THE CLINICAL FEATURES OF ELDERLY-ONSET RHEUMATOID ARTHRITIS

Transcription

1 THE CLINICAL FEATURES OF ELDERLY-ONSET RHEUMATOID ARTHRITIS A Comparison With Younger-Onset Disease of Similar Duration CHAD L. DEAL, ROBERT F. MEENAN, DON L. GOLDENBERG, JENNIFER J. ANDERSON, BURTON SACK, ROBERT S. PASTAN, and ALAN S. COHEN Patients with elderly-onset rheumatoid arthritis (EORA) may represent a clinical subset of individuals who differ prognostically and therapeutically from patients with younger-onset disease (YORA). In order to test this hypothesis, we reviewed the records of 212 patients with rheumatoid arthritis and grouped them according to age at onset above or below 60 years old. Seventy-eight EORA patients and 134 YORA patients with disease duration of 510 years were used for a comparison of presenting features and disease outcome. Abrupt onset occurred somewhat more frequently in EORA, but was not associated with a significantly different clinical course than was an insidious pcesentation in this older group. There were no differences between the EORA and YORA groups in terms of mean initial joint score, although the scores for the YORA group had wider variation. An initial clinical pcesentation resembling polymyalgia rheumatica (PMR) was 4 From the Arthritis Center of Boston University, the Departments of Medicine, University and Boston City Hospitals, and the Thorndike Memorial Laboratory, Boston City Hospital, Boston, Massachusetts. Supported by Multipurpose Arthritis Center grant AM and Arthritis Training grant AM from the National Institutes of Health, a Clinical Research Center grant from the Arthritis Foundation, and an institutional grant from the Massachusetts Chapter of the Arthritis Foundation. Chad L. Deal, MD: Assistant Professor of Medicine; Robert F. Meenan, MD, MPH: Associate Professor of Medicine; Don L. Goldenberg, MD: Associate Professor of Medicine; Jennifer J. Anderson, PhD: Assistant Research Professor of Medicine (Biostatistics); Burton Sack, MD: Associate Clinical Professor of Medicine; Robert S. Pastan, MD: Clinical Instructor of Medicine; Alan S. Cohen, MD: Conrad Wesselhoeft Professor of Medicine. Address reprint requests to Chad L. Deal, MD, Case Western Reserve University, Lakeside Hospital, 2073 Abington Road, Cleveland, OH Submitted for publication September 18, 1984; accepted in revised form March times as frequent in EORA. Elderly patients were less likely to have subcutaneous nodules or rheumatoid factor at disease onset. At the final examination, the EORA patients had lower joint scores and higher health assessments despite similar courses of treatment. These outcome differences persisted when patients with PMRlike presentations were excluded. Multivariate analyses indicated that joint scores and disease duration made important contributions to a better outcome of EORA, whereas PMR presentation and abrupt onset did not. After an adjustment was made for these 4 features, age at onset was an important contribution to joint score outcome. These results confirm the existence of important differences in onset, clinical features, and prognosis between patients with EORA and those with YORA. Longitudinal studies of rheumatoid arthritis (RA) patients have identified historic, clinical, and laboratory findings which have important associations with disease severity and outcome. For example, the presence of rheumatoid factor (seropositive RA) identifies a group of patients with greater functional disability, greater erosive disease, and higher frequency of use of remittive agents (1-3). Sex, race, type of onset, rheumatoid nodules, antinuclear antibodies, and HLA type have also been identified as important determinants of disease outcome (4-7). However, age as a determinant of disease course and outcome is controversial. Several reports have claimed that patients whose rheumatoid arthritis begins later in life (generally defined as after age 60) represent a clinical subset of patients with important differences in presentation, disease manifestations, and prognosis compared with patients who develop RA at a younger age, while other studies have found no significant differences (8-13). Arthritis and Rheumatism, Vol. 28, No. 9 (September 1985)

2 988 DEAL ET AL The present study attempts to clarify this issue by a direct comparison of the presentation and course of elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA). It differs from previous studies of EORA in that the sample size is larger, the subjects are drawn from both academic and community settings, and adjustments were made for differences in duration of disease. The results confirm that there are important differences in presentation and clinical features between EORA and YORA patients. With controlling for disease duration and initial severity of disease, there are still important differences in outcome between the 2 groups, despite highly similar courses of treatment. The results support the theory that age at onset is an important factor which influences outcome in rheumatoid arthritis. PATIENTS AND METHODS We reviewed the charts of 300 patients with RA who were being followed by 4 rheumatologists in the Boston area. Two rheumatologists based at a medical center were following 55 (20%) of the subjects, and 2 rheumatologists based in the community were following 245 patients (80%). All 300 patients met the American Rheumatism Association (ARA) criteria for definite or classic RA (14). Subjects were categorized as having EORA or YORA based on a cutoff age of 60. We chose this cutoff since it represented the modal age used by previous investigators who have studied EORA. A subset of RA patients with disease duration of 510 years at the time of the study was selected for analysis. This allowed analysis of groups with similar disease duration since duration has a major impact on clinical manifestations and the outcome of RA (15). The final study sample comprised 212 subjects (71% of the total), including 134 (62%) of the YORA group and 78 (95%) of the EORA group. A standard chart review procedure was used to collect data on the following variables from the initial and most recent routine visit (final visit) of each subject: type of onset, duration of morning stiffness (>30 minutes), presence of constitutional symptoms, subcutaneous nodules, joint score, and laboratory values (including complete blood count, erythrocyte sedimentation rate, and rheumatoid factor [RF]). Type of onset was classified as abrupt if the arthritis reached full intensity within 2 weeks of the initial symptoms. Joint score was based on a 0-3 grading of swelling or tenderness for 13 joints or joint groups (proximal interphalangeal/metacarpophalangeal, wrists, elbows, shoulders, acromioclavicular, sternoclavicular, temporomandibular, toes, ankles, knees, hips, cervical spine, lumbar spine). A positive joint on either side of the body resulted in a score of I (mild), 2 (moderate), or 3 (severe) for that joint pair or group, so that the score for joint count could potentially range from For example, severe tenderness in all 10 metacarpal joints resulted in a joint score of 3, as did bilateral severe tenderness of the wrists. Gradings for involved joints had been recorded in a large majority of patient charts; therefore these scores could be used directly. When exact grades were not available, scores were estimated from the narrative description. One author (CLD) reviewed all charts. Data were also collected on all medications, including those used before the first visit and those used during the interval between the initial and most recent visit. Information was gathered on whether the patient had had orthopedic surgery and whether each patient had been referred for physical and/or occupational therapy. Overall outcome was analyzed by using 4 measures of current status at the time of the final visit. These included the physician s overall assessment, the patient s overall assessment, the patient s ARA functional class, and the joint score. The first 2 estimates were based on a 5-point scale ranging from very good to very poor. The attending physicians estimated the patient s functional class using the standard 4-point scale (16). Statistical analysis focused on differences between groaps, using t-tests and analysis of covariance for continuous variables and chi-square tests for categorical variables. The t-tests were based on assumptions of independent samples and unequal variances, utilizing 2-tailed probability. The chi-square tests used Yates correction. Since multiple comparisons were performed, P values for a Type I error were given for each reported difference, and the Bonferroni adjustment was applied to maintain an overall significance level of 0.05 (17). The analysis of covariance was used to adjust for presenting features that would possibly confound differences in final joint count between the age at onset groups. The significance of these factors was tested using t-tests. The factors adjusted for included initial joint count, which has a quadratic relationship with final joint count in these data. Both linear and quadratic terms were included in adjusting for this variable (18). RESULTS The study group consisted of 53 men and 159 women. The age at disease onset in the EORA subjects was 67.2 * 4.7, mean * SD (range 60-84). The age at disease onset in the YORA patients was 46.6? 10.0, mean * SD (range 20-59). Seventy-eight subjects were 260 at disease onset, 134 were <60 at disease onset; 34 patients were between the ages of 55 and 60, and another 26 were between 50 and 55. Women were predominant in both groups. The average disease duration for both groups was 4.8 +: 3.1 years (4.06 * 2.87 for EORA; 5.25 * 3.17 for YORA). Presenting symptoms. An abrupt onset, with symptoms peaking in days to weeks, occurred nearly twice as often in the EORA group (Table 1) (27% versus 14%; P = 0.024). Morning stiffness >30 minutes duration was extremely common in both groups. There were no major differences in constitutional symptoms such as fatigue (P = 0.185) or weight loss (P = 0.169). The groups were also very similar in terms of selected extraarticular symptoms of RA,

3 ~ ~~ ~ ELDERLY-ONSET AND YOUNGER-ONSET RA 989 Table 1. Presenting features of elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA)* EORA YORA Presenting feature (n = 78) (n = 134) P Age at onset (years) Disease duration at initial visit (years) % female Positive family history Abrupt onset Morning stiffness PMR symptomst Total joint score Small joint disease5 Large joint disease Subcutaneous nodules Erythrocyte sedimentation rate Positive rheumatoid factor 67.2 lr % % 21% 27% 85% 23% 4.9 f % 14% 6% 52 f % 46.6 f f % 24% 14% 90% 5% 5.1 f % 8% 20% 40.7 f % O.OOO$ t $ * Values are mean? standard deviation unless designated otherwise. t Polymyalgia rheumatica (PMR) symptoms are positive if all 4 of the following are true: involvement of shoulders or hips, nodules absent, erythrocyte sedimentation rate >50, latex fixation titer <1:20. i: For significance at the 0.05 level, with 13 comparisons, the Bonferroni adjustment requires a P value of s0.05/13 = Small joint disease = involvement of fingers and toes; large joint disease = involvement of shoulders and hips. including dry eyes (7% overall), dry mouth (4% overall), and carpal tunnel syndrome (10% overall). The initial physical examination revealed significant differences between the 2 groups. The joint score at the time of the first visit was nearly identical in the 2 groups (4.9? 2.6 in the EORA group versus 5.1 * 3.6 in the YORA group; P = 0.696), but the spread of values was much greater in the YORA group (P < in the F-distribution test for equality of variances). The pattern of joint involvement was also somewhat different. Younger patients were twice as likely to have small joint disease with evidence of active inflammation in both finger and toe joints (46% versus 24%; P = ). Elderly patients tended to have more arthritis in both hips and shoulders (14% versus 8%; P = 0.186). This latter difference was primarily due to a higher frequency of shoulder synovitis in the EORA group (56% versus 36%; P = 0.060). The EORA patients were less apt to have subcutaneous nodules present at the initial examination (6% versus 20%; P = 0.016). The EORA group had a higher initial erythrocyte sedimentation rate (ESR) (P = 0.008) and was significantly less likely to have a positive RF (P = 0.001). There were no differences between groups in initial hematocrit value (mean = 38.7), white blood cell count (7,500/mm3), or frequency of antinuclear antibody (ANA) titers >1: 16 (7.5%). A polymyalgia rheumatica (PMR)-like presentation was defined as the presence of all 4 of the following features: shoulder or hip synovitis, absence of nodules, ESR >50 mm/hour, and negative RF. All of these features were present in 23% of the EORA group, compared with 5% of the YORA group (P = ). This difference was also apparent if the number of PMR features was treated as a continuous variable. In this case, the EORA group exhibited 2.5 k 1.1 of the features, whereas the YORA group had only (P < 0.001). Therapy. The younger and older patients were treated in a similar fashion (Table 2). The percentage of patients in each group who were treated with any second line drug was virtually identical. While steroids were used somewhat more often by the elderly-onset group, gold injections, penicillamine, and cytotoxic drugs tended to be used more frequently by the younger patients. These differences, however, were relatively small and none were clearly significant. The younger group was only slightly more likely to have undergone orthopedic surgery or occupational therapy, but was substantially more likely to have received physical therapy. The overall picture, however, is one of nearly identical treatment during the first years of disease, regardless of the age at onset. Outcome. The outcome of the elderly-onset group assessed at the last visit was significantly better than that of the younger-onset group across a variety of pertinent disease outcome measures (Table 3). The elderly subjects were much less apt to experience morning stiffness or fatigue, despite the fact that fatigue tended to be more common in the EORA group at the initial visit. Table 2. Treatment of elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA) EORA, % YORA, % Treatment (n = 78) (n = 134) P Second line drugs H ydroxychloroquine Gold injections Oral steroids Penicillamine Cytotoxic drugs* Orthopedic surgery Physical therapy Occupational therapy * Includes azathioprine, cyclophosphamide, chlorambucil, or methotrexate.

4 990 DEAL ET AL Table 3. Outcome comparison of elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA) in subjects with disease duration of 10 years or less* All subjects Excluding polymyalgia rheumatica presentations EORA YORA EORA YORA Outcome measure (n = 78) (n = 134) P (n = 61) (n = 127) P Disease duration (years) Morning stiffness Fatigue Total joint score Small joint disease$ Large joint disease No active joints Subcutaneous nodules Erythrocyte sedimentation rate Physician assessment Patient assessment5 ARA functional class I or f % 3% % 1% 36% 17% 24.4 f % 84% 94% 5.25 t % 15% 3.3 f % 4% 20% 33% % 70% 83% t f % 3% % 2% 38% 28% 25.1 f % 82% 94% 5.4 f % 15% 3.3 & % 3% 20% 35% 27.9 t % 70% 82% ~~ ~~ * Values are mean 2 standard deviation unless designated otherwise. t For significance at the 0.05 level, with 12 comparisons, the Bonferroni adjustment requires a P value of <0.05/12 = $ Small joint disease = involvement of fingers and toes; large joint disease = involvement of shoulders and hips. Percent rated as good or very good. 7 ARA = American Rheumatism Association t The total joint score at the time of the final visit was significantly lower in the older group ( versus 3.3 k 3.3; P < O.OOO), and the EORA patients were much more apt to have no active joints (36% versus 20%; P = 0.031). Eighty-seven percent of the EORA group and 74% of the YORA group had a lower final joint score compared with their initial joint score (P = 0.039). The younger patients continued to have more evidence of small joint involvement, while the older patients no longer exhibited a high frequency of large joint disease, The YORA group had nearly twice the frequency of subcutaneous nodules (17% versus 33%; P = 0.018). It is possible that the better outcome in the EORA group related to shorter disease duration. However, a separate analysis of subjects (EORA [n = 541 and YORA [n = 691) with disease duration 55 years, in which the mean duration of disease was nearly identical in the 2 groups (2.50 years in the EORA group versus 2.56 years in the YORA group; P = 0.85), continued to reveal better outcome in the EORA patients. Laboratory findings in the 2 groups were quite similar. The final ESR in the elderly-onset subjects was found to be nearly equal that of the younger-onset patients, despite the fact that the initial ESR in the older group had been significantly higher. Positive RF and positive ANA were present in both groups, but these observations were frequently omitted at the last visit. The tendency of the elderly-onset group to do better on specific disease activity measures is reflected in the fact that they also fared much better in terms of overall disease status estimates. Patients in the elderly-onset group were much more likely to rate their status as good or very good than were patients in the YORA group, and were more likely to be rated in the top 2 functional classes by their physicians. These outcome analyses were repeated after eliminating the 24 subjects (17 EORA, 7 YORA) who had a PMR-like presentation. These results are shown in Table 3. Only minor alterations in the differences previously described between EORA and YORA groups were seen in this revised analysis. For example, morning stiffness was no longer significantly less common in the EORA group. However, the outcome for the EORA group continued to be substantially better on all measures even when the subjects with a PMR-like presentation were omitted. Because of the tendency of elderly patients to have a much better outcome compared with younger patients, an attempt was made to identify the factors which might account for this difference. One possible explanation is that abrupt-onset disease is more common in the elderly-onset group, and that this type of RA tends, in general, to have a better outcome. Another plausible explanation is that the PMR-like disease frequently found in the elderly is, in fact, polymyalgia rheumatica rather than RA and therefore could be expected to have a more benign course. In addition to the differences in presenting

5 ELDERLY-ONSET AND YOUNGER-ONSET RA 99 1 Table 4. Analysis of variance of joint score outcome in elderly-onset and younger-onset rheumatoid arthritis groups All subjects* Excluding polymyalgia rheumatica (PMR) presentations? Standardized Standardized Regression regression Regression regression Factorkovariate coefficient coefficient P coefficient coefficient P Age at onset Initial joint score I Linear effect Quadratic effect Disease duration Mode of onset PMR symptomst Sex I Constant * n = 187; adjusted R2 = t n = 174; adjusted RZ = t PMR symptoms are positive if all 4 of the following are true: involvement of shoulders or hips, nodules absent, erythrocyte sedimentation rate >50, rheumatoid factor titer < 1 : 20. characteristics, there were differences between the EORA and YORA groups in the distribution of both initial joint counts and the total duration of disease, each of which is related to outcome. In order to adjust for each of these factors in a comparison of the 2 groups, they were included as covariates in an analysis of variance using final joint count as the outcome (dependent) variable. The results of the multivariate outcome analyses are presented in Table 4. (The left half of the table includes all subjects, the right half excludes those with PMR-like presentations. There were no substantial differences between the 2 sets of findings). Initial joint score and disease duration were strongly associated with the joint score outcome, whereas the associations of abrupt onset and PMR features were not. After covariance adjustment for each of these factors and for sex, there remained a linear effect of age at onset (-0.04 t 0.01). This translates to a final joint score 1 unit lower for every 28.5-year increase in age at onset. Analysis using the 4 PMR components (RF titer <1:20, absent nodules, ESR >50, hip or shoulder arthritis) as separate variables indicated that none made a significant contribution to outcome. The lower frequency of seropositive RA in the EORA group did not explain the better outcome, since the inclusion of a separate RF term in the regression equation at either the 1 : 20 or 1 : 80 titer level showed no effect. Additional covariance analysis using 3 alternative outcome measures (absence of active joints at final visit, physician s assessment, and patient assessment) also demonstrated no effect for the PMR variables. A comparison of joint outcomes for the YORA and EORA groups was then carried out within each of Table 5. Stratified comparison of joint outcome in elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA) groutx Final joint scores Stratum* EORA Y ORA Difference P Low joint score/short duration Low joint score/long duration High joint score/short duration High joint score/long duration t (n = 28) 1.92? 2.54 (n = 12) 2.20 * 2.61 (n = 25) (n = 12) 1.57 * (n = 46) 2.31 * (n = 26) (n = 22) (n = 36) * Low joint score if initial count 54; high joint score if >4. Short duration if duration at last visit 55 years; long duration if >5 years. t Values expressed as mean standard deviation.

6 992 DEAL ET AL 4 strata, based on low or high initial joint score (<4 or :>4) and long or short disease duration (<5 years or >5 years), since these strata were more uniform than the data set as a whole, with respect to these 2 measures. The results shown in Table 5 confirm that joint count outcome differences between the YORA and EORA groups depend on disease duration and on initial joint scores. These results indicate that, given low joint score and short duration, or any of the other 3 combinations of these 2 factors, EORA joint outcomes were always better than YORA outcomes. For those with low initial joint score, for example, there were final joint count differences of 0.64 (P = 0.042) and 0.39 P = 0.663) for the shorter and longer duration strata, respectively. For those with higher initial joint counts, the corresponding outcome differences were 2.80 (P = 0.002) and 2.00 (P = 0.034). Thus the outcome differences between EORA and YORA groups are particularly dependent on initial joint score. DISCUSSION The results of this study provided new information with which to address 2 important questions regarding elderly-onset rheumatoid arthritis. The first question is the longstanding one of whether elderlyonset rheumatoid arthritis is a substantially different disease than that found in younger people. Our results revealed that there are important differences in presentation and outcome when these 2 groups are compared. The second question is whether these differences between EORA and YORA are solely due to the fact that much of what is diagnosed as elderly-onset RA is actually polymyalgia rheumatica, a disease with a substantially more favorable outcome than rheumatoid arthritis. Our findings were not consistent with this view, and tended to support the theory that elderly-onset rheumatoid arthritis is a reasonably distinct entity. The issue of whether RA is different when it appears in different age groups has been debated since Cecil first claimed that RA in the aged did not differ essentially from that seen in younger patients (13). Subsequent studies, however, have tended to support the notion that there are differences between elderlyonset and earlier-onset disease. Differences in the frequencies of such features as abrupt onset, sex ratios, proximal joint involvement, and erythrocyte sedimentation rate have been noted (10-12). The researchers, however, have disagreed on the prognostic implications of elderly-onset disease. Most studies of EORA have not included a direct YORA comparison group and thus their conclusions are open to serious question. However, 2 of them have, as in the present study, involved a YORA comparison group (9,lO). Both used age 60 as a criterion for EORA, and both concluded that there were differences in presentation between EORA and YORA and a more favorable prognosis in the former. In each of these studies, however, there was a disconcerting difference in disease duration between the 2 groups, with YORA patients having substantially longer durations. The more favorable outcome of the older group may have been primarily due to this factor alone. The present study provides stronger evidence of the favorable outcome in EORA by using a direct comparison group with similar disease duration, by more directly addressing the potential for misdiagnosis of PMR as EORA, and by analyzing the impact of selected factors in a multivariate analysis. Our study clearly indicated that there are important differences in the presenting symptoms and in the initial physical and laboratory findings in patients with EORA. Our results confirm the finding that abrupt disease onset is substantially more frequent in elderly-onset disease and that large joint disease, particularly of the shoulder girdle, tends to be more common. In contrast, small joint disease is significantly more frequent in the younger-onset group. The younger patients, in general, have a condition that more closely resembles classic rheumatoid arthritis, with small joint disease, subcutaneous nodules, and positive latex fixation titers. A substantial proportion of the elderly-onset group have features which are similar to those found in polymyalgia rheumatica. Our results support the argument that the outcome of elderly-onset rheumatoid arthritis is substantially better than that of disease beginning at a younger age. In our large group, better outcomes for the elderly were found in all parameters including symptoms, joint evaluations, erythrocyte sedimentation rate changes, and overall assessment by both physicians and patients. These more favorable outcomes occurred despite the fact that the 2 groups were treated in a highly similar fashion in terms of drug, surgery, and other therapies. Why does EORA seem to have a more favorable prognosis? The presence of rheumatoid factor has been a fairly consistent predictor of more severe disease in patients with rheumatoid arthritis when compared with patients without RF (2,3). Fewer EORA patients had seropositive disease (48%, versus 76% of YORA

7 ELDERLY-ONSET AND YOUNGER-ONSET RA 993 patients; P = ), but analysis using rheumatoid factor as a variable revealed that it did not account for a better outcome. Although females have been shown to have less favorable outcomes (4), sex ratios were not substantially different in our study. Race also did not differ significantly between the 2 groups. Abrupt onset, which occurred more frequently in those with EORA, did not account for a better prognosis. There are conflicting findings in the literature on the effect of abrupt onset on disease outcome. Corrigan found that 26% of EORA patients had an abrupt onset with severe constitutional symptoms and that all experienced disease remission within 18 months (8). Terkeltaub (lo), in contrast, found no significant joint count differences between patients with acute and those with nonacute disease in either EORA or YORA groups. Both studies used 48 hours or less as the dividing point between acute and nonacute disease, whereas the present study used 2 weeks. These digerences in definition may account for some of the disagreement about the effect of this factor on outcome. Our findings confirm the notion that features of PMR are quite common in EORA (14,19-24), but we believe that the patients with these features in this study had true rheumatoid arthritis with features of polymyalgia, rather than having PMR which was confused with rheumatoid arthritis. Diagnosis was made by rheumatologists, each subject met the ARA criteria for definite or classic RA, no patient had symptoms of temporal arteritis or required a temporal artery biopsy during an average disease duration of 5 years, and the percentage of patients with PMR features who were treated with steroids during this period was not significantly different from the percentage of other patients so treated (74% versus 62%; P = 0.33). Most importantly, the presence of PMR features had minimal impact on outcome when we controlled for age at onset. If a substantial number of these elderly individuals actually had PMR with articular manifestations, one would expect the better outcome to be explained in large part by the presence of the PMR features. This did not prove to be the case. Separate outcome analyses in which all subjects with PMR presentations were excluded showed nearly identical findings regarding the differences between the EORA and YORA groups. The effect of PMR features on outcome found in this study is consistent with the interpretation that elderly-onset RA is a relatively distinct subset of this disease which frequently exhibits PMR-like features and which tends to have a better prognosis. This is not to say that all or even most elderly patients who present with a combination of RA and PMR features actually have EORA. The clinician must remember that PMR is a frequent rheumatic problem in the elderly and that it may mimic RA in some cases. In summary, the study has reviewed the clinical features and outcome of patients with elderly-onset rheumatoid arthritis, and has compared them with a younger-onset group with similar disease duration. Our findings support the theory that EORA is a definable clinical subset of rheumatoid arthritis which differs in initial clinical features and subsequent outcome from that found in younger subjects. Age at onset appears to be an important determinant of prognosis in RA, and this is not solely attributable to the possible effects of abrupt onset or PMR-like disease. REFERENCES 1. Ragan C, Farrington E: The clinical features of rheumatoid arthritis. JAMA 181: , Cats A, Hazevoet HM: Significance of positive tests for rheumatoid factor in the prognosis of rheumatoid arthritis. Ann Rheum Dis 29: , Mongan ES, Atwater EC: Comparison of patients with seropositive and seronegative rheumatoid arthritis. Med Cliii North Am , Feigenbaurn SL, Masi AT, Kaplan SB: Prognosis in rheumatoid arthritis: a longitudinal study of newly diagnosed younger adult patients. Am J Med 66: , Masi AT, Maldonado-Cocco JA, Kaplan SB, Feigenbaum SL, Chandler RW: Prospective study of the early course of rheumatoid arthritis in younger adults: comparison of patients with and without rheumatoid factor positivity at entry and identification of variables correlating with outcome. Semin Arthritis Rheum 5: , Alarc6n GS, Koopman WJ, Acton RT, Barger BO: Seronegative rheumatoid arthritis: a distinct immunogenetic disease? Arthritis Rheum , Terkeltaub R, DeCary F, Esdaile J: An immunogenetic study of older age onset rheumatoid arthritis. J Rheumato1 11: , Corrigan AB, Robinson RG, Terenty JR, Dick-Smith JB, Walters D: Benign rheumatoid arthritis of the aged. Br Med J 1: , Ehrlich GE, Katz WA, Cohen SH: Rheumatoid arthritis in the aged. Geriatrics 25: , Terkeltaub R, Esdaile J, Decary F, Tannenbaum H: A clinical study of older age rheumatoid arthritis with comparison to a younger onset group. J Rheumatol 10: , 1983

8 994 DEAL ET AL 1 1. Moesmann G: Clinical features in subacute rheumatoid arthritis in old age. Acta Rheum Scand 14: , Brown JW, Somes DA: The onset of rheumatoid arthritis in the aged. J Am Geriatr SOC 10: , Cecil RS, Krammerer WH: Rheumatoid arthritis in the aged. Am J Med 13: , Ropes MW, Bennett GA, Cobb S, Jacox R, Jessar RA: 1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 9: , Healey LA: Polymyalgia rheumatica and the American Rheumatism Associatisn criteria for rheumatoid arthritis. Arthritis Rheum 26: , Steinbrocker 0, Traeger CH, Batterman RC: Therapeutic guidelines for rheumatoid arthritis. JAMA 140: , Cupples LA, Heeren T, Schatzkin A, Colton T: Multiple testing of hypotheses in comparing two groups. Ann Intern Med 100: , Cochran WG: Planning and Analysis of Observational Studies. New York, John Wiley & Sons, 1983, pp Dimant J: Rheumatoid arthritis in the elderly, presenting as polymalgia rheumatica. J Am Geriatr SOC 27: , Ehrlich GE: Polymyalgia rheumatica. JAMA , Chuang TY, Hunder GG, Ilstrup DM, Kurland LT: Polymyalgia rheumatica: 9 10-year epidemiologic and clinical study. Ann Intern Med 97: , Bruk MI: Articular and vascular manifestations of polymyalgia rheumatica. Ann Rheum Dis 26: , O Duffy JD, Wahner HN, Hunder GG: Joint imaging in polymyalgia rheumatica. Mayo Clin Proc , O Duffy JD, Hunder GG, Wahner HW: A follow-up study of polymyalgia rheumatica: evidence of chronic axial synovitis. J Rheumatol 7: , 1980