Introduction

Rheumatoid arthritis (RA) is an autoimmune disease leading to joint damage, disability, and decreased quality of life [1]. As the advent of biological agents and novel disease-modifying anti-rheumatic drugs (DMARDs), “treat to target” and “tight control” strategies were proposed, aiming at clinical remission [2, 3]. The preliminary criteria for remission in RA were established by American College of Rheumatology (ACR) in 1981 [4]. Thereafter, a series of different scoring systems were consequently developed, including the Disease Activity Score of 28 joints (DAS28), the Simplified Disease Activity Index (SDAI), and the Clinical Disease Activity Index (CDAI) [5]. In 2001, the ACR and European League Against Rheumatism (EULAR) redefined RA remission for clinical trials based on two definitions, the Boolean definition and the SDAI [6]. Currently, DAS28 < 2.6, SDAI ≤ 3.3, CDAI ≤ 2.8, and the ACR/EULAR Boolean definition are most widely recognized criteria [7]. However, disagreement still exists concerning the reliability of these criteria for defining remission and predicting prognosis of RA patients [7–11].

The cross-sectional remission rates of 4.5-49 % in RA patients have been reported in previous studies using different definitions [12–21], which varies remarkably with different populations. To date, few studies have investigated the remission of RA in China. Our previous study based on a small cohort in two hospitals has reported RA remission rates of 15.4–38.2 % [21]. Yet considering the relatively small sample size, a multi-center study is warranted to demonstrate the overall remission condition across the country.

Previous studies have identified several factors as potential predictors of remission in RA patients, including male sex, disease duration, non-smoker, favorable Health Assessment Questionnaire (HAQ) scores, absence of rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA), and regular treatment [22–26]. However, the magnitude of the association between these factors and remission is controversial [23, 27], and it is unknown whether these factors are associated with RA remission in Chinese patients.

In this study, we conducted a cross-sectional on-site questionnaire study in 28 tertiary hospitals in China to assess the prevalence of remission in RA patients according to the DAS28, the SDAI, the CDAI, and the Boolean definitions of remission. Furthermore, we investigated the potential determinants of RA remission, in order to provide evidence for clinical treatment strategies for Chinese RA patients.

Materials and methods

Patients

The study was carried out from July 2009 to January 2012 at the rheumatology outpatient clinics of 28 hospitals in China. Five hundred and thirteen patients who fulfilled the 1987 American College of Rheumatology (ACR) classification criteria [28] were recruited, of which 27 (5.3 %) were excluded because of incomplete data for calculating the disease activity scores using the four remission criteria. Therefore, a total of 486 patients were included in the data analysis.

Data collection

Questionnaires were administered through face-to-face interviews in each center by trained physicians. Data collected included: (1) demographics: sex, age, smoking status; (2) clinical features of RA: duration of disease, 28 tender joint counts (TJC28), 28 swollen joint counts (SJC28), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), titers of rheumatoid factor (RF), and anti-cyclic citrullinated peptides (anti-CCP), patient and physician visual analogue scale (VAS) global scores of disease activity (PtGlobal and PhGlobal), and interstitial lung disease confirmed by X-Ray or CT scan; and (3) treatments: previous and present use of DMARDs (methotrexate [MTX], hydroxychloroquine [HCQ], leflunomide [LEF], and sulfasalazine [SSZ]), biological agents (adalimumab, etanercept, infliximab, and rituximab), and glucocorticoids (GCs, including oral, intravenous, intramuscular and intra-articular administration of prednisone and similar compounds).

Definitions of remission

The data were analyzed for remission according to the following definitions: (1) DAS28 (0.56 √[TJC28] + 0.28 √[SJC28] + 0.70 ln[ESR] + 0.14 [PtGlobal]) <2.6; (2) SDAI (TJC28 + SJC28 + CRP [mg/dL] + PtGlobal + PhGlobal) ≤3.3; (3) CDAI (TJC28 + SJC28 + PtGlobal + PhGlobal) ≤2.8; and (4) ACR/EULAR Boolean remission criterion (TJC28, SJC28, CRP [mg/dL], and PtGlobal all ≤1). For all VAS, we used a 0–10-cm scale.

Statistical analysis

The distributions of continuous variables were examined. Normally distributed continuous variables were presented as means and standard deviations (SDs), while skewed distributed continuous variables were depicted as medians and interquartile ranges (IQRs). Absolute and relative frequencies were reported for categorical variables. Remission rates were calculated based on each of the four definitions, presented as percentages with 95 % confidence intervals (95 % CIs). Kappa statistics were used to assess agreement between remission criteria. Kappa values were interpreted as follows: <0 as no agreement, 0–0.20 as slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.0 as almost perfect agreement [29].

In order to explore the predictive value of baseline characteristics for remission, all patients were classified into a remission group or a non-remission group according to the four different definitions. Comparisons of the demographics, clinical characteristics, and treatments between the two groups were performed using Student’s t tests for normally distributed continuous variables, Mann-Whitney U tests for skewed continuous variables, and the chi-square tests for categorical variables. A p value less than 0.05 was considered statistically significant. Additionally, backward stepwise logistic regression models were applied separately for each definition of remission to identify independent determinants of RA remission (deletion criterion p > 0.05). Variables with p values less than 0.10 in univariate analyses were included into the multivariate analyses as possible explanatory variables. All the analyses were conducted using SPSS version 20.0 for Windows.

Results

Demographic and clinical characteristics of RA patients

A total of 486 patients were included in this study (Table 1). Of these, 394 (81.1 %) were female. The mean (SD) age was 52.1 (14.3) years, with median (IQR) disease duration of 42 (9–109) months. 75.0 % of patients were RF positive, and 88.6 % were anti-CCP positive. For treatment of disease, 77.7 % of the patients have taken synthetic and/or biological DMARDs. The proportions of patients treated with MTX, LEF, HCQ, SSZ, and biological agents were 53.3, 50.0, 16.9, 13.8, and 9.1 %, respectively. In addition, 50.8 % of patients have received regular treatment with DMARDs for more than 6 months and 31.1 % of patients for more than 1 year. GCs in any form had been taken by 45 % of the patients and were currently used in 22.4 % of the patients. Oral GCs had ever been used in 36.7 % patients, including 20.9 % on low doses ≤15 mg/day, and the remaining 15.8 % tapered from the maximum doses ≤50 mg/day. Nineteen patients had used intravenous methylprednisolone of 40–80 mg/day, including 3 having received pulse therapy. Intramuscular or intra-articular compound betamethasone was ever used in 10.2 % of the patients to relieve articular symptoms.

Table 1 Characteristics of RA patients
Full size table

Remission rates according to different criteria

Different definitions yielded similar remission rates. The remission rates (95 % CIs) according to different definitions were as follows: 8.6 % (6.1–11.1) for DAS28, 8.4 % (6.0–10.9) for SDAI, 8.2 % (5.8–10.7) for CDAI, and 6.8 % (4.5–9.0) for Boolean remission. Although the remission rate was the lowest by the Boolean definition, the difference among these four remission rates was not statistically significant (p = 0.707, Fig. 1). The best agreement was between SDAI and CDAI (κ = 0.93), which might attribute to the similarity of these definitions. Boolean definition showed excellent agreement with SDAI (97.9 %, κ = 0.85, p < 0.001) and CDAI (97.7 %, κ = 0.84, p < 0.001). Agreement between DAS28 and other definitions was substantial (κ = 0.65–0.67).

Fig. 1
figure 1

Remission rates of RA patients (n = 486) from 28 tertiary hospitals in China, according to different definitions. p = 0.707 among different remission definitions. See methods for details on remission definitions

Full size image

Univariate analysis

We divided RA patients into a remission group and a non-remission group based on the four definitions. Univariate analysis indicated that several clinical and therapeutic characteristics were related to remission, including lower HAQ scores (p < 0.001), RF negative (p < 0.01), anti-CCP negative (p < 0.05), and treatment with MTX (p < 0.05) and HCQ (p < 0.01). Younger age was statistically significantly associated with DAS28 and Boolean remission (p = 0.031 and p = 0.039). Moreover, patients in remission were more likely to be male, have shorter disease durations, have fewer cases of concomitant interstitial lung disease, and receive more treatment with LEF, SSZ, or biological agents, and less treatment with GCs, yet the differences were not significant between the two groups (Table 2).

Table 2 Univariate analyses of the predictors of RA remission according to different definitions
Full size table

Multivariate analysis

Backward stepwise logistic regression analyses were employed to identify factors independently associated with RA remission (Table 3). Low HAQ scores and treatment with HCQ were independently associated with RA remission in all the models (p < 0.001 and p < 0.01). Absence of anti-CCP was an independent determinant of DAS28, SDAI, and Boolean remission (p < 0.01) and showed a trend toward significance in the CDAI model (p = 0.080). MTX was only independently associated with Boolean remission with a moderate p value of 0.054. Age was associated with DAS28 and Boolean remission in the univariate analysis, but it was not included in the multivariate models. Regardless of the definitions of remission, RF was not an independent risk factor for non-remission (Table 3). However, if RF was put in the models instead of anti-CCP, it turned out to be independently associated with Boolean remission (OR 0.39, 95 % CI 0.18–0.86, p = 0.020), yet it was not significant in the models of DAS28, SDAI, and CDAI remission criteria.

Table 3 Backward stepwise logistic regression models for RA remission according to different definitions
Full size table

Disease duration and remission

Statistical analysis failed to prove the association between disease duration and remission. However, according to the four remission definitions, the median disease durations were shorter in the remission groups than those in the non-remission groups, though the differences were not statistically significant (Table 2). In order to further investigate the association of disease duration and remission, we compared the remission rates of patients with different disease durations of ≤1 year, >1 and ≤3 years, and >3 years. The chi-square tests also showed no difference in remission rates among the three groups, regardless of the remission definitions (Fig. 2).

Fig. 2
figure 2

Remission rates in different disease duration groups. Chi-square tests showed no significant differences among patients with disease durations of less than 1 year, 1–3 years, and over 3 years, according to the DAS28 (p = 0.435), SDAI (p = 0.815), CDAI (p = 0.830), and Boolean (p = 0.674) definitions

Full size image

Discussion

Our results showed that the remission rates were 8.6 % (DAS28), 8.4 % (SDAI), 8.2 % (CDAI), and 6.7 % (Boolean) in Chinese RA patients. The remission rates determined by these four different definitions were not significantly different. In addition, a low HAQ score, anti-CCP positive, and HCQ were significant and independent determinants of RA remission. Younger age, RF negative, and treatment with MTX were associated with remission in univariate analyses. In the present study, we recruited RA patients from 28 rheumatology clinics covering different parts of China, and the sample size of our study was relatively large. Thus, our findings mostly reflected the real status of RA remission in clinical practice at the utmost.

In the present study, we used different definitions to determine the remission of RA in our Chinese cohort. The cross-sectional remission rates by different criteria were 6.8–8.6 %, lower than our previous results (15.4–38.2 %) [21]. In our previous study, patients were recruited from two top hospitals of the developed area in China, where patients were provided better health care. Therefore, the remission rates were expected to be higher than the average of the country.

Various studies have shown different remission rates in diverse rheumatoid arthritis populations [12–20]. The remission rates in the current study were similar to or lower than most results from other countries, a phenomenon that requires further explanations. One reason might be the less aggressive treatment for Chinese patients. According to a recent Qatari study with a high remission rate of 49 % by DAS28, 93 and 29 % of patients were on synthetic and biological DMARDs [12]. Total remission rates of 13.8 % by CDAI and 19.6 % by DAS28 were observed in the QUEST-RA study from 24 countries, and the remission rates in some countries even reached about 20–40 % [18]. In the QUEST-RA database, 88–100 % of patients had ever taken DMARDs, including 69–98 % treated with MTX [30]. However, in the present study, only 76.1 % of the patients have taken synthetic DMARDs, and 46 % of patients have taken MTX. Moreover, only 31.1 % of our patients have received regular treatment with DMARDs for more than 1 year (Table 1), indicating the insufficient medical care for RA patients in China. As for biological agents, they have not yet been widely used in China, due to the short time since they became available in clinical practice, as well as the economic issues [31]. Since clinical trials have provided evidence for the positive association of treatment and remission, and treatment-to-target is essential to the remission for RA patients [32, 33], this study underscores the need for more aggressive treatment with DMARDs in China to halt the disease progression and to increase the probability of RA remission.

According to this study, GCs have ever been used in 45.0 % of Chinese RA patients, and only 22.4 % of the patients are on current GCs treatment, similar to a previous report from China [34]. The low rate of GCs use in China represents a main difference compared with Western countries, where 42–72 % of the patients are on current oral GCs [12, 13, 17, 20, 30, 35]. The reason why GCs were less used in China requires further investigation. One explanation might be the more concerns about the adverse events of GCs in Chinese patients. Different races have discrepant behavioral models of “risk-seeking” or “risk-aversion” when given a treatment [36], and the patient and physician views on GCs might influence the administration of GC therapy [37]. In the realm of business, Chinese tend to make less risky decisions than Americans [38], and whether such risk-aversion behavior applies to RA patients and rheumatologists in China remains to be studied. Since low-dose GCs applied with DMARDs are beneficial for RA remission [39], the less use of GCs might partly explain the lower remission rate in China to a certain extent.

Moreover, although data were limited, the few studies of Asian patients have revealed lower remission rates compared to those from Europe and the US (4.5–9.7 % versus 10.3–31.4 % by DAS28) [12, 14, 15, 17–20]. Beyond treatment-related factors, genetic and socioeconomic differences might contribute to the disparities among different nations [30]. Further studies into the explanations are needed. In this study, the ACR/EULAR Boolean definition yielded the lowest remission rate (6.8 %) of RA, compared with DAS28 (8.6 %), SDAI (8.4 %), and CDAI (8.2 %), indicating that the Boolean definition was more rigorous. However, the differences between the remission rates by the Boolean definition and the other criteria were insignificant in this study, raising the question whether the new definition is stringent enough to be distinguished from the previous definitions in predicting disease outcome. Lillegraven et al. showed that, as with previous criteria, the new Boolean definition was unable to exclude all patients with further significant radiographic damage [40]. At the moment, the predictive value of the Boolean definition for disease outcome in our population is unknown, and therefore, further studies are needed.

Recent studies have reported that ACPA was an important prognostic predictor in RA patients [23, 26]. ACPA-positive RA was different from ACPA-negative RA [41, 42]. Our results indicated a significant and independent association between negative anti-CCP and RA remission. This is consistent with the results of recent studies of different populations by van der Woude et al. [22, 25]. In addition, our results also indicated that a low HAQ score was a significant and independent predictor for RA remission, which is also in line with previous findings in different populations [23, 25, 43].

Previous studies have reported that negative RF was positively related to RA remission [23, 26]. Our univariate analysis showed that the remission rates were significantly higher in RF-negative patients, regardless of the four different definitions, and that the presence of RF was markedly more common in non-remission patients than in remission patients. These results were in agreement with findings of Verstappen et al. [44]. However, other evidence showed that the presence of anti-CCP significantly compromised the influence of RF on remission in RA patients [43]. A similar result was observed in our multivariable logistic regression analyses of the remission rates by different definitions. Therefore, RF is not an independent risk factor for non-remission in our population.

Some previous studies have suggested that longer disease duration independently lowered the remission rate of RA [23, 25, 43, 45], while this influence of disease duration was not found in other studies [46–48], including ours. Our results indicated that RA patients who had the disease for a long time had the similar opportunity to achieve remission as patients with shorter disease durations, which may reassure RA patients that appropriate disease management should continue even in the presence of long-standing disease.

It has been reported that male sex, non-smoker, and the absence of extra-articular manifestations were associated with RA remission [23, 25]. Nonetheless, no significant effects of sex and smoking status were shown in our study. In this study, interstitial lung disease is not related with remission rates, and further study is needed to investigate the associations between other extra-articular manifestations and disease remission. At present, it is unclear whether ethnicity affects the influence of the above factors on RA remission.

Furthermore, our results showed that treatment with HCQ was strongly associated with disease remission according to different definitions (OR = 2.73–7.62, p < 0.05, Table 3). Patients with treatment of MTX were also more likely to experience remission. Although treatment with LEF, SSZ, or biological agents seemed to have raised remission rates, it was not confirmed with statistical significance. However, the lack of significance might result from the small number of patients receiving such treatment. As demonstrated above (Table 1), the number of patients who had undergone these therapies, especially biological agents, was relatively limited, and it is at risk to draw any conclusion simply based on such a small sample. Another consideration would be that the baseline characteristics of the patients before treatment might vary among different treatment groups. For example, monotherapy of HCQ is more likely to be given to patients with less disease activity [49]. Since this is a cross-sectional study, it is unclear whether remission was achieved as the effect of certain treatment, or certain treatment were likely given to patients with less disease activity and more tendency to achieve remission. Moreover, it was a retrospective study, and recall bias was difficult to avoid. As a result, further longitude studies are needed to assess the effects of medication in Chinese RA patients.

As is mentioned above, the remission rates differ markedly among countries, and various clinical studies have reached divergent conclusions about the determinants of RA remission. Whether socioeconomic and genetic characteristics of different populations play a critical part in explaining these differences remains to be investigated. Hence, comprehensive data from diverse populations are of great value. At present, there is a shortage of information about Asian patients compared to Western countries, and few researches have been made focusing on Chinese patients, which would be significant considering the large population. This is the first multi-center study on the remission of Chinese RA patients, and we hope our results will merit further researches.

In summary, our study indicated that the remission rate of RA patients in China was lower than those reported for Western populations. The ACR/EULAR Boolean definition seemed to be the most stringent in defining RA remission, although the remission rate determined by this definition was comparable to those by the DAS28, the CDAI, and the SDAI criteria. We identified a low HAQ score, absence of anti-CCP, and HCQ as independent predictors for RA remission. Younger age, absence of RF, and treatment with MTX were associated with increased probability of RA remission.