JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 65, NO.

12, 2015

ª 2015 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00

PUBLISHED BY ELSEVIER INC. http://dx.doi.org/10.1016/j.jacc.2015.01.028

ORIGINAL INVESTIGATIONS

Hospital Length of Stay and

Clinical Outcomes in Older STEMI
Patients After Primary PCI
A Report From the National Cardiovascular Data Registry

Rajesh V. Swaminathan, MD,* Sunil V. Rao, MD,y Lisa A. McCoy, MS,y Luke K. Kim, MD,* Robert M. Minutello, MD,*
S. Chiu Wong, MD,* David C. Yang, MD,* Paramita Saha-Chaudhuri, PHD,z Harsimran S. Singh, MD, MSC,*
Geoffrey Bergman, MD,* Dmitriy N. Feldman, MD*

ABSTRACT

BACKGROUND There has been a decline in hospital length of stay (LOS) after primary percutaneous coronary
intervention (PPCI) for ST-segment elevation myocardial infarction (STEMI).

OBJECTIVES The objective of this study was to examine whether shorter LOS is safe for older patients undergoing PPCI
for STEMI.

METHODS The study analyzed patients’ characteristics and 30-day outcomes by LOS (short, #3 days; medium, 4 to 5 days; long
>5 days; where LOS was the discharge date minus the admission date plus 1) among 33,920 patients with STEMI in the linked CathPCI
Registry-Centers for Medicare & Medicaid Services dataset who were $65 years of age and treated with PPCI from 2004 to 2009.

RESULTS Percents of patients in each category were as follows: 26.9%, 46.3%, and 26.8% for short, medium, and long
LOS, respectively. Patients with a long LOS were generally older, female, and had more comorbidities, including cardiogenic
shock and multivessel disease. Patients with a short LOS generally had higher ejection fraction and single-vessel disease.
There was no significant difference in 30-day all-cause mortality (hazard ratio [HR]: 1.00; 95% confidence interval [CI]: 0.74
to 1.34) or major adverse cardiac events (MACE) (death, readmission for myocardial infarction, unplanned revascularization:
HR: 1.03; 95% CI: 0.86 to 1.25) for medium versus short LOS. There was a significant increase in adjusted mortality (HR: 2.30;
95% CI: 1.72 to 3.07) and MACE (HR: 1.75; 95% CI: 1.44 to 2.12) for long versus short LOS. Patients with a very short LOS
(1 to 2 days) had significantly increased 30-day mortality and MACE compared with a 3- to 4-day LOS.

CONCLUSIONS Patients discharged as early as 48 h after PPCI have outcomes similar to patients who stay in the
hospital for 4 to 5 days. Early, but not very early (<48 h), discharge may be safe among selected older patients with
STEMI. (J Am Coll Cardiol 2015;65:1161–71) © 2015 by the American College of Cardiology Foundation.

From the *Weill Cornell Medical College, New York-Presbyterian Hospital, Greenberg Division of Cardiology, New York, New York;
yDuke Clinical Research Institute, Division of Cardiology, Duke University Medical Center, Durham, North Carolina; and the
zDepartment of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Canada. This research was
supported by the American College of Cardiology Foundation’s National Cardiovascular Data Registry (NCDR). The views expressed
in this article represent those of the authors and do not necessarily represent the official views of the NCDR or its associated pro-
fessional societies identified at www.ncdr.com. The CathPCI Registry is an initiative of the American College of Cardiology Foun-
dation and the Society for Cardiovascular Angiography and Interventions. Dr. Feldman is a consultant and Speakers Bureau member
for Abbott Vascular, Eli Lilly, Daiichi-Sankyo, Bristol-Myers Squibb, and Pfizer. Dr. Bergman is a proctor for transcatheter aortic valve
replacement. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript’s audio summary by JACC Editor-in-Chief Dr. Valentin Fuster.
You can also listen to this issue’s audio summary by JACC Editor-in-Chief Dr. Valentin Fuster.

Manuscript received November 24, 2014; revised manuscript received December 22, 2014, accepted January 13, 2015.
1162 Swaminathan et al.
Hospital Length of Stay and Outcomes in STEMI
H
ABBREVIATIONS ospital length of stay (LOS) is a
AND ACRONYMS critical measure of health care effi-
ciency, yet it must be carefully
ACC = American College
of Cardiology
balanced with health care quality. Reducing
LOS is driven by financial pressures in the
AHA = American
Heart Association setting of decreasing reimbursements or
CABG = coronary artery bundled payment models proposed by the
bypass grafting Centers for Medicare & Medicaid Services
CMS = Centers for Medicare & (CMS) (1). In contrast, longer LOS allows for
Medicaid Services improved patient-centered care and transi-
IABP = intra-aortic tion to the outpatient setting, which may
balloon pump
result in lowered readmission rates and
ICD-9-CM = International
improved health outcomes (2). These issues
Classification of Diseases-Ninth
Revision-Clinical Modification are particularly germane to patients present-
LAD = left anterior descending
ing with an ST-segment elevation myocardial
infarction (STEMI), in whom in-hospital
LOS = length of stay
monitoring is important to detect arrhyth-
MACE = major adverse
cardiac event(s) mias and mechanical complications of myo-
MI = myocardial infarction
cardial infarction (MI), optimize secondary
prevention, and provide education regarding
PCI = percutaneous
coronary intervention risk factor modification, medication adher-
PPCI = primary percutaneous ence, and rehabilitation. Although primary
coronary intervention percutaneous coronary intervention (PPCI)
STEMI = ST-segment elevation is associated with improved survival and
myocardial infarction reduced risk of arrhythmic and mechanical
complications compared with fibrinolytic therapy
(3–7), there remains a substantial risk for subacute
stent thrombosis and heart failure. In the current
era of PPCI, there has been a decline in LOS in
the United States after STEMI, with significant
geographic variation in discharge practice patterns
(8). Median LOS is shortest in the United States
compared with other countries (9). Whether these
shorter hospital stays are safe remains unclear.
SEE PAGE 1172
In the current study, we used the linked national
CathPCI Registry with the CMS claims data to
examine LOS of older patients with STEMI and asso-
ciated 30-day clinical outcomes, including mortality
and combined major adverse cardiovascular events
(MACE: mortality, readmission for MI, or unplanned
revascularization) during the period from 2004 to
2009. We hypothesized that early discharge was as
safe as longer LOS with respect to clinical outcomes at
30 days.
METHODS
DATA COLLECTION. The CathPCI Registry, co-
sponsored by the American College of Cardiology
and the Society for Cardiovascular Angiography and
Interventions, is a large, national registry of pa-
tients undergoing diagnostic cardiac catheterizations
JACC VOL. 65, NO. 12, 2015
MARCH 31, 2015:1161–71
and/or percutaneous coronary intervention (PCI) that
captures w85% of PCI procedures performed at
>1,400 hospitals. Details of the registry, data collec-
tion, and monitoring have been previously described
(10–13). This study used version 3.0 of the CathPCI
Registry, which is the latest version linked to CMS
claims. Procedure codes were used to identify index
procedures in the Medicare files, which were then
linked to the CathPCI Registry by using indirect
identifiers. Linking rules were applied using a hier-
archical matching algorithm (14,15), until a match was
achieved. The feasibility of linkage to CMS claims has
been previously described (14), and it has been used
by other analyses of CathPCI data that sought long-
term outcomes (15). The Institutional Review Board
of Duke University Medical Center in Durham, North
Carolina approved the study after determining that
the study met the definition of research not requiring
informed consent.
STUDY POPULATION, DEFINITIONS, AND OUTCOME
MEASURES. We included patients who had electro-
cardiographic evidence of ST-segment elevations
on presentation to emergency departments at
acute care hospitals with PPCI capabilities and who
subsequently underwent PPCI from January 6, 2004
to December 31, 2009. We excluded the following
patients: those <65 years of age or >90 years of age
(n ¼ 1,208); STEMI transfers from centers without
PPCI capabilities; patients receiving thrombolysis
(n ¼ 1,030); patients who were discharged to
extended care, other facilities, or nursing homes (n ¼
4,297); and patients with in-hospital death (n ¼ 3,736)
or coronary artery bypass grafting (CABG) (n ¼ 1,862).
The final study population totaled 33,920 patients
who were discharged home alive. LOS was defined by
the standard CMS definition, as reported in LOS
published data (16), as the discharge date minus the
admission date plus 1. In other words, a LOS of 3 is
actually 2 days after the day of admission/PCI (e.g.
Wednesday discharge after Monday admission/PCI).
Given the discrete nature of LOS, we stratified LOS
into approximate tertiles: short (#3 days), medium
(4 to 5 days), and long (>5 days). We also separately
analyzed the group of patients with very short LOS
(1 to 2 days) relative to other LOS days.
Our primary outcomes were post-discharge 30-day
clinical events stratified by LOS, including mortality
and MACE (defined as a composite of mortality,
readmission for MI, or unplanned repeat revascular-
ization). Secondary outcomes included the incidence
of 30-day readmission for MI, unplanned revascular-
ization, and bleeding events (accounting for death as
a competing factor). All-cause mortality was obtained
from the claims data. Readmissions for MI and
JACC VOL. 65, NO. 12, 2015 Swaminathan et al. 1163
MARCH 31, 2015:1161–71 Hospital Length of Stay and Outcomes in STEMI

unplanned revascularization that occurred during

F I G U R E 1 Study Sample Selection Flow Diagram
follow-up were obtained from claims data using the
International Classification of Diseases-Ninth Revi-
Patients with STEMI with admission linked to CMS
sion-Clinical Modification (ICD-9-CM) diagnosis code (n = 84,050)
410.X1 (MI) and ICD-9-CM procedure codes 36.00,
36.06, 36.07, 36.09, and 36.10 to 36.19 (revasculari-
zation). Unplanned repeat revascularization was Exclude if not admitted through emergency department, received
defined as readmission with a PCI or CABG procedure thrombolytic therapy, age <65 or >90 years, in–hospital death, in–hospital
CABG, discharge other than home, or not Fee For Service Eligible
and a principal discharge diagnosis that was clearly
(n = 33,920)
not consistent with an elective readmission (17):
specifically, heart failure, acute MI, unstable angina,
arrhythmia, or cardiac arrest. Readmissions for bleed-
ing were identified from claims data using the
following ICD-9-CM diagnosis codes: 430 to 432, 578.X, Length of stay ≤3 days Length of stay 4–5 days Length of stay >5 days
(n = 9,135) (n = 15,704) (n = 9,081)
719.1X, 423.0, 599.7, 626.2, 626.6, 626.8, 627.0, 627.1,
786.3, 784.7, or 459.0.
STATISTICAL ANALYSIS. For descriptive analyses,
Primary Endpoints:
baseline characteristics, procedural characteristics, in- All-cause mortality, MACE (death, readmission
hospital adverse events, and hospital characteristics for MI, or unplanned revascularization)
were compared among patients grouped by LOS. Contin-
uous variables are presented as medians with 25th and Secondary Endpoints:
75th percentiles, and categorical variables are presented as Readmission for MI, unplanned repeat
revascularization, bleeding
frequencies (percents). Differences among patient groups
were compared using Pearson chi-square tests for cate-
gorical variables and Mann-Whitney Wilcoxon rank sum CABG ¼ coronary artery bypass graft; CMS ¼ Centers for Medicare & Medicaid Services;
test for continuous variables. MACE ¼ major adverse cardiac event(s); MI ¼ myocardial infarction; STEMI ¼ ST-segment
elevation myocardial infarction.
Estimates of the event rates for primary endpoints
at 30 days post-discharge were determined on the
basis of weights that were functions of Kaplan-Meier
estimating equations was used. The generalized
censoring estimates; the log-rank test was used to
estimating equations method was used to account for
assess differences by LOS group. The cumulative
within-hospital clustering (20), because patients at
incidence differences for time-to-event clinical out-
the same hospital are more likely to receive similar
comes were assessed using Gray’s method (18). Both
the unadjusted and adjusted hazard ratios (HRs) were
F I G U R E 2 Distribution of LOS for the Overall Study Population
estimated for comparison of LOS groups along with a
95% confidence interval (CI) on the basis of the 30
sandwich estimator of standard errors to account for
within-hospital correlation. The adjusted analysis
25
included the variables adopted in the validated
CathPCI Registry all-cause mortality model as cova-
20
riates, which consisted of age, race, sex, body mass
index, glomerular filtration rate, presence of cardio-
Percent

genic shock at admission, medical history (i.e., 15

congestive heart failure, diabetes, valvular surgery,
cerebrovascular disease, peripheral vascular disease, 10
chronic lung disease, dialysis, PCI), highest-risk
lesion characteristics, New York Heart Association 5
(NYHA) functional class, and pre-procedure intra-
aortic balloon pump (IABP), ejection fraction (19), as
0
well as in-hospital bleeding, vascular complications, 0 10 20 30
need for transfusions, and hospital characteristics. Post–Procedure LOS (days)
To examine factors associated with longer LOS
(defined as greater than the median LOS of 4 days), LOS ¼ length of stay.
a multivariable logistic regression with generalized
1164 Swaminathan et al. JACC VOL. 65, NO. 12, 2015

Hospital Length of Stay and Outcomes in STEMI MARCH 31, 2015:1161–71

T A B L E 1 Baseline Characteristics of Patients, Procedures, In-Hospital Adverse Events, and Hospitals*

Overall LOS #3 days LOS 4-5 days LOS >5 days

(N ¼ 33,920) (n ¼ 9,135) (n ¼ 15,704) (n ¼ 9,081) p Value

Patient characteristics
Age, yrs 73.0 (69.0–79.0) 72.0 (68.0–78.0) 73.0 (68.0–79.0) 76.0 (70.0–81.0) <0.001
Female 13,524 (39.87) 3,037 (33.25) 6,181 (39.36) 4,306 (47.42) <0.001
Nonwhite race 3,126 (9.21) 673 (7.37) 1,410 (8.98) 1,043 (11.48) <0.001
Prior MI (>7 days) 6,281 (18.52) 1,723 (18.86) 2,723 (17.34) 1,835 (20.21) <0.001
Prior congestive heart failure 1,950 (5.75) 332 (3.63) 703 (4.48) 915 (10.08) <0.001
Diabetes 7,869 (23.20) 1,917 (20.99) 3,470 (22.10) 2,482 (27.33) <0.001
Prior renal failure 1,359 (4.01) 248 (2.71) 497 (3.16) 614 (6.76) <0.001
Cerebrovascular disease 3,391 (10.00) 742 (8.12) 1,429 (9.10) 1,220 (13.43) <0.001
Peripheral vascular disease 3,023 (8.91) 711 (7.78) 1,270 (8.09) 1,042 (11.47) <0.001
Chronic lung disease 4,778 (14.09) 1,103 (12.07) 1,971 (12.55) 1,704 (18.76) <0.001
Hypertension 24,049 (70.90) 6,342 (69.43) 10,971 (69.86) 6,736 (74.18) <0.001
Tobacco history 15,245 (44.94) 4,194 (45.91) 7,074 (45.05) 3,977 (43.79) <0.001
Dyslipidemia 20,933 (61.71) 5,787 (63.35) 9,601 (61.14) 5,545 (61.06) <0.001
Prior PCI 7,103 (20.94) 2,072 (22.68) 3,104 (19.77) 1,927 (21.22) <0.001
Prior CABG 3,361 (9.91) 963 (10.54) 1,454 (9.26) 944 (10.40) 0.001
Congestive heart failure 3,122 (9.20) 477 (5.22) 1,118 (7.12) 1,527 (16.82) <0.001
Cardiogenic shock 2,373 (7.00) 235 (2.57) 768 (4.89) 1,370 (15.09) <0.001
Symptom onset to admission time <0.001
#6 h 27,074 (79.82) 7,391 (80.91) 12,663 (80.64) 7,020 (77.30)
>6 h and #12 h 2,947 (8.69) 738 (8.08) 1,357 (8.64) 852 (9.38)
>12 h 3,548 (10.46) 905 (9.91) 1,546 (9.84) 1,097 (12.08)
Procedure/in-hospital adverse events
Contrast volume, ml 200.0 (150.0–250.0) 185.0 (145.0–240.0) 200.0 (150.0–250.0) 200.0 (150.0–267.0) <0.001
IABP insertion 2,225 (6.56) 66 (0.72) 629 (4.01) 1,530 (16.85) <0.001
LVEF, % 47.0 (40.0–55.0) 50.0 (45.0–60.0) 48.0 (40.0–55.0) 40.0 (30.0–50.0) <0.001
Percutaneous entry location <0.001
Femoral 33,324 (98.24) 8,943 (97.90) 15,455 (98.41) 8,926 (98.29)
Radial 323 (0.95) 123 (1.35) 133 (0.85) 67 (0.74)
Multivessel disease 19,859 (58.55) 5,041 (55.18) 9,759 (62.14) 5,943 (65.44) <0.001
High-risk (type C) lesion 20,864 (61.51) 5,162 (56.51) 8,363 (61.02) 5,173 (64.72) <0.001
Bifurcation lesion 3,679 (10.85) 908 (9.94) 1,714 (10.91) 1,057 (11.64) 0.001
Culprit artery segment <0.001
Left main artery 151 (0.45) 31 (0.34) 69 (0.44) 51 (0.56)
Proximal LAD 5,773 (17.02) 1,051 (11.51) 2,570 (16.37) 2,152 (23.70)
Pre-procedure TIMI flow grade <0.001
0 (no flow) 20,498 (60.43) 4,784 (52.37) 9,737 (62.00) 5,977 (65.82)
1 4,417 (13.02) 1,253 (13.72) 2,034 (12.95) 1,130 (12.44)
2 4,914 (14.49) 1,609 (17.61) 2,197 (13.99) 1,108 (12.20)
3 (complete flow) 3,960 (11.67) 1,455 (15.93) 1,667 (10.62) 838 (9.23)
PCI procedure success 31,320 (92.33) 8,681 (95.03) 14,648 (93.28) 7,991 (88.00) <0.001
Periprocedural MI 276 (0.81) 55 (0.60) 83 (0.53) 138 (1.52) <0.001
Cardiogenic shock 505 (1.60) 27 (0.30) 134 (0.90) 344 (4.46) <0.001
Congestive heart failure 815 (2.65) 26 (0.30) 170 (1.17) 619 (8.19) <0.001
Cerebrovascular accident/stroke 68 (0.20) 2 (0.02) 13 (0.08) 53 (0.58) <0.001
Renal failure 167 (0.51) 2 (0.02) 19 (0.12) 146 (1.72) <0.001
Any bleeding 1,918 (5.65) 128 (1.40) 633 (4.03) 1,157 (12.74) <0.001
Any vascular complication 323 (0.95) 23 (0.25) 115 (0.73) 185 (2.04) <0.001
Continued on the next page

treatment relative to patients in other hospitals (i.e.,
within-center correlation for response). The adjusted
analysis included covariates from the CathPCI Regis-
try mortality model as well as in-hospital complica-
tions (i.e., bleeding, need for transfusions, vascular
complications, renal failure).
To evaluate mortality and MACE rates of the very
short LOS group relative to other LOS days further, we
also performed a propensity-matched analysis to
account for differences in case mix between the very
short LOS group (LOS 1–2) and other LOS days
(LOS 3–4) more robustly. We used the “gmatch” macro
JACC VOL. 65, NO. 12, 2015 Swaminathan et al. 1165
MARCH 31, 2015:1161–71 Hospital Length of Stay and Outcomes in STEMI

T A B L E 1 Continued

Overall LOS #3 days LOS 4-5 days LOS >5 days

(N ¼ 33,920) (n ¼ 9,135) (n ¼ 15,704) (n ¼ 9,081) p Value

Hospital
Number of beds 390.0 (278.0–531.0) 366.0 (256.0–510.0) 394.0 (283.0–537.0) 410.0 (300.0–560.0) <0.001
University hospital 2,270 (6.69) 463 (5.07) 1,092 (6.95) 715 (7.87) <0.001
Hospital region <0.001
West 5,942 (17.52) 2,064 (22.59) 2,483 (15.81) 1,395 (15.36)
Northeast 4,373 (12.89) 705 (7.72) 2,312 (14.72) 1,356 (14.93)
Midwest 10,833 (31.94) 3,236 (35.42) 4,929 (31.39) 2,668 (29.38)
South 12,731 (37.53) 3,125 (34.21) 5,959 (37.95) 3,647 (40.16)
Number of annual PCI cases 659.90 (404.19–1,083.80) 628.65 (378.37–978.81) 663.88 (405.31–1,083.80) 700.68 (425.83–1,166.40) <0.001

Values are median (IQR) or n (%). *Baseline characteristics stratified by LOS categories.
CABG ¼ coronary artery bypass graft surgery; IABP ¼ intra-aortic balloon pump; IQR ¼ interquartile range; LAD ¼ left anterior descending; LOS ¼ length of stay; LVEF ¼ left ventricular ejection fraction;
MI ¼ myocardial infarction; PCI ¼ percutaneous coronary intervention; TIMI ¼ Thrombolysis In Myocardial Infarction.

from the Mayo Clinic, which performs greedy matching
of cases to control subjects (LOS 3–4 vs. LOS 1–2) within a
pre-specified caliper having a width of 0.2 times the
SD of the logit of the propensity score. In the propensity
score model, we adjusted for the variables in the NCDR
mortality model, hospital characteristics, and additional
patient and procedural factors including in-hospital
complications of bleeding, transfusions, vascular com-
plications, and renal insufficiency. To estimate the LOS
3–4 versus LOS 1–2 effect on outcomes among the pro-
pensity matched sample, we present the odds ratio (OR)
from a logistic regression model stratified by matched
pair. We were able to match 85% of the patients with
a very short LOS within the pre-specified caliper for
analysis.
To assess LOS trends over time, we calculated
percentiles of the distribution of LOS by discharge
year. A p value <0.05 was considered significant for
all tests. The Duke Clinical Research Institute per-
formed all statistical analyses using SAS software
version 9.2 (SAS Institute, Cary, North Carolina).
RESULTS
patients with longer LOS were older and more
frequently had prior MI, heart failure, diabetes, renal
failure, cerebrovascular disease, peripheral vascular
disease, chronic lung disease, hypertension, and
shock. In addition, symptom onset to admission time
was generally shorter in short and medium LOS
groups (18.0% and 18.5%, respectively) compared
with the long LOS group, in which presentation >6 h
from symptom onset was more frequent (21.5%).
There were also differences in procedural characteris-
tics and in-hospital adverse events. Patients in the short
LOS group were more likely to have preserved left ven-
tricular function (median ejection fraction 50%), non–left
anterior descending (LAD) culprit vessel, TIMI (Throm-
bolysis In Myocardial Infarction) flow grade 1 to 3 before
PCI, and were more likely to have a successful PCI proce-
dure. Patients in the long LOS group had more multiorgan
derangements and anatomic features associated with
worse outcomes, such as the following: multivessel dis-
ease, complex (type C) coronary lesions, LAD culprit
vessel, TIMI flow grade 0 before PCI, shock or heart failure,
IABP placement, and in-hospital bleeding or vascular
complications.

BASELINE CHARACTERISTICS BY LOS. During the HOSPITAL, GEOGRAPHIC VARIATION, AND TRENDS
study period, 33,920 patients with STEMI pre- IN LOS. Patients in the short LOS group were more
sented to 1,028 sites after inclusion and exclusion likely to be treated at smaller hospitals with fewer
criteria were applied to the linked dataset, where beds and annual PCI cases. Discharging patients
the matching algorithm successfully linked 70% of with STEMI early was also more commonly prac-
eligible patients (Figure 1). The distribution of LOS is ticed in hospitals in the West and Midwest regions
displayed in Figure 2, and the proportion of patients compared with the South and Northeast (Table 1).
in each LOS category was as follows: short LOS The median LOS over the 6-year period was
(26.9%, n ¼ 9,135), medium LOS (46.3%, n ¼ 15,704), 4 days; however, the relative frequency of patients
and long LOS (26.8%, n ¼ 9,081). in each LOS category changed with a trend toward
There were differences in baseline demographics earlier discharges over time. In 2005, 24% of
among the 3 examined groups, with short and discharges in older patients with STEMI were in
medium LOS groups having fewer comorbidities patients with a short LOS; this percent increased
compared with the long LOS group (Table 1). Overall, to 30% by 2009. Conversely, 47% and 29% of
1166 Swaminathan et al. JACC VOL. 65, NO. 12, 2015

Hospital Length of Stay and Outcomes in STEMI MARCH 31, 2015:1161–71

discharges in patients with STEMI were in patients
with medium and long LOS, respectively, and
these percents decreased to 45% and 25% by 2009,
respectively.
PRIMARY OUTCOMES. The 30-day mortality and
MACE rates were 0.9% and 1.9% for short LOS, 1.0%
and 2.2% for medium LOS, and 3.5% and 5.0% for long
LOS, respectively (p < 0.0001) (Figures 3A and 3B). A
multivariable
demonstrated similar adjusted rates of death and
MACE within 30 days for medium versus short LOS
(mortality, HR: 1.00; 95% CI: 0.74 to 1.34; MACE, HR:
1.03; 95% CI: 0.86 to 1.25).
SECONDARY OUTCOMES. The incidence of read-
missions for MI at 30 days was 1.0% for short LOS
and 1.1% for medium LOS, but it was significantly
higher at 1.6% for long LOS (p ¼ 0.001) (Figure 3C).
The rates of 30-day readmissions for unplanned
Cox regression analysis (Table 2) revascularizations were not significantly different
across LOS categories (0.7% for short LOS and

F I G U R E 3 Cumulative Incidence of Primary and Secondary Outcomes

A 0.05
LOS >5 B 0.05 LOS >5
LOS >5

Cumulative Incidence of MACE

LOS 4-5 LOS 4-5
Cumulative Incidence of Death

LOS <=3 LOS <=3

0.04 0.04
LOS >5
0.03 0.03

LOS 4-5
0.02 0.02
LOS <=3
LOS 4-5
0.01 0.01
LOS <=3

0.00 0.00
9080 8855 8703 8584 9079 8758 8567 8426
15704 15520 15356 15207 15703 15386 15157 14959
9135 9006 8908 8807 9135 8929 8796 8665
0 10 20 30 0 10 20 30
Days Days
C D 0.0150
0.0150 LOS >5 LOS >5 LOS >5
LOS 4-5 LOS 4-5
Cumulative Incidence of Revasc

LOS <=3 0.0125 LOS <=3

0.0125
Cumulative Incidence of MI

LOS 4-5 LOS >5

0.0100 0.0100
LOS <=3 LOS 4-5

0.0075 0.0075
LOS <=3

0.0050 0.0050

0.0025 0.0025

0.0000 0.0000
9079 8767 8579 8442 9079 8793 8613 8480
15703 15396 15177 14983 15703 15413 15191 14997
9135 8931 8801 8671 9135 8954 8826 8694
0 10 20 30 0 10 20 30
Days Days

E 0.0150
LOS >5
LOS 4-5
Cumulative Incidence of Bleed

0.0125 LOS <=3

0.0100

0.0075

0.0050 LOS >5

LOS 4-5
0.0025
LOS <=3

0.0000
9080 8827 8647 8508
15704 15471 15273 15085
9135 8983 8864 8736

0 10 20 30
Days

30-day clinical outcomes stratified by LOS. (A) All-cause mortality. (B) MACE. (C) Readmission for MI. (D) Unplanned revascularization.
(E) Bleeding event. Abbreviations as in Figures 1 and 2.
JACC VOL. 65, NO. 12, 2015 Swaminathan et al. 1167
MARCH 31, 2015:1161–71 Hospital Length of Stay and Outcomes in STEMI

1.0% for both medium and long LOS; p ¼ 0.23)

T A B L E 2 Association of LOS on Time-to-Event Within 30 Days*
(Figure 3D). The frequency of bleeding events
requiring readmission after initial discharge was Unadjusted HR Adjusted HR
Outcome LOS (95% CI) p Value (95% CI) p Value
relatively low, but it increased across all LOS cate-
Death 4–5 days vs. #3 days 1.16 (0.87–1.55) 0.32 1.00 (0.74–1.34) 0.98
gories: 0.2%, 0.4%, and 0.6% for short, medium, and
>5 days vs. #3 days 3.98 (3.08–5.15) <0.001 2.30 (1.72–3.07) <0.001
long LOS, respectively (p < 0.001) (Figure 3E). MACE 4–5 days vs. #3 days 1.16 (0.96–1.40) 0.12 1.03 (0.86–1.25) 0.73
>5 days vs. #3 days 2.65 (2.22–3.15) <0.001 1.75 (1.44–2.12) <0.001
FACTORS ASSOCIATED WITH LONGER LOS. Table 3
displays independent predictors of longer LOS *Unadjusted and adjusted hazard ratios for death and MACE between medium versus short LOS and long versus
(>4 days) after controlling for baseline characteris- short LOS.
CI ¼ confidence interval; HR ¼ hazard ratio; LOS ¼ length of stay; MACE ¼ major adverse cardiac event(s).
tics, variables from the NCDR mortality model, and
in-hospital complications. The factors associated with
longer LOS included pre-operative IABP, transfusion, remained significantly lower for LOS 3–4 versus
vascular complication, cardiogenic shock, renal in- LOS 1–2.
sufficiency, and bleeding. Hospitals located in the
West and Midwest regions were 24% and 28% more DISCUSSION
likely to have short versus longer LOS than hospitals
located in the South, respectively, whereas hospitals In this large, contemporary national registry, we
in the Northeast had longer LOS versus hospitals in found that a majority of older patients with STEMI
the South (see Table 3). (46%) had a medium LOS, whereas 27% of patients
had a short LOS. After adjustment, patients with
VERY SHORT LOS AND OUTCOMES. Very short LOS
short LOS had similar mortality and MACE rates as
(1 to 2 days), which represents discharge on the day
those with medium LOS. These data suggest that
of PCI or the day after PCI (similar to the practice for
elective PCIs) was infrequent and present in 1,244
patients (3.7%). The relationship between distribu- T A B L E 3 Association of Selected Covariates on LOS*

tion of LOS and clinical outcomes (MACE) is shown in

LOS >4 Days vs. #3 Days
Figure 4 and reveals that very short LOS is associated
OR (95% CI) p Value
with significantly worse 30-day clinical outcomes. A
Vascular complication 3.88 (2.88–5.22) <0.001
“U-shaped” curve for MACE events was noted with a Bleeding complication 2.61 (2.30–2.95) <0.001
nadir at LOS 3–4, where overall unadjusted 30-day Transfusion 3.94 (2.24–8.35) <0.001
MACE rates were similar for LOS of 3 or 4 days. Pre-operative IABP 4.32 (3.47–4.47) <0.001
Compared with LOS 3–4 (Online Table 1), patients in Cardiogenic shock 3.01 (2.73–3.32) < 0.001

the very short LOS group were of similar age, but Highest-risk lesion
Proximal LAD vs. other 1.58 (1.48–1.69) <0.001
more commonly they were male with hypertension,
Left main vs. other 2.23 (1.55–3.23) <0.001
hyperlipidemia, chronic lung disease, diabetes, and
Renal insufficiency 2.68 (2.15–3.34) <0.001
earlier PCI or CABG. Patients in the very short
Pre-procedure TIMI flow grade 0 1.21 (1.14–1.29) <0.001
LOS group were more likely to have TIMI flow grade Age (per 10-year increase) 1.46 (1.41–1.51) <0.001
1 to 3 pre-PCI and a lower frequency of LAD culprits, Cerebrovascular disease 1.28 (1.18–1.39) <0.001
cardiogenic shock, any bleeding, or vascular compli- Prior congestive heart failure 1.50 (1.34–1.67) <0.001
cations. The PCI procedural success rate was similar. Peripheral vascular disease 1.07 (0.98–1.17) 0.11

Guideline-directed post-STEMI discharge medica- Number of hospital beds 1.06 (1.03–1.09) <0.001
(per 100 increase)
tions including antiplatelet agents, beta-blockers, Chronic lung disease 1.37 (1.28–1.47) <0.001
and statins were less frequently prescribed in pa- University hospital 0.93 (0.78–1.11) 0.41
tients in the very short LOS group. Compared with Prior PCI 0.86 (0.82–0.91) <0.001
very short LOS, patients with a 3- to 4-day LOS had White race 0.83 (0.77–0.89) <0.001
>50% reduction in adjusted mortality (OR: 0.47; Male 0.76 (0.73–0.80) <0.001
No diabetes vs. IDDM 0.61 (0.55–0.68) <0.001
95% CI: 0.27 to -0.82 for LOS 3 vs. LOS 1–2; OR:
Region
0.46; 95% CI: 0.27 to 0.79, for LOS 4 vs. LOS 1–2)
West vs. South 0.76 (0.67–0.86) <0.001
and a more than 40% reduction in MACE (OR:
Northeast vs. South 1.29 (1.12–1.50) <0.001
0.59; 95% CI: 0.40 to 0.86 for LOS 3 vs. LOS 1–2; Midwest vs. South 0.72 (0.65–0.80) <0.001
OR: 0.56; 95% CI: 0.38 to 0.82 for LOS 4 vs. LOS
1–2) (Table 4). Similarly, among propensity-matched *ORs of factors associated with prolonged LOS versus short LOS.
IDDM ¼ insulin-dependent diabetes mellitus; OR ¼ odds ratio; other
cohorts, the risk of death (OR: 0.41; 95% CI: 0.21 to abbreviations as in Tables 1 and 2.
0.81) and MACE (OR: 0.52; 95% CI: 0.33 to 0.80)
1168 Swaminathan et al. JACC VOL. 65, NO. 12, 2015

Hospital Length of Stay and Outcomes in STEMI MARCH 31, 2015:1161–71

reductions in LOS nationally may be accompanied by

F I G U R E 4 Association of 30-Day MACE by Each LOS Day and
Frequency of Distribution of LOS
unintended consequences (23). Therefore, it is critical
to examine the safety and understand the effects of
5 5.0 % 10,000 such changes, particularly among Medicare benefi-
N = 9,936
N = 9,081 ciaries with substantial comorbidities. It is reassuring
that in broadly representative older patients at more
4 N = 7,891 8,000 than 1,000 centers in the United States, 30-day mor-
tality and MACE outcomes were similarly low among
patients with short and medium LOS. It remains un-

# of Patients
3 2.9 % 2.9 % 6,000
known whether shorter LOS and reduced hospital
% MACE

N = 5,768
costs align with effects on care continuity and pa-
tients’ preferences.
2 4,000
1.7 % 1.8 %
We found that patients with long LOS (27%) had
the highest rate of 30-day mortality and MACE. Given
1 2,000 the observational nature of our study, these data
N = 1,244 should not be interpreted to mean that keeping a
patient with STEMI in the hospital beyond 5 days
0 0 would translate into poor outcomes. Rather, this
LOS 1-2 LOS 3 LOS 4 LOS 5 LOS>5 group was a sicker cohort of patients with substantial
% MACE # of patients
comorbidities and in-hospital complications relative
to those with short and medium LOS, and thus these
“U-shaped” curve demonstrating high MACE with very short LOS, lower and equivalent patients would be expected to have more read-
MACE for an LOS of 3 to 4 days, and highest MACE for patients with a long LOS.
missions within 30 days. An early discharge strategy
Abbreviations as in Figures 1 and 2.
would apply only to patients who do not develop
post-procedural complications, and recuperation at

selected patients with STEMI who do not develop post- home would attenuate nosocomial infection and

procedural complications may be eligible for an ear- delirium, to which older patients are prone.

lier discharge without an increase in 30-day adverse Patients with short LOS had less complicated pre-

events (Central Illustration). Furthermore, from 2004 sentations. The safety of early discharge for such pa-

to 2009, the proportion of patients undergoing ear- tients is in line with previous prospective studies

lier discharges increased. This pattern may reflect (24,25), which were limited by small sizes and

the increasing safety of PCI in general (21), the use of single-center experiences. The PAMI-II (Primary An-

bleeding avoidance strategies (22), and the financial gioplasty in Myocardial Infarction -11) study from 1998

pressures for hospitals to reduce inpatient LOS. implemented a strategy of early identification of low-

In contrast to established guidelines for “door-to- risk patients with STEMI to receive accelerated care

balloon” times, current guidelines do not provide in a non–intensive care unit setting with hospital

criteria for determining an optimal LOS or “balloon- discharge on day 3 without noninvasive testing or

to-door” time for patients with STEMI. Large-scale traditional care (24). The study showed that the ac-

changes in the way post-STEMI care is delivered and celerated group had similar in-hospital and 6-month
outcomes, yet these patients were discharged 3 days
earlier than the traditional group (4.2  2.3 days vs.
T A B L E 4 Association of Each LOS Day on Time-to-Event Within 30-Days*
7.1  4.7 days; p ¼ 0.0001). Similarly, the Safe-Depart
Length of Stay Unadjusted HR Adjusted HR trial (25), a small, single-center study, identified 54
Outcome (Days) (95% CI) p Value (95% CI) p Value
low-risk patients with STEMI on the basis of the
Death 3 vs. 1–2 0.49 (0.28–0.84) 0.01 0.47 (0.27–0.82) 0.01
Zwolle Primary PCI Index (26), after these patients
4 vs. 1–2 0.51 (0.30–0.87) 0.01 0.46 (0.27–0.79) <0.01
underwent primary or rescue PCI, and found that an
5 vs. 1–2 0.88 (0.52–1.50) 0.63 0.67 (0.39–1.15) 0.14
early discharge strategy with rapid follow-up was
$6 vs. 1–2 2.21 (1.35–3.61) <0.01 1.26 (0.75–2.10) 0.38
MACE 3 vs. 1–2 0.60 (0.41–0.88) 0.01 0.59 (0.40–0.86) 0.01 feasible.
4 vs. 1–2 0.61 (0.41–0.89) 0.01 0.56 (0.38–0.82) <0.01 An important finding of our study is that very early
5 vs. 1–2 1.01 (0.70–1.46) 0.96 0.82 (0.57–1.20) 0.30 discharge (same-day or overnight stay after PCI) is
$6 vs. 1–2 1.72 (1.22–2.43) <0.01 1.13 (0.79–1.61) 0.50 associated with an increase in 30-day mortality and
MACE. Risks associated with PCI, such as acute stent
*Unadjusted and adjusted hazard ratios for death and MACE between each LOS day and very short LOS.
Abbreviations as in Table 2.
thrombosis, bleeding, and renal failure, are highest
within the first 24 to 48 h (27); thus, longer observation
JACC VOL. 65, NO. 12, 2015 Swaminathan et al. 1169
MARCH 31, 2015:1161–71 Hospital Length of Stay and Outcomes in STEMI

CENTRAL I LLU ST RAT ION Hospital LOS and Clinical Outcomes in Older Patients With STEMI After Primary PCI

HOSPITAL LENGTH EXPECTED PRIMARY OUTCOME EXPECTED SECONDARY OUTCOME

OF STAY POST PCI TYPICAL PATIENT (30-day mortality and MACE rates) (Incidence of 30-day readmission)

Very early discharge None High 30-day mortality and N/A

(same-day or major adverse cardiac events (MACE).
overnight Risks associated with PCI, such as
stay after PCI) acute stent thrombosis, bleeding, and
renal failure, are highest within the
first 24 to 48 hours

Short • Symptom onset to admission time ≤6 hours 0.9% and 1.9% • Myocardial infarction: 1%
(≤3 days after PCI) • Preserved LV function prior to PCI • Unplanned revascularizations: 0.7%
• Without comorbidities • Bleeding event: 0.2%
• Without post-procedural complications
• Patients tend to be younger

Medium Similar to Short 1.0% and 2.2% • Myocardial infarction: 1.1%

(4-5 days after PCI) • Unplanned revascularizations: 1%
Similar rates as short suggesting • Bleeding event: 0.4%
that patients who do not develop
post-procedural complications may
be eligible for an earlier discharge
without an increase in 30-day
adverse events

Long • Symptom onset to admission time >6 hours
(>5 days after PCI) • Without preserved LV function prior to PCI
• With comorbidities
• With post-procedural complications
• Patients tend to be older
3.5% and 5.0% • Myocardial infarction: 1.6%
• Unplanned revascularizations: 1%
This should not be interpreted to mean • Bleeding event: 0.6%
that keeping a patient with STEMI in
the hospital beyond 5 days would
translate into poor outcomes; this
group is a sicker cohort of patients

Swaminathan, R.V. et al. J Am Coll Cardiol. 2015; 65(12):1161–71.

A summary of 30-day mortality and major adverse cardiac events (MACE) rates stratified by length of stay (LOS) categories. LV ¼ left ventricular; PCI ¼ percutaneous
coronary intervention; STEMI ¼ ST-segment elevation myocardial infarction.

may be required to avert these complications.
Furthermore, optimal post-MI medications were less
frequently prescribed on discharge in patients with
very short hospital stays. The current data highlight
the importance of implementing standardized pro-
tocols for STEMI discharge and post-discharge follow-
up in patients with very short LOS, in addition to the
need for further clinical evidence that a very early
discharge strategy can be safely applied in contempo-
rary practice. From a hospital policy standpoint, this
finding supports justification for “inpatient” moni-
toring and reimbursement under the recently intro-
duced CMS “2-midnight rule.” The concept and safety
of an “outpatient STEMI PCI” when the patient stays
<2-midnights would require further investigation and
policy discussion because clearly there would be a
disconnect between the level of monitoring required
and the time spent in the hospital.
We also found geographic and hospital-level vari-
ation that affected LOS. Hospitals with smaller
numbers of beds had greater proportion of patients in
the short LOS group, a finding that may reflect pres-
sure to expedite bed availability. Of interest, hospi-
tals in the West and Midwest had significantly shorter
LOS compared with the Northeast and South. Such
variation has been seen in earlier studies for STEMI
(8) and elective PCI (28), and it may be a reflection of
hospital discharge policy or institutional culture,
unrelated to medical or logistical factors. This finding
highlights potential areas for improvement and cost
savings within certain regions.
STUDY LIMITATIONS. First, because the CathPCI regis-
try accounts for only in-hospital events, we linked
the database to CMS claims data to obtain 30-day
clinical outcomes. This limited our population to pa-
tients older than 65 years and subjects the data to
coding bias and possibly missing events. However,
older patients are at higher risk for in-hospital com-
plications after PPCI (29–31), so it is likely that a
higher proportion of younger patients would be
1170 Swaminathan et al. JACC VOL. 65, NO. 12, 2015

Hospital Length of Stay and Outcomes in STEMI MARCH 31, 2015:1161–71

eligible for earlier discharge. Second, our study question that needs to be addressed in future
period ended in 2009, after which further evolution studies, particularly those that use chart reviews to
in care models for patients with STEMI, advances in examine why certain patients have a very short
bleeding avoidance strategies including increased use hospital LOS after STEMI.
of radial access for STEMI (32), and use of newer-
CONCLUSIONS
generation drug-eluting stents may have led to
continued reductions in LOS with a positive impact
Nearly one-half of the older patients with STEMI in
on 30-day outcomes. Third, although our robust sta-
the United States who are undergoing PPCI have a 4-
tistical approach using validated risk models and
to 5-day LOS, yet patients discharged earlier have
propensity matching accounted for some of the po-
similar 30-day mortality and MACE. These data sug-
tential bias inherent to observational analyses, there
gest that early discharge (but not within 48 h after
remains risk of unmeasured confounders. Fourth,
PPCI) may be safe among selected older patients un-
some patients after STEMI may return to the hospital
dergoing PPCI for STEMI who do not develop in-
within 30 days for elective or “staged” revasculariza-
hospital complications.
tion of a nonculprit artery. To minimize counting this
as an unplanned revascularization, we defined repeat
REPRINT REQUESTS AND CORRESPONDENCE: Dr.
revascularization as readmissions with PCI or CABG
Rajesh Swaminathan, Cardiac Catheterization Labo-
that were associated with a principal discharge
ratories, Weill Cornell Medical College, 520 East 70th
diagnosis that was clearly not consistent with an
Street, Starr-431 Pavilion, New York, New York 10021.
elective readmission. This methodology has inherent
E-mail: rvs9001@med.cornell.edu.
limitations because it is based on administrative
data; however, it is currently the most reliable al-
PERSPECTIVES
gorithm available and has been utilized in other
published registry data. For this reason, we also
COMPETENCY IN SYSTEMS-BASED PRACTICE:
analyzed each component of the combined MACE
The importance of in-hospital monitoring for patients
separately as secondary endpoints. Furthermore,
with MI and appropriate reasons for variation from the
miscoding the second admission, given that it is
mean 4- to 5-day inpatient stay on the basis of risk
linked so closely to the index admission for STEMI,
factors and comorbidities should be made clear to
may have increased readmission event rates; how-
patients, family members, other caregivers, and
ever, miscoding would likely be similar across the
payers. Earlier discharge (still no <48 h after PCI)
various LOS categories. Finally, we could not ac-
may be safe in selected older patients undergoing
count for patients who may have very short stays as
uncomplicated primary PCI.
a result of leaving against medical advice, a factor
that may also portend worse outcomes secondary to
TRANSLATIONAL OUTLOOK: Early discharge
medication noncompliance. These factors, in combi-
strategies could be developed that promote safe
nation with the overall small sample size of the very
transitions to the outpatient setting, optimize effi-
short LOS group, preclude any definitive conclusions
ciency, and enhance patient-oriented outcomes.
about the potential safety of a very early discharge
strategy. Our analysis therefore raises an important

REFERENCES

1. Mechanic R, Tompkins C. Lessons learned pre-
paring for Medicare bundled payments. N Engl J
Med 2012;367:1873–5.
2. Jencks SF, Williams MV, Coleman EA. Rehospi-
talizations among patients in the Medicare fee-
for-service program. N Engl J Med 2009;360:
1418–28.
3. Andersen HR, Nielsen TT, Rasmussen K, et al.
A comparison of coronary angioplasty with fibri-
nolytic therapy in acute myocardial infarction.
N Engl J Med 2003;349:733–42.
4. Busk M, Maeng M, Rasmussen K, et al. The
Danish multicentre randomized study of fibrino-
lytic therapy vs. primary angioplasty in acute
myocardial infarction (the DANAMI-2 trial): out-
come after 3 years follow-up. Eur Heart J 2008;
29:1259–66.
5. Widimsky P, Budesinsky T, Vorac D, et al. Long
distance transport for primary angioplasty vs. im-
mediate thrombolysis in acute myocardial infarc-
tion: final results of the randomized national
muticentre trial—PRAGUE-2. Eur Heart J 2003;24:
94–104.
6. Widimsky P, Bilkova D, Penicka M, et al. Long-
term outcomes of patients with acute myocardial
infarction presenting to hospitals without cathe-
terization laboratory and randomized to immedi-
ate thrombolysis or interhospital transport for
primary percutaneous coronary intervention: five
years’ follow-up of the PRAGUE-2 Trial. Eur Heart
J 2007;28:679–84.
7. Schomig A, Kastrati A, Dirschinger J, et al. Cor-
onary stenting plus platelet glycoprotein IIb/IIIa
blockade compared with tissue plasminogen acti-
vator in acute myocardial infarction: Stent versus
Thrombolysis for Occluded Coronary Arteries in
Patients with Acute Myocardial Infarction Study
investigators. N Engl J Med 2000;343:385–91.
8. Chin CT, Weintraub WS, Dai D, et al. Trends and
predictors of length of stay after primary percu-
taneous coronary intervention: a report from the
CathPCI Registry. Am Heart J 2011;162:1052–61.
JACC VOL. 65, NO. 12, 2015
MARCH 31, 2015:1161–71
9. Kociol RD, Lopez RD, Clare R, et al. Interna-
tional variation in and factors associated with
hospital readmission after myocardial infarction.
JAMA 2012;307:66–74.
10. Brindis RG, Fitzgerald S, Anderson HV,
Shaw RE, Weintraub WS, Williams JF. The Amer-
ican College of Cardiology-National Cardiovascular
Data Registry (ACC-NCDR): building a national
clinical data repository. J Am Coll Cardiol 2001;37:
2240–5.
11. Weintraub WS, McKay CR, Riner RN, et al. The
American College of Cardiology National Data-
base: progress and challenges. American College
of Cardiology Database Committee. J Am Coll
Cardiol 1997;29:459–65.
12. Messenger JC, Ho KK, Young CH, et al. The
National Cardiovascular Data Registry (NCDR) data
quality brief: the NCDR Data Quality Program in
2012. J Am Coll Cardiol 2012;60:1484–8.
13. NCDR CathPCI Registry v4.4 Coder’s Data
Dictionary. 2011. Available at: https://www.ncdr.
com/WebNCDR/docs/public-data-collection-docu
ments/cathpci_v4_codersdictionary_4-4.pdf?
sfvrsn¼2. Accessed August 5, 2014.
14. Brennan JM, Peterson ED, Messenger JC, et al.
Linking the National Cardiovascular Data Registry
CathPCI Registry with Medicare claims data. Circ
Cardiovasc Qual Outcomes 2012;5:134–40.
15. Rao SV, Dai D, Subherwal S, et al. Association
between periprocedural bleeding and long-term
outcomes following percutaneous coronary inter-
vention in older patients. J Am Coll Cardiol Intv
2012;5:958–65.
16. Bueno H, Ross JS, Wang Y, et al. Trends in
Length of Stay and Short-term Outcomes Among
Medicare Patients Hospitalized for Heart Failure,
1993–2006. JAMA 2010;303:2141–7.
17. Krumholz HM, Normand S-LT, Keenan PS, et al.
Hospital 30-Day Acute Myocardial Infarction
Readmission Measure: Methodology. Report pre-
pared for the Centers for Medicare & Medicaid
Services. June 9, 2008. Available at: https://www.
qualitynet.org/dcs/ContentServer?c¼Page&page
name¼QnetPublic%2FPage%2FQnetTier4&cid¼
1219069855841. Accessed August 5, 2014.
18. Gray JR. A class of K-sample tests for com-
paring the cumulative incidence of a competing
risk. Ann Stat 1988;16:1141–54.
19. Peterson ED, Dai D, DeLong ER, et al. Contem-
porary mortality risk prediction for percutaneous
coronary intervention: results from 588,398 pro-
cedures in the National Cardiovascular Data Regis-
try. J Am Coll Cardiol 2010;55:1923–32.
20. Zeger SL, Liang KY. An overview of methods
for the analysis of longitudinal data. Stat Med
1992;11:1825–39.
21. Rao SV, Hess CN, Dai D, Green CL, Peterson ED,
Douglas PS. Temporal trends in percutaneous
coronary intervention outcomes among older pa-
tients in the United States. Am Heart J 2013;166:
273–81.
22. Marso SP, Amin AP, House JA, et al. Associa-
tion between use of bleeding avoidance strategies
and risk of periprocedural bleeding among pa-
tients undergoing percutaneous coronary inter-
vention. JAMA 2010;303:2156–64.
23. Baker DW, Einstadter D, Husak SS, Cebul RD.
Trends in postdischarge mortality and read-
missions: has length of stay declined too far? Arch
Intern Med 2004;164:538–44.
24. Grines CL, Marsalese DL, Brodie B, et al. Safety
and cost-effectiveness of early discharge after
primary angioplasty in low risk patients with acute
myocardial infarction: PAMI-II Investigators. Pri-
mary Angioplasty in Myocardial Infarction. J Am
Coll Cardiol 1998;31:967–72.
25. Kotowycz MA, Cosman TL, Tartaglia C, Afzai R,
Syal RP, Natarajan MK. Safety and feasibility of
early hospital discharge in ST-segment elevation
myocardial infarction: a prospective and random-
ized trial in low-risk primary percutaneous coro-
nary intervention patients (the Safe-Depart Trial).
Am Heart J 2010;159:117.e1–6.
26. De Luca G, Suryapranata H, van’t Hof AW,
et al. Prognostic assessment of patients with acute
Swaminathan et al. 1171
Hospital Length of Stay and Outcomes in STEMI
myocardial infarction treated with primary angio-
plasty: implications for early discharge. Circulation
2004;109:2737–43.
27. Rao SV, Ohman EM. Anticoagulant therapy for
percutaneous coronary intervention. Circ Car-
diovasc Interv 2010;3:80–8.
28. Rao SV, Kaltenbach LA, Weintraub WS, et al.
Prevalence and outcomes of same-day discharge
after elective percutaneous coronary intervention
among older patients. JAMA 2011;306:1461–7.
29. Wennberg DE, Makenka DJ, Sengupta A, et al.
Percutaneous transluminal coronary angiography
in the elderly: epidemiology, clinical risk factors
and in-hospital outcomes. The Northern New En-
gland Cardiovascular Disease Study Group. Am
Heart J 1999;137:639–45.
30. Shaw RE, Anderson HV, Brindis RG, et al.
Development of a risk adjustment mortality model
using the American College of Cardiology-National
Cardiovascular Data Registry (ACC-NCDR) experi-
ence: 1998–2000. J Am Coll Cardiol 2002;39:
1104–12.
31. Halkin A, Singh M, Nikolsky E, et al. Prediction
of mortality after primary percutaneous coronary
intervention for acute myocardial infarction: the
CADILLAC risk score. J Am Coll Cardiol 2005;45:
1397–405.
32. Feldman DN, Swaminathan RV, Kaltenbach LA,
et al. Adoption of radial access and comparison of
outcomes to femoral access in percutaneous cor-
onary intervention: an updated report from the
national cardiovascular data registry (2007–2012).
Circulation 2013;127:2295–306.
KEY WORDS length of stay, myocardial
infarction, older patients, outcomes,
primary percutaneous coronary intervention,
ST-segment elevation myocardial infarction
A PP END IX For a supplemental table, please
see the online version of this article.