scores of all joints erosions are not significantly associated with functional ability. On the other hand, JSN in osteoarthritis is mostly seen in the PIPs (within the SHS score). and not that frequently in the joints assessed in the wrist,19 which may thus have influenced the relationship with the total JSN, but does not play a major role in the dominant effect of the wrist on HAQ.

We did not take large joint damage into account, which is an important determinant of physical disability.20 However, since large and small joint damage are closely related to each other,20 ,21 the effect of that exclusion may be limited. It is also likely that in this cohort with relatively little damage progression in the small joints, there is even less damage in the large joints.

The correlation between narrowing and erosions scores (per joint) also influences their independent relationship with physical functioning. Separate erosion scores and narrowing scores show a substantial correlation within this dataset (ρ=0.70), but the joint groups per feature correlate significantly as well (range ρ=0.21 to ρ=0.57). Especially, narrowing in the feet was strongly related with erosions in the feet, and narrowing in the wrist with erosions in the wrist. This may explain why in combined models including both features, the effect of one factor seems to dominate over the other, although both could provide relevant information.

Strong points include that we have not only assessed the damage features separately, but also included several joints groups, to see whether these may explain the relationships that were found. Also, we were able to include data of patients followed longitudinally over a period of 5 years. We believe that our patient population with recent RA, treated in a tight control setting with limited joint damage, represents the RA patients of this and future decades. A final strong point, since joint damage does not follow a Gaussian distribution, is that we used a GEE model, which is relatively robust against violations of normality.

Because the wrist is an important determinant for disability, we may need to focus on the prevention of structural damage especially there. There is evidence that local synovitis is associated with damage progression in the same joint.22 It has also been suggested that intra-articular corticosteroid injections suppress joint inflammation and damage progression.23 Further research is needed to determine if local treatment in combination with effective systemic treatment has additional benefits to halt local damage progression and prevent disability.

In conclusion, in a large cohort with recent onset of RA, treated with a tight control treatment strategy over 5 years, and limited joint damage progression, the relation of such joint damage progression with functional ability as measured with the HAQ concentrates around erosions in the wrist. This may have consequences for evaluation of treatment success as well as for localised treatment strategies.


We would like to thank all patients as well as the following rheumatologists (other than the authors) who participated in the Foundation for Applied Rheumatology Research (all locations are in The Netherlands): WM de Beus (Medical Center Haaglanden, Leidschendam); C Bijkerk (Reinier de Graaf Gasthuis, Delft); MHW de Bois (Medical Center Haaglanden, The Hague); H Boom (Spaarne Hospital, The Hague); M de Buck (Medical Center Haaglanden, Leidschendam); G Collée (Medical Center Haaglanden, The Hague); JAPM Ewals (Haga Hospital, The Hague); AH Gerards (Vlietland Hospital, Schiedam); RJ Goekoop (Haga Hospital, The Hague); YPM Goekoop-Ruiterman (Haga Hospital, The Hague); BAM Grillet (Zorgsaam, Terneuzen); JHLM van Groenendael (Franciscus Hospital, Roosendaal); KH Han (Medical Center Rijnmond-Zuid, Rotterdam); L Lard (Medical Center Haaglanden, Leidschendam); H van der Leeden (retired); WF Lems (VUMC, Amsterdam); MF van Lieshout-Zuidema (Spaarne Hospital, Hoofddorp); PAHM van der Lubbe (Vlietland Hospital, Schiedam); C Mallée (Kennemer Gasthuis, Haarlem); ETH Molenaar (Groene Hart Hospital, Gouda); M van Oosterhout (Groene Hart Hospital, Gouda); AJ Peeters, MD (Reinier de Graaf Gasthuis, Delft); N Riyazi (Haga Hospital, The Hague); AA Schouffoer (Groene Hart Hospital, Gouda); PEH Seys (retired); PBJ de Sonnaville, MD (Oosterschelde Hospital, Goes); I Speyer, MD (Bronovo Hospital, The Hague); KSS Steen, MD (Kennemer Gasthuis, Haarlem); GM Steup-Beekman (Bronovo Hospital, The Hague); JPh Terwiel, MD (retired); AE Voskuyl, MD (VU Medical Center, Amsterdam); ML Westedt, MD (Bronovo Hospital, The Hague); S ten Wolde, MD (Kennemer Gasthuis, Haarlem); D van Zeben, MD (Sint Franciscus Gasthuis, Rotterdam). We would also like to thank all other rheumatologists and trainee rheumatologists who enrolled patients in this study, and all research nurses for their contributions.


View Abstract

Supplementary materials

  • Web Only Data

    Files in this Data Supplement:


  • Contributors Design of the study: MVK, HKR, DMFMvdH, TWJH, PSJMK, WFL and CFA. Data acquisition: LD, NBK, MVK, HKR, TWJH, PSJMK, WFL and CFA. Data analysis and interpretation: RK, LD, NBK, DMFMvdH, TWJH and CFA. Writing and revising the manuscript: RK, LD, NBK, MVK, HKR, DMFMvdH, TWJH, PSJMK, WFL and CFA.

  • Funding This study was funded by a grant from the Dutch College of Health Insurances (College Voor Zorgverzekeringen) with additional funding provided by Schering-Plough BV and Janssen Biologics BV The authors, not the sponsors, were responsible for the study design, trial management, data collection, data analysis and preparation of the manuscript.

  • Competing interests DMFMv/dH, TWJH, WFL and CFA received speakers’ fees from various pharmaceutical companies (less than US$10 000 per year).

  • Patient consent Obtained.

  • Ethics approval The study was approved by the medical ethics committees of the participating hospitals, and all patients gave written informed consent.

  • 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.