Search tips
Search criteria 


Logo of rheumatologyLink to Publisher's site
Rheumatology (Oxford). 2009 June; 48(6): 686–690.
Published online 2009 April 24. doi:  10.1093/rheumatology/kep054
PMCID: PMC2722796

Evaluation of composite measures of treatment response without acute-phase reactants in patients with rheumatoid arthritis


Objectives. To evaluate composite measures of response without acute-phase reactants in RA patients. Specifically, Clinical Disease Activity Index (CDAI)-derived response criteria were compared with the European League Against Rheumatism (EULAR) response criteria, and the modified ACR (mACR) response criteria were compared to the ACR response criteria.

Methods. Data from 10 108 RA patients enrolled in the Consortium of Rheumatology Researchers of North America registry were examined, including 649 patients initiating DMARD therapy. CDAI cut-off points for disease activity levels and responses were derived using receiver operating characteristic curves with the DAS28 and EULAR response criteria as gold standards. The κ-statistics were applied to assess agreement between CDAI-derived and EULAR-defined responses, as well as ACR20 and ACR50 with mACR20- and mACR50-defined responses, respectively.

Results. For the components of the EULAR response, the derived CDAI cut-off points for DAS28 levels of 3.2 and 5.1 were 7.6 and 19.6, respectively. The derived CDAI cut-off points were 4.3 and 10.0 for DAS28 changes of 0.6 and 1.2, respectively. There were moderate to substantial agreements between CDAI-derived and EULAR responses (κ = 0.57–0.71). Agreement of ACR20 and ACR50 with mACR20 and mACR50 responses, respectively, was excellent (κ = 0.88–0.95).

Conclusions. Agreement between composite measures of response without acute-phase reactants and standard measures ranged from moderate to excellent. The mACR20 and mACR50 criteria as well as CDAI-derived response criteria, can serve as composite measures of response in clinical practice and research settings without access to acute-phase reactants.

Keywords: Rheumatoid arthritis, Acute-phase reactants, Response criteria


RA is a chronic inflammatory disease which often causes joint pain, bone destruction and disability. In the past decade, use of DMARDs including biological agents has significantly improved treatment outcomes as documented in randomized controlled trials (RCTs) [1–3]. These RCTs have relied upon the ACR or the European League Against Rheumatism (EULAR) response criteria as composite measures of response to demonstrate efficacy.

However, assessing treatment responses in clinical practice using standard composite measures of response can be challenging. Both the ACR and EULAR response criteria utilizing the disease activity 28-joint score (DAS28) can entail complicated calculations and require acute-phase reactant values including CRP or ESR [4–6]. For logistical reasons, results of these laboratory tests are most often not available at the time of the provider–patient interaction, when treatment decisions are typically made. Thus, simplified indices such as the Clinical Disease Activity Index (CDAI), which do not require an acute-phase reactant, have been developed for use in clinical practice settings [7].

It has been suggested that measurement of acute-phase reactants may add little to composite disease activity indices for RA patients [8,9]. Indeed, there has been growing interest in the application of the CDAI for measuring disease activity status and modified ACR (mACR) response criteria, both of which do not require laboratory testing. Several studies have validated the CDAI as a composite measure of disease activity status; however, to our knowledge, only one published manuscript to date has developed cut-points for CDAI-derived response to therapy [10,11]. Similarly, only a single study to our knowledge has evaluated mACR20 and mACR50 as composite outcome measures [9]. We, therefore, sought to evaluate two measures of response that did not require acute-phase reactants using a large cohort of patients from a multi-centred observational registry within the USA [12]. Specifically, we examined correlations between CDAI-derived response cut-points and EULAR responses, as well as the ACR20 and ACR50 responses with mACR20 and mACR50 in the study population.


Data sources and data collection

Between October 2001 and September 2006, a total of 76 rheumatology practices enrolled over 13 000 patients with RA in the Consortium of Rheumatology Researchers of North America (CORRONA) registry. The CORRONA registry has been previously described [13,14]. Data are collected from both patients and their treating rheumatologists using questionnaires, which gather information on disease severity and activity (including components of ACR and EULAR response criteria), medical comorbidities, use of medications including DMARDs and adverse events [15]. Follow-up assessments are requested at 3-month intervals and completed during routine clinical encounters. Approvals for participation in the CORRONA registry are obtained from respective institutional review boards of participating academic sites and a central institutional review board for private practice sites.

To assess disease activity, we identified RA patients in the CORRONA registry with complete ACR core measures including tender joint count (TJC) for 28 joints, swollen joint count (SJC) for 28 joints, patient global assessment (PGA), evaluator global assessment (EGA), patient-reported pain and physical function using the modified HAQ (mHAQ) score and ESR as the acute-phase reactant (n = 10 108). There were fewer patients in the registry with results for CRP than ESR values; thus, ESR was selected as the acute-phase reactant for the analyses. The cohort was randomly divided into ‘training’ (n = 4982) and ‘test’ (n = 5126) sets for development and evaluation of cut-off points defining disease activity using the CDAI score.

To assess treatment responses, we identified RA patients who initiated therapy with a biological or non-biological DMARD and at least one follow-up study visit between 3 and 12 months after initiation. Data for all components of the ACR and EULAR response criteria had to be available for the initiation as well as the follow-up visit (n = 649). This cohort was also randomly divided into ‘training’ (n = 306) and ‘test’ (n = 343) sets for the development and evaluation of cut-off points defining response using the CDAI score.

Measures of disease activity

CDAI is calculated by adding the sum of TJC and SJC as well as PGA and EGA using a 10-cm visual analogue scale [CDAI = SJC + TJC + PGA (in cm) and EGA (in cm)]. Unlike the DAS28, it does not require an acute-phase reactant test result. The DAS28 is calculated as follows: DAS28 = 0.56 × √(28TJC) + 0.28 × √(28SJC) + 0.70 × log (ESR) + 0.014 × PGA. DAS28 and CDAI represent a continuous scale; cut-off points have been described to delineate categories of disease activity status [8,10,16].

Measures of treatment response

EULAR response criteria classify patients into three groups (non-responders, moderate response and good response) based on the individual change in DAS and the level of DAS attained [17]. The CDAI-derived response criteria were calculated in a manner analogous to the EULAR response. ACR20 criteria define responses based on [gt-or-equal, slanted]20% improvement in TJC and SJC and in three of the following: EGA, PGA, pain, mHAQ and ESR or CRP. The previously proposed mACR20 scoring excludes acute-phase reactants and is calculated based on [gt-or-equal, slanted]20% improvement in TJC and SJC as well as two (mACR202 of 4) or three of the four remaining criteria (mACR203 of 4); similarly [gt-or-equal, slanted]50% improvement is used for mACR502 of 4 and mACR503 of 4 [9].

Assessment of disease activity

Overall correlations between CDAI and DAS28 were assessed using the Pearson correlation coefficient. When evaluating correlations based on disease activity (low, moderate and high) the κ-statistic was used. CDAI disease activity cut-offs were derived using standard methodology using the training set, based on receiver operating characteristic (ROC) curves and the Youden index [18], which identified the cut-off point with the maximum sensitivity and specificity [10,16]. To assess the precision of the estimates, we generated the range of cut-off points corresponding to values greater than or equal to the lower limit of the 95% CI of the Youden index [18]. The first CDAI cut-off point to differentiate low disease activity from moderate or high disease activity used the previously published DAS28 cut-off point of 3.2 [6,19]. The second CDAI cut-off point for differentiating moderate vs high disease activity levels was a DAS28 of 5.1. CDAI scores were compared with DAS28 levels using the κ-statistic based on both the derived and previously published cut-off points [10].

Assessment of response to therapy

Correlations between changes in CDAI and DAS28 scores were assessed using the Pearson correlation coefficient in patients initiating new DMARD therapy. As is the standard method, CDAI-derived treatment responses were derived using the training set based on ROC curves and Youden index [10,16,18]. The cut-off points were determined based on previously published cut-off points for change in DAS28 levels of 0.6 and 1.2 [6,19]. The CDAI-derived responses were calculated using both the change and the values reached in a manner analogous to the derivation of the EULAR responses. Using these newly created cut-off points as well as the previously published cut-off points, CDAI-derived responses were compared with EULAR responses using the test set based on two comparisons using the κ-statistic [11]. First comparisons were made based on three categories of response: ‘non-responder’ vs ‘moderate’ response vs ‘good’ response. Subsequent comparisons were between ‘non-responder’ vs ‘moderate or good’ responses. For both comparisons, we examined agreement based on low or moderate and high disease activity based on the DAS28.

Using the total (n = 649) and test (n = 343) populations initiating new DMARD therapy, mACR202 of 4 and mACR203 of 4 as well as mACR502 of 4 and mACR503 of 4 were compared with ACR20 and ACR50 responses, respectively, using the κ-statistic. The agreement between the measures was then further evaluated based on disease activity (low or moderate and high disease activity based on the DAS28). We did not compare the modified ACR70 to the standard ACR70 response, because there were too few patients who achieved ACR70 responses in the study cohorts.


Characteristics of the two populations, namely the entire study cohort and those initiating new DMARD therapy are described in Table 1. The mean age of the study population was 60 ± 14 years, 76% were female and the mean disease duration was ~11 years. Patients initiating new DMARD treatment had higher measures of disease activity, by both CDAI and DAS28 scores. Patients with high disease activity were more likely to be classified as responding to therapy based on the EULAR response, ACR20 and ACR50 criteria.

Table 1.
Demographics and disease characteristics of the study population

Correlation of measures of disease activity

In the entire study population, correlation between DAS28 and CDAI was 0.84 (Pearson). Using previously published cut-off points for both DAS28 and CDAI, the population was divided into low, moderate and high disease activity [5,10]. There was 74% agreement between DAS28 and CDAI scores compared with an expected 39% agreement resulting in a κ of 0.58.

Using the training set from the entire study cohort, cut-off points were determined for CDAI using DAS28 (Table 2). Results for CDAI cut-off points can be found in Table 2, which are similar to those previously published [10]. Patients were then categorized based on DAS28 (low vs moderate vs high) disease activity in the test set. Using the derived cut-off points, there was 75% agreement compared with an expected agreement of 37% (κ = 0.60) (Table 3). Results were similar when published cut-off points were utilized [10]. When patients were divided into those with low vs moderate or high levels of disease activity, the agreement improved (κ = 0.69).

Table 2.
Determination of CDAI cut-off points to assess disease activity
Table 3.
Agreement between the CDAI and DAS28 to categorize disease activity levels

Correlation of measures of treatment response

Correlations between changes in CDAI and DAS28 were examined in those initiating new DMARD treatment (n = 649) and found to be 0.87. As a sensitivity analysis, there was little change when selecting only patients with moderate or severe disease activity (correlation = 0.88). Using the training data set in new DMARD initiators, cut-off points were determined for CDAI-derived responses using DAS28 changes of 0.6 and 1.2 [5]. These cut-off points can be found in Table 4 and were compared with the previously published values [11]. Patients were then categorized based on level of EULAR response (no vs moderate vs good responses) in the test set. Using the derived cut-off points, there was 75% agreement compared with an expected agreement of 41% (κ = 0.57)—with diminished agreement when published cut-off points were used (Table 5) [11]. Agreement improved with high disease activity (derived κ = 0.68 and published κ = 0.60 as compared to derived κ = 0.50 and published κ = 0.47, in those with low or moderate disease activity). When patients were divided into no vs moderate or good responses, agreement improved (κ = 0.71). Agreement was greater in those with high disease activity (κ = 0.76 vs 0.66 in low or moderate disease activity). Agreement between mACR20 and mACR50 measures with standard ACR20 and ACR50 responses was excellent (κ = 0.88–0.95) (Table 6). There was little change in the results when agreement was examined separately in those with low or moderate and high disease activity (κ = 0.82–0.97).

Table 4.
Determination of CDAI-derived response cut-off points to assess treatment response
Table 5.
Agreement between the CDAI-derived response and EULAR response
Table 6.
Agreement between the mACR and standard ACR response criteria


This study evaluated composite measures of treatment response without acute-phase reactants with standard EULAR and ACR response criteria in a large, multi-centred observational RA cohort. Using CDAI-derived cut-offs for disease activity levels, there were moderate to substantial agreements between CDAI-derived and EULAR-defined responses. As might be expected, mACR20 and mACR50 responses were highly correlated with those defined by ACR20 and ACR50 response criteria. Taken together, these findings provide two alternative approaches for measuring treatment response in both observational research and clinical settings when acute-phase reactant measurements are not always available.

With respect to disease activity states, this work confirms that of others showing that CDAI correlates with DAS28 and the cut-off points derived in this study were similar as well [8]. Aletaha and Smolen [8] found substantial agreement between CDAI and DAS28 (weighted κ = 0.70), which was slightly higher than the agreement in these analyses (using their methodology, the weighted κ in this population was 0.63). This suggests that CDAI cut-points for stratifying disease activity states in patients without acute-phase reactant results can indeed be applied in both clinical practice and observational research settings.

The actual numerical value for cut-off points for change in CDAI corresponding to DAS28 changes of 1.2 were somewhat lower than those from a recently published study; these findings are likely due to differences in the study populations and references used [11]. Ranganath et al. [11] developed cut-off points in a cohort of DMARD-naïve, RF-positive early RA patients. The patient population in this study differed as they had long-standing disease (mean disease duration of 11 years). While the cut-off points derived in this study were comparable with those previously published, validation of our derived cut-off points in an independent cohort would be of value and strengthen our results.

Goldman et al. [9] similarly found that patterns of improvement defined by mACR20 scores without acute-phase reactants were consistent with accepted ACR20 scores. Comparing responses to therapy in both MTX-naïve early and DMARD-refractory late RA patients, they observed that both the mACR203 of 4 and mACR202 of 4 discriminated clinical improvement for etanercept vs placebo. Various other modifications of the ACR response criteria have already begun to be implemented in observational registries missing individual components [20]. The findings in the current study, coupled with the work of Goldman et al. [9], provide convincing evidence that mACR responses can be applied to measure treatment responses in patients when acute-phase reactant measurements are not available.

In the current study, correlations between responses defined by mACR with accepted ACR response criteria were greater than those between CDAI- and EULAR-defined responses. In part, this may be due to fundamental differences between the instruments. Modified ACR criteria represent the same calculation as ACR responses, but with one or two measures removed, including ESR or CRP. In contrast, calculating CDAI is markedly different from DAS28, which is integral to defining EULAR responses. Although in these analyses mACR20 and mACR50 were superior to CDAI in assessing responses to treatment, there are circumstances when the CDAI-derived response criteria may be preferable to mACR, because it also incorporates an assessment of the final disease activity state. The importance of including the final disease activity state in a composite measure of response has been previously described [21].

While acute-phase reactant measurements provide valuable clinical information in many patients, they are often not available at the time of the clinical evaluation. Moreover, excluding acute-phase reactant measurements from composite measures simplifies and streamlines the evaluation of disease activity and treatment responses, and may be what is required to convince rheumatologists to use such measures in daily practice. Goldman et al. [9] have suggested that mACR20 could help physicians who do not conduct laboratory tests in their offices to guide treatment decisions, as well as facilitate group comparisons of changes in disease status. Based on the findings of the current study, utilization of CDAI-derived response criteria may be another alternative for clinical practice. In contrast, others have asserted that the ACR20, and by extension the mACR20, should be employed only for analysis of RCTs (for which the ACR response criteria were developed) or longitudinal databases and are not useful in the clinical management of individual patients [22]. Indeed, further evaluation of the benefits and limitations of integrating CDAI and mACR in regular clinical practice in other populations may be needed, but the present evidence strongly suggests that alternative approaches without acute-phase reactants can be implemented. Moreover, these findings are also relevant for observational studies of DMARD and biological agent effectiveness.

The disease activity in this population, while consistent with other published US cohorts from clinical practice settings, was lower than those seen in patients participating in clinical trials [23,24]. To address this potential limitation, we also examined the influence of disease activity on the agreement between the CDAI-derived response with the EULAR response, and the mACR and standard ACR response criteria based on disease activity. In addition, we compared those with acute-phase reactant results to those without in both the total population and initiators of DMARD therapy in our registry in terms of demographic and disease activity characteristics. The only significant difference was the proportion of patients with serological positivity among those who initiated a DMARD (80% in those with an ESR vs 58% in those without). Disease activity as measured by the CDAI was not statistically different in those with and without an ESR in both the populations.

In summary, we have demonstrated good to excellent agreements between both mACR and CDAI-derived responses with well-accepted ACR and EULAR composite measures, which include acute-phase reactant measurements. These findings have important implications. In clinical settings without access to results from acute-phase reactant testing, clinicians may choose to use the mACR- and CDAI-derived response to drive treatment decisions. Additionally, these tools may also be used for epidemiological and outcomes research using disease registry data where there may be missing laboratory results.

An external file that holds a picture, illustration, etc.
Object name is kep054b1.jpg


Funding: J.D.G. was supported by Grant Number K23AR054412 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases and a Clinical Translational Research Award in Rheumatoid Arthritis from the Arthritis Foundation. L.R.H. was supported by Grant Number K23AR053856 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases.

Disclosure statement: J.K. is president of CORRONA from where the data were derived. G.R. has a research contract through the University of Massachusetts with CORRONA. V.S. is a consultant to Abbott Immunology, Allergan, Almirall, Amgen Corporation, AstraZeneca, Biogen Idec, CanFite, Centocor, Chelsea, Cypress Biosciences, Inc, Eurodiagnostica, Fibrogen, Forest Laboratories, Genentech, Human Genome Sciences, Incyte, Jazz Pharmaceuticals, Lexicon Genetics, Lux Biosciences, Merck Serono, Novartis Pharmaceuticals, NovoNordisk, Noxxon Pharma, Ono Pharmaceuticals, Pfizer, Rigel, Rigen, Roche, Sanofi-Aventis, Savient, Schering-Plough, SKK, UCB and Wyeth. V.S. is also on advisory boards for Abbott, Amgen, Biogen Idec, Bioseek, CanFite, Centocor, Chelsea, Crescendo, Eurodiagnostica, Forest, Incyte, Novartis, Pfizer, Rigel, Roche, Savient, Schering-Plough, UCB and Wyeth. J.D.G. has received honoraria for advisory boards from BMS, Centocor, Genentech, Roche, UCB and receives grant support for research from BMS. He also serves as Chief Scientific Officer for CORRONA. All other authors have declared no conflicts of interest.


1. Bathon JM, Martin RW, Fleischmann RM, et al. A comparison of etanercept and methotrexate in patients with early rheumatoid arthritis. N Engl J Med. 2000;343:1586–93. [PubMed]
2. Goekoop-Ruiterman YPM, de Vries-Bouwstra JK, Allaart CF, et al. Clinical and radiographic outcomes of four different treatment strategies in patients with early rheumatoid arthritis (the BeSt study): a randomized, controlled trial. Arthritis Rheum. 2005;52:3381–90. [PubMed]
3. Grigor C, Capell H, Stirling A, et al. Effect of a treatment strategy of tight control for rheumatoid arthritis (the TICORA study): a single-blind randomised controlled trial. Lancet. 2004;364:263–9. [PubMed]
4. Felson DT, Anderson JJ, Boers M, et al. American College of Rheumatology. Preliminary definition of improvement in rheumatoid arthritis. Arthritis Rheum. 1995;38:727–35. [PubMed]
5. Prevoo ML, van 't Hof MA, Kuper HH, et al. Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis. Arthritis Rheum. 1995;38:44–8. [PubMed]
6. van Gestel AM, Prevoo ML, van 't Hof MA, et al. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis. Comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism Criteria. Arthritis Rheum. 1996;39:34–40. [PubMed]
7. Zatarain E, Strand V. Monitoring disease activity of rheumatoid arthritis in clinical practice: contributions from clinical trials. Nat Clin Pract Rheumatol. 2006;2:611–8. [PubMed]
8. Aletaha D, Nell VPK, Stamm T, et al. Acute phase reactants add little to composite disease activity indices for rheumatoid arthritis: validation of a clinical activity score. Arthritis Res Ther. 2005;7:R796–806. [PMC free article] [PubMed]
9. Goldman JA, Xia HA, White B, et al. Evaluation of a modified ACR20 scoring system in patients with rheumatoid arthritis receiving treatment with etanercept. Ann Rheum Dis. 2006;65:1649–52. [PMC free article] [PubMed]
10. Aletaha D, Smolen J. The Simplified Disease Activity Index (SDAI) and the Clinical Disease Activity Index (CDAI): a review of their usefulness and validity in rheumatoid arthritis. Clin Exp Rheumatol. 2005;23(5 Suppl. 39):S100–8. [PubMed]
11. Ranganath VK, Yoon J, Elashoff DA, et al. Comparison of composite measures of disease activity in an early sero-positive rheumatoid arthritis cohort. Ann Rheum Dis. Advance Access published May 21, 2007, doi.10.1136./ard.2006.065839. [PMC free article] [PubMed]
12. Mierau M, Schoels M, Gonda G, et al. Assessing remission in clinical practice. Rheumatology. 2007;46:975–9. [PubMed]
13. Greenberg JD, Bingham CO, 3rd, Abramson SB, et al. Effect of cardiovascular comorbidities and concomitant aspirin use on selection of cyclooxygenase inhibitor among rheumatologists. Arthritis Rheum. 2005;53:12–17. [PubMed]
14. Kremer J. The CORRONA database. Ann Rheum Dis. 2005;64(Suppl.4):iv37–41. [PMC free article] [PubMed]
15. Felson DT, Anderson JJ, Boers M, et al. The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. The Committee on Outcome Measures in Rheumatoid Arthritis Clinical Trials. Arthritis Rheum. 1993;36:729–40. [PubMed]
16. Aletaha D, Ward MM, Machold KP, et al. Remission and active disease in rheumatoid arthritis: defining criteria for disease activity states. Arthritis Rheum. 2005;52:2625–36. [PubMed]
17. van Gestel AM, Anderson JJ, van Riel PL, et al. ACR and EULAR improvement criteria have comparable validity in rheumatoid arthritis trials. American College of Rheumatology European League of Associations for Rheumatology. J Rheumatol. 1999;26:705–11. [PubMed]
18. Youden WJ. An index for rating diagnostic tests. Cancer. 1950;3:32–5. [PubMed]
19. van Gestel AM, Haagsma CJ, Van Riel PLCM. Validation of rheumatoid arthritis improvement criteria that included simplified joint counts. Arthritis Rheum. 1998;41:1845–50. [PubMed]
20. Kievit W, Fransen J, Oerlemans AJM, et al. The efficacy of anti-TNF in rheumatoid arthritis, a comparison between randomised controlled trials and clinical practice. Ann Rheum Dis. 2007;66:1473–8. [PMC free article] [PubMed]
21. Aletaha D, Funovits J, Keystone EC, et al. Disease activity early in the course of treatment predicts response to therapy after one year in rheumatoid arthritis patients. Arthritis Rheum. 2007;56:3226–35. [PubMed]
22. Ranganath VK, Khanna D, Paulus HE. ACR remission criteria and response criteria. Clin Exp Rheumatol. 2006;24(6 Suppl. 43):S14–21. [PubMed]
23. Greenberg JD, Kishimoto M, Strand V, et al. Tumor necrosis factor antagonist responsiveness in a United States rheumatoid arthritis cohort. Am J Med. 2008;121:532–8. [PMC free article] [PubMed]
24. Sokka T, Pincus T. Eligibility of patients in routine care for major clinical trials of anti-tumor necrosis factor alpha agents in rheumatoid arthritis. Arthritis Rheum. 2003;48:313–8. [PubMed]

Articles from Rheumatology (Oxford, England) are provided here courtesy of Oxford University Press