Background

The risk of coronary artery disease (CAD) is known to be elevated in patients with rheumatoid arthritis (RA). A retrospective cohort study found a 39–58% increased risk of major cardiovascular events in patients with RA in comparison with the general population [1], and a meta-analysis of observational studies found that the risk of cardiovascular events was 48% higher in patients with RA [2]. One possible reason for this is the systemic chronic inflammation in patients with RA contributing to the pathway of atherosclerosis, from plaque formation to the rupture of plaque and thrombosis [3, 4]. Furthermore, in the development of both RA and atherosclerosis, the same proinflammatory cytokines, i.e., interleukin (IL)-6 and tumor necrosis factor-alpha (TNF-α), are at work [3]. The release of IL-6 and TNF-α is stimulated by chronic systemic inflammation, and these proinflammatory cytokines which increase vascular calcification and accelerate atherosclerosis progression, from early atheroma formation to thrombus development, may play a primary role in increasing the risk of CAD in patients with RA [5,6,7].

The conventional treatment for RA includes nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroids, disease-modifying antirheumatic drugs (DMARDs), and TNF antagonists. These commonly used drugs have an anti-inflammatory mechanism. Typically, the combined use of NSAIDs and DMARDs is effective for treating patients with RA. However, certain NSAIDs and corticosteroids are known to increase the risk of CAD. A meta-analysis found that the risk of cardiovascular events increased by 18 and 47% in patients with RA who were taking NSAIDs and corticosteroids, respectively [8]. Conversely, DMARDs have been found to have a protective effect regarding CAD. A previous report indicated that DMARDs, particularly methotrexate, were associated with an 18% reduction in the risk of CAD [9]. Apart from its anti-inflammatory effect, methotrexate has some adverse effects, such as nausea, vomiting, hepatotoxicity, nephrotoxicity, and myelosuppression [10]. Subesinghe and Scott (2015) revealed that the most common reason for the permanent discontinuation of methotrexate was its adverse effects [11]. Some patients discontinued the conventional treatment because due to its adverse effects, which could induce disease reactivation. In contrast, most of complementary and alternative medicine (CAM) was associated with only minor adverse effects [12]. Accordingly, patients have tended to seek CAM for pain relief or symptom management in recent times [13].

Patients with RA were observed increasingly to use both CAM and conventional medicine in parallel. In the United States, a population-based survey found that 41% of patients with RA used some kind of CAM [14]. Eighty-two percent of RA patients reportedly use CAM in Korea, and traditional Asian medicine is the most widespread type of treatment [15]. In Taiwan, nearly 30% of RA patients have been reported to use Chinese herbal products (CHP), and certain herbs in commonly prescribed formulae, such as Paeonia lactiflora Pall. (Shao-Yao) and Platycladus orientalis (Linn.) (Huang-Po), are known to have an anti-inflammatory effect [16, 17]. Kuei-Chih-Shao-Yao-Chih-Mu-Tang, the formula used in clinical practice for RA, has shown promising anti-inflammatory effects in lowering the erythrocyte sedimentation rate (ESR) and alleviating morning stiffness [18]. Paeonia lactiflora Pall. (Shao-Yao), in addition to its anti-inflammatory properties, has also been implicated in the prevention of hepatotoxicity in concurrent methotrexate use. A randomized controlled trial found that patients with RA who received total glucosides of peony in combination with leflunomide and methotrexate had lower rates of hepatotoxicity [19, 20]. However, despite the promising anti-inflammatory effect of CHP reported in previous studies, the association between the effect of CHP and risk of CAD remains unknown. Thus, this present study aimed to investigate the risk of CAD in patients with RA who were using CHP and conventional medicine in parallel.

Methods

Data source

This study used reimbursement claims data from the Taiwan National Health Insurance (NHI) program implemented in March 1995. The NHI program is a single-payer, compulsory universal health insurance program that has offered comprehensive medical care coverage to 99% of the Taiwanese population and has contracts with 97% of the country’s hospitals and clinics (http://www.nhi.gov.tw/english/index.aspx). The National Health Insurance Research Database (NHIRD) has documentation regarding every medical treatment reimbursed by the NHI program, including traditional Chinese medicine treatments.

The data sets of the study comprised registry information for beneficiaries, ambulatory and inpatient care claims, and the Registry for Catastrophic Illness of NHIRD. We used ambulatory and inpatient care records of patients between 1997 and 2010 from the Registry for Catastrophic Illness to identify study subjects for follow-up until the end of 2011. As previously described, RA is statutorily included in the catastrophic illness category with solid evaluation criteria [21]. The ambulatory care claims record contain an individual’s sex, date of birth, visit dates, and International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for three primary diagnoses. Inpatient claims contain ICD-9-CM codes for principal diagnosis and up to four secondary diagnoses. The ICD-9-CM codes include physicians who practice Chinese medicine. According to the principle of data protection of NHIRD, any data in the NHIRD that could be used to identify patients or care providers, including medical institutions and physicians, are scrambled before being sent to the National Health Research Institutes for database construction and then further scrambled before being released to each researcher. Theoretically, it is impossible to query the data in isolation alone to identify individuals at any level by using this database. The requirement regarding informed consent in the present study was waived. This study was approved by the Institutional Review Board of the China Medical University (CMUH104-REC2–115).

Study population selection

This was a retrospective cohort study, in which the population was followed up between January 1, 1997 and December 31, 2011. Patients diagnosed with RA (ICD-9-CM code 714.0) between January 1, 1997 and December 31, 2010 formed our study cohort and were followed up until December 31, 2011. Patients aged < 18 years and those who had either withdrawn from insurance coverage or been diagnosed with CAD (ICD-9-CM code 410–414) and stroke (ICD-9-CM 430–438) before the first diagnosis date of RA were excluded from the study, including patients whose follow-up period was < 6 months. In total, 22,353 patients with RA during the period 1997–2010 were identified (Fig. 1).

Fig. 1
figure 1

Flow recruitment chart of patients with rheumatoid arthritis (RA) from the Registry Database for Catastrophic Illness Certificate. After excluding patients not fitting the inclusion criteria, both the CHP and non-CHP groups comprised 4889 patients after 1:1 matching by sex, age, index year, and initial diagnosis year of RA

Full size image

The participants use of CHP, in parallel with their conventional treatments from the date of their initial RA diagnosis, were allocated to the CHP group. Those RA patients who were using conventional medicine without receiving any CHP were assigned to the non-CHP group. The index date for the non-CHP group was randomly selected between the initial RA diagnosis date and the study’s endpoint. The date on which CHP was first used to treat RA was used as the index date for the CHP group. Prior to the matching, there were 5742 patients in the CHP group and 16,611 patients in the non-CHP group. A 1:1 frequency matching was applied for sex, age (in 5-year increments), index year, and initial RA diagnosis year. Following the frequency matching, both the CHP and non-CHP groups comprised 4889 patients with RA.

Primary outcome

The primary outcome was CAD (ICD-9-CM codes 410–414) diagnosed by a cardiologist during the 15-year follow-up period with at least three outpatient claims or one hospital claim. All eligible patients were followed up from the index date to the date of diagnosis of CAD, withdrawal from the NHI program, or the study endpoint (December 31, 2011), whichever occurred first.

Characteristics, comorbidities, and drugs

The sociodemographic variables included age and sex. Comorbidities were identified using ICD-9-CM codes from the database of outpatient and inpatient visits. The baseline comorbidities were defined as disease occurring within 1 year before the index date, with at least three associated outpatient claims or at least one hospital claim. Baseline comorbidities included diabetes mellitus (ICD-9-CM code: 250), hypertension (ICD-9-CM codes: 401–405), hyperlipidemia (ICD-9-CM code: 272), chronic obstructive pulmonary disease (ICD-9-CM codes: 491–496), and end-stage renal disease (ICD-9-CM code: 585). We also considered medication use, including NSAIDs, corticosteroids, DMARDs and TNF antagonists from the initial RA diagnosis date to the study endpoint.

Statistical analysis

The continuous variables were reported as the mean and standard deviation, whereas the categorical variables were reported as numbers and percentages. Differences in the proportions and means were evaluated using the chi-square test or t-test. The Cox proportional hazards model was used to estimate the hazard ratio (HR) of CAD after the use of CHP, controlling for potential confounding factors including age, sex, diabetes mellitus, hypertension, hyperlipidemia, COPD, ESRD and RA related medicine usage. For categorical covariates, the Kaplan–Meier and log-rank tests were performed. A P value < 0.05 was considered statistically significant. In this study, SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA), was used for the statistical analysis.

Results

Following the 1:1 matching, there were more than six times more females than male patients in both groups. Regarding in the aspect of comorbidities, the proportion of hypertension was significantly lower in the CHP group than in the non-CHP group but, by contrast, the proportion of COPD was higher in the CHP group than in the non-CHP group. Moreover, the proportion of DMARDs and TNF-antagonist use in the CHP group was significantly higher than that in the non-CHP group. The mean (median) follow-up period was 5.38 (4.94) years and 4.88 (4.34) years for the CHP group and non-CHP group, respectively (Table 1).

Table 1 Characteristics of patients with rheumatoid arthritis classified according to the use of Chinese herbal products
Full size table

A total of 568 CAD cases were identified during the follow-up period, with an incidence rate of CAD of 8.06 per 1000 person-years in the CHP group and 14.93 per 1000 person-years in the non-CHP group. Univariate and multivariate Cox proportional hazards models were estimated according to the risk of CAD using HR and a 95% confidence interval (CI) in the cohort of patients with RA in the CHP and non-CHP groups. The Kaplan-Meier analysis demonstrated that the cumulative incidence of CAD was significantly lower in the CHP group than in the non-CHP group (log-rank test, p < 0.0001) (Fig. 2). Patients in the CHP group were more likely to have a decreased risk of CAD (crude HR: 0.55, 95% CI: 0.46–0.65) compared with the non-CHP group. Following the multivariate adjustment, the HR remained the same (aHR: 0.59, 95% CI: 0.5–0.71) (Table 2).

Fig. 2
figure 2

Cumulative incidence of CAD between the CHP group and the non-CHP group. The cumulative incidence of CAD in the CHP group is significantly lower than in the non-CHP group (log-rank test, p < .0001). Abbreviations: CAD, coronary artery disease. CHP, Chinese herbal products

Full size image
Table 2 Incidence rates, hazard ratios, and confidence intervals of coronary artery disease in patients with rheumatoid arthritis using and not using Chinese herbal products, stratified by sex, age, comorbidities, and drug used
Full size table

Stratified by sex, the incidence rate of CAD in the CHP group was 7.53 per 1000 person-years in women and 11.54 per 1000 person-years in men, whereas, in the non-CHP group, it was 14.07 per 1000 person-years in women and 21.04 per 1000 person-years in men. The risk of CAD was significantly lower among women (0.59-fold, 95% CI: 0.49–0.72) and men (aHR: 0.55, 95% CI: 0.36–0.85) in the CHP group compared the non-CHP group. In the 40–59 and > 60-year age groups, the risk of CAD was lower in the CHP group than in the non-CHP group (Table 2). Patients with RA who received CHP treatment were found to have a significantly lower adjusted HR regarding the risk of CAD (Table 3). In the CHP group, patients who used CHP for > 180 days had a lower risk of CAD than those who used CHP for < 30 days (aHR: 0.64, 95% CI: 0.43–0.95).

Table 3 Hazard ratios and 95% confidence intervals of coronary artery disease risk associated with cumulative number of days of CHP use among patients with rheumatoid arthritis
Full size table

The three most commonly prescribed herbs were Corydalis ambigua Cham. & Schltdl. (Yen-Hu-So), Spatholobus suberectus Dunn (Chi-Hsieh- Teng), and Morus alba L. (Sang-Chih), and the three most commonly prescribed formulae were Kuei-Chih-Shao-Yao-Chih-Mu-Tang, Tang-Kuei-Nien-Tung-Tang, and Shu-Ching-Huo-Hsieh-Tang. The Cox proportional hazards models showed that Kuei-Chih-Shao-Yao-Chih-Mu-Tang, Tang-Kuei-Nien-Tung-Tang, Shu-Ching-Huo-Hsieh-Tang, Tu-Huo-Chi-Sheng-Tang, and Chia-Wei-Hsiao-Yao-San were associated with a reduced risk of CAD. (Table 4).

Table 4 Hazard Ratios and 95% confidence intervals of coronary artery disease (CAD) risk associated with type of herbal formulae among RA patients
Full size table

Discussion

This population-based cohort study was performed as preparation for an evaluation of the association between CHP application and the risk of CAD in RA patients. We found that the risk of CAD in patients with RA who used CHP in combination with conventional treatments was 41% lower than that in patients who used conventional treatments only (aHR: 0.59, 95% CI: 0.5–0.71). Of the several commonly prescribed formulae, Kuei-Chih-Shao-Yao-Chih-Mu-Tang, Tang-Kuei-Nien-Tung-Tang, and Shu-Ching-Huo-Hsieh-Tang were associated with a lower risk of CAD. Moreover, patients using CHP for > 180 days had a 36% reduced risk of CAD compared with patients using CHP for < 30 days (aHR: 0.64, 95% CI: 0.43–0.95). The results of our study indicate a strong relationship between the use of CHP and a reduced risk of CAD in patients with RA. A possible explanation of this is that commonly prescribed CHP treatments might have anti-inflammatory effects [18,19,20]. The proinflammatory cytokines associated with RA, which is an autoimmune and chronic inflammatory disorder, are the same as those seen in the development of atherosclerosis, which is thought to be the main driver of CAD [3, 5, 6]. The use of CHP in combination with other therapies might block the inflammation process and thereby lower the risk of CAD.

In both groups, there were 6.5 times more women than men, and 85% of the patients were under the age of 60. According to previous research, the prevalence of RA is higher among woman than men, and the sex ratio is around 4.5:1 [22]. Moreover, women are more prone then men to use Chinese herbal medicine in Taiwan [23], which may explain the difference in the respective number of female and male participants in our study. Furthermore, we observed that the incidence rate of CAD in male patients is higher than that in female patients. One possible reason may be the effect of hormones. Compared with men, the incidence rate of CAD is significantly lower in premenopausal women because of the protective effect of endogenous estrogens [24]. The majority of the female patients are under 60-years old, which equates to the pre-menopause period. Estrogens are reported to reduce the atherosclerotic process and lower the effect of total cholesterol. The protective mechanisms account for the lower incidence rate of CAD in female patients compared to male patients with rheumatoid arthritis.

Some commonly prescribed herbs and herbal formulae have been found to lower the risk of CAD. One possible reason is that these frequently prescribed formulae and herbs may have anti-inflammatory effects. Kuei-Chih-Shao-Yao-Chih-Mu-Tang, which originated in Shang-Han-Lun and was the most commonly used formula in our study, can alleviate morning stiffness and lower the ESR, a nonspecific inflammatory factor [18]. The individual herbs contained in Kuei-Chih-Shao-Yao-Chih-Mu-Tang, such as Paeonia lactiflora Pall. (Shao-Yao) and Anemarrhena asphodeloides Bge (Chih-Mu), have also been found to have potential anti-inflammatory effects. Chen (2011) found that Paeonia lactiflora Pall. (Shao-Yao) suppressed TNF-α-induced chemokine production by blocking nuclear factor-κB and the extracellular signal-regulated kinase pathway in human dermal microvascular endothelial cells [25]. Moreover, Chen (2013) documented that Paeonia lactiflora Pall. (Shao-Yao) had anti-inflammatory effects by lowering the ESR and CRP levels and a protective effect against hepatotoxicity in a combination treatment regimen with leflunomide and methotrexate after 24 weeks of treatment [19]. Anemarrhena asphodeloides Bge (Chih-Mu) was found to have cardioprotective effects on isoproterenol-induced myocardial infarction in rats [26]. These results suggest that pretreatment with Anemarrhena asphodeloides Bge (Chih-Mu) extracts reversed the elevation of IL-6 and TNF-α induced by isoproterenol. Additionally, certain herbs in Tang-Kuei-Nien-Tung-Tang were also found to have a potential anti-inflammatory effect. Hang (2018) demonstrated that Scutellaria baicalensis Georgi (Huang-Chin) could lower CRP and blood lipids in patients with RA and CAD [27]. Other herbs, such as Angelica sinensis (Oliv.) Diels (Tang-Kuei) and Sophora flavescens Aiton (Ku-Tsen), were also found to have inhibitory effects on TNF-α in an animal study [28]. Thus, commonly used CHPs might suppress the production of IL-6 and TNF-α, the same proinflammatory cytokines active in the process of atherosclerosis, to reduce the risk of CAD.

In the CHP group, patients who had used CHP for > 180 days were found to have a lower risk of CAD than those who had used CHP for < 30 days. A possible explanation for this is that CHP requires at least 180 days to achieve a steady protective effect in patients with RA. The cumulative number of days of CHP use was similar to the treatment duration of methotrexate recommended by the European League Against Rheumatism [29]. Moreover, the results obtained from the minimum number of days of CHP use were consistent with those of previous randomized controlled trials. Patients with RA taking Paeonia lactiflora Pall. (Shao-Yao) extracts combined with treatment with leflunomide and methotrexate for 24 weeks had lower CRP levels and ESR [19]. Additionally, after 24 weeks of treatment with Salvia miltiorrhiza Bunge (Tan-Tsen) and Pueraria lobata (Willd.) Ohwi (Ko-Ken), there was an improvement in the carotid intima-media thickness in coronary patients [30]. RA is characterized by chronic systemic inflammation, and patients using CHP for > 180 days may achieve an improved anti-inflammatory effect that reduces the risk of CAD.

Finally, several important limitations of our study must be considered. First, certain factors, such as disease activity, disease severity, smoking, body mass index, diet, exercise habits, and stress levels, all of which are associated with CAD, were not measured and were unavailable from our database. However, previous study documented that the use of TNF-antagonists and DMARDs may decrease the risk of CAD [31, 32] Thus, we applied a Cox Proportional Hazards Model by controlling for the available potential confounding factors of CAD, including age, gender, comorbidities and medication, to make the two groups comparable. Second, there was a possibility of misclassification bias in the days of cumulative use of CHP. We presumed that all of the CHP prescribed was taken by the patients, but the actual dosage taken may have been overestimated due to noncompliance. Due to the impact of medication adherence, the protective effect of CHP against CAD may be underestimated. Third, the use of over-the-counter CHP was not recorded in the NHIRD. The database used in our study contained only records of the CHP prescribed by licensed physicians practicing traditional Chinese medicine; thus, the rate of CHP use may have been underestimated.

Conclusion

This study revealed an association with decreased CAD risk in RA patients receiving CHP in combination with conventional treatments. In particular, RA patients who took CHP for > 180 days had a lower risk. Additionally, Kuei-Chih-Shao-Yao-Chih-Mu-Tang, Tang-Kuei-Nien-Tung-Tang, and Shu-Ching-Huo-Hsieh-Tang might be associated with a lower risk of CAD. Future randomized controlled trials are required to clarify the causal association between the use of CHP and the risk of CAD.