a b s t r a c t

Objectives: This study aimed to investigate the association between time to percutaneous coronary intervention (PCI) and hospital mortality in non-ST-elevation myocardial infarction (NSTEMI).

Methods: Adult patients with NSTEMI were enrolled from November 2007 to December 2012 at 28 emergency departments (EDs) in Korea, excluding those who met the following criteria: age less than 20 years, PCI not per- formed or performed after 72 hours, cardiac arrest at ED presentation, and unknown outcome. Exposure variable was defined as early PCI (b 6 hours after ED arrival) and late PCI group (>= 6 hours). The primary outcome was hos- pital mortality. The adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) for late vs early PCI on mortality were calculated in original data set and propensity score-matched data set using Multivariable logistic regression models with/without interaction term (PCI group and time from symptom to ED arrival within 12 hours, or S2D).

Results: A total of 4363 patients were analyzed as early (n = 1109) and late (n = 3254) PCI groups. The mortality rates were 2.4%, 5.4%, and 1.5% for the total, early, and late PCI groups, respectively. Adjusted ORs (95% CIs) of late PCI for hospital mortality were 0.36 (0.22-0.61) in the original cohort and 0.29 (0.27-0.48) in the propensity score- matched cohort, respectively. Adjusted ORs (95% CIs) in the propensity score-matched subset were 0.28 (0.17-0.45) in the short S2D group and 0.50 (0.18-1.37) in the long S2D group, respectively.

Conclusions: Percutaneous coronary intervention earlier than 6 hours after ED presentation was associated with higher hospital mortality than PCI 6 hours later in NSTEMI. However, the effect disappeared in the long S2D group.

(C) 2015

Introduction

In patients with non-ST-elevation acute coronary syndrome (NSTE ACS), which includes non-ST-elevation myocardial infarction (NSTEMI), an early invasive strategy with a coronary angiography was associated with a reduced rate of refractory ischemia and better out- come in high-risk patients [1]. Recent guidelines recommend an early invasive strategy within the first 24 hours after hospital presentation

? This study was financially supported by the Center for Disease Control and Prevention of Korea (Korea CDC; 2008-2012).

* Corresponding author at: Department of Emergency Medicine, Seoul National University Hospital, 101 Daehak-Ro Jongno-Gu 110-744, Seoul Korea. Tel.: +82 2 2072 0854;

fax: +82 2 741 7855.

E-mail addresses: [email protected] (T.G. Kim), [email protected] (S.D. Shin), [email protected] (K.J. Song), [email protected] (Y.J. Lee), [email protected] (E.J. Lee), [email protected] (Y.S. Ro), [email protected] (K.O. Ahn).

for patients with NSTE ACS who have been selected for an initial inva- sive strategy, particularly among those at high risk [2]. However, there is no evidence that a benefit or hazard is derived by angiography and intervention that are performed within the first few hours of hospital admission [3,4].

A potential benefit from an earlier intervention might be the preven- tion of further Ischemic events that could occur while the patient is awaiting the procedure [5]. Stabilization at first and followed by later intervention strategy might be safer and more beneficial. One small, randomized clinical trial reported lower in-hospital peak troponin I values in patients who underwent an immediate percutaneous coro- nary intervention (PCI; <= 6 hours) compared with values in patients who had undergone an early PCI (<=72 hours) [6]. Moreover, the degree of Troponin elevation is known to independently predict mortality and morbidity in NSTE ACS [7].

This study aimed to determine the association between time to PCI from admission to emergency department (ED) and hospital outcomes in patient presenting with NSTEMI.

http://dx.doi.org/10.1016/j.ajem.2015.06.046

0735-6757/(C) 2015

Methods

This study was approved by the institutional review board of the study site and the Korea Centers for Disease prevention and Control (CDC).

Study setting

A prospective, Multicenter observational study was conducted at 28 EDs in South Korea, from November 1, 2007, to December 31, 2012. The participating hospitals were all academic, teaching tertiary hospitals in which every ED was staffed with emergency physicians and at which emergency PCI is not always available 24 hours per day and 365 days per year. All EDs have a protocol for emergency PCI or thrombolysis for patients with ST-segment elevation myocardial infarction . However, the protocol among hospitals that included a Reperfusion strategy was not standardized for patients with NSTEMI.

Study population

Adult patients 20 years or older who were diagnosed as having NSTEMI were enrolled, and the diagnoses of NSTEMI were made by the characteristic change of cardiac biochemical markers that are associated with Ischemic symptoms, with or without the development of pathologic Q waves on the electrocardiogram (ECG) and/or ECG changes indicative of ischemia but not STEMI. The exclusion criteria were as follows: (1) an age less than 20 years, (2) cardiac arrest at ED presentation, (3) coronary intervention not performed, (4) patients receiving PCI after 72 hours after admission to ED, and (5) unknown outcome (eg, transferred from ED).

Data collection

The data were collected according to preexisting registry of Korea CDC, which had been established in reference to the Cardiology Audit and Registration Data Standard [8]. The collected primary data included baseline characteristics (ie, sex, age, socioeconomic information, comor- bidities, smoking, and alcohol drinking), use of the emergency medical service, onset of symptoms, elapsed time information (ie, time from the onset of the last symptom to PCI and time from ED presentation to

PCI), ECG findings, cardiac biomarkers, diagnosis, and ED disposition. The secondary data included methods of reperfusion, hospital treat- ment (ie, oxygen, ?-blocker, heparin, aspirin, morphine, and clopidogrel bisulfate), cardiopulmonary resuscitation (CPR), and outcome at discharge.

The primary data were collected by the emergency physician or res- idents. The secondary data were collected by the research coordinator in each hospital. The research coordinators were employed by the hospital and financially supported by the Korea CDC. After collecting the primary data, the coordinator reviewed all of the information under the supervi- sion of the study site principle investigator and was responsible for case data entry on the Korea CDC’s Web-based data server. The secondary data were traced for hospital course and entered into the Web-based data server after the patient had been discharged.

The data management team in data coordination center, which was independently operated to maintain Data quality, comprised emergency physicians, cardiologists, statistical experts, and epidemiologists for all of the study periods. For each hospital data set, the data were subjected to quality control measures, and feedback was performed monthly. The study advisory committee also comprised cardiologists and emergency physicians who reviewed the collected variable information and demographics.

The patients were categorized into 2 groups: (1) early PCI group who received the PCI within 6 hours after arrival to ED and (2) late PCI group who received PCI 6 to 72 hours after arrival to ED.

Outcomes

The primary outcome was in-hospital mortality.

Statistical analysis

We conducted multivariable logistic regression analyses for both the full original cohort and the propensity score-matched subsets. The pro- pensity score is the probability of receiving treatment for a patient with specific prognostic factors. Within propensity score strata, covariates in “early PCI” and “late PCI” groups are similarly distributed. Propensity- based matching is used to select cases and control that have similar combinations of confounders. We calculated propensity scores were to a maximum of 10 decimal places. Patients receiving early PCI (cases)

Fig. 1. Study population.

were matched to the closest control (late PCI) in each model whose pro- pensity score differed by less than 1 x 10^-9. There was no overlapping of control cases.

The independent variables for multivariable logistic regression model were individual and socioeconomic factors (ie, age, sex, educa- tion level, and insurance), comorbidity (diabetes mellitus, hyperten- sion, dyslipidemia, renal failure, history of cardiovascular and neurovascular disease), behavior factors (ie, smoking, alcohol, and

exercise), history of reperfusion therapy (ie, PCI and coronary artery by- Female pass surgery [CABG]), date/time of ED visit, use of emergency medical Age (y)	1263 3100	28.9 71.1	300 809	27.1 72.9	963 2291	29.6 70.4	.129
services and Interhospital transfer, clinical presentation (ie, chief 20-45	304	7.0	89	8.0	215	6.6
concerns, cardiogenic shock, initial ECG findings, and time from 45-60	1318	30.2	354	31.9	964	29.6
symptom onset to arrival to ED), hospital treatment (ie, oxygen, 60-75	1884	43.2	457	41.2	1427	43.9
N 75	857	19.6	209	18.8	648	19.9
?-blocker, heparin, aspirin, morphine, Plavix, and CPR). Median (IQR)	64	54-73	64	53-72	64	54-73	.197
The adjusted odds ratios (ORs) with 95% confidence intervals (95% CIs) Education level for both original cohort and propensity score-matched subset were Middle school or less	1899	43.5	457	41.2	1442	44.3	.085
estimated by a multivariable logistic regression model with the inclusion High school or more	2062	47.3	556	50.1	1506	46.3
of covariates used in the univariate analysis. Unknown	402	9.2	96	8.7	306	9.4	.247

Table 1

Demographic findings of the original cohort Category Original cohort

All Early PCI Late PCI P

n % n % n %

All 4363 100.0 1109 100.0 3254 100.0

Sex .106

Male

Insurance

We categorized the patients with 2 groups as patients with shorter	National health	4186	95.9	1072	96.7	3114	95.7
duration of symptom to ED arrival (short S2D group) and patients	insurance
with longer duration of symptom to ED arrival (long S2D group) by 12	Medial aid program	177	4.1	37	3.3	140	4.3

hours. We developed interaction model between the PCI group and	Exercise Nonexercise	3136	71.9	823	74.2	2313	71.1	.024
the S2D group for hospital mortality using original cohort and propensi-	Exercise	956	21.9	211	19.0	745	22.9
ty score-matched subset. We compared the ORs (95% CIs) using multi-	Unknown	271	6.2	75	6.8	196	6.0
variable logistic regression model with interaction term. The	Smoke							.021

confounding factors are same as multivariable logistic regression models without interaction term.

No	1840	42.2	493	44.5	1347	41.4
Current smoker	1646	37.7	424	38.2	1222	37.6
Ex-smoker Alcohol No	877 2858	20.1 65.5	192 759	17.3 68.4	685 2099	21.1 64.5	.017
Drinker Season, ED visit	1505	34.5	350	31.6	1155	35.5	.400

Results

3.1. Patient characteristics	Spring (March-May)	1169	26.8	270	24.3	899	27.6
	Summer	1017	23.3	248	22.4	769	23.6

Of 7605 patients, 4363 were eligible to enroll into early PCI (n = 1109) and late PCI (n = 3254) groups (Fig. 1). Table 1 lists the baseline demo- graphics of the original study population by 2 PCI groups. Male (71.4%) and adults aged 60 to 75 years were dominant. Significant differences

Sunday 557		12.8	73	6.6	484	14.9
Monday	761	17.4	192	17.3	569	17.5
Tuesday	694	15.9	201	18.1	493	15.2
Wednesday	619	14.2	155	14.0	464	14.3
Thursday	595	13.6	183	16.5	412	12.7
Friday	534	12.2	218	19.7	316	9.7
Saturday	603	13.8	87	7.8	516	15.9
Hour, ED visit							b.001
0-05	583	13.4	88	7.9	495	15.2
5-12	1238	28.4	438	39.5	800	24.6
12-17	1566	35.9	447	40.3	1119	34.4
18-23 Medical history	976	22.4	136	12.3	840	25.8

(June-August) Fall (September- November)

Winter

(December-February)

1050 24.1 275 24.8 775 23.8

1127 25.8 316 28.5 811 24.9

were found in behavior risks (exercise status, smoke, and alcohol con- sumption), date/time of ED visit, prevalence of dyslipidemia and chronic kidney disease, history of CABG, and use of emergency medical service.

Table 2 shows the clinical characteristics of original cohort by both PCI groups. The early vs late PCI group in original cohort showed a sig- nificantly higher proportion of cardiogenic shock (5.0% vs 1.6%) and higher ST elevation at J point (13.8% vs 6.5%). Higher proportion of treat- ment (oxygen, aspirin, Plavix, morphine, and CPR) in the early PCI group and lower proportion of treatment (heparin and ?-blocker) in the late PCI group were found in original cohort.

Table 3 showed demographic findings of 2 PCI groups in the propen-

sity score-matched subset. Both PCI groups showed similar distribution	Diabetes mellitus	1309	30.0	317	28.6	992	30.5	.233
for all risk factors in both groups. Table 4 compares both groups with	Hypertension	2336	53.5	568	51.2	1768	54.3	.072
similar proportion of clinical parameters, except mortality.	Dyslipidemia	541	12.4	118	10.6	423	13.0	.040

Weekday, ED visit b.001

Hospital mortality rates were significantly different in the early and late PCI groups in the original cohort (5.0% vs 1.5%, P b .001) and propen- sity score-matched subsets (4.9% vs 1.7%). Fig. 2 shows the hospital	Chronic kidney disease Coronary heart disease	231 1073	5.3 24.6	46 266	4.1 24.0	185 807	5.7 24.8	.048 .586
mortality according to time to PCI from arrival to ED. The hospital mor- tality rates were higher (5.1% and 5.0%) during the first 6 hours and then decreased to 1.0% to 1.5% after 24 hours of time to PCI from ED arrival.	Cerebrovascular disease Previous coronary intervention	380	8.7	87	7.8	293	9.0	.237

3.2. Multivariable logistic regression analysis without interaction term

PCI	479	11.0	128	11.5	351	10.8	.487
CABG	94	2.2	13	1.2	81	2.5	.009
Emergency medical	650	14.9	190	17.1	460	14.1	.016

regression model for original cohort and matched subset. Table 5 shows the unadjusted and adjusted ORs of each group. Compared with early PCI group, the adjusted ORs (95% CIs) of the late PCI groups for hospital mortality were 0.36 (0.22-0.61) in the original cohort and

	services use
All of the potential covariates were used in the multivariable logistic Transferred from 2141 49.1 564 50.9 1577 48.5			.169

0.29 (0.27-0.48) in the propensity score-matched cohort, respectively.

other hospital

Table 2

Clinical findings of the original cohort Category Original cohort

All Early PCI Late PCI P

Table 3

Demographic findings of propensity score-matched subset Category Propensity score-matched subset

All Early PCI Late PCI P

	n	%		n	%		n	%				n	%		n	%		n	%
All Time from symptom to	4363	100.0		1109	100.0		3254	100.0	.449		All Sex Female	2068 583	100.0 28.2		1034 283	100.0 27.4		1034 300	100.0 29.0	.406
ED (h)											Male	1485	71.8		751	72.6		734	71.0
0-3	1387	31.8	359		32.4	1028		31.6	Age (y) .833
3-6	798	18.3	218		19.7	580		17.8		20-45		153	7.4	81		7.8	72		7.0
6-9	382	8.8	99		8.9	283		8.7		45-60		644	31.1	324		31.3	320		30.9
9-12	252	5.8	57		5.1	195		6.0		60-75		882	42.6	433		41.9	449		43.4
N 12	1554	35.6	376		33.9	1168		35.9		N 75		389	18.8	196		19.0	193		18.7
Median	5.6	2.2-19.0	5.3		2.1-17.9	5.7		2.2-19.4	.347	Median (IQR)		64	53-72	64		53-72	64		53-72	.928
(IQR), h Chief concern									.142	Education level Middle school or less		889	43.0	428		41.4	461		44.6	0.290
Chest pain	3834	87.9	986		88.9	2848		87.5		High school or more		999	48.3	517		50.0	482		46.6
Dyspnea	304	7.0	63		5.7	241		7.4		Unknown		180	8.7	89		8.6	91		8.8

Syncope and	225	5.2	60	5.4	165	5.1 Insurance .891
others								National health	1998	96.6	998	96.5	1000	96.7
Shock at presentation, yes	107	2.5	55	5.0	52	1.6	b.001	insurance Medial aid program Exercise	70	3.4	36	3.5	34	3.3	.131
ECG at ED							b.001	Nonexercise	427	20.6	195	18.9	232	22.4
LBBB/	152	3.5	37	3.3	115	3.5		Exercise	1508	72.9	772	74.7	736	71.2
uncertain ST elevation	365	8.4	153	13.8	212	Unknown 133 6.4 67 6.5 66 6.4 6.5 Smoke .204
(J point)							No		907	43.9	458	44.3	449	43.4
ST	828	19.0	224	20.2	604	18.6	Current smoker		770	37.2	396	38.3	374	36.2
depression >= 0.5 mm T inversion	405	9.3	105	9.5	300	9.2	Ex-smoker Alcohol No		391 1396	18.9 67.5	180 698	17.4 67.5	211 698	20.4 67.5	1.000
(>= 3 mm)							Drinker		672	32.5	336	32.5	336	32.5
T inversion	374	8.6	100	9.0	274	8.4 Season, ED visit .073
(>= 1 mm)							Spring (March-May)		545	26.4	247	23.9	298	28.8
Normal/ nonspecific	2239	51.3	490	44.2	1749	53.7	Summer (June- August)		464	22.4	235	22.7	229	22.1
Treatment Fall (September- 496										24.0	255	24.7	241	23.3
within 24 h Oxygen	3414	78.2	933	84.1	2481	76.2	b.001	November) Winter (December-	563	27.2	297	28.7	266	25.7
Nitroglycerin	2957	67.8	735	66.3	2222	68.3	.216	February)
Aspirin	3860	88.5	1000	90.2	2860	87.9	.040 Weekday, ED visit .574
Plavix	3907	89.5	1013	91.3	2894	88.9	.024	Sunday	135	6.5	72	7.0	63	6.1
Morphine	1154	26.4	345	31.1	809	24.9	b.001	Monday	399	19.3	185	17.9	214	20.7
Heparin	3885	89.0	945	85.2	2940	90.4	b.001	Tuesday	363	17.6	186	18.0	177	17.1
?-Blocker	1686	38.6	395	35.6	1291	39.7	.017	Wednesday	283	13.7	149	14.4	134	13.0
CPR	53	1.2	25	2.3	28	0.9	b.001	Thursday	343	16.6	173	16.7	170	16.4
Hospital	106	2.4	56	5.0	50	1.5	b.001	Friday	360	17.4	183	17.7	177	17.1

mortality

Saturday 185 8.9 86 8.3 99 9.6

Hour, ED visit .759

LBBB, Left bundle-branch block.	0-05	174	8.4	87	8.4	87	8.4
	5-12	807	39.0	399	38.6	408	39.5
	12-17	835	40.4	414	40.0	421	40.7
3.3. Multivariable logistic regression analysis with interaction term	18-23	252	12.2	134	13.0	118	11.4
	Medical history

Table 6 shows the unadjusted and adjusted ORs (95% CIs) of late PCI vs early PCI across the S2D groups in the original cohort and propensity score-matched subset. Compared with the early PCI group, the adjusted ORs (95% CIs) of the late PCI groups for hospital mortality in the original cohort were 0.21 (0.12-0.39) in the short S2D group and 0.44 (0.20- 0.93) in the long S2D group. The adjusted ORs (95% CIs) in the propen- sity score-matched subset were 0.28 (0.17-0.45) in the short S2D group and 0.50 (0.18-1.37) in the long S2D group.

Discussion

In this study, the patients who had undergone PCI in less than 6 hours showed the highest hospital mortality than did the later PCI group in NSTEMI patients presenting to 28 EDs of South Korea. These findings were similar in both the original cohort and propensity score-matched subset.

The Intracoronary Stenting with Antithrombotic Regimen Cooling Off trial randomized patients with NSTE ACS into early (b 6 hours from

Diabetes mellitus Hypertension Dyslipidemia Chronic kidney disease

Coronary heart disease Cerebrovascular disease

618	29.9	304	29.4	314	30.4	.631
1070	51.7	533	51.5	537	51.9	.860
238	11.5	111	10.7	127	12.3	.270
83	4.0	46	4.4	37	3.6	.313
515	24.9	253	24.5	262	25.3	.647
168	8.1	80	7.7	88	8.5	.520
243	11.8	123	11.9	120	11.6	.838
28	1.4	13	1.3	15	1.5	.704
344	16.6	177	17.1	167	16.2	.555
1011	48.9	522	50.5	489	47.3	.147

Previous coronary intervention PCI

CABG

Emergency medical services use

Transferred from other hospital

Table 4

Clinical findings of propensity score-matched subset Category Propensity score-matched subset

All Early PCI Late PCI P

	n	%		n	%		n	%
All Time from symptom to ED (h)	2068	100.0		1034	100.0		1034	100.0	.186

presentation) or deferred (3-5 days from presentation) coronary angi- ography intervention groups and showed less incidence of large myo- cardial infarction or death at 30 days [5]. In The Timing of Intervention in Acute Coronary Syndromes (TIMACS) trial, patients with NSTE ACS were randomized into early intervention (coronary angiography b 24 hours from randomization) or delayed intervention (N 36 hours from randomization) groups. The TIMACS trial did not report a significant dif- ference in 6-month composite of death, myocardial infarction, or stroke, but a superior outcome was observed in high-risk patients who had

0-3	662	32.0	340	32.9	322	31.1	been randomized into the early intervention group [1]. The Angioplasty
3-6	375	18.1	203	19.6	172	16.6	to Blunt the Rise of Troponin in Acute Coronary Syndromes Randomized
6-9	184	8.9	93	9.0	91	8.8	for an Immediate or Delayed Intervention study compared an angiogra-

9-12	115	5.6	55	5.3	60	5.8
N 12	732	35.4	343	33.2	389	37.6
Median (IQR) Chief concern	5.6	2.1-20.6	5.3	2.0-17.6	6.1	2.1-23.9	.058 0.066
Chest pain	1821	88.1	921	89.1	900	87.0
Dyspnea	136	6.6	55	5.3	81	7.8
Syncope and others	111	5.4	58	5.6	53	5.1
Shock at	77	3.7	38	3.7	39	3.8	.908

presentation, yes

ECG at ED .797

LBBB, left bundle-branch block.

phy and intervention performed immediately upon presentation, with an intervention performed on the following working day, and the re- sults did not show a significant difference in the peak Troponin I levels between the 2 groups [3]. Therefore, recent guidelines from the American College of Cardiology/American Heart Association recommended an early invasive strategy in selected patients within 24 hours of their hospital presentation [2].

Our study population included patients with NSTEMI, not NSTE ACS, which is a broader spectrum of acute ischemic heart disease. All patients were admitted to the ED. These characteristics are different from those reported in previous studies. Therefore, we may interpret the findings differently from previous studies.

LBBB/ uncertain ST elevation	61 241	2.9 11.7	35 123	3.4 11.9	26 118	2.5 11.4
(J point) ST	402	19.4	207	20.0	195	18.9
depression >= 0.5 mm T inversion	201	9.7	98	9.5	103	10.0
(>= 3 mm) T inversion	182	8.8	91	8.8	91	8.8
(>= 1 mm)
Normal/ nonspecific Treatment	981	47.4	480	46.4	501	48.5
within 24 h Oxygen	1730	83.7	860	83.2	870	84.1	.552
Nitroglycerin	1387	67.1	686	66.3	701	67.8	.483
Aspirin	1837	88.8	931	90.0	906	87.6	.081
Plavix	1882	91.0	943	91.2	939	90.8	.759
Morphine	619	29.9	309	29.9	310	30.0	.962
Heparin	1783	86.2	889	86.0	894	86.5	.750
?-Blocker	739	35.7	376	36.4	363	35.1	.551
CPR	34	1.6	18	1.7	16	1.5	.730
Hospital mortality	69	3.3	51	4.9	18	1.7	b.001

There looks to be many reasonable explanations for higher mortality in the early intervention group than in the delayed intervention group. One possible explanation of this outcome is vascular and thrombotic vulnerability in combination with a periprocedural event in an acute setting. Prolonged medical treatment with antiplatelets and anticoagu- lants may decrease the angiographic Thrombus burden and facilitate thrombus resolution in patients with NSTE ACS [9,10], and the presence of a thrombus in an angiogram is associated with a higher rate of abrupt closure, myocardial infarction, and an in-hospital repeat revasculariza- tion [11]. It is also possible that the patients in the early PCI group underwent PCI earlier according to clinical judgment because they had some characteristics, such as hemodynamic or electrical instability; such patients may be more likely to have an unfavorable outcome than others [12].

These controversial findings may be from study design issue and from incomplete adjustment for all confounding factors related to out- comes. Previous randomized controlled trials included acute coronary syndrome which included unstable angina to NSTEMI. Our study en- rolled the only NSTEMI that was diagnosed by criteria at discharge. We extracted propensity score-matched subset to compare the effect of late PCI with early PCI using a much more similar patient group.

Fig. 2. Mortality rates according to time to PCI from ED arrival.

Table 5 Multivariable logistic regression of late PCI vs early PCI for hospital mortality in NSTEMI in the original cohort and propensity score-matched subset

Dataset Group Total Death Unadjusted Adjusted^a

n n % OR 95% CI OR 95% CI

Original cohort	All	4363	106	2.4
	Early PCI Late	1109 3254	56 50	5.0 1.5	1.00 0.29	0.20	0.43	1.00 0.28	0.17	0.45
Propensity score-	PCI All	2068	69	3.3
matched subset	Early PCI	1034	51	4.9	1.00			1.00
	Late PCI	1034	18	1.7	0.34	0.20	0.59	0.22	0.11	0.44

^a Adjusted for age, sex, education level, insurance, diabetes mellitus, hypertension, dyslipidemia, renal failure, history of cardiovascular and neurovascular disease, smoking, alcohol, and exercise, previous PCI, previous CABG, date/time of ED visit (season, weekday, day time), use of emergency medical services and interhospital transfer, chief concerns, cardiogenic shock, initial ECG findings, time from symptom onset to arrival to ED, hospital therapy of oxygen, ?-blocker, heparin, aspirin, morphine, Plavix, and CPR.

Table 6 Multivariable logistic regression of late PCI vs early PCI for hospital mortality in NSTEMI in the original cohort and propensity score-matched subset using interaction model be- tween the PCI group and the S2D group

or electrical instability, which are associated with a worse outcome, may undergo PCI earlier. Third, it is uncertain whether all of the enrolled patients had been treated in a similar manner to that currently used in clinical practice. The exact type and amount of the drugs that were ad- ministered were not recorded in our registry. Finally, we could only compare hospital mortality. In our study, we could not determine whether the time from admission to PCI affects longer-term mortality or other kinds of outcomes (e., stroke or myocardial infarction).

6. Conclusion

In patients with NSTEMI who undergo PCI, a PCI earlier than 6 hours after hospital presentation is an independent predictor of higher in- hospital mortality than undergoing PCI 6 hours later. The effect disappeared in patients with longer duration of symptom to ED arrival than 12 hours.

References

1. Mehta SR, Granger CB, Boden WE, Steg PG, Bassand JP, Faxon DP. Early versus delayed invasive intervention in acute coronary syndromes. N Engl J Med 2009;360:2165-75.
2. 2011 ACCF/AHA focused update of the guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction (updating the 2007 guideline): a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2011;123:2022-60.
3. Montalescot G, Cayla G, Collet JP, Elhadad S, Beygui F, Le Breton H. Immediate vs de- layed intervention for acute coronary syndromes: a randomized clinical trial. JAMA

   Data set Time from symptom

   to ED arrival

   Unadjusted Adjusted^a

   OR 95% CI OR 95% CI

   2009;302:947-54.

   Sorajja P, Gersh BJ, Cox DA, McLaughlin MG, Zimetbaum P, Costantini C. Impact of delay to angioplasty in patients with acute coronary syndromes undergoing invasive

   Original cohort S2D <= 12 h 0.21 0.13 0.35 0.21 0.12 0.39

   S2D N 12 h 0.50 0.26 0.97 0.44 0.20 0.93

   Propensity score- S2D <= 12 h 0.20 0.10 0.45 0.28 0.17 0.45

   matched subset S2D N 12 h 0.73 0.31 1.71 0.50 0.18 1.37

   ^a Adjusted for age, sex, education level, insurance, diabetes mellitus, hypertension, dyslipidemia, renal failure, history of cardiovascular and neurovascular disease, smoking, alcohol, and exercise, previous PCI, previous CABG, date/time of ED visit (season, weekday, day time), use of emergency medical services and inter-hospital transfer, chief concerns, cardiogenic shock, initial ECG findings, time from symptom onset to arrival to ED, hospital therapy of oxygen, ?-blocker, heparin, aspirin, morphine, Plavix, and CPR.

   Both the original cohort and the matched subset showed similar find- ings. However, interaction model by time from symptom to ED arrival showed a different effect size. Late PCI was beneficial in patients who ar- rived earlier than 12 hours after symptom onset, while not beneficially longer in patients who arrived at ED after 12 hours after symptom onset. This finding means that the patient group with Short duration symptom may be more beneficial when the patient receive PCI after a stabilization period longer than 12 hours.

   Limitations

   There are several limitations of our study. First, risk stratification scores, such as the Thrombolysis in Myocardial Infarction risk score [13] or Global Registry of Acute coronary events score [14], were not considered. To minimize this limitation, we established a multivariate logistic regression model that incorporated as many of the covariates as possible that are included in TIMI and GRACE risk score models. Second, we could not fully evaluate the factors that had actually affected the time at which the PCI was performed. As men- tioned above, patients who have characteristics such as hemodynamic

   management: analysis from the ACUITY (Acute Catheterization and Urgent Inter- vention Triage strategY) trial. J Am Coll Cardiol 2010;55:1416-24.

   Neumann FJ, Kastrati A, Pogatsa-Murray G, Mehilli J, Bollwein H, Bestehorn HP. Eval- uation of prolonged antithrombotic pretreatment (“cooling-off” strategy) before in- tervention in patients with unstable coronary syndromes: a randomized controlled trial. JAMA 2003;290:1593-9.

4. Sciahbasi A, Madonna M, De Vita M, Agati L, Scioli R, Summaria F. Comparison of im- mediate vs early invasive strategy in patients with first acute non-ST-elevation myocardial infarction. Clin Cardiol 2009;33(10):650-5.
5. Jolly SS, Shenkman H, Brieger D, Fox KA, Yan AT, Eagle KA. Quantitative troponin and death, cardiogenic shock, cardiac arrest and new heart failure in patients with non- ST-segment elevation acute coronary syndromes (NSTE ACS): insights from the Global Registry of acute coronary events. Heart 2011;97:197-202.
6. Flynn MR, Barrett C, Cosio FG, Gitt AK, Wallentin L, Kearney P. The Cardiology Audit and Registration Data Standards (CARDS), European data standards for clinical car- diology practice. Eur Heart J 2005;26(3):308-13.
7. Tricoci P, Lokhnygina Y, Berdan LG, Steinhubl SR, Gulba DC, White HD. Time to cor- onary angiography and outcomes among patients with high-risk non ST-segment el- evation acute coronary syndromes: results from the SYNERGY trial. Circulation 2007;116(23):2669-77.
8. Zhao XQ, Theroux P, Snapinn SM, Sax FL. Intracoronary thrombus and platelet glycoprotein IIb/IIIa receptor blockade with tirofiban in unstable angina or non-Q- wave myocardial infarction. Angiographic results from the PRISM-PLUS trial (Platelet receptor inhibition for ischemic syndrome management in patients limited by unstable signs and symptoms). PRISM-PLUS Investigators. Circulation 1999; 100(15):1609-15.
9. Van den Brand M, Laarman GJ, Steg PG, De Scheerder I, Heyndrickx G, Beatt K. As- sessment of coronary angiograms prior to and after treatment with abciximab, and the outcome of angioplasty in refractory unstable angina patients. Angiographic results from the CAPTURE trial. Eur Heart J 1999;20(21):1572-8.
10. Singh M, Reeder GS, Ohman EM, Mathew V, Hillegass WB, Anderson RD, et al. Does the presence of thrombus seen on a coronary angiogram affect the outcome after percutaneous coronary angioplasty? An Angiographic Trials Pool data experience. J Am Coll Cardiol 2001;38(3):624-30.
11. Antman EM, Cohen M, Bernink PJ, McCabe CH, Horacek T, Papuchis G. The TIMI risk score for unstable angina/non-ST elevation MI: a method for prognostication and therapeutic decision making. JAMA 2000;284(7):835-42.
12. Granger CB, Goldberg RJ, Dabbous O, Pieper KS, Eagle KA, Cannon CP, et al. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003;163(19):2345-53.