Does the age of onset of rheumatoid arthritis influence phenotype?: a prospective study of outcome and prognostic factors

Transcription

1 Rheumatology 1999;38: Does the age of onset of rheumatoid arthritis influence phenotype?: a prospective study of outcome and prognostic factors C. T. Pease, B. B. Bhakta, J. Devlin and P. Emery Rheumatology and Rehabilitation Research Unit, University of Leeds, 36 Clarendon Road, Leeds LS2 9NZ, UK Abstract Objective. To identify factors affecting prognosis in patients with late-onset rheumatoid arthritis (RA). Methods. A total of 400 patients with RA fulfilling the American College of Rheumatology criteria for diagnosis were prospectively recruited from two hospital rheumatology centres. Of these patients, 214 had disease onset above age 65 yr (LORA) and 186 below age 65 yr ( YORA). Follow-up clinical, functional, laboratory and radiological assessments were compared. The Ritchie articular index (RAI) and joint erosions were used as markers of disease activity and damage, respectively. Disability was assessed using the Stanford Health Assessment Questionnaire (HAQ). Results. At median follow-up of 3.6 yr, the frequency of joint erosions was similar ( YORA, 51.6%; LORA, 54.2%). The remission rate was greater in the LORA group (YORA, 20.4%; LORA, 45.8%, P < 0.01). Factors associated with the development of erosions were: IgM rheumatoid factor (RF) seropositivity [odds ratio (OR) = 4.24, 95% confidence interval (CI) 2.56, 6.94], HLA DR4 (OR = 2.07, 95% CI 1.28, 3.35) and elevated inflammatory markers (OR = 1.81, 95% CI 1.04, 3.14). Continuous steroid use >3 months for the LORA group was associated with increased erosions (OR = 4.09, 95% CI 1.81, 9.27). LORA patients (OR = 2.99, 95% CI 1.77, 5.02) were more likely to go into remission and IgM RF-seropositive patients less likely to go into clinical remission (OR = 0.47, 95% CI 0.28, 0.77). Female patients with a high HAQ score at presentation experienced a poor functional outcome (female OR = 3.01, 95% CI 1.59, 5.68; high HAQ OR = 3.02, 95% CI 1.98, 4.62). Conclusion. LORA can be as damaging as classical RA and joint erosions are often observed at presentation. Being RF seropositive, DR4 positive, and having elevated inflammatory markers at onset, were associated with poor radiological outcome irrespective of age of onset. Being female and having marked disability at presentation were associated with poor functional outcome in both groups. These findings suggest that treatment approaches used in classical YORA should be instituted with equal vigour in patients with LORA. KEY WORDS: Rheumatoid arthritis, Age at onset, Prospective study, Remission, Erosions. The proportion of people aged 65 yr or older in the UK in 1992 was 15.8% of the total population with the projected proportion increasing to 19.4% by the year 2026 [1]. In Sweden, the overall prevalence of rheumatoid arthritis (RA) in the elderly has been reported to be 2.7% [2]. RA has been reported to decline with age (perhaps due to reduced survival of patients with RA), a prevalence of 3 4% being noted in the yr age group compared with a prevalence of % in those aged 75 yr or older [2, 3]. In the new millennium, late- Submitted 15 October 1997; revised version accepted 16 October Correspondence to: C. T. Pease, Department of Rheumatology, The General Infirmary, Great George Street, Leeds, UK. onset RA (LORA) is likely to be an increasing health problem. Evidence differs regarding the presentation, severity and prognosis of RA in relation to age of disease onset. Certain features such as explosive onset, polymyalgic presentation, weight loss and early shoulder joint involvement have been reported to be more frequent in RA presenting in the elderly [4 21]. Outcome in LORA has been reported to be favourable [6, 8, 10, 14, 16], similar [5, 7, 13] or worse [4, 9, 11, 12, 17, 19] when compared to younger-onset RA (YORA). Only a few of these studies directly compared patients with LORA to those with YORA [14, 16 20, 22]. Only two of these studies [17, 19] included prospective longitudinal followup. Both concluded that patients with LORA had a poorer outcome in terms of radiological damage, British Society for Rheumatology

2 Age of onset of RA and outcome 229 continued disease activity [19] and decline in functional status [19]. Seronegativity for rheumatoid factor ( RF) both in patients with LORA and YORA has been associated with a milder course [20, 22]. Prospective study of the outcome of patients with LORA is important so that early treatment strategies with disease-modifying drugs can be justified. The aims of this prospective study were (1) to compare the clinical, radiological and functional outcome of patients presenting with RA above and below the age of 65 yr and (2) to identify predictors of poor clinical and radiological outcome within the whole cohort. Materials and methods The ARA criteria for disease remission were difficult to apply to the LORA group because of co-existing diseases causing fatigue and joint pain (e.g. osteoarthritis). Therefore, disease remission was defined as: clinically inactive disease (RAI = 0, and absence of soft-tissue swelling in joints and tendon sheaths) documented on two clinic visits at least 3 months apart. This definition included therapeutic remission as some of the patients were still on disease-modifying anti- rheumatic drugs (DMARDs). The baseline and follow- up clinical, laboratory, radiological and functional assessments were documented using a standardized pro- forma and data collated on a dedicated computerized database. Statistical analysis Since 1989, patients who presented with recent-onset RA and who fulfilled the revised 1987 American College Baseline assessments between patients in the YORA and of Rheumatology (ACR) diagnostic criteria for RA LORA groups were compared to identify differences in were recruited from two rheumatology services (Leeds clinical and laboratory features at presentation. Clinical, and Birmingham) for this study. LORA was defined as radiological and functional data were compared between those developing RA after their 65th birthday, YORA the two groups at the final evaluation ( last out-patient was defined as those developing RA between the ages visit) to determine any differences between outcome. 18 and 65 yr. In our study, we chose age 65 as the Some of the data had a skewed frequency distribution dividing line between YORA and LORA to be compatstatistical analysis were used ( x2 analysis for dichotom- and therefore median values, ranges and non-parametric ible with retirement age in the UK. In addition, this cut-off allows separation of patients developing LORA ous variables and Mann Whitney U-test for continuous from the YORA patients, in whom peak age of onset variables). Multivariate analysis using the logistic regres- (females) is reported to be between 55 and 64 yr [23] sion model was undertaken to determine the relative based on hospital referrals, a figure which is similar to contribution of dichotomous variables (gender, age of that reported by Garrod [24]. onset, HLA DR4, seropositivity for IgM RF, myalgia, The following clinical information was recorded from anaemia, elevated acute-phase reactants, HAQ at prehistory and examination: gender, age at onset of arthdevelopment of joint erosions. sentation, steroid use) in predicting remission and the ritis, rapidity of onset, pattern and symmetry of joint involvement, possible precipitating events, duration of early morning stiffness, weight loss, myalgia (as found Results in polymyalgia rheumatica), sicca syndrome, Raynaud s A total of 422 consecutive patients who were seen by disease, peripheral limb oedema and family history of the rheumatology service within 2 yr of onset of RA RA. The Ritchie articular index ( RAI) was used to were included in this study. The date of onset of RA assess the severity of joint involvement. A visual ana- was defined as the date when the patient could first logue scale (VAS) was used to assess pain. Grip strength recall developing joint pain and swelling. Twenty-two (mmhg) of both hands was measured using a Martin- patients had insufficient follow-up data to be included type vigorimeter [25]. This is a modified sphygmomano- in the analysis (16 patients had follow-up data <1 yr, meter linked to a manometer. This measurement was of whom three had died, four had moved out of the used as a proxy for one aspect of the global impairment locality and two had refused follow-up). The remaining of hand movement due to synovitis of the hands and 400 patients were followed up for a minimum of 1 yr wrist. Functional status was measured using the [median follow-up 3.6 yr (range 1 7)]. A total of 214 Stanford Health Assessment Questionnaire in 274 patients with LORA and 186 patients with YORA were patients (YORA, n = 186; LORA, n = 88). The follow- recruited for this study. ing laboratory data were recorded: haemoglobin ( Hb), All patients were seen within 25 months after the C-reactive protein levels (CRP; normal: <10) and eryth- onset of their first symptoms (median 5.8 months, range rocyte sedimentation rate ( ESR; normal in males: 1 week 2.08 yr). The baseline clinical features and func- <12 mm/first h, normal in females: <19 mm/first h) in tional ability for the two groups of patients ( YORA 274 patients and plasma viscosity (normal: <1.73) in and LORA) are shown in Table 1. The baseline labora- 116 patients were used as markers of inflammatory tory and radiological data are shown in Table 2. activity; IgM RF by nephelometry (positive if titre 60 Comparison of the radiological, clinical and functional IU ); antinuclear antibody (ANA) using indirect immuno- outcome of the two groups is shown in Table 3. fluorescense; serum uric acid (SUA). Plain radiography of the hands, wrists and feet was used to identify the Clinical, functional and laboratory data at onset frequency of joint erosions, and these used as markers The gender ratio between the two groups was similar for disease damage. (LORA group, 65.4% female; YORA group, 67.8%

3 230 C. T. Pease et al. TABLE 1. Clinical and functional data at the first assessment of YORA (group 1) and LORA (group 2); median values are given Group 1 Group 2 n = 186 n = 214 Female-to-male ratio 2.1 : 1 1.9:1 Age at onset (yr) 51.5 (19 64) 72.8 (65 90) Time to presentation (months) 7.2 ( ) 4.2 ( ) Pattern of arthritis at onset Monoarthritisa 41 (22.0%) 26 (12.8%) Symmetrical polyarthritis 108 (58.1%) 171 (79.9%) RS3PE presentationa 0 (0.0%) 24 (11.2%) Weight loss 17 (9.1%) 48 (22.4%) Duration of EMS (h)a Myalgic onseta 1 (0.5%) 28 (13.1%) Ritchie articular index 10 (0 50) 12 (0 52) HAQa 1.13 (0 2.87) 1.81 (0 3.0) Grip strength (mmhg)a Left hand 174 (S.D., 79) 138 (S.D., 69) Right hand 175 (S.D., 81) 135 (S.D., 70) VAS pain score 4.4 (S.D., 2.76) 4.6 (S.D., 2.72) EMS, early morning stiffness. adenotes statistical significance, P < female). The median age of onset of RA in the YORA TABLE 2. Laboratory and radiological data at the first assessment of group was 51.5 (19 64) yr and in the LORA group 72.8 YORA (group 1) and LORA (group 2); median values are given (range ) yr. Although symmetrical poly- Group 1 Group 2 arthritis was the commonest pattern of joint involvement n = 186 n = 214 seen at presentation in both groups, monoarthritis at onset was nearly twice as common in the YORA group Rheumatoid factora (60 IU ) 105 (56.5%) 92 (42.9%) Haemoglobin (g/dl )a (P < 0.01). Weight loss and myalgia were more frequent ( ) ( ) at onset of arthritis in the LORA patients. The RAI at Anaemic at presentationa 55 (29.6%) 87 (42.6%) baseline was similar between the two groups. Both the Raised acute-phase reactants 135 (72.6%) 161 (75.2%) HAQ score and grip strength were relatively worse in Antinuclear factor (100 IU ) 13 (6.9%) 11 (5.1%) Joint erosions 38 (20.4%) 43 (20.1%) the LORA group when compared with the YORA (n = 186) (n = 179) group; a finding most likely explained on the basis of HLA status ageing rather than differences directly attributable to Dr1 34 (18.3%) 37 (20.7%) development of RA at a later age. There was no Dr4 77 (41.4%) 82 (45.8%) difference in VAS estimation of pain between the two Dr10 0 (0.0%) 3 (1.7%) Dr14 28 (15.0%) 16 (8.9%) groups. Joint erosions of the hands, wrists or feet (plain radiography) were found in 20.4% of YORA patients and in 20.1% of LORA patients. Twenty-four out of adenotes statistical significance, P < LORA patients presented with features of the TABLE 3. Comparison of clinical and radiological outcome of the RS3PE syndrome remitting seronegative symmetrical whole cohort at last follow-up synovitis with pitting oedema [ 26]. No such patients Young-onset group Late-onset group were identified in the YORA group. Laboratory data (n = 186) (n = 214) are shown in Table 2. YORA patients were more likely to be seropositive for IgM RF at presentation (YORA, RAI (median)a 6.0 (0 48) 1.0 (0 51) 56.5%; LORA, 42.9%; P < 0.05). Antinuclear factor RAI = 0a 38 (20.4%) 98 (45.8%) Joint erosions 96 (51.6%) 116 (54.2%) (100 IU ) was found in 6.9% of YORA patients and (n = 186) (n = 92) 5.1% of LORA patients. Patients with LORA had a HAQ (median) 0.88 (0 3.0) 0.88 (0 2.88) higher frequency of anaemia at the baseline visit than HAQ = 0 32 (17.2%) 13 (14.1%) the YORA patients (P < 0.05). A total of 72.6% of Grip strength (mean) YORA patients and 75.2% of LORA patients had Left hand 193 (S.D., 84) 176 (S.D., 96) Right hand 192 (S.D., 86) 174 (S.D., 95) elevated acute-phase reactants at baseline. There was no VAS pain score (mean) 3.5 (S.D., 2.6) 2.8 (S.D., 2.8) significant difference in frequencies of HLA DR1, DR4, DR10 and DR14 between the two groups. adenotes statistical significance, P < Outcome RAI at last follow-up for the YORA group (6.0, range Patients in both groups were prospectively followed up 0 48) was significantly higher than that for the LORA for at least 1 yr with the median follow-up for patients group (1.0, range 0 51, P < 0.01), as shown in Table 3. with LORA being 3.4 yr (range yr) and for A total of 45.8% of the LORA group compared with patients with YORA 4.0 yr (range ). The median 20.4% of the YORA group had clinically inactive disease

4 Age of onset of RA and outcome 231 at final follow-up (P < 0.01). The median duration of reached statistical significance as predictors of remission active disease in patients who went into clinical remission or the development of joint erosions. was longer in the YORA group [ YORA, 2.0 yr The median disability level, as measured by the HAQ ( ); LORA, 1.34 yr ( )]. Seropositive score at last follow-up, was similar for the LORA and patients (remission rate: YORA, 20%; LORA, 29%) and YORA groups ( Table 3). Grip strength and VAS pain seronegative YORA patients (remission rate 16.1%) were scores improved from baseline in both groups, although less likely to go into remission compared with seronega- there was no significant difference between the two tive LORA patients (remission rate 59.8%). There was groups in the amount of improvement. The logistic no difference in the overall HAQ scores at last follow- regression model was used to calculate OR for those up between the two groups (LORA, 0.88, range 0 3.0; factors at presentation that predicted poor functional YORA, 0.88, ). outcome (defined as an HAQ score at follow-up of Over a median follow-up period of 30.6 (range greater than one). Patients at presentation were deemed ) months, 54.2% of the LORA patients had joint to have limited functional loss if their HAQ scores were erosions of either the hands, wrist or feet, which is one or less and deemed to have marked disability if comparable with the frequency of joint erosions in the their HAQ score was greater than one. This allowed the YORA patients (51.6%) ( Table 3). HAQ score at presentation to be entered as a dichotom- For the whole cohort, the relative contribution of ous variable in the model. The overall predictive ability potential prognostic indicators at presentation in determining of this model was 69.2%. This analysis showed that (1) the likelihood of developing joint erosions and patients with an HAQ score greater than one at presenta- entering remission is presented as odds ratios (OR) tion were more likely to fall into the poor functional (Table 4). Age at onset of arthritis, baseline IgM RF outcome category (OR = 3.02, 95% CI 1.98, 4.62) and seropositivity, gender, presence of HLA DR1, HLA (2) women were also more likely to have a poorer DR4, myalgia, elevated acute-phase reactants, anaemia outcome (OR = 3.01, 95% CI 1.59, 5.68). at onset and weight loss were entered as dichotomous In order to explore these findings, further regression covariates into the logistic regression analysis model analysis was undertaken separately for YORA and using forward stepwise selection (P < 0.05) with the LORA patients. In the YORA group, being female presence of joint erosions and clinically inactive disease (OR = 2.96, 95% CI 1.37, 6.39) and having a high (as defined above) as the dependent variables. Age of baseline HAQ score (OR = 2.36, 95% CI 1.46, 3.82) onset of arthritis was dichotomized using age 65 yr or were predictors of poor functional outcome. In the above as the cut-off point for inclusion in the LORA LORA group, however, gender did not reach statistical group. The overall predictive ability of this approach significance (P = 0.09), but a high HAQ score was 70.6% for joint erosions and 72.7% for remission, (OR = 7.42, 95% CI 2.44, 22.49) and IgM RF seroposi- confirming the validity of using this model for this tivity (OR = 8.17, 95% CI 2.09, 31.91) predicted poor data set. functional outcome. None of the other covariates (pres- This analysis showed that: (1) LORA patients are ence of HLA DR1 or DR4, raised inflammatory markers more likely to go into clinical remission (OR = 2.99, at presentation) entered into the logistic regression 95% CI 1.77, 5.02); (2) patients with baseline IgM RF model reached statistical significance as predictors of seropositivity were more likely to develop joint erosions poor functional outcome. (OR = 4.24, 95% CI 2.56, 6.94), and less likely to go into clinical remission (OR = 0.47, 95% CI 0.28, 0.77); Treatment (3) patients with elevated acute-phase reactants at base- A total of 69.4% of the YORA group and 57.9% of the line were more likely to develop joint erosions (OR = LORA group received at least one DMARD. Table , 95% CI 1.04, 3.14); (4) patients with the HLA shows the types of DMARDs used, with no differences DR4 subtype were more likely to develop joint erosions apparent between the two groups. Systemic steroid (OR = 2.07, 95% CI 1.28, 3.35). None of the other treatment for >3 months was used in 16.7% of YORA covariates entered into the logistic regression model patients and 31.8% of LORA patients. In most cases, TABLE 4. Odds ratios for those dichotomous variables predicting the development of joint erosions and remission (P 0.05) Development of joint erosions Remission OR 95% CI OR 95% CI Age at onset 65 yr a Female Elevated inflammatory markers 1.81a HLA DR4 positive 2.07a Rheumatoid factor positive 4.24a a Steroid use 2.00a adenotes statistical significance, P < 0.05.

5 232 C. T. Pease et al. TABLE 5. DMARD treatment received by patients Young-onset group Late-onset group (% of total cohort) (% of total cohort) Number treated with DMARD Types of DMARD used Hydroxychloroquine Sulphasalazine i.m. Gold Penicillamine Methotrexate Azathioprine Cyclosporin A Systemic steroids (excludes intra-articular and short-term oral steroids: 3 months) the maximum dose of prednisolone initially given was development of erosions is similar to that suggested by 12.5 mg daily. The doses used in LORA and YORA van der Heijde et al. [19] and again points to the need patients were similar. The increased steroid usage at for early intervention. presentation in the LORA group may have been a In our study, there was no difference in female prepon- contributing factor to the observed differences in out- derance with age of onset (females: YORA, 67.7%; come between the two age groups. To explore this LORA, 65.5%). This is in accordance with some reports assertion further, steroid use was entered as a dicho- [17, 20], but in contrast to others [19, 21]. Our data are tomous predictor into the above regression model to in accordance with prevalence data reported by analyse its effect on therapeutic remission and the devel- Bergstrom et al. [ 2]. Constitutional features (weight opment of erosions. Steroid use was defined as regular loss), myalgia, increased duration of early morning oral steroids taken for >3 months. For the whole stiffness and impaired function as measured by the HAQ cohort, this analysis showed that use of steroids was and grip strength were more common in the LORA associated with the development of erosions (OR = 2.0, patients. LORA patients also tended to present early 95% CI 1.1, 3.63). Further regression analysis of the and frequently needed admission at disease onset YORA and LORA groups separately showed that this because of inability to cope at home. As a presenting steroid effect was confined to the LORA patients feature, monoarthritis was more common in the (OR= 4.09, 95% CI 1.81, 9.27). No association between YORA group. steroid use and the development of erosions was found Over half the patients in both groups had developed for the YORA (OR = 0.77, 95% CI 0.30, 1.77). Further erosions by a median follow-up period of 3.6 yr. Similar analysis of the whole cohort showed that steroid use did proportions of seropositive patients in each group not increase the frequency of remission (OR = 0.91, developed joint erosions ( YORA, 72.8%; LORA, 95% CI 0.51, 1.67). 62.5%). However, in those that were seronegative for RF, 40% of LORA patients developed erosions as Discussion compared with 25% of the YORA group. The duration of disease, baseline HAQ score, RAI and acute-phase The notion that LORA is a milder disease than classic reactant levels were comparable between the seropositive RA, and therefore requires less aggressive treatment, is and seronegative patients. One factor that may explain not supported by our results. Most of the earlier studies the increased proportion of joint erosions in the seronega- into LORA have either been retrospective, cross- tive LORA group compared with the seronegative sectional or not directly compared with YORA patients. YORA group is the possibility that elderly cartilage is Only two studies have prospectively compared patients more susceptible to damage from synovial inflammation. with YORA and those with LORA (follow-up of up to Continuous steroid use for >3 months in LORA 6 yr in one and 2 yr in the other [17, 19]). Both these patients was also identified as an independent factor studies report a poorer prognosis for patients presenting associated with joint erosions in this LORA group of with LORA. Our study suggests that LORA requires patients. This association of long-term steroid use and early and appropriate DMARD treatment similar to frequency of joint erosions may relate to their more that given to patients with YORA. However, the higher frequent use in patients with already poor prognosis remission rate in the seronegative LORA patients suggests disease with difficult to control symptoms. Pre-existing that treatment in this group of patients may be cartilage damage from osteoarthritis may also increase given for a shorter duration (to optimize the benefits of the likelihood of joint damage following synovial treatment vs the adverse reactions) when compared with inflammation from RA. seropositive patients. One-fifth of patients in both The disease in patients developing LORA was more groups were found to have joint erosions on plain likely to remit than that in the YORA group, and the radiography at initial presentation. The timing of the duration of disease was shorter in LORA patients in

6 Age of onset of RA and outcome 233 remission. The use of steroids may have a role in 6. Isemein L, Redon M. La polyarthrite chronique evolutive inducing clinical remission in a subgroup of LORA après l age de 55 ans. Rev Rhum Mal Osteoartic patients presenting with RA who, although fulfilling 1953;20: ACR criteria for diagnosis, nevertheless have other 7. Dordick JR. Rheumatoid arthritis in the elderly. J Am Geriatr Soc 1956;4: features such as acute peripheral limb oedema suggesting 8. Oka M, Kytila J. Rheumatoid arthritis with the onset in RS3PE, a syndrome that seems to predict good out- old age. Acta Rheumatol Scand 1953;3: come [26]. 9. Evers A. Die im alter auftretende primär chronische The level of disability during follow-up arises from a polyarthitis. Z Rheumaforsch 1956;24: combination of co-morbid conditions, ongoing joint 10. Adler E. Rheumatoid arthritis in old age. Isr J Med Sci inflammation and joint damage. As expected, the LORA 1966;2: patients had higher baseline HAQ scores; however, 11. Brown JW, Somes DA. The onset of rheumatoid arthritis despite this increased disability at onset, the functional in the aged. J Am Geriatr Soc 1967;10: Ebert H. Der verlauf der im alter auftretenden chronoutcome can be significantly improved with appropriate isch-rheumatischen polyarthritis. Z Alternsforsch 1967; treatment, as shown by the overall reduction in HAQ 20: at follow-up (median initial HAQ 1.81; median follow- 13. Moesmann G. Clinical features in subacute rheumatoid up HAQ 0.88; P < 0.01). The predictors of disability arthritis in old age. Acta Rheumatol Scand that emerged from this study are consistent with other 1968;14: reports, in that a high HAQ score at presentation and 14. Ehrlich GE, Katz WA, Cohen SH. Rheumatoid arthritis female gender are predictors of poor functional outcome in the aged. Geriatrics 1970;25: [27, 28]. While HLA DR4 was associated with the 15. Corrigan AB, Robinson RG, Terenty TR, Dick-Smith JB, development of joint erosions, there was no association Walters D. Benign rheumatoid arthritis of the aged. Br Med J 1974;i: with disability, in keeping with previous evidence [ 29]. 16. Terkeltaub R, Esdaile J, Decary F, Tannenbaum H. A In the whole cohort, RF seropositivity was lower than clinical study of older age rheumatoid arthritis with that reported by other investigators [19]. RF seropositiv- comparison to a younger onset group. J Rheumatol ity in the LORA patients (42.9%) was less than in the 1983;10: YORA group (56.5%). This is in contrast to findings of 17. Ferraccioli GF, Cavalieri F, Mercandati M, Conti G, van der Heijde et al. (LORA, 89%; YORA, 78%) [19], Viviano P, Ambanelli U. Clinical features, scintiscan char- but similar to others [14, 17, 21]. This difference might acteristics and X-ray progression of late onset rheumatoid reflect the type of patients referred to the individual arthritis. Clin Exp Rheumatol 1984;2: Deal CL, Meenan RF, Goldenberg DL, Anderson JJ, units or differences in the sensitivity of the tests (nephelo- Sack B, Pastan RS et al. The clinical features of elderly metry vs ELISA). The frequency of HLA DR4 in the onset rheumatoid arthritis A comparison with younger LORA patients in our study (45.8%) is similar to that onset disease of similar duration. Arthritis Rheum reported by van der Heijde et al. (52%) [19]. 1985;28: In conclusion, this study shows that LORA can be as 19. van der Heijde DMFM, van Reil PLCM, van Leeuwen damaging as classical RA and that joint erosions are MA, van t Hof MA, van Rijswijk MH, van de Putte LBA. often observed at presentation to the rheumatology Older versus younger onset rheumatoid arthritis: results service. RF positivity, HLA DR4 and baseline elevated at onset and after 2 years of a prospective follow-up study acute-phase reactants are all associated with an increased of early rheumatoid arthritis. J Rheumatol 1991; 18: frequency of erosions in both groups of patients. 20. van Schaardenburg D, Hazes JMW, de Boer A, Although LORA patients at presentation are more Zwinderman AH, Meijers KAE, Breedveld FC. Outcome disabled, function can nevertheless be significantly of rheumatoid arthritis in relation to age and rheumatoid improved. These findings suggest that the current treat- factor at diagnosis. J Rheumatol 1993;20: ment approaches used in patients with classical YORA 21. Inoue K, Shichikawa K, Nishioka J, Hirota S. Older age should be instituted with equal vigour in patients with onset rheumatoid arthritis with or without osteoarthritis. LORA. Ann Rheum Dis 1987;46: van Schaardenburg D, Breedveld FC. Elderly onset rheumatoid arthritis. Semin Arthritis Rheum 1994;23: References 23. Symmons DPM, Barrette EM, Bankhead C, Scott DGI, Silman AJ. The incidence of rheumatoid arthritis in the United Kingdom: Results from the Norfolk Arthritis 1. Office of Population Censuses and Surveys National Register. Br J Rheumatol 1994;33: population projections (1992 based). Government 24. Garrod AE. A contribution to the nervous origin of Publications Statistical Service, April rheumatoid arthritis. Med Chir Trans 1888;71: Bergstrom G, Bjelle A, Sundh V, Svanborg A. Joint 25. Desrosiers J, Hebert R, Bravo G, Dutil E. Comparison of diseases at ages 70, 75 and 79 years A cross sectional the Jamar dynamometer and the Martin vigorimeter for comparison. Br J Rheumatol 1986;25: grip strength measurements in a healthy elderly popula- 3. Lawrence JS. Prevalence of rheumatoid arthritis. Ann tion. Scand J Rehabil Med 1995;27: Rheum Dis 1961;20: Bhakta BB, Pease CT. Late onset rheumatoid arthritis: is 4. Schnell A. The clinical features of rheumatic infection in pitting oedema of the hands a good prognostic indicator? the aged. Acta Med Scand 1941;106: Br J Rheumatol 1997;36: Cecil RL, Kammerer WH. Rheumatoid arthritis in the 27. Young A. Short term outcomes in recent onset rheumatoid aged. Am J Med 1951;10: arthritis. Br J Rheumatol 1995;34:79 86.

7 234 C. T. Pease et al. 28. van der Heide A, Jacobs JWG, Haanen HCM, Bijlsma 29. Olsen NJ, Callahan LF, Brooks RH, Nance EP, Kaye JJ, JWJ. Is it possible to predict the first year extent of pain Stastny P et al. Associations of HLA-DR4 and radioand disability for patients with rheumatoid arthritis? graphic severitiy in rheumatoid arthritis. Am J Med J Rheumatol 1995;22: ;84: