ARTHRITIS & RHEUMATISMVol. 43, No. 2, February 2000, pp 386–389© 2000, American College of Rheumatology

SCORES FOR FUNCTIONAL DISABILITY IN PATIENTS WITHRHEUMATOID ARTHRITIS ARE CORRELATED AT HIGHER LEVELS

WITH PAIN SCORES THAN WITH RADIOGRAPHIC SCORES

TUULIKKI SOKKA, ANNALIISA KANKAINEN, and PEKKA HANNONEN

Objective. To analyze correlations of functionaldisability scores with other measures of clinical status,in particular, Larsen radiographic scores and painscores, in patients with rheumatoid arthritis (RA).

Methods. The functional capacity of 141 patientswith RA (102 women, 39 men; median age 57 years;median disease duration 11.8 years; 83% rheumatoidfactor positive) was assessed according to the StanfordHealth Assessment Questionnaire (HAQ). Other vari-ables studied included Larsen scores for radiographicdamage of the small joints of the hands, wrists, and feet,pain scores by visual analog scale (VAS), Disease Activ-ity Scores, general health scores by VAS, and BeckDepression Inventory (BDI) scores.

Results. The Spearman correlation coefficientcomparing HAQ and Larsen scores was 0.277 (P 50.001) and between HAQ and pain scores 0.652 (P <0.001). In regression analysis, pain scores explained41.4% of the variation in HAQ scores, normalizedLarsen scores explained 7.3%, and BDI scores explained5.5%; other variables were not significant in the model.

Conclusion. Functional capacity scores of pa-tients with RA are correlated at higher levels with painscores than with radiographic scores of small joints.

Functional disability and anatomical joint dam-age are among the primary long-term consequences ofrheumatoid arthritis (RA). Analyses of correlations be-tween radiographic scores for the hands and feet and

functional capacity scores have ranged from 0.21 to 0.68.Some of these correlations were statistically significant(1–6), while others were not (7–9). Therefore, furtheranalysis of a possible association between radiographicdamage and functional disability appeared to be ofinterest, particularly an examination of whether othervariables might provide greater explanation of the vari-ation in functional disability scores than the radio-graphic scores.

PATIENTS AND METHODS

A total of 141 patients with RA were evaluated atJyvaskyla Central Hospital from November 1997 to June 1998.Only patients with a disease duration of .3 years wereincluded in the present study, since many patients with earlydisease had low radiographic scores associated with extensiveuse of disease-modifying antirheumatic drugs (DMARDs) atour hospital (10). Patients who had recent surgery and thosewho had severe comorbidities, both of which may distortmeasurements of functional capacity, were excluded from thisstudy.

Jyvaskyla Central Hospital is the only rheumatologycenter in the region. The population of the region is 260,000.All new patients with RA are referred to this center fordiagnostic studies and initiation of therapies. Most patientswith severe RA visit the outpatient clinic regularly, approxi-mately every 3 months. A database was assembled whichincluded cross-sectional demographic measures, treatments,and outcomes of patients with RA. The Ethics Committee ofthe Jyvaskyla Central Hospital gave their approval for thestudy. All patients gave informed consent.

All patients met the American College of Rheumatol-ogy (formerly, the American Rheumatism Association) 1987revised criteria for RA (11). The median duration of diseasewas 11.8 years, with lower and upper quartiles of 7.5 and 21.1years; the median age of the patients was 57 years. Of the 141patients, 102 (72%) were female, and 115 (82%) were sero-positive for rheumatoid factor at least once during the disease.Symptoms prior to diagnosis had been present for ,16 monthsin 75% of the patients.

All patients had been treated aggressively, with contin-uous use of DMARDs in accordance with the “sawtooth”strategy (12). A total of 100 patients (71%) were treated with

Supported by grants from Central Finland Health CareDistrict, Kuopio University Hospital (EVO funding), and Muikkusa-atio, Finland.

Tuulikki Sokka, MD, Pekka Hannonen, MD, PhD: JyvaskylaCentral Hospital, Jyvaskyla, Finland; Annaliisa Kankainen, MSc, PhD:University of Jyvaskyla, Jyvaskyla, Finland.

Address reprint requests to Tuulikki Sokka, MD, Departmentof Medicine, Jyvaskyla Central Hospital, FIN-40620 Jyvaskyla, Fin-land.

Submitted for publication April 28, 1999; accepted in revisedform October 12, 1999.

386

prednisolone (median dosage 5 mg/day) for at least 1 month atsome time during disease. Twenty-six patients had 1 or severaltotal joint replacement procedures of the knees, hips, and/orshoulders (Table 1).

The patients’ functional capacity was assessed accord-ing to the Stanford Health Assessment Questionnaire (HAQ),which was scored from 0 to 3 points. Pain and general healthwere assessed according to 100-mm horizontal visual analogscales (VAS) (0 5 no pain or excellent general health, and100 5 most severe pain or poorest imaginable general health).Patients completed the short version of the Beck DepressionInventory (BDI) (range of scores from 0 to 39, with scores $8indicating depression). Current disease activity was assessedaccording to the 28-joint–based Disease Activity Score (DAS),ranging from 0 to 10 points (13,14). The DAS is calculated asfollows:

DAS 5

0.56 3 ÎTJC 1 0.28 3 ÎSJC 1 0.70 3 ln~ESR! 1 0.014 3 GH

where TJC 5 tender joint count, SJC 5 swollen joint count,ESR 5 erythrocyte sedimentation rate, and GH 5 generalhealth.

Radiographic damage of the small joints (first throughfifth metacarpophalangeal joints, second through fifth meta-tarsophalangeal joints, and the wrists) was assessed accordingto the Larsen score, ranging from 0 to 100 points. Arthritis wasjudged to be seropositive if rheumatoid factor was present anytime during the disease (Table 1).

The data were analyzed using SPSS/PC1 software(SPSS, Chicago, IL). Spearman’s correlations between vari-ables were computed. Regression analysis was performed with

HAQ scores as the dependent variable; independent variablesincluded age, sex, serologic status (positive/negative), durationof disease, Larsen score, pain scores by VAS, DAS, and BDIscores (dichotomized as ,8 and $8).

After a square root transformation of 2 variables(Larsen score and duration of disease), all continuous variableswere found to be normally distributed. Both intercept andslope dummy variables of the dichotomized variables weretested. Only variables that contributed significantly to themodel are reported. The Mann-Whitney test was used tocompare central values of unpaired samples. P values less than0.05 were considered statistically significant.

RESULTS

HAQ scores were correlated statistically signifi-cantly with VAS pain scores (r 5 0.652, P , 0.001), DAS(r 5 0.450, P , 0.001), and Larsen scores (r 5 0.277, P5 0.001). Larsen scores were correlated statisticallysignificantly with HAQ scores, DAS (r 5 0.332, P ,0.001), and disease duration (r 5 0.636, P , 0.001), but

Table 2. Explanatory variables for findings on the Health Assess-ment Questionnaire, by linear regression analysis (n 5 141)*

Variable

Explanation

B SE Rate (%) P

Pain by VAS 0.0176 0.002 41.4 0.000Larsen score (normalized) 0.0719 0.015 7.3 0.000BDI (intercept dummy

variable)0.368 0.117 5.5 0.002

Total 54.2

* B 5 coefficient; SE 5 standard error; VAS 5 visual analog scale;BDI 5 Beck Depression Inventory.

Figure 1. Scatter plot of the Larsen score versus the Health Assess-ment Questionnaire (HAQ) score for each patient studied (n 5 141).The curve is a lowess (locally weighted scatterplot smoothing) regres-sion line.

Table 1. Demographic and clinical characteristics and treatment ofthe study patients*

No. of patients 141Age, median (IQR) years 57 (50, 65)No. (%) female 102 (72)No. (%) seropositive 115 (82)Duration of symptoms before diagnosis, median

(IQR) months (n 5 110)7 (4, 16)

Disease duration after diagnosis, median (IQR)years

11.8 (7.5, 21.1)

HAQ, median (IQR) score (0–3 scale) 0.75 (0.25, 1.25)Larsen score, median (IQR) (0–100 scale) 25 (7, 51)Pain by VAS, median (IQR) (0–100 mm) 40 (21, 55)DAS, median (IQR) (1–10 scale) 3.8 (3.3, 4.7)BDI score $8, no. (%) of patients 19 (14)No. of patients treated with DMARDs 141No. (%) of patients treated with combinations

of DMARDs105 (75)

No. of DMARD periods, median (IQR) 6 (3, 8)Duration of DMARD/combination treatment,

median (IQR) years (n 5 122)8.3 (5.3, 13.1)

No. (%) of patients ever treated withprednisolone

100 (71)

No. (%) of patients with replacement of largerjoints

26 (18)

* IQR 5 lower and upper quartiles; HAQ 5 Health AssessmentQuestionnaire; VAS 5 visual analog scale; DAS 5 Disease ActivityScore; BDI 5 Beck Depression Inventory; DMARD 5 disease-modifying antirheumatic drug.

FUNCTIONAL DISABILITY AND PAIN IN RA 387

not with pain (r 5 0.008, P 5 0.929) or age (r 5 0.010,P 5 0.906). Although the correlation of the HAQ andLarsen scores was statistically significant, the r value of0.277 indicates considerable variation between thescores (see Figure 1).

In the regression analysis, the pain score was theprimary explanatory variable for the HAQ scores, ac-counting for 41.4% of the variation. Larsen scorescontributed only 7.3% to explaining the variation of theHAQ scores, and the BDI contributed 5.5% (Table 2).Other tested variables did not contribute statisticallysignificantly to the model.

DISCUSSION

The primary observation of the present study isthat functional disability is only weakly associated withscores for radiographic damage of the small joints inthese patients with longstanding RA. Although correla-tions between these variables in this and other studieshave been found to be statistically significant (1–6),explanatory power calculated from the pairwise correla-tions remains low. For example, if the pairwise correla-tion coefficient between the HAQ and Larsen scores is0.30, then the Larsen score explains only 9% of thevariation in HAQ scores, and a correlation coefficient of0.50 produces an explanatory power of only 25%. Fur-thermore, weak correlation coefficients become statisti-cally significant with large study samples.

The HAQ has been shown to be valid and reliablefor the assessment of functional status in patients withRA, and it is the best variable for predicting workdisability and mortality (15). In the regression model,;50% of the variation of the HAQ scores was explainedby pain, radiographic scores, and BDI scores, whilecurrent disease activity, age, and duration of disease,serologic features, and sex did not contribute statisticallysignificantly to the model. It is likely that the lack ofsignificance of certain variables in the model may be aresult of multicollinearity. DAS scores were correlatedsignificantly with pain scores, and Larsen scores werecorrelated significantly with age. Nonetheless, one-halfof the variation of the HAQ scores remains unexplainedby available measures. It is possible that further expla-nation might be found in motivational and psychologicalfactors that can be captured by the HAQ, such ashelplessness and lack of exercise, but which are notcollected in the daily course of typical rheumatologiccare.

We used the short version of the BDI to assessdepression. The BDI, as other self-reported depression

scales, includes items concerning fatigue, loss of appe-tite, and inability to work, which are recognized asreflecting findings in RA as well as in depression.However, scores $8 (of 39), which we used as a cutpoint, are regarded as mirroring at least moderatedepression. In the linear model, the independent explan-atory power of the BDI was 5.5%.

It has been suggested that anatomical damage ofjoints leads to functional declines in RA (16). However,in our study of patients with longstanding RA, painscores, rather than radiographic scores, were the mostpowerful explanatory variables for variation in the HAQscores. Similar findings have been reported in a study ofpatients with early RA (6), suggesting that a weakrelationship between radiographic scores and HAQscores may be seen throughout the course of RA.Furthermore, it appears that pain reflects mostly clinicaldisease activity but is weakly, if at all, correlated withradiographic scores in early and late RA.

Only 25% of the patients in our study had aLarsen score that was at least 50% of the maximum, afinding that may reflect our policy of aggressive treat-ment with DMARDs; 1,276 (80%) of 1,588 patient-yearsincluded DMARD therapy. Furthermore, 26 patients(18%) had undergone knee, hip, or shoulder jointreplacement surgeries, and this might have improvedtheir HAQ scores. It is possible that dissociation ofHAQ scores and pain scores versus HAQ scores andradiographic scores may reflect a relative preservation ofjoints with DMARD therapy, and correlations might behigher in patients with more extensive radiographicdamage.

In conclusion, our results confirm previous stud-ies that indicate that scores for functional disability areonly weakly associated with radiographic scores and areexplained more by pain scores. These findings suggestthat functional disability and radiographic damage mayreflect different aspects of the long-term progression ofRA, as has been suggested by Kirwan (16). Furtheranalyses of the dissociation of functional capacity andpain scores versus radiographic scores may lead togreater understanding of pathogenic mechanisms in thelong-term consequences of RA.

ACKNOWLEDGMENTS

The authors gratefully acknowledge Drs. Mikko Ha-kola, Ilppo Palvimaki, and Jari Ahonen for their work incollecting the clinical data.

388 SOKKA ET AL

REFERENCES

1. Guillemin F, Briancon S, Pourel J. Functional disability in rheu-matoid arthritis: two different models in early and establisheddisease. J Rheumatol 1992;19:366–9.

2. Pincus T, Callahan L, Brooks RH, Fuchs HA, Olsen NJ, Kaye JJ.Self-report questionnaire scores in rheumatoid arthritis comparedwith traditional physical, radiographic, and laboratory measures.Ann Intern Med 1989;110:259–66.

3. Hakala M, Nieminen P, Manelius J. Joint impairment is stronglycorrelated with disability measured by self-report questionnaires:functional status assessment of individuals with rheumatoid arthri-tis in a population based series. J Rheumatol 1993;21:64–9.

4. Kaarela K, Sarna S. Correlations between clinical facets of out-come in rheumatoid arthritis. Clin Exp Rheumatol 1993;11:643–4.

5. Mottonen T, Paimela L, Leirisalo-Repo M, Kautiainen H, IlonenJ, Hannonen P. Only high disease activity and positive rheumatoidfactor indicate poor prognosis in patients with early rheumatoidarthritis treated with ‘sawtooth’ strategy. Ann Rheum Dis 1998;57:533–9.

6. Van Leeuwen MA, van der Heijde DMFM, van Rijswijk MH,Houtman PM, van Riel PLCM, van de Putte LBA, et al. Interre-lationship of outcome measures and process variables in earlyrheumatoid arthritis: a comparison of radiologic damage, physicaldisability, joint counts, and acute phase reactants. J Rheumatol1994;21:425–9.

7. Wolfe F, Sharp JT. Radiographic outcome of recent-onset rheu-matoid arthritis: a 19-year study of radiographic progression.Arthritis Rheum 1998;41:1571–82.

8. Regan-Smith MG, O’Connor GT, Kwoh CK, Brown LA, Olm-stead EM, Burnett JB. Lack of correlation between the Steinbro-

cker staging of hand radiographs and the functional health statusof individuals with rheumatoid arthritis. Arthritis Rheum 1989;32:128–33.

9. Fex E, Jonsson K, Johnson U, Eberhardt K. Development ofradiographic damage during the first 5-6 years of rheumatoidarthritis: a prospective follow-up study of a Swedish cohort. Br JRheumatol 1996;35:1106–15.

10. Sokka T, Kautiainen H, Mottonen T, Hannonen P. Work disabilityin rheumatoid arthritis 10 years after the diagnosis. J Rheumatol1999;26:1681–5.

11. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

12. Fries JF. Reevaluating the therapeutic approach to rheumatoidarthritis: the ‘sawtooth’ strategy. J Rheumatol 1990;17:12–5.

13. Van der Heijde DMFM, van ’t Hof MA, van Riel PLCM, vanLeeuwen MA, van Rijswijk MH, van de Putte LBA. A comparisonof validity of single variables and composed indices for measuringdisease activity in rheumatoid arthritis. Ann Rheum Dis 1992;51:177–81.

14. Prevoo MLL, van ’t Hof MA, Kuper HH, van Leeuwen MA, vande Putte LBA, van Riel PLCM. Modified disease activity scoresthat include twenty-eight–joint counts: development and valida-tion in a prospective longitudinal study of patients with rheuma-toid arthritis. Arthritis Rheum 1995;38:44–8.

15. Pincus T, Callahan L. Quantitative measures to assess, monitorand predict morbidity and mortality in rheumatoid arthritis.Baillieres Clin Rheumatol 1992;6:161–91.

16. Kirwan JR. Conceptual issues in scoring radiographic progressionin rheumatoid arthritis. J Rheumatol 1999;26:720–5.

FUNCTIONAL DISABILITY AND PAIN IN RA 389