Coronary Catheterization and Percutaneous Coronary Intervention in China: 10-Year Results From China PEACE-Retrospective CathPCI Study

Abstract

IMPORTANCE

The use of coronary catheterization and percutaneous coronary intervention (PCI) is increasing in China, but there are no nationally representative assessments of the quality of care and outcomes of patients undergoing these procedures.

OBJECTIVE

To assess quality of care and outcomes of patients undergoing coronary catheterization and PCI.

DESIGN, SETTING, AND PARTICIPANTS

We used a two-stage random sampling strategy to create a nationally representative sample of 11 241 patients undergoing coronary catheterization and PCI at 55 urban Chinese hospitals in 2001, 2006, and 2011.

MAIN OUTCOMES AND MEASURES

Patient characteristics, treatment patterns, quality of care and outcomes associated with these procedures and changes over time.

RESULTS

Between 2001 and 2011, national rates of hospitalizations for coronary catheterization and PCI increased by 17 fold and 21 fold, respectively. More than half of stable patients undergoing coronary catheterization had non-obstructive coronary artery disease; which did not change over time (2001: 60.3% [95% CI, 56.1–64.5%]; 2011: 57.5% [95% CI, 55.8–59.3%], P_trend=.05). The proportion of PCI procedures performed via radial approach increased from 3.5% (95% CI, 1.7–5.3%) in 2001 to 79.0% (95% CI, 77.7–80.3%) in 2011 (P_trend<.001). The use of drug-eluting stents (DES) increased from 18.0% (95% CI, 14.2–21.7%) in 2001 to 97.3% (95% CI, 96.9–97.7%) in 2011 (P_trend<.001), largely due to increased use of domestic DES. The median length of stay decreased from 14 days (IQR 9–20) in 2001 to 10 days (IQR 7–14) in 2011 (P_trend <.001). In-hospital mortality did not change significantly, but both adjusted risk of any bleeding (OR 0.53 [95% CI 0.36–0.79], P_trend <.001) and access bleeding (OR 0.23 [95% CI 0.12–0.43], P_trend <.001) were decreased between 2001 and 2011. The medical record lacked documentation needed to calculate commonly used process metrics including door to balloon times for primary PCI, and the prescription of evidence-based medications at discharge.

CONCLUSIONS AND RELEVANCE

Although the use of catheterization and PCI in China has increased dramatically, we identified critical quality and documentation gaps that represent opportunities to improve care. Our findings can serve as a foundation to guide future quality improvement initiatives in China.

Introduction

In China, the dramatic increase in the burden of ischemic heart disease has been accompanied by expanded use of coronary catheterization and percutaneous coronary intervention (PCI).^1–4 This growth has occurred during a period of dynamic change within the Chinese healthcare system, characterized by substantial investments in healthcare infrastructure and broader availability of health insurance coverage. Today, China faces a challenge in ensuring that invasive cardiovascular procedures are performed appropriately, safely and efficiently. However, the rapid growth in procedural volume has not been accompanied by efforts to systematically monitor the quality of care and outcomes of patients undergoing these procedures, due in part to limited resources to support quality improvement activities.⁵ The absence of national assessment on the practice of interventional cardiology remains an important barrier to the identification of gaps in care and the implementation of quality improvement efforts.

To provide a comprehensive overview of coronary angiography and PCI in China over the past decade, we conducted the China PEACE-Retrospective Study of Coronary Catheterization and Percutaneous Coronary Intervention (China-PEACE-Retrospective CathPCI Study). The study identified a representative sample of 55 urban hospitals performing PCI and abstracted the medical records of 11 241 patients who underwent coronary angiography or PCI during 2001, 2006 and 2011.⁶ We assessed patient characteristics, treatment patterns, quality of care, and in-hospital outcomes of patients undergoing coronary catheterization and PCI in China and their change over time.

Methods

Study Design

The design of the China PEACE-Retrospective CathPCI Study has been published previously.⁶ In brief, we used a two-stage random sampling design to create a representative sample of patients undergoing coronary catheterization within urban regions, as catheterization capability is restricted almost exclusively to these regions, in 2001, 2006 and 2011. In the first stage, we divided urban areas into two study strata: Eastern-urban and Central/Western-urban. We randomly sampled among the highest-level hospitals in each stratum using a simple random sampling procedure. We selected representative hospitals from 2011 to reflect current practices and traced this cohort of hospitals back to 2006 and 2001 to identify hospitals in these time periods. In the second stage, we identified all eligible cases within each sampled hospital using systemic random sampling procedure. We doubled cluster sizes for 2011 to improve the precision of estimates on hospital-level treatment patterns and outcomes this year. We collected data through central abstraction of medical charts using standard data definitions. To ensure data quality, we monitored data abstraction by random auditing of 5% of the medical records, with accuracy exceeding 98%.⁶

Both the central ethics committee at the China National Centre of Cardiovascular Disease and the Human Investigation Committee of the Yale University School of Medicine approved the China PEACE-Retrospective CathPCI Study. Fifty collaborating hospitals accepted the central ethics committee approval, and five hospitals received approval from their local ethics committees.

Study Population

Patients undergoing coronary catheterization or PCI at participating hospitals for any indication were included in the study sample. For the analysis of in-hospital outcomes and the proportion of PCI procedures with creatinine and cardiac biomarkers assessed post-PCI, patients who were transferred out were excluded.

Variables and Definitions

We investigated the discharge diagnoses of patients undergoing coronary angiography and PCI. Patients with a discharge diagnosis consistent with coronary artery disease (CAD) but not acute coronary syndromes (ACS) were considered to have stable CAD. If the local diagnosis was not definitive, cardiologists at the coordinating center reviewed the medical record, to determine a discharge diagnosis.

We abstracted patient demographic, clinical characteristics and treatment patterns with regard to indication(s) for coronary catheterization and PCI; PCI-related characteristics and adjunctive therapies; and medications used during hospitalization and at discharge. We determined whether the medical record captured key variables needed to assess quality of PCI care including assessment of creatinine, cardiac biomarkers, contrast volume, procedural success, and the timeliness of primary PCI.⁷ In addition, we assessed whether medications prescribed at discharge were documented in the medical record, and the proportion of patients with discharge medication information discharged on aspirin, a thienopyridine, and a statin. Finally, among the patients undergoing PCI, we assessed in-hospital outcomes by collecting data on death, treatment withdrawal, a major adverse cardiac event (death, treatment withdrawal, stroke or repeat target vessel revascularization), bleeding events, and length of stay.^8,9 Bleeding events were classified as any bleeding, major bleeding, access bleeding, and bleeding requiring blood transfusion. The relevant definitions are described in eAppendix 1.

Statistical Analysis

We used percentages with 95% confidence intervals (CIs) to describe categorical variables and medians with interquartile ranges (IQRs) to describe continuous variables. To estimate nationally representative rates in each study year, we calculated the sampling weights proportional to the inverse sampling fraction of patients within each stratum, which equals the product of the sampling fraction of patients within each hospital and the sampling fraction of hospitals within each stratum, then standardized them within each year to the total sampling number of the year so that the comparison would not be affected by differences in sample size (eAppendix 2). We applied these standardized weights for all analyses. We examined changes in geographic and clinical characteristics, treatments patterns, unadjusted rate and adjusted odds ratio (OR) of outcomes, and quality metrics across different study years using the Cochran-Armitage trend test for the trend of binary variables and the Mann-Kendall trend test for trends of continuous variables. All trend tests were based on three time points (2001, 2006, and 2011).

To compare the adjusted risk of in-hospital death, death or treatment withdrawal, composite endpoints and bleeding events, we used time indicators for 2006 and 2011 as key explanatory variables using 2001 as the reference group. Given the relatively low rates of outcomes, we selected the variables that were both clinically and statistically significantly associated with the risk of adverse outcomes (eAppendix 3).^10,11 These variables included cardiogenic shock, ST segment elevation myocardial infarction (STEMI) versus non STEMI, estimated glomerular filtration (eGFR), gender and age. To account for clustering of patients within hospitals, we established multilevel logistic regression models with use of generalized estimating equations. We reported ORs and 95% CIs from the multilevel logistic regression related to the year indicators. The linear trend over time in the models was tested.

All comparisons were two-sided, with statistical significance defined as p value less than .05. Statistical analyses were performed using SAS software (Cary, NC), version 9.2, and R software, version 3.0.2.

Results

From a total of 833 urban hospitals, we sampled 70 hospitals and invited them to participate in the study. We excluded 15 hospitals because they did not provide inpatient services (n=5), had no capability for coronary catheterization or PCI (n=8), or declined participation (n=2). Among the remaining 55 hospitals, the availability of coronary angiography and PCI services varied over time (2001: 24 hospitals; 2006: 44 hospitals; 2011: 54 hospitals). Examination of patient databases from the 55 hospitals yielded 58731 eligible hospital admissions. We randomly identified a subset of 12 477 hospitalizations and acquired 11 900 medical records (95.4%). We excluded 659 ineligible hospitalizations providing a final sample of 11 241 hospitalizations with coronary catheterization (1000 in 2001, 2755 in 2006, and 7486 in 2011) of which 5460 (48.6%) cases included a PCI (eAppendix 4). After weighting these data to estimate national rates, the use of coronary angiography and PCI increased by 17 fold and 21 fold, respectively between 2001 and 2011 (coronary angiography: 26570 in 2001, 133 942 in 2006, and 452 784 in 2011; PCI: 9678 in 2001, 62 308 in 2006, and 208 954 in 2011) (Figure 1).

Figure 1. Hospital Admissions and PCI indication.

(A) Trends in hospital admissions for coronary catheterization and PCI (P_trend< .001); (B) Trends in the proportion of PCI procedures for stable CAD (P_trend< .001), UA (P_trend< .001), NSTEMI (P_trend< .001) and STEMI (P_trend= .004) among all the PCI procedures; (C) Trends in the proportion of primary PCI (P_trend= .07), PCI after fibrinolytic therapy (P_trend< .001) and late reperfusion ^a(P_trend<.001) among all the PCI procedures for patients with STEMI. CAD: coronary artery disease; NSTEMI: non ST- segment elevation myocardial infarction; PCI: percutaneous coronary intervention; STEMI: ST-segment elevation myocardial infarction; UA: unstable angina. ^a For patients who didn’t receive fibrinolytic therapy or primary PCI during the same admission.

Characteristics of Patients Undergoing Coronary Angiography

Between 2001 and 2011, there were modest changes in the characteristics of patients undergoing cardiac catheterization. Compared with 2001, higher proportions of patients in 2011 were female, aged ≥70 years, and had comorbid conditions including diabetes, hypertension, and prior PCI. However, the proportions of patients with renal dysfunction (estimated GFR ≤60) and prior myocardial infarction (MI) were lower. Among all the patients undergoing coronary catheterization, the proportion of cases with STEMI decreased while that of non-ST segment elevation myocardial infarction (NSTEMI) increased. Compared with 2001, a lower proportion of patients in 2006 and 2011 had non-obstructive CAD, and a higher proportion of patients were found to have left main disease. However, among stable patients, the proportion of patients with non-obstructive CAD did not change significantly over time (P_trend=.05) (Table 1).

Table 1.

Demographic, Clinical, Angiographic Characteristics of Patients Undergoing Catheterization (with or without PCI)

	2001 (n=1000)		2006 (n=2755)		2011 (n=7486)		Ptrend
	n	Weighted %	n	Weighted %	n	Weighted %
Demographics
Female	285	28.6(25.8–31.4)	826	32.2(30.5–34.0)	2588	35.7(34.6–36.7)	<.001
Age(years)	59(49, 66)		61(52, 69)		61(54, 70)		<.001
<60	509	51.1(48.0–54.2)	1244	47.6(45.7–49.4)	3263	43.5(42.3–44.6)	<.001
60–69	329	32.1(29.2–35.0)	832	29.8(28.1–31.6)	2279	31.7(30.6–32.7)	.51
70–79	157	15.9(13.6–18.2)	620	20.2(18.7–21.7)	1692	21.4(20.5–22.3)	<.001
≥80	5	0.8(0.2–1.3)	59	2.4(1.8–3.0)	252	3.5(3.0–3.9)	<.001
Co-morbidities
Hypertension	528	54.0(50.9–57.1)	1528	57.6(55.8–59.5)	4502	59.4(58.3–60.5)	.001
Dyslipidemia	576	55.5(52.4–58.5)	1736	65.6(63.9–67.4)	5015	66.7(65.7–67.8)	<.001
Diabetes	161	16.8(14.5–19.1)	567	20.4(18.9–21.9)	1702	22.7(21.8–23.7)	<.001
Current smoking	296	31.1(28.2–33.9)	817	30.5(28.8–32.2)	2207	30.3(29.3–31.4)	.64
Prior MI	240	25.4(22.7–28.1)	630	22.8(21.2–24.4)	1509	19.6(18.7–20.5)	<.001
Prior PCI	32	3.3(2.2–4.4)	159	6.2(5.3–7.1)	696	10.1(9.4–10.8)	<.001
Prior CABG	8	1.0(0.4–1.7)	11	0.4(0.2–0.7)	30	0.5(0.3–0.6)	.10
Prior stroke	56	6.9(5.3–8.4)	157	6.0(5.1–6.9)	491	6.5(5.9–7.1)	.93
Presentations at admission
eGFR (mL/min per1.73m2)	78.5(65.2, 92.9)		83.5(67.7, 99.4)		90.3(74.7, 107.1)		<.001
eGFR<60	135	13.6(11.5–15.8)	380	13.0(11.8–14.3)	707	9.1(8.5–9.8)	<.001
eGFR≥60	637	65.0(62.1–68.0)	2038	75.4(73.8–77.0)	6231	82.9(82.0–83.7)	<.001
eGFR unmeasured	228	21.3(18.8–23.9)	337	11.5(10.4–12.7)	548	8.0(7.4–8.6)	<.001
Heart failure at admission	44	3.9(2.7–5.1)	123	4.7(3.9–5.5)	295	3.5(3.1–3.9)	.05
Cardiogenic shock at admission	5	0.3(0.0–0.6)	19	0.5(0.2–0.8)	45	0.6(0.4–0.7)	.28
Cardiac arrest at admission	0	0.0(0.0–0.0)	1	0.0(0.0–0.1)	2	0.0(0.0–0.0)	.84
Extent of CAD
Non-obstructive CAD (<50%)	345	36.0(33.0–38.9)	747	30.4(28.7–32.1)	2022	28.4(27.4–29.4)	<.001
1-vessel disease	271	25.1(22.4–27.8)	676	22.9(21.3–24.4)	1866	24.7(23.7–25.6)	.46
2-vessel disease	225	22.1(19.6–24.7)	739	26.0(24.4–27.6)	1967	26.1(25.1–27.1)	.03
3-vessel disease	153	16.5(14.2–18.8)	586	20.5(19.0–22.0)	1621	20.7(19.8–21.6)	.01
Left main disease	20	2.2(1.3–3.2)	131	4.4(3.6–5.2)	399	6.2(5.6–6.7)	<.001
Diagnostic catheterization process metrics
Incidence of non-obstructive CAD a	300/522	60.3(56.1–64.5)	637/1129	61.2(58.4–64.1)	1638/2967	57.5(55.8–59.3)	.05
Discharge diagnosis
STEMI	219	20.4(17.9–22.9)	735	23.0(21.5–24.6)	1606	18.9(18.0–19.8)	.001
NSTEMI	19	1.7(0.9–2.5)	109	3.8(3.1–4.5)	409	4.9(4.4–5.4)	<.001
Unstable angina	240	26.4(23.7–29.1)	779	31.3(29.6–33.1)	2498	36.3(35.2–37.3)	<.001
Stable CAD	358	33.7(30.8–36.7)	785	27.8(26.1–29.5)	2066	27.0(26.0–28.0)	<.001
Non-CAD	164	17.7(15.4–20.1)	344	13.8(12.6–15.1)	899	12.7(11.9–13.4)	<.001

^aAmong the patients without acute coronary syndrome.

CABG: coronary artery bypass graft; CAD: coronary artery disease; eGFR: estimated glomerular filtration rate; MI: myocardial infarction; NSTEMI: non ST-segment elevation myocardial infarction; STEMI: ST-segment elevation myocardial infarction.

Characteristics of Patients Undergoing PCI

The characteristics of patients undergoing PCI showed similar changes over time to the overall study population (eAppendix 5). Of note, the proportion of patients with high-risk features, including cardiogenic shock and cardiac arrest at admission, did not change over time. Although the proportion of PCIs performed in the context of STEMI decreased, the proportion of patients undergoing primary PCI did not change significantly over time. In contrast, the proportions of PCIs performed on patients with NSTEMI increased over the study period (Figure 1).

PCI Procedural Characteristics

From 2001 to 2011, there was a significant increase in the proportion of PCI procedures through radial access. Among procedures with femoral access, vascular closure devices were used in less than 10% of cases, a proportion that did not change substantially from 2001 to 2011. The majority had a single vessel treated, and nearly a quarter of patients in 2011 underwent PCI of more than one vessel. The proportion of cases in which a left main lesion was treated increased slightly, and there were significant increases in the use of drug-eluting stent (DES) and domestic DES in particular. The use of fractional flow reserve and intravascular ultrasound was low in all time periods (Table 2). The proportions of patients who received a glycoprotein IIb/IIIa (GP IIb/IIIa) inhibitor and clopidogrel both increased from 2001 to 2011. The use of statins during hospitalization increased from 60.1% in 2001 to 94.6% in 2011 (eAppendix 6).

Table 2.

Procedure Characteristics of PCI

	2001 (n=419)		2006 (n=1476)		2011 (n=3961)		Ptrend
	Relative frequency	Weighted %	Relative frequency	Weighted %	Relative frequency	Weighted %
Vascular access
Femoral	348/419	85.8(82.5–89.2)	985/1476	60.3(57.8–62.8)	857/3961	19.4(18.2–20.6)	<.001
Radial	15/419	3.5(1.7–5.3)	456/1476	37.4(35.0–39.9)	3035/3961	79.0(77.7–80.3)	<.001
Brachial	1/419	0.6(0.0–1.3)	7/1476	0.4(0.1–0.8)	37/3961	0.6(0.3–0.8)	.75
Unrecorded	55/419	10.1(7.2–13.0)	28/1476	1.8(1.1–2.5)	32/3961	1.0(0.7–1.4)	<.001
Hemostasis technique a
Manual compression	289/348	84.0(80.1–87.8)	815/985	84.0(81.7–86.3)	610/857	73.1(70.2–76.1)	<.001
Sealant	1/348	0.3(0.0–0.8)	25/985	2.9(1.9–4.0)	14/857	1.4(0.6–2.2)	.65
Suture	6/348	0.9(0.0–2.0)	37/985	2.5(1.5–3.4)	42/857	4.2(2.9–5.6)	0.001
Other	0/348	0.0(0.0–0.0)	1/985	0.1(0.0–0.3)	2/857	0.3(0.0–0.7)	.19
Unrecorded	52/348	14.8(11.1–18.6)	107/985	10.5(8.6–12.4)	189/857	21.0(18.2–23.7)	<.001
No. of vessels treated during PCI
1-vessel	295	71.0(66.7–75.4)	1028	69.8(67.5–72.2)	2819	70.4(69.0–71.8)	.95
2-vessel	71	18.2(14.5–21.9)	325	24.2(22.0–26.4)	821	21.3(20.0–22.6)	.82
3-vessel	5	1.3(0.2–2.4)	59	3.1(2.2–4.0)	105	2.7(2.2–3.2)	.47
Multi-vessel	76	19.5(15.7–23.3)	384	27.3(25.0–29.6)	926	24.0(22.7–25.3)	.96
Unrecorded	48	9.5(6.7–12.3)	64	2.9(2.0–3.7)	216	5.6(4.9–6.3)	.82
Lesion locationb
Left main	2/517	0.4(0.0–1.0)	26/2089	1.3(0.8–1.8)	83/5410	2.1(1.7–2.5)	.001
Left anterior descending	256/517	47.7(43.4–52.0)	1012/2089	47.5(45.3–49.6)	2600/5410	47.7(46.4–49)	.91
Circumflex	95/517	19.3(15.9–22.7)	398/2089	20(18.3–21.7)	1067/5410	20(18.9–21.0)	.81
Ramus	4/517	1.0(0.2–1.9)	12/2089	0.6(0.3–1.0)	43/5410	0.9(0.6–1.1)	.71
Right coronary artery	150/517	29.9(26.0–33.9)	640/2089	30.7(28.7–32.7)	1651/5410	30.2(28.9–31.4)	.83
Bypass graft lesions	7/517	1.4(0.4–2.4)	27/2089	1.1(0.6–1.5)	58/5410	0.8(0.6–1.0)	.10
Intracoronary devices used
Aspiration Catheters	0	0.0(0.0–0.0)	3	0.3(0.0–0.6)	44	1.7(1.3–2.1)	<.001
Brachytherapy/Cutting balloon/Roblator	1	0.2(0.0–0.7)	1	0.0(0.0–0.1)	7	0.2(0.1–0.4)	.34
Flowire/IVUS/Pressure wire	1	0.2(0.0–0.6)	10	2.8(2.0–3.7)	13	0.8(0.5–1.1)	.009
Thrombectomy Device	3	0.4(0.0–1.1)	1	0.1(0.0–0.2)	1	0.1(0.0–0.2)	.11
Stent	267	69.4(65.0–73.8)	1291	90.4(88.9–91.9)	3445	89.4(88.5–90.4)	<.001
Stent typec
DES	72/408	18.0(14.2–21.7)	1942/2310	87.2(85.9–88.6)	6229/6399	97.3(96.9–97.7)	<.001
Domestic DES	2/72	1.6(0.0–4.4)	1213/1942	52.5(50.3–54.7)	4968/6229	74.8(73.7–75.8)	<.001
BMS	226/408	54.7(49.8–59.5)	271/2310	9.6(8.4–10.8)	32/6399	0.8(0.6–1.1)	<.001
Unrecorded	109/408	27.3(22.9–31.6)	94/2310	3.1(2.4–3.9)	131/6399	1.8(1.4–2.1)	<.001
The number of stents/patient	1.4(0.7)		1.7(0.9)		1.8(0.9)		<.001
Ad hoc PCI	295	74.3(70.1–78.5)	1237	88.3(86.7–90.0)	3422	89.1(88.1–90.0)	<.001

^aAmong the patients with femoral access;

^bAmong 8016 lesions treated;

^cAmong 9117 stents.

BMS: bare mental stent; DES: drug-eluting stent; IVUS: intravascular ultrasound; PCI: percutaneous coronary intervention.

PCI Quality Metrics

For most quality metrics, the information documented in the medical record was frequently incomplete. For example, in 2011, among patients undergoing primary PCI, only 3.0% and 0.5% of patients had the information about hospital arrival time and balloon dilation time, the key elements needed to determine their door to balloon time. Similarly, in 2011, 51.5% of PCIs lacked documentation about whether or not the procedure was successful, and 36.3% of PCI cases had no information about discharge medications documented. Documentation improved for a few metrics over time including the proportion of patients with a creatinine assessed prior to the procedure and documentations of contrast volume. However, documentation for most metrics, including proportion of primary PCI procedures recording hospital arrival time and balloon dilation time, as well as proportion of procedures with cardiac biomarkers assessed post PCI did not improve from 2001 to 2011 (Figure 2).

Figure 2. PCI Quality Metrics.

^aTrends in the proportion of primary PCI procedures with recording hospital arrival time (P_trend= .10); ^b Trends in the proportion of primary PCI procedures with recording balloon dilation time (P_trend= .86); ^c Trends in the proportion of PCI with missing procedural success indicators (P_trend=.03); ^d Trends in the proportion of successful procedures among PCIs with complete documentation of success indicators (P_trend< .001); ^e Trends in the proportion of PCI procedures with creatine assessed pre PCI (P_trend< .001)^{*; f} Trends in the proportion of procedures with creatine assessed post PCI (P_trend< .001)^{*; g} Trends in the proportion of procedures with cardiac biomarkers assessed post-PCI (P_trend= .64)^{*; h} Trends in the proportion of procedures with documentation of contrast volume (P_trend< .001); ⁱ Trends in the proportion of patients with missing all discharge medications(P_trend= .13); ^j Trends in the proportion of patients with documentation of statin (P_trend< .001);^k Trends in the proportion of patients with documentation of aspirin (P_trend< .001);^l Trends in the proportion of patients with documentation of thienopyridine (clopidogrel or ticlopidine) (P_trend< .001) ^#. PCI: percutaneous coronary intervention. ^* : For the first PCI procedure if more than one procedure during a hospitalization. ^# : Among patients with stents.

PCI Outcomes

The median hospital length of stay was 14 days (IQR 9–20) in 2001, 11 days (IQR 8–16) in 2006 and 10 days (IQR 7–14) in 2011, P_trend <.001. The unadjusted rates of most adverse outcomes did not change over the study period, including the mortality of patients undergoing primary PCI (eAppendix 7). However, the rates of both any bleeding and access bleeding were lower in 2011 than 2001. Similarly, in multivariable analysis, the adjusted risk of both any bleeding and access bleeding event was significantly lower in 2011 than 2001, while adjusted risks of other outcomes were similar across time periods (Figure 3).

Figure 3. Unadjusted Rate and Adjusted Odds Ratio of Adverse Outcomes of Patients Undergoing PCI.

Adjusted odds ratio (OR) of patient outcomes are shown along the horizontal axis with the vertical line demarking an OR of 1 (i.e., no difference from year 2001); estimates to the right (i.e., >1) are associated with higher risk of the outcome, while those to the left (i.e., <1) indicate lower risk of the outcome. The variables for risk adjustment include cardiogenic shock, STEMI versus non STEMI, eGFR, gender, and age. C= 0.77 for death, C= 0.76 for death or treatment withdrawal, C= 0.70 for composite complications, C= 0.64 for any bleeding, C= 0.63 for major bleeding, C= 0.69 for access bleeding, and C= 0.71 for blood transfusion. Composite endpoints: death or withdrawal, stroke or repeat target vessel revascularization. eGFR: estimated glomerular filtration rate; STEMI: ST-segment elevation myocardial infarction.

Discussion

The China-PEACE Retrospective CathPCI Study is the first large, nationally representative study of hospitals in China performing coronary angiography and PCI. We found that the dramatic increase in the rate of hospitalization for coronary catheterization and PCI reflected growth in procedures performed on patients with both stable CAD and ACS. There have been substantial changes in clinical practice, most notably the widespread adoption of radial access and the routine use of DES, especially domestic DES. Most importantly, our study identified opportunities to improve the quality of care delivered to patients undergoing these procedures. A substantial proportion of stable patients undergoing diagnostic catheterization had insignificant CAD. Moreover, the medical record of many patients undergoing PCI lacked documentation of important process measures needed to assess quality of care. These findings serve as a foundation to guide future efforts on quality improvement in the use of coronary catheterization and PCI in China.

Our findings suggest that there may be an opportunity to improve the selection of patients referred for elective coronary catheterization. Among stable patients undergoing coronary angiography, more than half had no hemodynamically significant stenosis, and this proportion did not change over time. This finding is not unique to China, with studies of countries with different healthcare systems demonstrating similarly low yield of coronary angiography.^12–14 We cannot determine this finding reflects limited access to appropriate non-invasive testing, patient preference, or other unmeasured factors. Regardless there may be an opportunity to improve care by prioritizing efforts to make sure that patients undergo appropriate risk stratification prior to the performance of coronary angiography.^15,16

We identified several important changes in the practice of interventional cardiology in China. First, radial artery access was widely adopted during this timeframe, increasing from 3.5% to 79.0%. The emergence of radial access as the preferred vascular access strategy is supported by the studies demonstrating higher patient satisfaction, reduced length of stay, and lower risk of procedural complications.^17–19 However, the use of radial access varies by country and region.²⁰ In the United States, for example, only 16% of PCIs were performed using radial access in 2011.²¹ Given the established benefits of radial access, understanding the factors that facilitated the rapid adoption of radial access in China may help promote adoption of radial PCI in countries that use radial access is less frequently.²²

Second, over the study period, physicians increasingly used medicated stents such that in 2011 virtually all (97.3%) implanted stents were DES. The proportion of PCI patients treated with a DES varies across countries. In the United States 70% of PCI patients receive a DES, compared with 78% in Australia and 93% in Korea.^12,23,24 Although the optimal proportional use of DES has not been established, the near universal use of DES in China may indicate an opportunity to align stent selection with patient preference, underlying risk of restenosis, cost, and suitability for long term dual anti-platelet therapy.^25,26 In addition, we found that the majority of DES were domestically produced. Although a number of small studies have suggested that these stents perform similarly to imported DES, conclusive evidence is lacking. ^27–31 The outcomes of patients treated with domestic DES is the focus of an ongoing multi-center retrospective study.³²

Previously, we identified significant gaps in the quality of care provided to AMI patients in China.³³ Our study builds on these findings, demonstrating that the need to improve care extends to patients with both stable and unstable cardiovascular disease. We found that the medical records often lacked the information needed to assess the quality of medical care. This gap may represent poor care or simply reflect poor documentation, but in the absence of this information, we cannot determine whether these procedures are appropriate, safe, effective, and timely. We believe that the development of a prospective national registry to collect information using standardized data definitions would promote quality improvement efforts by allowing hospitals to benchmark their performance. Such an effort would require a substantial investment by hospitals and clinicians, but would represent a key step towards the goal of creating a learning health care system.³⁴

In China, in-hospital mortality following PCI is lower than that reported in contemporary North American and European studies.^12,35–38 This finding may be associated with differences in case selection wherein patients at highest risk of adverse outcomes were not undergoing PCI. In China, many patients with STEMI do not received timely revascularization,³³ and in our study, the proportions of patients with cardiac arrest or cardiogenic shock were lower than those reported in the United States.¹² Although we did not observe substantial changes in the risk of death or treatment withdrawal from 2001 to 2011, we may have been underpowered to detect clinically important differences in patient outcomes over time. However, we did observe a lower risk of any bleeding and access bleeding events over time despite increasing use of GP IIb/IIIa inhibitors.

The findings of this study should be interpreted in context of several limitations. First, we abstracted clinical variables on the basis of documentation in medical records, and the completeness of documentation may not have been consistent either across hospitals or over time. Nevertheless, the use of a centralized data abstraction ensured that the review was consistent and accurately reflected what had been recorded in the medical record. Second, we were unable to collect information needed to identify elective procedures, and we had limited information about the use of adjunctive studies such as stress testing and coronary CT performed prior to the hospitalization. As such, we were unable to evaluate the appropriateness of diagnostic catheterization and PCI procedures.³⁹ Third, the diagnosis of unstable angina was based solely on discharge diagnosis in medical records. It is possible that the proportion of patients with unstable angina may represent an over diagnosis of unstable angina. However, this proportion is similar to that observed in the United States.¹² Finally, our study was limited to procedures performed in 2011 or earlier, and our findings may not reflect current practice patterns. Despite these limitations, the study has provided the most comprehensive overview of contemporary practice of interventional cardiology in China.

In summary, the use of catheterization and PCI in China has increased substantially from 2001 to 2011. However, changes in procedural volume have not been matched by the development of systems to ensure that the care delivered to these patients is evidence-based, safe, and efficient. Developing such a system is particularly important in a country with constrained resources and large variation among care providers. Although the Chinese Ministry of Health has initiated preliminary efforts to standardize the care of patients undergoing PCI, there is more work to be done.⁴⁰ A comprehensive strategy is needed, including development of a national system that prioritizes data collection, performance measurement, public reporting, and quality improvement.³⁴ Our findings can serve as a foundation to guide efforts to further improve the quality of care and allocation of resources not only for China, but also for other developing countries with a rapidly growing cardiovascular disease burden, limited medical resources, and dynamic healthcare systems.