What Do Self-Administered Joint Counts Tell Us about Patients with Rheumatoid Arthritis?

Transcription

1 What Do Self-Administered Joint Counts Tell Us about Patients with Rheumatoid Arthritis? Agusti n Escalante Objective. This report presents data from two sources showing that a self-administered joint count (SAJC) suitable for use in clinical settings provides information comparable with that of observerassessed joint counts. Methods. Patients were tested with a 1-page form containing a 40-joint mannequin on which they could mark their painful or swollen joints. The first sample of 110 patients was used to compare the SAJC with the tender or swollen joint counts (TJC or SJC) performed by a rheurnatologist and to a battery of clinical and laboratory measurements. The second sample consisted of 240 rheumatoid arthritis (RA) patients enrolled in a cohort study of RA outcomes, in whom the relationship between the SAJC and health-related quality of life measures was examined. Presented in abstract form at the 59th National Scientific Meeting of the American College of Rheumatology, San Francisco, CA, October 21-26, Supported in part by the Mexican American Medical Treatment Effectiveness Research Center, funded by Agency for Health Care Policy and R(:search grant # 1-U01-HS Dr. Escalante s work supported hy an award from the Robert Wood Johnson Minority Medical Faculty Development Program. Augustin Escalante, MD, Division of Clinical Immunology and Klieumatolog,y, The University of Texas Health Science Center at San Antonio. Address correspondence to Agustin Escalante, MD, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive. San Antonio, TX Submittcd for publication August 1, 1997; accepted in revised form November 25, by the American College of Rheumatology. 280 Results. Test-retest reliability of the SAJC was excellent (ri = 0.89), as was its agreement with the observer-assessed TJC (rj = 0.78). The SAJC was significantly correlated (P ) to pain on a 10- point scale (r = 0.33), the McGill Pain Questionnaire (r = 0.27), the pain subscale of the Arthritis Impact Measurement Scales (AIMS) (r = 0.32), the duration of morning stiffness (r = 0.27), and to the AIMS subscafes of physical function (r = 0.201, impact (r = 0.31), and global health (r = 0.29). The SAJC was inversely related to formal education (r = -0.32), but did not correlate significantly with the modified Health Assessment Questionnaire, walking velocity, grip strength, or erythrocyte sedimentation rate. The responsiveness of the SAJC was comparable with that of other measures commonly employed to assess RA outcomes. Either the SAJC or the TJC could be included alternatively in multivariate models to explain 7 of the 8 subscales of the Medical Outcomes Study Short Form-36 (SF-36) questionnaire. Conclusion. The SAJCis a reliable and responsive measure that agrees highly with the observerassessed TJC and is significantly associated to the health-related quality of life of patients with RA. Given its low cost and ease of administration, it is suggested that SAJC be included in future studies of RA outcome in routine clinical practice. Key words. Rheumatoid arthritis; Disease activity; Self-report; Articular index; Clinimetrics. INTRODUCTION The use of standard self-report questionnaires has been recognized as an important advance in rheuma /98/$5.00

2 Arthritis Care and Research Self-Administered Joint Count in RA 281 tology (1). A variety of these instruments is available for the study of rheumatoid arthritis (RA) and other rheumatic diseases, including the Arthritis Impact Measurement!kales (AIMS; 2) and the Stanford Health Assessment Questionnaire (3), among others. Condensed versions of some of these questionnaires have been developed as well, partly to encourage clinicians to employ them and to facilitate large scale observational studies of the outcome of RA in routine clinical practice (4). Most brief self-assessment questionnaires for the measurement of arthritis outcomes have focused on a patient s pain, attitudes, and ability to perform functional activities (2,3). Because the joints are the immediate target of the various forms of arthritis, there has also been interest in eliciting information about the status of joints from the patients themselves (5-12). IOf the studies published to date on self-administered joint counts, most have been able to demonstrate agreement between self-administered and observer-derived assessments of joint status (5,6,8,9,11), but some have not (1O,l2). In this study, 1 present inforination from two patient data sets in which a 1-page form was used for patients to mark their painful or swollen joints on a 40-joint mannequin. In the first data set, which was developed expressly to examine the properties of the selfadministered joint count (SAJC), the patients answers were compared with the joint exam performed by a rheumatologist and with concurrent measures of pain and functional status. The second patient data set consisted of the baseline evaluation of a cohort of RA patients enrolled in a longitudinal assessment of RA outcomes. This data set was employed to evaluate the associations between either the SAJC or the tender joint count (TJC) and measures of health-related quality of life (HRQOL). PATIENTS AND METHODS Patients. First data set. Consecutive patients with RA according to the criteria of the American College of Rheumatolagy (formerly the American Rheumatism Association; 13) were invited to participate in a study to validate a 40-joint self-administered articular mannequin. All patients were attending a public rheumatology referral clinic located at the University Health Center in San Antonio, Texas, for a scheduled physician visit between September 1993 and August These patients were asked to attend an outpatient clinical research center where all of the study assessments were conducted. Second data set. The second data set comprised patients meeting criteria for RA who were consecutively enrolled in a longitudinal study of RA outcomes. Baseline evaluations of this RA patient cohort were conducted between January and December 1996 in the outpatient rheumatology clinics of the University Health Center and the Audie L. Murphy Veterans Administration Hospital, both of them public facilities, and at the university clinic for patients with private medical insurance. Study procedures. Data from patients was collected in a similar fashion in both data sets presented, but the instruments utilized and personnel performing the observer-derived joint counts differed slightly. On arrival at the facility where the study was being conducted, patients in both data sets were handed a 1-page self-assessment questionnaire for completion in the waiting room. Patients were then assessed by a trained research nurse who administered a battery of standardized questionnaires and performed upper and lower extremity functional assessments. In the first data set, which was collected to validate the SAJC, articular examinations were conducted by the author, who is a rheumatologist; in the second data set, these evaluations were performed by a trained research nurse. Patients were assessed for the presence or absence of tenderness on pressure or motion and swelling in the following 48 joints: 2 temporomandibular (TMJ), 2 sternoclavicular (SC), 2 acromioclavicular (AC), 2 shoulders, 2 elbows, 2 wrists, 10 metacarpophalangeal (MCP), 2 interphalangeal (IP) of thumb, 8 proximal interphalangeal (PIP), 2 hips (tenderness only), 2 knees, 2 ankle/subtalar, and 10 metatarsophalangeal (MTP). Agreement between rheumatologist and research nurse in recording joint findings was measured after 3 training sessions in a sample of 10 patients with RA who were examined by both observers. Intraclass correlations equaled 0.88 for the tender joint count and 0.84 for the swollen joint count. A subsample of 20 consecutive patients in the first data set were asked to complete a second selfassessment questionnaire after the interviews for evaluation of its test-retest reliability. All patients in the first data set were asked to return for a second evaluation 3 months after the initial one to measure the self-assessed instrument s ability to detect change. Instruments. All patients in both data sets completed a self-administered questionnaire that contained the 8 items of the Modified Health Assessment Questionnaire (2), a question about the duration of morning stiffness, and a 10-point vi-

3 282 Escalante Vol. 11, No. 4, August 1998 Shoulder Shoulder Toe joints Knuckles Figure 1. Joint mannequin employed in this study. Patients were asked to mark involved joints using the following format: Please look at the figure below, and mark (X) the joints that are painful or swollen at this time. sual analog scale. The joint mannequin for selfassessment was placed on the reverse side of the 1-page sheet. It consisted of a professionally drawn frontal outline of the human figure, with hands and feet magnified and 40 joints circled. The circled joints were 2 shoulders, 2 elbows, 2 wrists, 10 MCP, 2 IP, 8 PIP, 2 knees, 2 ankles, and 10 MTP (Figure 1). Forty joints were included in the selfassessed joint count instead of the 48 joints included in the observer-assessed joint counts. The SC, AC, TMJ, and hip joints were not included in the self-assessed form because of the possibility that patients could confuse marks to identify these joints with marks for the shoulders (SC or AC), for headache (TMJ), or because swelling would not be expected (hips). Patients were asked to mark on the mannequin the joints that were painful or swollen. Time required by patients to complete the questionnaire ranged from 3 minutes to 10 minutes. A few patients had questions about the side of the body represented by the articular mannequin. Otherwise, patients did not receive any help from the study nurse while completing the self-assessment questionnaires, and most patients were able to complete the instrument without problems. A Spanish language version of the questionnaire was used for Spanish-speaking patients (14). After completion of the self-assessment questionnaire, patients were evaluated by the research nurse using a structured interview format. The nurse noted demographic, socioeconomic, and ethnic characteristics of patients. In the first data set, the research nurse also administered the pain, function, global health, and impact subscales of the AIMS (2) and the full McGill Pain Questionnaire (MPQ; 15). For the second data set, the research nurse administered the Medical Outcomes Study Short Form-36 (SF-36) questionnaire (16), which measures 8 dimensions of health-related quality of life: physical functioning, role-physical, role-emotional, social functioning, bodily pain, mental health, vitality, and general health. Following the research nurse s interview, grip strength was measured using a Jamar hand-held dynamometer (J.A. Preston Corp., Jackson, MI), with results expressed as the mean value of 3 measurements for each hand in kilograms. Walking velocity was measured over a distance of 50 feet and expressed in feet/second. Patients unable to walk were assigned a walking velocity of 0.0 feetlsecond. The Westergren erythrocyte sedimentation rate (ESR) was used as a laboratory measure of disease activity. The most recent hemoglobin level was recorded from the medical records. Stafistical analysis. Descriptive statistics were calculated for all study variables. Paired t-tests and xz were used for comparisons of continuous and categorical variables between the SAJC and observerassessed joint counts. Test-retest reliability and agreement between the SAJC and the observerassessed TJC or swollen joint count (SJC) were measured by means of the kappa statistic for each individual joint (17), and by the intraclass correlation coefficient (Ti) for the total joint counts (18). For all comparisons between self- and observer-assessed joint counts, only the 40 joints shared by both these measures were employed. The concurrent validity of the SAJC was quantified by comparison with other study variables using Pearson correlations. Responsiveness of the variables in the study was measured in the first data set, based on the improvement or worsening in the patients observer-assessed TJC and SJC between the first and second visits (19). Patients were considered to have improved if there was a 50% reduction in both the observer-assessed TJC and the SJC between the first and second visit, and worsened if there was a 50% increase in the same indices in the said time period. Guyatt s Responsiveness Score (20) was then calculated for each of the

4 Arthritis Care and Research Self-Administered Joint Count in RA OTJC msajc y 0.30 w u Y 9 $ Y INDIVIDUAL JOINTS Figure 2. Frequency of involvement for each of 40 joints by the self-administered joint count (SAJC) or the observerassessed tender joint count (TJC). Significant differences were observed in the left (L) knee, left fifth metacarpophalangeal (MCP) joint, right (R) shoulder, and bilateral second through fifth metatarsophalangeal (MTP) joints. PIP = proximal interphalangeal. measured variables, using the formula provided by Deyo et a1 (21): Responsiveness score Mean change (among improved or worsened patients - Standard deviation of change among stable patients Results from the second data set were employed to measure the contributions of either the SAJC or the TJC to the HRQCIL of RA patients. This was done by constructing stepwise multivariate regression models in which the scales of the SF-36 questionnaire were used as dlependent variables, and either the SAJC or the observer-assessed TJC were alternately entered as independent variables together with other nonquestionnaire-derived variables. Variables remained in the multivariate models only if their F statistic was equal to or greater than For these models, variables whose regression coefficients were significantly related to the SF-36 scales are shown, as is the explanato.ry power, or R2, of each model. All analyses were performed on a personal computer using the Stata Statistical Software package (22). RESULTS Patient populations. A total of 110 patients were evaluated in the first, or validation, data set. Seven of these (6.3%) were excluded from the present analysis because they completed the self-assessed joint mannequin inaccurately by placing marks outside of the joint areas, marking joint areas as a group, or in other ways that made scoring of the mannequin impossible. The remaining 103 patients had an average age of 51.2 years (range ); 77% were women, 86.4% were Mexican-American, 7.7% were African-American, and 5.8% were European-American. Spanish was the language of choice in 28.1%, and 77.7% were United States-born. Their average number of years of formal education was 9.4 (median 10, range 0-17). Clinically, 89.3% of the patients were positive for rheumatoid factor, and 55.3% had rheumatoid nodules. Using the Steinbrocker functional classification (23), 9% of the patients were in functional class I, 68% were in class 11, 20% were in class 111, and 3% were in class IV. The second RA patient data set contained 149

5 284 Escalante Vol. 11, No. 4, August ' INDIVIDUAL JOINTS Figure 3. Agreement between self-administered joint count and observer-assessed tender joint count in rating individual joints as affected or not affected. Kappa values above 0.40 were obtained for 23 of the 40 joints. L = left; R = right; MCP = metacarpophalangeal; PIP = proximal interphalangeal; MTP = metatarsophalangeal. women and 91 men. Their mean age was 55.1 years (range ), and their educational level averaged 9.7 years (range 0-17). Seventy-one percent were Mexican-American, 21.7% were European- American, 5 YO were African-American, and 2.1 % were Asian. Positive rheumatoid factor was present in 84.89'0, and nodules were detected in 40.1%. Their Steinbrocker functional classes were as follows: 32.6% were in class I, 47.3% were in class 11, 15.1% were in class 111, and 5% were in class IV. Test-retest reliability of the self-assessed joint count. Test-retest reliability of the SAJC was evaluated in the initial 20 patients enrolled in the first data set. These patients were asked to complete a second self-assessment questionnaire, after the research nurse's and rheumatologist's assessments had been completed. The minimum time interval between the first and the second questionnaires was 90 minutes. The number of joints marked on the first questionnaire was not significantly different from the number marked in the second (mean '-+ SD = versus 12.7? 10.6, P > 0.05), and there was excellent agreement between the two (Ti = 0.89). The test-retest reliability for each individual joint was also high, with kappa values for 31 of the 40 joints included in the mannequin. The joints with lower kappa values included the right second MCP joint (kappa = 0.74), the right second through fifth MTP joints (kappas from 0.30 to 0.68), and the left first through fifth MTP joints (kappas from 0.43 to 0.72). Agreement between self-assessed and physicianassessed joint counts. This analysis included only the joints that were present in both observer- and self-assessed joint counts, and thus excluded the temporomandibular, sternoclavicular, acromioclavicular, and hip joints. The SAJC and the physician-assessed TJC did not differ significantly in terms of the total number of joints involved (17.2 f 10.4 versus 16.0 f 10.0, P > 0.05), and agreement between the two joint indices was correspondingly high (ri = 0.78). The frequency of involvement of individual joints using the SAJC was not significantly different from the TJC in 29 of the 40 joints (72.5%) assessed with both methods (Figure 2). Significant differences between the SAJC and the TJC in the frequency of involvement of individual joints were observed in the left knee (60% versus 44%, xz = 5.62, P < 0.05), left fifth MCP joint (40% versus 2670, x2 = 4.3, P < 0.05),

6 Arthritis Care and Research Self-Administered Joint Count in RA 285 Table 1. Relationship between the self-assessed joint count and concurrent variables in 103 patients with rheumatoid arthritis Vari.able* Demographics Age Disease duration Years of education -0.33t Functional status Steinbrocker functional class 0.09 MHAQ 0.17 AIMS physical function 0.20* Pain measures 10-point pain scale 0.33t McGill Pain Questionnaires AIMS pain 0.32t Global measures Physician global asessment of severity AIMS global arthritis impact 0.31* AIMS global health Morning stiffness Articular indices Tender joint count Swollen joint count Performance measures Grip strength Walking velocity Erythrocyte sedimentation rate * MHAQ = Modified Health Assessment Questionnaire: AIMS = Arthritis Impact Measurement Scales. t P I P Number of words chosen from the McGill Pain Questionnaire. 91 P right shoulder 1(20?'0 versus 50%, x2 = 4.25, P < 0.05), and bilateral second through fifth MTP joints (x2 > 5.71 and P < 0.05 for each of these 8 joints). Agreemlent between patient and physician in rating individual joints as tender or nontender was fair, with kappa values for 23 of the 40 joints (Figure 3). In contrast, there was a significant difference between the SAJC and the physician-assessed SJC ( versus 7.8? 6.0, P ), arid the agreement between the two indices was low (Ti = 0.31). Relationship between self-assessed joint count and nonarticular measures of patient status. Correlations between the SAJC and other variables measured in the study are shown in Table 1. In general, correlations were modest, but they were in the expected direction, and the sample size was sufficient to reach statistical significance for most of them. The strongest correlations observed were with the measures of pain: the 10-point pain scale (r = 0.33, P = ), the number of words selected and the pain r rating index of the MPQ (r = 0.27, P = and r = 0.22, P = 0.03, respectively), and the AIMS pain scale (r = 0.32, P = 0.001). There was also a significant negative correlation between the self-assessed joint count and the number of years of formal education (r = -0.33, P = ). The AIMS general health and the physician-assessed global arthritis severity also correlated significantly with the SAJC, as did the duration of morning stiffness (Table 1). However, there was no correlation with measures of patient function, either questionnaire- or performance-based, nor with the ESR, the only laboratory measure we employed. Responsiveness of self-assessed joint count. All of the 56 patients in the first data set who showed up for their followup evaluation were able to complete the 40-joint mannequin correctly. Four of them (7.3%) had a 50% or greater improvement in their physician-assessed TJC and in their SJC, and 5 (9.1%) had a 50% worsening in both of these variables. The remaining 47 (83.9%) were considered stable because they had less than 50% improvement or worsening in their physician-assessed TJC and SJC. The magnitude of the change in the study variables among improved, stable, and worsened patients is shown in Table 2, as is Guyatt's Responsiveness Score for each one. Most of the variables included in this analysis showed a satisfactory ability to detect improvement or worsening in a patient's clinical condition, as defined in this study. For example, patients whose conditions were stable as defined by their TJC and SJC had a change of only t 14.5 kilograms in grip strength, while worsened patients had a change of kilograms. The Guyatt score for detecting worsening grip strength ( ) thus equals Since the observer-assessed TJC and the SJC were the variables employed to define clinical change, their responsiveness scores are the highest. The only variable with a low responsiveness score is the walking velocity, which had a poor performance in responding to improved clinical condition (Table 2). Associations between articular counts and health-related quality of life. The second patient data set was used to examine the relative contribution of the SAJC to the HRQOL of RA patients, compared with the contribution of the observer-assessed TJC (Table 3). Separate multivariate models were constructed entering either the SAJC or the observerassessed TJC as independent variables simultaneously with age, sex, European-American versus non-european-american ethnicity, years of formal

7 286 Escalante Vol. 11, No. 4, August 1998 Table 2. Responsiveness of study variables among 56 patients with rheumatoid arthritis Change between first and second visit* Variable1 Stable Worsened Guyatt Improved Guyatt (n = 47)$ (n = 5)* scores (n = 4)* scores Articular indices Observer-assessed Tender joints Swollen joints Self-assessed Painful or swollen joints Other self-assessed variables MHAQ 10-point pain scale Morning stiffness Performance tests Grip strength (kg) Walking velocity (ftlsec) AIMS scales Physical function Pain Impact Global health -0.1 t t t t t Z t ? t t ? ? t ? ? ? t t ? t t t t f 4.5 * 50% increase or decrease in both number of tender and number of swollen joints. t MHAQ = Modified Health Assessment Questionnaire: AIMS = Arthritis Impact Measurement Scales. * Values shown are means 5 standard deviations. 5 Mean change in improved or worsened patients standard deviation of change among stable patients pi Numbers of tender or swollen joints ? t t t t t t t t ? ? ? education, body mass index, Steinbrocker functional class, grip strength, walking velocity, and hemoglobin level. These nonquestionnaire-derived measurements were selected as co-variables to avoid contamination with the questionnaire-derived SF-36 subscales. Each of the 8 scales of the SF-36 questionnaire was used as a dependent variable in 16 separate multivariate models. Significant coefficients in these models are shown in Table 3. The relative contribution of the SAJC was similar to that of the observer-assessed TJC in 7 of the 8 subscales of the SF-3 6 questionnaire. For example, in a model including the SAJC, a significant negative association was observed between the physical functioning subscale of the SF-36 and the SAJC (regression coefficient = -0.32, [95% confidence interval = to -0.09]), the Steinbrocker functional class (-8.29, [-12.35, -4.23]), the grip strength (0.58 [0.28, 0.89]), and the walking velocity (5.27 [2.36, 8.18]), with this model explaining 42% of the variance in physical functioning (Table 3). If the observer-assessed TJC was substituted in place of the SAJC in the same model, the TJC remained significant (-0.36 [-0.58, -0.15]), as did the Steinbrocker class (-8.31 [-12.32, -4.29]), the grip strength (0.45 [0.12, 0.781), and the walking velocity (5.33, L2.47, 8.19]), but the variance ex- plained increased to 44% (Table 3). For all models, the regression coefficients of the SAJC tended to be smaller than the coefficients of the observer-assessed TJC, as did the explanatory power of the models containing the SAJC. The only discrepancy between the SAJC and the observer-assessed TJC in explaining SF-36 subscales was for mental health, with which TJC was significantly associated, while the SAJC was not. DISCUSSION The information presented demonstrates 4 aspects of the performance of the SAJC. First, the SAJC agrees highly with observer-derived measures of joint tenderness and less so with joint swelling, as recorded by a rheumatologist. Second, the SAJC is highly reproducible, suggesting that it is a reliable clinical measure. Third, the SAJC responds satisfactorily to clinically significant changes in joint status and does so in a way that is comparable with the responsiveness of other variables commonly employed to assess patients with RA. Fourth, the SAJC and the TJC are both related to the HRQOL of RA patients in similar ways. The above information pro-

8 N 01 U Table 3. Multivariate regression models showing associations of self- or observer-assessed joint counts with Medical Outcomes Study Short Form-36 (SF-36) health-related quality of life scales Dependent variablest Independent Social variables* Physical function Role-physical Role-emotional functioning Bodily pain Mental health Vitality General health Self Observer Self Observer Self Observer Self Observer Self Observer Self Observer Self Observer Self Observer Articular index Age Sex White ethnicity Formal education BMIS Functional class Grip (lbs) Velocity (ft/sec) Hemoglobin R' * Variables shown were entered in a backward stepwise selection process to examine their combined association with SF-36 subscales. t Each of the 8 subscales of the SF-36 questionnaire was used as a dependent variable in separate regression models. Either the self-assessed joint count or the observer-assessed tender joint count were alternatively entered in separate regression models. Only the values of regression coefficients significant at P are shown. $ BMI = body mass index (weight in kilograms + height in meters'). Steinbrocker functional class.

9 288 Escalante Vol. 11, No. 4, August 1998 vides evidence of the satisfactory performance of the SAJC as a clinical outcome measure for RA patients. Although this study revealed very good agreement between physician and patient in rating the sum of the joints as tender or painful, important discrepancies at the level of individual joints were noted, such as the right shoulder and the second through fifth MTP joints bilaterally. It is possible that the right shoulder was more likely than subsequent joints to be confused with the contralateral side, because patients asked for clarification on the sides of the mannequin only after they had begun completing the questionnaire. To any reader wishing to use the mannequin in Figure 1, I suggest marking clearly the left and right sides. Disagreements in the MTP joints may be due to the small size of these joints, making them more difficult to localize by patients; also, they are perhaps more susceptible to tenderness on pressure than other joints. Counting the number of painful, tender, swollen, or otherwise involved joints is a traditional way of assessing the status of patients with RA (24,25). Recent efforts to refine the clinical methods of assessing rheumatoid joints have consisted mainly of reducing the number of joints to assess (26). The ultimate goal of these efforts is to reduce the amount of time necessary to perform the joint counts, rendering feasible the collection of this information from busy clinical environments. This would permit the use of information from these clinical settings to generate information about arthritis and its outcome (1). An alternative method of gathering information about patients joints is to bypass physicians, nurses, and other trained observers altogether and go directly to the patient. This approach has been very successful in generating data about patients functional status (27), attitudes (28), and pain (29), among other outcomes. Prior studies have, in fact, examined the use of self-administered questionnaires to assess patients joints. The first of these was the study by Stewart and colleagues showing that patients could accurately assess the pain or inflammation of their own joints on a self-administered mannequin (6,7). Stewart and colleagues study was followed by reports by Mason et a1 (5), Abraham et a1 (8), Stucki et a1 (9), Hewlett et a1 (lo), Hanly et a1 (12), and Prevoo et a1 (11). In general, these studies have all shown fair to good levels of agreement between patient-assessed and physician- or trained observer-assessed joint tenderness, but lower levels of agreement for joint swelling (12). The studies that have compared a self-assessed joint count with laboratory measures of inflammation have had mixed results, with one study showing significant associations between the two (9), but others showing no correlation (6,10,11). The present analysis echoes most prior studies in that the SAJC agreed highly with the physician-assessed TJC, less so with the SJC, and not at all with the ESR. The poor association found in most studies between self-assessed joint counts and laboratory tests such as the ESR or C-reactive protein has prompted some investigators to question the usefulness of self-report articular indices for assessing RA patients (10). However, it should be noted that in those studies, as in the present one, the observer-assessed joint counts did not correlate well with laboratory measures either (6,10,11). One contribution of the present study is the measure of responsiveness of the SAJC and other variables provided by Guyatt s Responsiveness Score, which measures the magnitude of change in a given variable among improved (or worsened) patients, compared with the normal variation in the same variable among stable patients (20). To calculate the responsiveness score, a strict definition of clinical change was used, similar to definitions employed in controlled clinical trials (19). To be considered improved or worsened, a patient must have a 50% or greater reduction, or increase, both in the number of tender and in the number of swollen joints, as measured by a rheumatologist. Using these criteria, the SAJC compared favorably with other variables commonly employed in RA to measure outcome by detecting improvement or worsening in the clinical status of patients (Table 2). One advantage of Guyatt s Responsiveness Score is that it permits estimates of the sample size requirements to detect a significant change with any given instrument, if that instrument s responsiveness score is known, using a published table (19). Using Guyatt s table, only 10 to 20 patients would be needed to reflect a significant change in articular indices using the SAJC (20). What do self-assessed joint counts tell us about RA patients? It seems clear that these self-report indices reflect articular pain and/or tenderness much better than they reflect joint swelling. The tender joint count is an important, well-established measure of disease activity in RA (25), which has been employed in numerous clinical trials (19) and is at the top of the list of the American College of Rheumatology core set of disease activity measures in RA (30). The ability to measure a variable that closely parallels the TJC using a simple 1-page self-assessed form opens the possibility of recording this variable in large-scale studies of the outcome of RA in routine

10 Arthritis Care and Research Self-Administered Joint Count in RA 289 clinical settings, without the need for specialized personnel. It is doubtful that the SAJC will ever take the place of careful physician- or trained observer-derived articular assessments in controlled clinical trials where there are usually sufficient resources to justify the time required for a detailed observer-derived articular assessment. Likewise, it is unlikely that any self-assessed measure will ever substitute for an experienced clinician s assessment in the diagnosis and management of RA. Rather, the proper place for the SAJC presented in this article, and other similar self-assessed articular scores, is in the busy, highvolume clinic, where the time of the trained personnel who would ordinarily perform a formal joint count is at a premium. In such a setting, a brief Self-administered questionnaire such as the SAJC can enable the recording of an important disease outcome (1,4,27). If a formal joint count is still desired, but resources are limited, trained personnel can concentrate on measuring swollen joints, which may be more important in determining joint destruction in RA, while joint tenderness is monitored with the SAJC or a similar self-administered measure. The SAJC would thus add to the brief self-administered instruments that are already widely used to study RA outcomes. Some limitations of the present analysis are acknowledged a priori: most of the analyses presented are of a cross-sectional nature, and the only longitudinal data shown stems from the one-half of the patients in the first data set who attended a followup visit. Furthermore, no specific therapeutic interventions were given, and thus the clinical changes measured in the first patient data set reflect spontaneous variation in clinical status among RA patients. However, it could be argued that a therapeutic intervention would result in clinical changes of a greater magnitude, which would probably enhance the performance of the SAJC measuring change. Finally, the majority of our patients were recruited from publicly funded clinics, and thus their education and other socioeconomic indicators are low, and many do not speak English. It is possible that different results, perhaps even more encouraging that the ones presented, would be obtained with patients of higher socioeconomic status. In summary, thle present study has shown that reliable information about a RA patient s joints can be obtained with a 3-page form containing a mannequin on which the patients mark their painful or swollen joints. This information agrees strongly with the tender joint count, but not with the swollen joint count, both performed by a rlieumatologist, and correlates signifi- cantly with standardized measures of pain, functional status, and HRQOL. Preliminary evidence of the selfassessed joint count s responsiveness to change over time has been presented as well. The author acknowledges research nurses Rachel Villarreal and Sabrina Smith for their assistance with patient rec:riiitment, administration of questionnaires, functional assessments, and articular evaluations; Drs. Michael Fischbach, Cindy Mulrow, Helen Hazuda, and Michael Lichtenstein from the author s institution, and Dr. C. Kent Kwoh from Case Western Reserve University tor their useful comments about the manuscript to1 1990;17: REFERENCES Wolfe F, Pincus T. Standard self-report questionnaires in routine clinical and research practice-an opportunity for patients and rheumatologists. J Rheumatol 1991 ;18: Meenan RF, Gertman PM, Mason JH. Measuring health status in arthritis: the Arthritis Impact Measurement Scales. Arthritis Rheum 1980; 23 : Fries JF, Spitz P, Kraines RG, Holnian HR. Measurement of patient outcome in arthritis. Arthritis Rheum 1980;2 3: Pincus T, Summey JA, Soraci SA Jr, Wallston KA, Hummon NP. Assessment of patient satisfaction in activities of daily living using a modified Stanford Health Assessment Questionnaire Arthritis Rheum 1983;26: Mason JH, Anderson JJ, Meenan RF, Haralson KM, Lewis-Stevens D, Kaine JL. The Rapid Assessment of Disease Activity in Rheumatology (RADAR) questionnaire: validity and sensitivity to change of a patient self-report measure of joint count and clinical status. Arthritis Rheum 1992; 35: Stewart MW, Palmer DG, Knight RG. A self-report articular index measure of arthritic activity: investigations of reliability, validity and sensitivity. J Rheuma- Stewart MW, Palmer DG, Knight RG, Highton J. A self-report articular index: relationship to variations in mood and disease activity measures. Br J Rheumatol 1993;32: Abraham N, Blackmon D, Jackson JR, Bradley LA, Lorish CD, Alarc6n GS. Use of self-administered joint counts in the evaluation of rheumatoid arthritis patients. Arthritis Care Res 1993;6: Stucki G, Stucki S, Briihlman P, Maus S, Michel BA. Comparison of the validity and reliability of self-reported articular indices. Br J Rheuniatol 1995;34: Hewlett SE, Haynes J, Shepstone L, Kirwan JR. Rheumatoid arthritis patients cannot accurately report signs of inflammatory activity. Br J Rheumatol 1995; 34:

11 290 Escalante Vol. 11, No. 4, August Prevoo MLL, Kuper IH, van t Hof MA, van Leeuwen MA, van de Puite LBA, van Riel PLCM. Validity and reproducibility of self-administered joint counts. A prospective longitudinal followup study in patients with rheumatoid arthritis. J Rheumatol1996;23: Hanly JG, Mosher D, Sutton E, Weerasinghe S, Theriault D. Self-assessment of disease activity by patients with rheumatoid arthritis. J Rheumatol 1996;23: Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988; 31 : Escalante A, Galarza-Delgado D, Beardmore TD, Baethge BA, Esquivel-Valerio J, Marines AL, et al. Cross-cultural adaptation of a brief outcome questionnaire for Spanish-speaking arthritis patients. Arthritis Rheum 1996;39: Melzack R. The McGill Pain Questionnaire: major properties and scoring methods. Pain 1975;1: Ware JE Jr. SF-36 health survey: manual and interpretation guide. Boston: The Health Institute, New England Medical Center; Fleiss JL. Measuring agreement between two judges on the presence or absence of a trait. Biometrics 1975;31: Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psycho1 Bull 1979;86: Paulus HE, Egger MJ, Ward JR, Williams HJ, and the Cooperative Systematic Studies of Rheumatic Disease Group. Analysis of improvement in individual rheumatoid arthritis patient treated with disease-modifying antirheumatic drugs, based on the findings in patients treated with placebo. Arthritis Rheum 1990;33: Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis 1987;40: Deyo RA, Diehr P, Patrick DL. Reproducibility and responsiveness of health status measures: statistics and strategies for evaluation. Control Clin Trials 1991; 12 : 142 S-58s. 22. Stata Statistical Software. Release 4.0. College Station (TX): Stata; Steinbrocker 0, Traeger CH, Batterman RC. Therapeutic criteria for rheumatoid arthritis. JAMA 1949;140: Lansbury J, Haunt DD. Quantitation of the manifestations of rheumatoid arthritis. 4. Area of joint surfaces as an index to total joint inflammation and deformity. Am J Med Sci 1956;232: Ritchie DM, Boyle JA, McInnes JM, Jasani MK, Dalakos TG, Grieveson P, et al. Clinical studies with an articular index for the assessment of joint tenderness in patients with rheumatoid arthritis. QJM 1968;37: Fuchs HA, Brooks RH, Callahan LF, Pincus T. A simplified twenty-eight-joint quantitative articular index in rheumatoid arthritis. Arthritis Rheum 1989;32: Pincus T, Callahan LF, Brooks RH, Fuchs HA, Olsen NJ, Kaye JJ. Self-report questionnaire scores in rheumatoid arthritis compared with traditional physical, radiographic and laboratory measures. Ann Intern Med 1989;110:259-66, 28. Nicassio PM, Wallston KA, Callahan LF, Herbert M, Pincus T. The measurement of helplessness in rheumatoid arthritis. The development of the Arthritis Helplessness Index. J Rheumatol 1985;12: Callahan LF, Brooks RH, Summey JA, Pincus T. Quantitative pain assessment for routine care of rheumatoid arthritis patients, using a pain scale based on activities of daily living and a visual analog pain scale. Arthritis Rheum 1987;30: Felson DT, Anderson JJ, Boers M, Bombardier C, Chernoff M, et al. The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. Arthritis Rheum 1993;36: