RAPID3 (Routine Assessment of Patient Index Data 3) severity categories and response criteria: Similar results to DAS28 (Disease Activity Score) and CDAI (Clinical Disease Activity Index) in the RAPID 1 (Rheumatoid Arthritis Prevention of Structural Damage) clinical trial of certolizumab pegol†
Dr. Keystone has received consultancy fees and/or honoraria (less than $10,000 each) from Abbott, Biotest, Bristol-Myers Squibb, Centocor, Genentech, Merck, Nycomed, and Pfizer (more than $10,000 each), from Amgen, AstraZeneca, F. Hoffmann-LaRoche, and UCB, and has received research funding from Abbott, Amgen, AstraZeneca, Bristol-Myers Squibb, Centocor, F. Hoffmann-LaRoche, Genzyme, Merck, Novartis, Pfizer, and UCB.
New York University School of Medicine and New York University Hospital for Joint Diseases, New York
Dr. Bergman has received consultancy fees, speaking fees, and/or honoraria from Genentech, Pfizer, Takeda, and UCB (less than $10,000 each) and from Abbott (more than $10,000) and research funding from Abbott and Bristol-Myers Squibb.
To compare categories for activity/severity according to the Disease Activity Score 28-joint count (DAS28), the Clinical Disease Activity Index (CDAI), and the Routine Assessment of Patient Index Data 3 (RAPID3), an index without formal joint counts calculated in 5 versus >100 seconds, as well as the European League Against Rheumatism (EULAR)– DAS28 and the RAPID3 response criteria, in the Rheumatoid Arthritis Prevention of Structural Damage (RAPID 1) clinical trial of certolizumab pegol (CZP).
Post hoc analyses were performed using correlations, cross-tabulations, and kappa statistics. Patients (treated with CZP plus methotrexate [MTX] or placebo plus MTX) were classified at baseline and at 52 weeks as high, moderate, low activity/severity or remission, according to the DAS28 (>5.1, >3.2 to ≤5.1, 2.6 to ≤3.2, <2.6 [total range 0–10]), the CDAI (>22, >10 to ≤22, >2.8 to ≤10, ≤2.8 [total range 0–76]), and RAPID3 (>12, >6 to ≤12, >3 to ≤6, ≤3 [total range 0–30]), as well as for good, moderate, and poor EULAR-DAS28 and proposed RAPID3 response criteria.
All measures were correlated significantly: RAPID3 with DAS28 and CDAI (rho > 0.7), higher than erythrocyte sedimentation rate with C-reactive protein level (rho = 0.47). At 52 weeks, DAS28, CDAI, and RAPID3 low activity/remission was seen in 30%, 44%, and 42% of CZP-treated patients versus 3%, 7%, and 10% of control patients. Good, moderate, and poor EULAR-DAS28 responses were seen in 30%, 51%, and 19% of CZP-treated patients versus 3%, 28%, and 70% of control patients, and for RAPID3 in 39%, 30%, and 32% of CZP-treated patients versus 8%, 16%, and 76% of control patients. Kappa and weighted kappa values ranged from 0.36–0.53, indicating fair to moderate agreement.
RAPID3, DAS28, and CDAI give similar results to distinguish CZP patients from controls in the RAPID 1 clinical trial. DAS28 is specific for clinical trials; RAPID3 appears pragmatically useful for usual care.
Quantitative assessment of rheumatoid arthritis (RA) requires indices, since single measures are not sufficient to characterize clinical status in all individual patients (1). The most widely used index is the Disease Activity Score 28-joint count (DAS28), which includes a 28 tender joint count, a 28 swollen joint count, an acute-phase reactant laboratory test (erythrocyte sedimentation rate [ESR] or C-reactive protein [CRP] level), and a patient global estimate of status, calculated (using a calculator or an excellent web site [http://www.das-score.nl/www.das-score.nl]) according to a complex formula (2, 3). The Clinical Disease Activity Index (CDAI) (4) eliminates a need for complex calculations, as well as a laboratory test, which frequently is unavailable and/or uninformative in many patients (5). Both DAS28 and CDAI require a formal joint count, the most specific measure for RA (6). However, formal joint counts are not performed at most visits to a rheumatologist unless required for clinical research or reimbursement (7).
An index without formal swollen or tender joint counts, the Routine Assessment of Patient Index Data 3 (RAPID3) (8), includes only the 3 patient-reported RA core data set measures: physical function, pain, and patient global estimate. RAPID3 can be calculated in 5–10 seconds on a Multidimensional Health Assessment Questionnaire (MDHAQ) compared to 42 seconds on average for the Health Assessment Questionnaire (HAQ), 90–95 seconds for a formal 28-joint count, 106 seconds for the CDAI, and 114 seconds for the DAS28 (9). RAPID3 is correlated significantly with DAS28 and CDAI in usual care (8), in clinical trials involving adalimumab (10) and abatacept (11), with results similar to these indices to distinguish active from control treatments.
RAPID3 would appear desirable to provide quantitative assessment of patient status in usual clinical care, particularly when a formal quantitative joint count is not performed. Further data concerning associations of RAPID3 compared to DAS28 and CDAI in clinical trials would be informative to clinicians in consideration of collection in usual care. In this study, we analyze the Rheumatoid Arthritis Prevention of Structural Damage (RAPID 1) clinical trial of certolizumab pegol (CZP) plus methotrexate (MTX) versus MTX-only control, to compare patients classified according to 4 categories of disease activity or severity, i.e., high, moderate, and low activity, and remission, for the DAS28, the CDAI, and RAPID3, as well as for categories for RAPID3 versus the recognized European League Against Rheumatism (EULAR)–DAS28 improvement criteria (12).
Significance & Innovations
Routine Assessment of Patient Index Data 3 (RAPID3) is an index of only patient-reported measures, completed by patients in a waiting room and calculated on a Multidimensional Health Assessment Questionnaire by a health professional in 5 seconds versus >100 seconds for the Disease Activity Score 28-joint count (DAS28) and the Clinical Disease Activity Index (CDAI), which require formal joint counts.
RAPID3 is correlated significantly with DAS28 and CDAI in the Rheumatoid Arthritis Prevention of Structural Damage (RAPID 1) clinical trial, including identification of patients who have high, moderate, or low disease activity, or remission at the conclusion of the trial.
The RAPID3 score identifies good, moderate, and poor responses in the RAPID 1 clinical trial similarly to European League Against Rheumatism DAS response criteria.
RAPID3, by providing informative clinical data toward better rheumatologic care and patient outcomes, can overcome the problem that laboratory tests (with many limitations) usually are the only quantitative data in the medical records of most rheumatoid arthritis patients in usual care.
MATERIALS AND METHODS
RAPID 1 clinical trial.
Post hoc analyses were conducted of the RAPID 1 clinical trial. Details of the RAPID 1 trial have been described in previous reports (13). Briefly, 982 patients who had prior incomplete responses to MTX were randomized to treatment with either 200 mg (393 patients) or 400 mg (390 patients) of CZP plus MTX every other week (after initial treatment with 400 mg at weeks 0, 2, and 4) or placebo plus MTX (199 patients) (13). Patients who were nonresponders at both week 12 and week 14, according to the American College of Rheumatology 20% (ACR20) improvement criterion (14), were withdrawn at 16 weeks and had the option of enrolling in an open-label extension study in which they received CZP 400 mg every 2 weeks.
The RAPID 1 trial was conducted at 147 centers worldwide between February 2005 and October 2006. The institutional review board or ethics committee at each participating center approved the study protocol. All patients gave their written consent, and the study was conducted in accordance with the principles of the Declaration of Helsinki. For the post hoc analyses presented here, data sets were analyzed by a single author (KL); all other authors reviewed only aggregate results prepared by the same author (KL).
Measures and indices.
The 7 RA core data set measures, i.e., tender joint count, swollen joint count, physician global estimate of status, ESR, patient self-report of physical function (the disability index [DI] of the HAQ), pain (patient-reported on a pain visual analog scale [VAS]), and patient global estimate, as well as CRP level, and 3 indices, DAS28, CDAI, and RAPID3, were analyzed (Table 1). RAPID3 scores were based on the 20-item HAQ physical function scale rather than the 10-item MDHAQ physical function scale. The DAS28 includes a 28 tender joint count, 28 swollen joint count, acute-phase reactant laboratory test (ESR or CRP level), and a patient global estimate of status (2, 3); the index score is calculated using the formula:
The CDAI includes a 28 swollen joint count, 28 tender joint count, physician global estimate of status, and patient global estimate of status (4); the index is calculated at the site by the rheumatologist or associate. RAPID3 includes only the 3 patient-reported RA core data set measures, i.e., physical function, pain, and patient global estimate of status (8, 11), and the index is calculated at the site by the rheumatologist or associate.
Table 1. Features of the 3 indices (DAS28, CDAI, and RAPID3) that assess patients with rheumatoid arthritis*
DAS28 = Disease Activity Score 28-joint count; CDAI = Clinical Disease Activity Index; RAPID3 = Routine Assessment of Patient Index Data 3; TJC = tender joint count; SJC = swollen joint count; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; PATGL = patient global estimate of status; NA = not available (formal criteria not published).
Core data set measures
0.56 × √TJC28
0.28 × √SJC28
Physician global estimate
ESR or CRP level
0.70 × ln(ESR)
Patient physical function
0.014 × PATGL
>3.2 to ≤5.1
>10.0 to ≤22.0
>6.0 to ≤12.0
2.6 to ≤3.2
>2.8 to ≤10.0
>3.0 to ≤6.0
0 to <2.6
0 to ≤2.8
0 to ≤3.0
(Decrease >1.2 units AND end point score <3.2)
(Decrease >3.6 units AND end point score <6)
(Decrease >1.2 units AND end point score ≥3.2) OR (decrease 0.6–1.2 units AND end point score ≤5.1)
(Decrease >3.6 units AND end point score ≥6) OR (decrease 1.8–3.6 units AND end point score ≤12)
(Decrease <0.6 unit) OR (decrease 0.6–1.2 units AND end point score >5.1)
(Decrease <1.8 units) OR (decrease 1.8–3.6 units AND end point score >12)
The criteria for high, moderate, and low activity and remission, based on the analyses of several candidate values (10), are presented in Table 1 for DAS28 (15), CDAI (4), and RAPID3 (8, 11). In earlier reports, 0–30 RAPID3 scores were recoded to 0–10, dividing by 3. However, the raw 0–30 score requires ∼5 seconds compared to the 10 seconds for a recoded score, and 0–30 is now recommended (16).
Results were calculated for EULAR-DAS28 response categories (15) and compared to the proposed RAPID3 response categories, also based on several candidate values (8, 11) as presented in Table 1. Patients were compared in the 3 categories of the 2 indices.
Spearman's rank order correlations were computed for all 7 RA core data set measures and 3 indices (DAS28, CDAI, and RAPID3) at baseline, 52 weeks, and change from baseline to 52 weeks. The proportion of patients in the categories of high, moderate, low activity, and remission according to DAS28-ESR, CDAI, and RAPID3 were compared at baseline and 52 weeks, using cross-tabulations. Patient responses in the trial according to DAS28 and proposed RAPID3 response criteria were compared with cross-tabulations for the proportion of good, moderate, and poor responses. Only 52-week data are presented in this report; other data indicate similar results. Statistical significance was analyzed using kappa and weighted kappa statistics (17).
Correlation of RA core data set measures and indices.
Spearman's correlations of all RA core data set measures and indices at baseline and at 52 weeks, and the change in values between baseline and 52 weeks, were all statistically significant (P < 0.001). At 52 weeks, all correlation coefficients for all patients were ≥0.21 (for CRP level versus tender joint count and for ESR versus swollen joint count), and all were ≥0.43 for all measures other than laboratory tests (Table 2). RAPID3 was correlated significantly with DAS28 (rho = 0.73) and CDAI (rho = 0.70) (Table 2 and Figure 1), with which 1 of 3 measures (patient global estimate) is shared. DAS28 and CDAI, which share 3 of 4 measures, are correlated at rho = 0.90. The physician global estimate, which is not included in RAPID3 or DAS28, is correlated similarly at rho = 0.75 with RAPID3 and rho = 0.77 with DAS28 (and rho = 0.78 with CDAI, which does include the physician global estimate). The correlations of ESR with CRP level (rho = 0.47), and DAS28-ESR with CRP level (rho = 0.37), were lower than for RAPID3 with DAS28 and CDAI. Correlations of DAS28 and CDAI with patient physical function (reported as HAQ DI) and pain (reported on a pain VAS), neither of which is included in these indices, are rho > 0.5, which reflects correlations of RAPID3 with DAS28 and CDAI.
Table 2. Correlation of rheumatoid arthritis core data set measures and 3 derived indices, DAS28, CDAI, and RAPID3, at 52 weeks in the RAPID 1 clinical trial, in all 982 patients*
DAS28 = Disease Activity Score in 28 joints; CDAI = Clinical Disease Activity Index; RAPID3 = Routine Assessment of Patient Index Data 3; RAPID 1 = Rheumatoid Arthritis Prevention of Structural Damage; TJC28 = tender joint count; SJC28 = swollen joint count; VAS = visual analog scale (for patient self-report of pain); PATGL = patient global estimate of status; DOCGL = physician global estimate of status; HAQ DI = disability index of the Health Assessment Questionnaire (for patient self-report of physical function); CRP = C-reactive protein; ESR = erythrocyte sedimentation rate.
One or more measures are shared between indices compared, or measure is included in index compared.
DAS28 scores for 31% of the CZP patients were classified as high activity, 39% as moderate activity, 13% as low activity, and 18% as remission, compared to 78%, 19%, 1%, and 2%, respectively, in the control patients (Table 3). Values of RAPID3 at 52 weeks of high, moderate, low severity, and remission, respectively, were 28%, 31%, 17%, and 25% in the patients randomized to CZP, compared to 68%, 22%, 7%, and 3% in the control patients. Among all the subjects, DAS28 high, moderate, low activity, and remission were seen in 40%, 35%, 10%, and 14%, respectively, compared to 36%, 29%, 15%, and 20% according to RAPID3 (Table 3).
Table 3. RAPID3 versus DAS28 disease activity/severity categories at 52 weeks in certolizumab subjects, control subjects, and all subjects in the RAPID 1 clinical trial*
Values are the number (percentage). Line and column totals may exceed 100% due to rounding. RAPID3 = Routine Assessment of Patient Index Data 3; DAS28 = Disease Activity Score 28-joint count; RAPID 1 = Rheumatoid Arthritis Prevention of Structural Damage.
Overall, 71% of the patients who had high DAS28 activity had high severity, and 96% had high or moderate severity, according to RAPID3. Furthermore, 64% of the patients classified as in remission by DAS28 were also classified as in remission, and 85% as low severity or remission, according to RAPID3 (Table 3). Kappa and weighted kappa values for all patients were 0.36 and 0.53, respectively, including 0.31 and 0.49 for CZP subjects, and 0.33 and 0.42 for control subjects, with levels indicating fair to moderate agreement (Table 3).
Comparisons of RAPID3 and CDAI.
CDAI high, moderate, low activity, and remission were seen respectively in 31%, 26%, 28%, and 15% of CZP patients at 52 weeks, compared to 79%, 14%, 6%, and 2% of control patients (Table 4). RAPID3 high, moderate, low severity, and remission were seen respectively in 28%, 31%, 17%, and 25% of the CZP patients at 52 weeks, compared to 68%, 22%, 7%, and 3% of control patients (Table 4). Among all patients, high, moderate, low severity, and remission were seen respectively in 40%, 24%, 24%, and 12% of patients according to CDAI, versus 36%, 29%, 15%, and 20% according to RAPID3.
Table 4. RAPID3 versus CDAI disease activity/severity categories at 52 weeks in certolizumab subjects, control subjects, and all subjects in the RAPID 1 clinical trial*
Values are the number (percentage). Line and column totals may exceed 100% due to rounding. RAPID3 = Routine Assessment of Patient Index Data 3; CDA1 = Clinical Disease Activity Index; RAPID 1 = Rheumatoid Arthritis Prevention of Structural Damage.
Overall, 70% of patients classified as high activity by CDAI had high RAPID3 scores, and 94% had high or moderate RAPID3 scores at 52 weeks. Among the patients classified as in remission according to CDAI, 84% were in remission and 99% were in low activity or remission according to RAPID3 (Table 4). Kappa and weighted kappa values for CDAI versus RAPID3 were 0.38 and 0.56 in CZP-treated patients, 0.40 and 0.51 in the control patients, and 0.41 and 0.60 in all patients, indicating fair to moderate agreement (Table 4).
Response criteria at 52 weeks.
EULAR-DAS28 good, moderate, and poor responses were seen, respectively, in 30%, 51%, and 19% of the CZP patients, compared to 39%, 30%, and 32% according to RAPID3 (Figure 2). In the control patients, good, moderate, and poor responses were seen, respectively, in 3%, 28%, and 70% according to DAS28, compared to 8%, 16%, and 76% by RAPID3 (Figure 2). In all patients with DAS28 good responses, 74% had RAPID3 good responses, and 96% had good or moderate RAPID3 responses, while 85% of all patients with poor responses by DAS28 had poor responses, and 98% had poor or moderate responses, by RAPID3 (Figure 2). Kappa values for improvement criteria for all patients, CZP patients, and control patients were 0.41, 0.36, and 0.37, respectively, and weighted kappas were 0.53, 0.46, and 0.46, indicating moderate to good agreement.
The findings in this study confirm evidence from clinical trials of adalimumab (18) and abatacept (11) that RAPID3 scores provide similar information to DAS28 and CDAI at the conclusion of a clinical trial. DAS28-ESR was correlated with RAPID3 at rho = 0.73, compared to rho = 0.37 with log(CRP). The correlation of physician global estimate with DAS28, CDAI, and RAPID3 was high, and virtually identical (rho = 0.75–0.78). The correlations of RAPID3 with DAS28 and CDAI were greater than the correlations seen for ESR versus CRP level. This report extends earlier observations with new information that proposed RAPID3 improvement criteria provide comparable data to EULAR-DAS28 improvement criteria. The data also reconfirm that CZP has substantially greater efficacy than MTX alone in incomplete responders to MTX.
A careful joint examination is required for a diagnosis of RA, and the joint count is the most specific measure to assess status and possible improvement or worsening (19). Therefore, a formal count of tender and swollen joints to incorporate into a DAS28 and/or CDAI appears appropriate in clinical trials to assess efficacy of active versus control treatments. However, formal joint counts are not performed at most usual care visits to a rheumatologist (7), and RAPID3 on an MDHAQ can be scored in 5 seconds compared with the 100 seconds it takes for a DAS28 or CDAI, which is a substantial difference (16). The data in this report suggest that RAPID3 may be quite adequate for usual clinical care.
Some patients clearly are “misclassified” in different categories for RAPID3, DAS28, and CDAI, as with almost all measures and indices in clinical medicine. For example, ∼15% of patients with elevated CRP levels have a normal ESR, and vice versa (Table 2). A correlation of rho = 0.51 of ESR and CRP level is seen in a clinical database (8), which is similar to the level at 52 weeks (rho = 0.47) in this study. Any measure or index can serve only as a guide, and not as an absolute indicator for clinical decisions. For example, an elevated ESR in an RA patient whose value was normal a few months earlier may indicate infection, malignancy, or a flare of RA. A complete medical history and physical examination, as well as appropriate laboratory assessment and ancillary studies, must be performed at every clinical encounter to provide optimal information for clinical decisions; any index, whether it is the DAS28, CDAI, or RAPID3, is only one component.
One particular concern may involve patients with extensive damage or high levels of noninflammatory symptoms (such as in fibromyalgia), who may have high RAPID3 scores without necessarily having high inflammatory activity (20). Nonetheless, this type of problem pertains to all indices. For example, a typical patient with fibromyalgia may have a swollen joint count of 0, a tender joint count of 28, an ESR of 18 mm/hour, and a patient global score of 10 (or 100); this patient's DAS28 score of 6.4 and CDAI score of 38 would suggest high inflammatory activity, although there are no swollen joints and the ESR is in the normal range. This phenomenon again emphasizes the importance of interpretation of any index by a knowledgeable physician.
The only quantitative data collected at most encounters of RA patients with rheumatologists are laboratory tests. Even if a RAPID3 score were less informative than a DAS28 score, it may be considerably more valuable to have 80% of optimal quantitative clinical data in all patients (21), rather than only laboratory tests and narrative descriptions of medical history and the physical examination information in medical records. Finally, collection of RAPID3 scores in no way prevents a clinician from assessing a DAS28, CDAI, or any other quantitative measure regarded as important, since the RAPID3 adds only 5 seconds to the doctor's effort at the encounter (16).
The RAPID3 may present additional advantages to a clinician beyond a quantitative estimate of severity of clinical status, particularly when a multidimensional HAQ is completed prior to the patient seeing the physician, thereby preparing the patient for a visit. A quantitative RAPID3 score is then available for the physician before the patient visit, rather than during the visit when conversation between doctor and patient would be interrupted to perform a joint count. A RAPID3 score can be completed when the patient is seen by any rheumatologist, another physician or health professional, or even at home or another location, a feature that does not apply to a formal joint count. The availability of a RAPID3 score of, for example, 18 versus 2, or even as simple a matter as a pain score of, for example, 8 versus 2, enhances communication at the visit. Furthermore, the review of systems and medical history on the multidimensional HAQ generally saves considerable time during the visit (22). The collection of a RAPID3 score also helps to prepare the patient for the visit (21).
RAPID3 scores were reported initially on a 0–10 scale, after conversion from a 0–30 scale (8, 9, 11). However, recent formal studies have indicated that the RAPID3 on a 0–30 scale can be scored in ∼5 seconds rather than 10 seconds when conversion to a 0–10 scale is required (16). Therefore, it is recommended that RAPID3 on a 0–30 scale be the standard measure for usual clinical care.
In summary, these studies document that the RAPID3 can provide similar data to the DAS28 and CDAI, in less than 10% of the time. The only quantitative data collected at most visits in usual rheumatology care at this time are laboratory tests, which are often not available at the time of the visit, and frequently give false-positive or false-negative results. A careful, nonquantitative joint examination accompanied by a quantitative RAPID3 score would appear to be a useful advance for usual rheumatology care.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Pincus had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Pincus, Yazici, Bergman.
Acquisition of data. Keystone, Luijtens.
Analysis and interpretation of data. Pincus, Furer, Keystone, Yazici, Bergman, Luijtens.
ROLE OF THE STUDY SPONSOR
The work reported herein was an investigator-initiated study supported in part by a research grant to Dr. Pincus from UCB. UCB had designed the initial clinical trial (previously reported) and produced the analyses of the data according to a design proposed by the authors based on previous studies of clinical trials involving other biologic agents and usual clinical care. UCB played no role in the design of the analyses in the present study, involving post hoc analyses of data acquired in the initial clinical trial sponsored by UCB. Dr. Luijtens, Senior Biostatistician at UCB, was involved in the analysis and interpretation of data, writing of the report, and the decision to submit the manuscript for publication, in her capacity as an author. The content was submitted to the UCB publication office for review of accuracy prior to submission. However, all content of the present report was initiated by and under control of the authors.