Risk of Developing Coronary Artery Disease in Patients with Type 2 Diabetes Receiving Traditional Chinese Medicine Therapy

Abstract

Objectives:

To compare risk of developing coronary artery disease (CAD) between diabetic patients receiving Traditional Chinese Medicine (TCM) therapy and those treated by Western medicine (WM).

Methods:

This prospective cohort study included 13,655 diabetic patients receiving solely TCM and 435,165 patients treated exclusively by WM. Study patients were identified from Taiwan's National Health Insurance (NHI) ambulatory claims in 2000–2001. These patients were then linked to the 2000–2008 NHI inpatient claims, searching for possible new onset of hospitalization for CAD. A Cox proportional hazard model and logistic regression model were used to assess the hazard ratio of CAD admission and odds ratio (OR) of higher rates of admission for CAD in relation to TCM.

Results:

During 9 years of follow-up, 2607 diabetic patients with TCM were hospitalized for CAD, representing a cumulative incidence rate of 19.1% and an incidence density of 50.5 per 1000 person-years. The corresponding figures for patients treated by WM were 24.1% and 72.7 per 1000 person-years. Compared with the patients treated by WM, those treated by TCM were associated with a slightly reduced, but insignificantly, adjusted OR of CAD admission (0.96; 95% confidence interval, 0.92–1.01). Moreover, the adjusted OR for a higher rate (≥0.212 admission per person-year) of CAD admission for the patients with TCM was also insignificantly decreased at 0.97.

Conclusions:

After adjustment for prior co-morbidity score, risk or rate of CAD admission did not significantly differ between diabetic patients receiving TCM therapy and those treated by WM, suggesting that TCM is as efficacious as WM in preventing diabetes from being complicated with CAD.

Introduction

Diabetes mellitus (DM) is an epidemic disease worldwide. Marked changes in human health behaviors and lifestyle have resulted in increasing incidence and prevalence of DM.¹ The prevalence of diabetes for all age groups worldwide was estimated to be 2.8% in 2000 and is predicted to be 4.4% in 2030.² The number of people with diabetes is also increasing because of population growth, aging, urbanization, and increasing prevalence of obesity and physical inactivity.² The potential for increases in the number of patients with diabetes is greatest in Asia.³

In Traditional Chinese Medicine (TCM), diabetes is called Xiaoke disease, which is usually a consequence of obesity. The complications of Xiaoke disease include stroke, carbuncle, and foot gangrene.^4,5 In Chinese, Xiao means “losing body weight” and Ke means “thirsty,” which are similar to the symptoms of diabetes, including losing body weight in the presence of increased drinking, eating, and urination. In the TCM theory, the treatment of diabetes should be focused on replenishing yin (fluid) and evacuating fire (heat) from the body.⁶ Additionally, diabetes is usually associated with inflammation, and there is also evidence of anti-inflammatory and antioxidant effects from TCM therapies.^7,8

TCM, when given to relieve symptoms, has gained widespread popularity among diabetic patients.⁹ Several possible mechanistic pathways with which TCM might present the potential efficacy in the management of diabetes have been proposed. These include increased insulin secretion, improvement in insulin sensitivity, enhanced glucose uptake by adipose and muscle tissues, inhibition of glucose absorption from intestine, inhibition of glucose production from hepatocytes, and anti-inflammatory activities.¹⁰ A recent Taiwanese study reported that among the top 10 most frequently prescribed herbal formulae, four remedies—zhi-bo-di-huang-wan, qi-ju-di-huang-wan, ji-sheng-shen-qi-wan, and ba-wei-di-huang-wan—are derivative formulae of liu-wei-di-huang-wan. In other words, liu-wei-di-huang-wan and its derivatives were the most common herbal formulae prescribed by TCM doctors for the treatment of diabetes in Taiwan.¹¹ Despite that, this particular study¹¹ still argued the need for well-conducted clinical trials to further evaluate the efficacy of liu-wei-di-huang-wan on patients with type 2 diabetes.

Despite the evidence showing efficacy of TCM as a therapeutic option for diabetes, little is known about the relative effectiveness of TCM and Western medicine (WM) in treating diabetes. This large-scale population-based cohort study was performed to compare the risk of developing complicated cardiovascular disease between diabetic patients who received solely TCM therapy and those treated only by WM.

Materials and Methods

Source of data

Data analyzed in this study were retrieved from four datasets of the National Health Insurance Research Database (NHIRD) provided by the Bureau of National Health Insurance (BNHI). These datasets included Ambulatory Care Visit Claims, Inpatient Expenditures by Admissions, Registry for Contracted Medical Facilities, and Registry for Beneficiaries. The NHIRD provides all inpatient and ambulatory medical claims for all residents in Taiwan. The coverage rate of the National Health Insurance (NHI) from 2000 to 2007 had increased steadily from 96.1% to 98.6%.¹² To ensure the accuracy of claim files, the BNHI performed experts' review on a random sample of every 50–100 ambulatory and inpatient claims quarterly, and false reports of diagnosis would face fine penalty from the BNHI.¹³ Therefore, information obtained from the NHIRD is believed to be complete and accurate. The Review Committee of the National Health Research Institutes approved access to research data.

Study design and data linkage

This is a prospective cohort study that identified from Ambulatory Care Visit Claims patients coded with diabetes diagnoses by an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code of 250 or A-code 181 in 2000. To be an eligible participant, the individual identified must experience an additional one or more diabetes diagnoses within the subsequent 12 months after his or her first diabetic diagnosis in 2000. Moreover, the first and last outpatient visits during the 12-month period had to be separated by at least 30 days to avoid accidental inclusion of miscoded patients.¹⁴ The study patients who had a history of hypertensive and cardiovascular diseases (ICD-9-CM codes: 401–405, 410–414, 420–429) were excluded to ensure the accuracy of incident rates of coronary artery disease (CAD) estimated in the present study. Also excluded were patients who had been admitted because of diabetic ketoacidosis (ICD-9-CM code: 250.1), with the discharge diagnosis of insulin-dependent DM (ICD-9-CM code: 250×1, 250×3), and those who received a catastrophic certificate for type 1 DM. Those who had received both TCM and WM therapies (n=3881) were further excluded. On the basis of such criteria, a total of 448,820 individuals were eligible for inclusion as the study participants, most likely to be patients with type 2 DM. Information on sex and age at the first outpatient visit in 2000 was retrieved.

All study patients were categorized into two groups according to specialty of the physician who provided ambulatory care services for the patient's first and last ambulatory care visit in 2000–2001. The two study groups included patients with type 2 diabetes receiving solely TCM therapy (n=13,655) and those treated exclusively by WM (n=435,165). To account for potential confounding by treatment indication, all study patients were traced back for prior disease histories (inpatient/outpatient claims) in 1997–1999, and the Charlson comorbidity score was calculated to indicate the patient's level of comorbidity. The Charlson score is a method of summarizing 17 categories of comorbidities based on the ICD diagnoses codes of individual patients using administrative data, such as medical claims. Each comorbidity category has an associated weight, based on the adjusted risk of 1-year mortality, and the sum of all the weights will yield a single index score (i.e., Charlson score) for a patient. A Charlson score of 0 indicates that no comorbidities were found. Higher Charlson scores indicate higher medical resource use and higher risk of poor prognosis.^15,16

The two study groups were thereafter linked to the inpatient data (i.e., Inpatient Expenditures by Admissions) of 2000–2008, searching for individual's possible hospitalizations with primary or secondary diagnoses of coronary artery diseases (CAD), including acute myocardial infarction (ICD-9-CM code: 410), ischemic heart disease (ICD-9-CM code: 411–414), and coronary revascularization procedures (ICD-9-CM code: 36.0, 36.01, 36.02, 36.05, 36.06, 36.1, 36.10–36.19). The study cohort was also linked to the Registry for Contracted Medical Facilities and Registry for Beneficiaries for the information on characteristics of medical institutions (accreditation level and type) and study subjects (insurance premium, geographic districts and urbanization level of residential area), respectively.

Statistical analysis

The characteristics of patients and medical institutions were compared between the two study groups. Incidence density (ID) of CAD admission was calculated for the two study groups as the ratio of number of new-onset CAD admissions to the total person-years (PYs) observed. The PY was calculated as the difference between date of the first ambulatory care visit for diabetes in 2000 and date at end of follow-up, which is either the date of withdrawal (including death) from the NHI program or date of study termination (i.e., the last day of 2008).

Risk of CAD admission and number of CAD admissions per PY were the two primary outcomes in this study. Risk of CAD admission was measured as a binary variable (i.e., yes/no), while the number of CAD admissions per PY was further multinominally categorized, using a median value (0.212 CAD admissions per PY) as the cutoff point, into a lower rate (>0 to <0.212 CAD admissions per PY) and a higher rate (≥0.212 CAD admissions per PY). A Cox proportional hazard model was used to estimate the hazard ratio (HR) and 95% confidence interval (CI) of encountering CAD admission. A multinomial logistic regression model was further used to estimate the odds ratios (ORs) with 95% CIs of the lower and higher rates, respectively, of CAD in relation to TCM.

The relative risk estimates estimated from both regression models were adjusted for the characteristics of patients and medical institutions listed in Table 1. Two regional variables were adjusted for because there is a clear urban-rural difference in accessibility to medical care in Taiwan.¹⁴ SAS software (version 9.3; SAS Institute, Cary, NC) was used for the statistical analyses. A p-value <0.05 was considered to indicate a statistically significant difference.

Table 1.

Characteristics of Study Patients According to Treatment Received

Variable ^a	WM	TCM
Age
<45 y	42,062 (9.6)	2032 (14.8)
45–64 y	198,756 (45.7)	7623 (55.8)
>65 y	194,203 (44.7)	3999 (29.4)
Mean age (y)	61.2±12.9	56.9±12.2
Sex
Female	229,680 (52.8)	6332 (46.4)
Male	205,175 (47.2)	7316 (53.6)
Insurance premium
Dependent	119,693 (27.5)	3223 (23.6)
1∼<18,000 NTD	101,540 (23.3)	2822 (20.7)
18,000∼<30,000 NTD	170,816 (39.3)	5516 (40.4)
≥30,000 NTD	43,116 (9.9)	2094 (15.3)
Mean premium (NTD)	13,726.9±14,283.6	16,799.4±15804.5
Charlson score
1	299,603 (68.9)	12,025 (88.1)
2	96,736 (22.2)	1114 (8.1)
≥3	38,826 (8.9)	516 (3.8)
Mean score	1.5±0.8	1.2±0.6
Accreditation of medical institution
Medical center	111,133 (25.6)	1653 (12.1)
Regional hospital	104,841 (24.1)	1515 (11.1)
Local hospital	122,300 (28.1)	1664 (12.2)
Clinics	96,669 (22.2)	8821 (64.6)
Geographic area
Northern	190,547 (44.7)	5029 (37.4)
Central	101,282 (23.8)	3626 (27.0)
Southern	121,370 (28.5)	4425 (32.9)
Eastern	12,799 (3.0)	364 (2.7)
Urbanization status
Metropolis	183,995 (43.1)	6207 (46.1)
Satellite city/town	111,073 (26.0)	3431 (25.5)
Rural area	131,901 (30.9)	3813 (28.4)
Total	435,165 (100.0)	13,655 (100.0)

Inconsistency between total population and population summed for geographic areas and urbanization status was due to missing information. Values expressed with a plus/minus sign are the mean±standard deviation.

WM, Western medicine; TCM, Traditional Chinese Medicine; NTD, New Taiwan Dollar (1 USD is approximately 30 NTD).

Results

Table 1 shows the characteristics of study patients and medical institutions. Patients receiving WM therapy and those receiving TCM therapy differed in nearly all variables listed in Table 1. The diabetic patients in the TCM group tended to be younger (56.9 years versus 61.2 years), to have more men than women (53.6% versus 47.2%), to pay a higher insurance premium, and have a lower average Charlson score (1.2 points versus 1.5 points). Distributions of medical institution and geographic areas also substantially differed between the two groups. More than 64% of the study patients with TCM were treated in clinics compared with only 22.2% of the WM group. Additionally, only 37.4% of the study patients in the TCM group were employed or living in northern Taiwan versus 44.7% of the patients in the WM group.

During nearly 9 years of follow-up, 115,667 PYs and 3,501,176 PYs were accumulated for the TCM and WM groups, respectively. The number of CAD admissions was 5844 for the TCM group, representing an ID of 50.5 admissions per 1000 PYs. The ID for the WM group was estimated at 72.2 admissions per 1000 PYs based on 252,791 admissions (Table 2).

Table 2.

Incidence Densities of Admission due to Coronary Artery Diseases for Diabetic Patients According to Treatment Received

Therapy group	Observed person-years ^a	No. of admissions ^b	Incidence density ^c : estimate (95% CI)
WM	3,501,176	252,791	72.2 (72.2–72.3)
TCM	115,667	5,844	50.5 (50.4–50.5)
Total	3,616,943	25,8635	71.5 (29.5–29.6)

Person-years observed between date of the first ambulatory care visit for diabetes in 2000 and date of study termination (termination of health policy or December 31, 2008).

Discharge with acute myocardial infarction (International Classification of Diseases, Ninth Revision Clinical Modification [ICD-9-CM] code: 410), ischemic heart disease (ICD-9-CM code: 411–414), or coronary revascularization procedures (ICD-9-CM code: 36.0, 36.01, 36.02, 36.05, 36.06, 36.1, 36.10–36.19).

Per 1000 person-years.

CI, confidence interval.

A total of 2607 diabetic patients in the TCM group were admitted with a primary or secondary diagnosis of CAD, representing a cumulative incidence rate (CIR) of 19.1% (2607/13,655) during the study period. The corresponding CIR for the diabetic patients with WM was estimated at 24.1% (105,007/435,165). Compared with the patients in the WM group, the patients receiving TCM therapy had a significantly reduced HR for CAD onset (0.74; 95% CI, 0.71–0.78). However, the significant reduction in HR for CAD onset disappeared as the potential confounders were considered in the multiple regression model (adjusted HR, 0.96; 95% CI, 0.92–1.01) (Table 3).

Table 3.

Hazard Ratio for Admission due to Coronary Artery Disease in Relation to Traditional Chinese Medical Care Among Diabetic Patients

	CAD admission ^a		Crude estimate	Adjusted estimate ^b
Therapy group	No	Yes	HR (95% CI)	HR (95% CI)
WM	330,158	105,007	1.00	1.00
TCM	11,048	2,607	0.74 (0.71–0.78)	0.96 (0.92–1.01)
Total	341,206	107,614

Discharge with acute myocardial infarction (ICD-9-CM, code: 410), ischemic heart disease (ICD-9-CM code: 411–414), or coronary revascularization procedures (ICD-9-CM code: 36.0, 36.01, 36.02, 36.05, 36.06, 36.1, 36.10–36.19).

Adjustment for age, sex, insurance premium, Charlson score, medical institution accreditation, geographic area, and urbanization.

HR, hazard ratio.

Table 4 shows the distributions of CAD admission frequency per PY for both groups. In the TCM group, 1192 (8.73%) patients experienced at least 0.212 CAD admissions per PY and 1420 (10.40%) experienced more than 0 to less than 0.212 CAD admissions per PY. The corresponding figures for the WM group were 53,580 (12.31%) and 51,556 (11.85%). Compared with the patients in the WM group, the patients receiving TCM therapy had an essentially null adjusted OR for experiencing higher frequency (i.e., >0.212 admissions per PY) of hospitalization due to CAD, with an adjusted OR of 0.97 (95% CI, 0.91–1.03).

Table 4.

Odds Ratios for Lower and Higher Frequencies of Coronary Artery Disease Admission in Relation to Traditional Chinese Medicine Care Received by Diabetic Patients

		Number of CAD ^a admissions per person-year
		0	>0 to <0.212 ^b		≥0.212
Therapy group	Patients (n)	Patients (n)	Patients (n)	AOR ^c (95% CI)	Patients (n)	AOR ^c (95% CI)
WM	435,165	330,029	51,556	1.00	53,580	1.00
TCM	13,655	11,043	1420	0.95 (0.90–1.01)	1192	0.97 (0.91–1.03)
Total	448,820	341,072

Including acute myocardial infarction (ICD-9-CM code: 410), ischemic heart disease (ICD-9-CM code: 411–414), or coronary revascularization procedures (ICD-9-CM code: 36.0, 36.01, 36.02, 36.05, 36.06, 36.1, 36.10–36.19).

Median frequency for all study patients.

Adjustment for age, sex, insurance premium, Charlson score, medical institution accreditation, geographic area, and urbanization. AOR, adjusted odds ratio.

Discussion

This nationwide 9-year cohort study revealed little difference in the risk for CAD onset between the WM and TCM groups. This is believed to be the first study that performed a head-to-head comparison of CAD risk between type 2 diabetes treated by WM and those receiving TCM therapy. The results supported previous findings that TCM holds great and unique potential in the management of metabolic syndrome, especially in the control of certain cardiovascular risk factors, such as glucose and lipid metabolism.¹⁷

A recent report estimated that 382 million people worldwide have diabetes and most will likely die of cardiovascular disease.¹⁸ This estimation was based on an observation that diabetes is an independent risk factor for atherosclerotic cardiovascular disease as well as heart failure, with a 5-fold increased risk of heart failure in women with diabetes and a 2.4-fold increased risk in men.^19
–21 Satisfactory glycemic control in patients with diabetes is associated with reduced risks for atherosclerotic cardiovascular disease and new-onset heart failure.^19,22 A Cochrane review indicated that metformin monotherapy showed a significant benefit for glycemia control, weight, dyslipidemia, and diastolic blood pressure. Additionally, metformin presents a strong benefit for hemoglobin A1c (HbA1c) compared with placebo and diet.²³ A recent review further suggested that in patients with type 2 diabetes who do not achieve the glycemic targets with metformin alone, dipeptidyl peptidase-4 (DPP-4) inhibitors can reduce HbA1c.²⁴

It has been widely believed that decreasing HbA1c levels with glucose-lowering medications in patients with diabetes would result in clinical benefits, including the reduction of atherosclerotic cardiovascular events.¹⁸ Reduction of HbA1c levels by glucose-lowering medications, such as metformin and DPP-4 in patients with diabetes has been used as a surrogate measure of clinical benefit of those mediations. Given the information above, the current study findings tended to indicate that the treatments provided by WM and TCM are equally effective in reducing the risk of developing CAD.

Several previous studies also reported safety and effectiveness of TCM therapies administered to patients with diabetes. An earlier Cochrane review included 66 randomized trials, involving 8302 participants and 69 different herbal medicines tested in the included trials. The review concluded that some herbal medicines show hypoglycemic effects in type 2 diabetes. However, the authors also called for careful interpretation of the study findings due to the low methodologic quality, small sample size, and limited number of many trials.²⁵ A recent controlled, double-blind, multicenter trial found that treatment with Xiaoke pill, a compound of Chinese herbs combined with glibenclamide, led to significant reduction in risk of hypoglycemia and similar improvements in reaching optimal glycemic control (i.e., HbA1c level <6.5%) after 48 weeks, as compared to glibenclamide in patients with type 2 diabetes and inadequate glycemic control.¹⁷

The current study had several methodologic strengths. First, the diabetic cohort and control group were collected from the NHI database, which is population based and is highly representative, allowing little room for recall and selection bias; there is also less likelihood of nonresponse and loss to follow-up of cohort members. Second, the advantage of using insurance claim data sets in clinical research is easy access to the longitudinal records for a large sample of demographically disperse patients, resulting in good generalizability of the study findings.

Despite the above strength, the study has several limitations. First, exclusive reliance on the claim data might have resulted in potential disease misclassification bias in this study. However, at least two diabetes diagnoses, with the first and the last visits more than 30 days apart, were used, which largely reduced the likelihood of disease misclassification.²⁶ Second, to determine type 2 DM, the study excluded patients who had been admitted because of diabetic ketoacidosis, those with the discharge diagnosis of insulin-dependent diabetes mellitus, and those who received a catastrophic certificate for potential type 1 DM. Nonetheless, some patients with type 2 DM may also have been excluded by this algorithm. The control group might also have consisted of new-onset or undiagnosed diabetes. Such misclassification bias, however, was likely to be non-differential, which might underestimate the true relative risks.²⁷ Moreover, type 1 diabetes constitutes only 1.8% of all diabetes in Taiwan,²⁸ which provides further reassurance that the study patients are likely to have type 2 DM.

Third, the national diabetes surveillance system was similar to that implemented in other countries. Information on potential confounders, such as body mass index or waist circumference, blood pressure, laboratory data, smoking, and family history of cardiovascular disease, was not available from the claims database; this also limits the specific interpretations of our study. Finally, there is a potential for exposure misclassification arising from the purchase of Chinese herbs outside of established medical institutions by the study patients in the WM group. Unfortunately, the patients could not be contacted directly about their use of herbs for diabetes because of anonymization of their identification numbers in the NHI claims. However, because the NHI system has comprehensive coverage and the copayment for prescriptions is universally 50 NTD (new Taiwan dollar) (approximately equal to US $1.5), which is generally less than the cost of herbs sold in Taiwan's markets, the likelihood that patients purchased herbs without a prescription would be low.²⁹ If the patients from the WM group did take some herbs outside of established medical institutions, the findings would underestimate the difference in treatment effect between TCM and WM.

In conclusion, this study found no significant difference in the incidence rate of CAD between diabetic patients treated by TCM and those receiving WM therapies, suggesting that TCM may be as efficacious as Western medicine in preventing patients with diabetes from developing CAD. Specific interpretations of the study results may be derived from further investigations looking for specific TCM treatment regimens that provide the best cardiovascular outcomes.

Footnotes

Acknowledgments

This study was supported by grants from the National Scientific Council and Sin-Lau Hospital (grant number NSC101-2314-B-006-076-MY3). The interpretation and conclusions contained herein do not represent those of the BNHI, Department of Health, or NHRI.

Author Disclosure Statement

No competing financial interests exist.

References

Zimmet

, Alberti

, Shaw

. Global and societal implications of the diabetes epidemic. Nature, 2001; 414:782–787.

Wild

, Roglic

, Green

, et al. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care, 2004; 27:1047–1053.

Amos

, McCarty

, Zimmet

. The rising global burden of diabetes and its complications: estimates and projections to the year 2010. Diabet Med, 1997; 14:S1–S85.

Wang

. Experience in clinical practice of doctrines in Huangdi Neijing: Part 1. Zhong Xi Yi Jie He Xue Bao, 2005; 3:486–488.

, Guo

, Li

, Wang

, Fu

, Wang

. Integration of traditional Chinese medicines and Western medicines for treating diabetes mellitus with coronary heart disease: a systematic review. J Altern Complement Med, 2013; 19:492–500.

, Zheng

, Bukuru

, De Kimpe

. Natural medicines used in the traditional Chinese medical system for therapy of diabetes mellitus. J Ethnopharmacol, 2004; 92:1–21.

Xie

, Du

. Diabetes is an inflammatory disease:evidence from traditional Chinese medicines. Diabetes Obes Metab, 2011; 13:289–301.

Poon

, Ong

, Cheung

. Review of the effects of the traditional Chinese medicine Rehmannia Six Formula on diabetes mellitus and its complications. J Diabetes, 2011; 3:184–200.

Fang

, Tsai

, Lai

, Yeh

, Wu

. The traditional Chinese medicine prescription pattern of endometriosis patients in Taiwan: a population-based study. Evid Based Complement Alternat Med, 2012; 2012:591391.

10.

, Kam

, Wong

, et al. Herbal medicines for the management of diabetes. Adv Exp Med Biol, 2012; 771:396–413.

11.

Huang

, Tsai

, Lai

, Hsu

. Prescription pattern of Chinese herbal products for diabetes mellitus in Taiwan: a population-based study. Evid Based Complement Alternat Med, 2013; 2013:201329.

12.

Chiu

, Lai

. Nationwide population-based epidemiologic study of systemic lupus erythematosus in Taiwan. Lupus, 2010; 19:1250–1255.

13.

Chen

, Liu

, Chen

, et al. Risks of breast and endometrial neoplasm in women with diabetes: a population-based cohort study. PLoS One, 2013; 8:e67420.

14.

Chen

, Ho

, Li

. Age and sex may significantly interact with diabetes on the risks of lower extremity amputation and peripheral revascularization procedures: evidence from a cohort of a half million diabetic patients. Diabetes Care, 2006; 29:2409–2414.

15.

Charlson

, Pompei

, Ales

, MacKenzie CRA new method of classifying prognostic comorbidity in longitudinal studies:development and validation. J Chronic Dis, 1987; 40:373–383.

16.

Deyo

, Cherkin

, Ciol

. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol, 1992; 45:613–619.

17.

, Tong

, Wang

, et al. Efficacy and safety of traditional Chinese medicine for diabetes:a double-blind, randomised, controlled trial. PLoS One, 2013; 8:e56703.

18.

Fonarow

. Diabetes medications and heart failure: recognizing the risk. Circulation., 2014; 130:1565–1567.

19.

Seshasai

, Kaptoge

, Thompson

, et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med, 2011; 364:829–841.

20.

Nichols

, Gullion

, Koro

, Ephross

, Brown

. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care, 2004; 27:1879–1884.

21.

Kannel

, Hjortland

, Castelli

. Role of diabetes in congestive heart failure: the Framingham study. Am J Cardiol, 1974; 34:29–34.

22.

Holman

, Sourij

, Califf

. Cardiovascular outcome trials of glucose-lowering drugs or strategies in type 2 diabetes. Lancet, 2014; 383:2008–2017.

23.

Saenz

, Fernandez-Esteban

, Mataix

, Ausejo

, Roque

, Moher

. Metformin monotherapy for type 2 diabetes mellitus. Cochrane Database Syst Rev, 2005; 3:CD002966.

24.

Karagiannis

, Paschos

, Paletas

, Matthews

, Tsapas

. Dipeptidyl peptidase-4 inhibitors for treatment of type 2 diabetes mellitus in the clinical setting: systematic review and meta-analysis. BMJ, 2012; 344:e1369.

25.

Liu

, Zhang

, Wang

, Grimsgaard

. Chinese herbal medicines for type 2 diabetes mellitus. Cochrane Database Syst Rev, 2004; CD003642.

26.

Chen

, Li

. Effect-modifications by age and sex on the risks of coronary artery disease and revascularization procedures in relation to diabetes. Diabetes Res Clin Pract, 2007; 75:88–95.

27.

Gordis

. Epidemiology. 2nd ed. Philadelphia: WB Saunders, 2000.

28.

Chuang

, Tsai

, Huang

, Tai

. The current state of diabetes management in Taiwan. Diabetes Res Clin Pract, 2001; 54:S55–S65.

29.

Lai

, Wang

, Chen

, Wang

. Population-based case-control study of Chinese herbal products containing aristolochic acid and urinary tract cancer risk. J Natl Cancer Inst, 2010; 102:179–186.