Author + information
- Received June 7, 2012
- Revision received September 26, 2012
- Accepted October 26, 2012
- Published online February 1, 2013.
- Eltigani Abdelaal, MD⁎,
- Sunil V. Rao, MD†,
- Ian C. Gilchrist, MD‡,
- Ivo Bernat, MD§,
- Adhir Shroff, MD, MPH∥,
- Ronald Caputo, MD¶,
- Olivier Costerousse, PhD⁎,
- Samir B. Pancholy, MD# and
- Olivier F. Bertrand, MD, PhD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Olivier F. Bertrand, Institut Universitaire de Cardiologie et de Pneumologie de Québec, 2725 Chemin Sainte-Foy. Québec, Québec, G1V4G5, Canada
Objectives This study sought to evaluate outcomes of same-day discharge (SDD) following percutaneous coronary intervention (PCI) versus overnight hospitalization (ON).
Background Although there are data on the safety and feasibility of SDD after PCI, ON continues to be prevalent.
Methods The Cochrane search strategy was used to search the PubMed database, EMBASE, and the Cochrane Library for relevant literature. Thirteen studies (5 randomized and 8 observational) of SDD after uncomplicated PCI versus ON met inclusion criteria. Data were pooled using a random effects model, and reported as odds ratios (OR) with their 95% confidence intervals (CI). The primary outcomes were incidence of total complications, major adverse cardiovascular events (MACE), and rehospitalization within 30 days after PCI.
Results A total of 13 studies, involving 111,830 patients were pooled. There was significant variation in the definition of outcomes across studies. For total complications, the strategy of SDD compared with ON after PCI had an estimated OR of 1.20 (95% CI: 0.82 to 1.74) in randomized and 0.67 (95% CI: 0.27 to 1.66) in observational studies. Similar results were found for MACE (randomized, OR: 0.99, 95% CI: 0.45 to 2.18; observational, OR: 0.59, 95% CI: 0.06 to 5.57) and rehospitalizations (randomized, OR: 1.10, 95% CI: 0.70 to 1.74; observational, OR: 0.62, 95% CI: 0.10 to 3.98) at 30 days post PCI.
Conclusions There is considerable heterogeneity across published studies comparing SDD with ON. This, coupled with the low event rate and wide corresponding CIs, suggest that an adequately powered multicenter randomized trial comparing SDD with ON would require a very large sample size (>17,000). Until such a trial is completed, SDD after uncomplicated PCI seems a reasonable approach in selected patients.
Since its first inception over 3 decades ago, percutaneous coronary intervention (PCI) has become the most commonly performed cardiac intervention worldwide, with over 50% performed in the United States. The rising number of PCI procedures worldwide, combined with the escalating cost of health care, has led to significant logistic and financial constraints on healthcare facilities. Given the increased emphasis on measures to decrease medical care expenditure while upholding clinical efficacy and patient safety, same-day discharge after PCI in selected patients may lead to reduced healthcare costs and increased patient satisfaction. Over the past decade, continuous refinements of PCI procedures with the advent of coronary stents, miniaturization of PCI equipment, and the evolution of adjuvant pharmacology have substantially increased the success, improved short- and long-term safety of this technique, and reduced post-procedural length of stay (1–4).
A number of observational and a few randomized studies have described the safety of same-day discharge (SDD) in selected groups of patients (5–17), adding to several single-center reports extending back to the mid -90s, which suggested the feasibility and safety of this strategy (18–27). Individual reports have also highlighted potential financial savings with this approach, primarily by avoiding the cost of an overnight stay (28–30). Despite this, the practice of SDD after elective, uncomplicated PCI remains considerably variable worldwide, ranging from near routine in some parts of Europe, Asia, and Canada, to being rarely adopted in the United States. Reasons for this are multifactorial, including concerns about patient safety and short-term clinical outcomes, in addition to a perceived lack of fiscal incentive due to the current reimbursement structure of the United States healthcare system.
Many of the studies comparing SDD with overnight stay (ON) have shown no difference in the outcomes that were measured. This is either because both strategies are clinically equivalent, or because of Type II error. Pooling of data from all the studies may provide a large enough sample size to better evaluate whether SDD and ON are comparable with respect to patient outcomes. Accordingly, we undertook a systematic review and comprehensive meta-analysis of all available data comparing SDD after PCI with ON, to evaluate the impact of this strategy on clinical outcomes.
Eligibility, search strategy, and data collection
We performed this review and meta-analysis with standard protocols recommended by the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) (31) and the Quality of Reporting of Meta-analyses (QUOROM) groups (32) for randomized trials, and the Cochrane Collaboration and Meta-analyses of Observational Studies in Epidemiology (MOOSE) group (33) for observational studies. The medical literature databases were scrutinized for reports of studies with SDD after PCI. We performed a search of the PubMed database, EMBASE, and the Cochrane Library, using key words, such as “same-day,” “same-day discharge,” “outpatient,” “day-case,” “day-care,” and “short stay” [AND] “PCI,” “PTCA,” “angioplasty,” and “coronary angioplasty” (last update: 15 April 2011). There was no date or language restriction for our selection of publication. References of selected studies and all abstracts from international cardiology meeting programs (American College of Cardiology, American Heart Association, European Society of Cardiology, Transcatheter Cardiovascular Therapeutics, and Euro-PCR) were searched for relevant data.
For the purpose of this review and meta-analysis, inclusion criteria for studies were: 1) both randomized controlled trials and observational studies comparing SDD and ON following PCI; with 2) clinical and outcome data for up to 30 days following the index PCI procedure available for both SDD and ON groups separately; and 3) intent-to-treat or protocol-driven analysis. Exclusion criteria were: 1) studies reporting only in abstract format or conference presentation with no access to full data or manuscript; 2) studies reporting only outcome data on SDD patients, with no ON group as a comparator; 3) studies reporting on chronology of events but patients kept hospitalized overnight; and 4) studies with mixed outcome data between diagnostic and interventional procedures, or mixed outcome data for both the SDD and ON groups, with no possibility of separating individual outcomes. When study results were reported in abstract format and subsequently in a full paper, only results from the published manuscript were considered. Three evaluators (E.A., O.F.B., and O.C.) performed literature searches, and 2 (E.A. and O.F.B.) individually reviewed relevant articles, and extracted data independently. Discrepancies between datasets were resolved by consensus, if necessary after contact with authors.
Classification of studies and outcome definitions
We classified the studies on the basis of randomized or observational design. The primary outcomes for this study-level meta-analysis were the incidence of total complications, major adverse cardiovascular events (MACE), and rehospitalization within 30 days following PCI.
The definitions of complications were study specific (Table 1). Overall, it represents a summation of both cardiac and noncardiac complications, encompassing access site, bleeding and vascular complications, need for blood transfusion, and cerebrovascular events. MACE was defined as death, myocardial infarction (MI), or repeat revascularization. The definition of MI was as per the study author's definition, and most commonly defined using both typical symptoms, electrocardiogram, and enzymatic criteria (creatine kinase-MB, troponin, or both). Repeat revascularization was defined as the need for repeat PCI or coronary artery bypass grafting. Rehospitalization was defined as repeat hospital admission within 30 days of intervention for any reason related to the index procedure.
There was no standardized definition for bleeding across all studies included herein, and the total number of bleeding events was taken directly from papers, as defined by the authors. The largest randomized study (Bertrand et al. ) in this analysis offered a standardized definition according to REPLACE-2 (Randomized Evaluation in PCI Linking Angiomax to Reduced Clinical Events) criteria for major bleeding and used this as a component of the study's 7-event primary endpoint. Others defined this as major bleeding requiring blood transfusion, or hemoglobin drop of ≥2 g/dl.
The definition covered a wide spectrum of access site complications commonly encountered in clinical practice, such as arteriovenous fistula, pseudoaneurysm, any complication requiring surgery, radial artery occlusion, local infection requiring antibiotics, moderate–severe hematoma, and major bleeding.
Data synthesis and statistical analysis
Information on study design, sample size, population demographic characteristics, coronary angiographic characteristics, access site for PCI, procedural adjuvant pharmacotherapy, procedural success rate, outcome, and follow-up data were extracted and entered into a data sheet using a standardized protocol. The absolute numbers of events for each outcome of interest in every study were extracted for the SDD and ON groups, and entered into a statistical software program (the Cochrane Collaboration's Review Manager [RevMan], version 5.1.20, Nordic Cochrane Centre, Copenhagen, Denmark), which was used to synthesize the results. Differences in study characteristics, inclusion and exclusion criteria, procedural variables, and practice patterns mean that the true effect from each study is likely to vary, and a fixed-effects meta-analysis model would not account for this between-study variation. We therefore used the DerSimonian and Laird random effects model, with weights calculated using the Mantel-Haenszel method to better account for the differences between studies. Dichotomous outcomes data were measured and reported as odds ratio (OR), with their 95% confidence intervals (CIs) for randomized and observational studies. We examined heterogeneity across studies with the Cochran's Q statistic (based on the pooled OR) and the I2 statistic test. An I2 value of <25% was considered low heterogeneity, 25% to 50% moderate, and a value >50% was considered substantial heterogeneity. Funnel plots were constructed and inspected visually for evidence of publication bias (data not shown). The weight of each trial on the overall results of meta-analysis outcome was calculated as a percentage of the number of patients in that given trial over the total number included in each outcome analysis.
A sensitivity analysis was performed according to study design (randomized and observational) and population size to determine the effect of the largest study on the outcome of the meta-analysis.
A total of 13 studies conducted between 1999 and 2011 were included in the meta-analysis (Fig. 1). Five studies were randomized trials, and 8 were observational studies. Twelve studies were single-center (n = 4,182 patients), and 1 was a large U.S. multicenter observational study (N = 107,018 patients). Three studies were conducted in the United States, 3 in Canada, 5 in Europe, and 2 in Asia. Baseline characteristics and procedural data of study populations are shown in Table 2. Our comprehensive review involved a total of 111,830 patients (2,039 enrolled in randomized trials, and 109,791 in observational studies), with 4,179 (3.7%) in the SDD, and 107,651 (96.3%) in the ON group. The mean age for the entire meta-analysis population was 62 ± 4.9 years. Of these, 64% were men, 32% were diabetic, 82% had systemic hypertension, 79% had dyslipidemia, 39% had prior PCI, and 23% had prior coronary artery bypass grafting. Patients in randomized trials (n = 2,039) were slightly younger (60 ± 2 years vs. 63 ± 6 years), and more often men (79% vs. 64.5%) than in observational studies.
Twelve studies (n = 110,825) examined elective PCI in stable angina, but a small proportion also included unstable angina (Braunwald Class 1 and 2). A single randomized trial (EASY [Early Discharge After Transradial Stenting of Coronary Arteries]) with 1,005 patients, examined SDD after PCI in unstable angina and acute coronary syndrome (ACS) with elevated troponin, which represented 25% of study population (12). Khater et al. (14) also included low-intermediate risk ACS patients (using Thrombolysis In Myocardial Infarction risk scoring) in their single-center study of 150 patients. A total of 96% were treated via the femoral approach, whereas transradial access was used in 3%, and a minority of 0.36% had brachial access. When the large observational study of Rao et al. (16) was excluded, radial access was used in 45% of the remaining population, reflecting the current lower utilization of radial access in the United States compared with Europe, Asia, and Canada. Multivessel PCI was performed in 16% of patients, whereas 37% had angiographic Type B or C lesions.
In 4 of 5 randomized, and 7 of 8 observational studies, the anticoagulant of choice in 100% of patients was unfractionated heparin (UFH). In the EASY randomized trial, all patients received abciximab (bolus only in SDD patients vs. bolus plus a 12-h infusion in the ON group) in addition to UFH (12). In the EPOS (Elective PCI in Outpatient Study) randomized trial (13) and in 2 of the observational studies, investigators permitted variable use of a glycoprotein IIb/IIIa inhibitor (GPI). In the large observational study of Rao et al. (16), 40.5% of patients received UFH, 50.6% bivalirudin, whereas 23.1% received a GPI.
Of the 2,039 patients in randomized trials, 1,023 (50.2%) were assigned to SDD after PCI, and 1,016 (49.8%) to ON. Of the 1,023 assigned to SDD, 851 (83%) were successfully discharged home the same day of PCI, whereas 172 (17%) developed an indication for extended observation, and crossed over to an overnight stay (Fig. 2). Reasons for this included procedural-related complications, such as major coronary dissection not suitable for stenting, in-lab transient vessel closure with hemodynamic collapse, intracoronary thrombus, pericardial effusion, procedural arrhythmia, wire perforation, chest pain, or issues with the access site (e.g., hematoma, delayed hemostasis). Very few were due to social reasons or to patient refusal after randomization.
Where reported, the average length of stay after PCI in the majority of studies ranged between 4 and 11 h for SDD, with ≥77% of patients sent home between 4 and 8 h post-procedure, and 1 day for ON, with very few patients kept longer for various clinical indications. Selection criteria for SDD varied significantly between individual studies (Table 3), with the majority agreeing on the following criteria: elective PCI for stable or unstable angina, low–moderate risk ACS, successful uncomplicated PCI, stable clinical condition post-PCI, and patients' willingness to go home with the ability to reach the emergency department within a reasonable time (40 to 60 min), and adequate social support.
In the randomized trials, the total number of complications in the SDD group was 67 of 1,023 (6.5%) versus 56 of 1,016 (5.5%) for the ON group (OR: 1.20, 95% CI: 0.82 to 1.74). The total number of complications reported in the SDD group in the observational data were 148 of 3,156 (4.7%) versus 10,272 of 106,629 (9.6%) for the ON group, (OR: 0.67, 95% CI: 0.27 to 1.66) (Fig. 3). To assess the influence of the Rao et al. study (16) with the largest population size, we performed a sensitivity analysis by removing this study from the forest plot. For the remaining pooled observational studies, total complications in the SDD group was 19 of 1,817 (1.0%) versus 23 of 950 (2.4%) in the ON group (OR: 0.4, 95% CI: 0.07 to 2.14). At 30 days, the incidence of MACE in the SDD versus the ON group in randomized trials was 13 of 1,023 (1.3%) versus 13 of 1,016 (1.3%) (OR: 0.99, 95% CI: 0.45 to 2.18), and in observational studies 10 of 1,817 (0.6%) versus 11 of 950 (1.2%) (OR: 0.59, 95% CI: 0.06 to 5.57) (Fig. 4). Finally, the incidence of rehospitalizations within 30 days after PCI between the SDD and ON groups in randomized trials was 41 of 1,023 (4.0%) versus 37 of 1,016 (3.6%) (OR: 1.10, 95% CI: 0.70 to 1.74) and in observational studies 131 of 1,645 (8.0%) versus 10,147 of 105,827 (9.6%) (OR: 0.62, 95% CI: 0.10 to 3.98), respectively (Fig. 5). When Rao et al. (16) was excluded, the incidence of rehospitalization in observational studies was 3 of 306 (1.0%) in the SDD versus 2 of 148 (1.4%) in the ON group (OR: 0.34, 95% CI: 0.01 to 22.57).
In this systematic review and comprehensive meta-analysis, we show that many carefully selected and risk-stratified groups of patients undergoing elective or ad hoc PCI for low–intermediate risk ACS have been managed successfully with an SDD strategy. Still, due to the low event rate, the significant variation in the definition of outcomes across studies, and wide confidence intervals around the pooled point estimates, a definitive resolution of a statistically significant hazard or benefit to the SDD approach cannot be determined based on the totality of presently available data.
With the advancement of operative techniques and surgical equipment over the past few years, a number of noncardiac disciplines have evaluated the benefits and safety of SDD (34–36). This practice has also been extended to selected vascular procedures performed under general anesthesia and to interventional peripheral, renal, and carotid angioplasty (37). With significant accomplishments in safety and success, PCI techniques have likewise progressed to the point that SDD after PCI is not uncommon in many countries, without reported issues of safety.
Hesitance in changing current practice could broadly be attributed to 4 potential challenges. First, the biggest hurdle seems to be the medicolegal concerns from the medical community, hospitals, and care providers, which relate to the safety of SDD. Second, there may be concerns on the part of physicians, nurses, and hospitals that an SDD strategy may not provide enough time to educate patients about their underlying disease and/or their PCI procedure. Third, there is a misconception that current payment systems, such as in the United States, provide a disincentive to hospitals and physicians to change practice and discharge patients the same day as their procedure. Fourth, some patients may be reluctant to accept an SDD strategy after uncomplicated PCI.
Despite the spreading adoption of SDD after PCI practices worldwide, the present comprehensive systematic review and meta-analysis of all available clinical data involving 111,830 patients cannot scientifically resolve the ultimate question: whether SDD after uncomplicated PCI in a selected subgroup of patients is as safe as or safer than ON. To scientifically resolve this question, assuming similar results for total complications as obtained in this meta-analysis and a noninferiority design according to the Blackwelder equation, a large sample size would be required. Indeed, if there is truly no difference between ON and SDD strategies, then 17,812 patients would be required to be 90% sure that the upper limit of a 1-sided 95% confidence interval will exclude a difference in favor of the ON strategy of more than 1%.
In this meta-analysis, and by definition, patients in randomized trials who were discharged the same day as their PCI had a similar risk profile to those hospitalized overnight. In particular, no clinically important differences in medical comorbidities, procedural characteristics, multivessel PCI, high-risk lesions, access site, or use of vascular closure devices were observed between the 2 groups (Table 2). By contrast, patients in observational studies who were discharged home the same day of their PCI were a highly selected population as evaluated by the operators. Although it is likely that individual centers have created their local protocols for SDD after PCI, our review highlights the lack of standardized selection criteria for this strategy and the lack of standardized definitions of outcomes used across studies. Existing recommendations in the Society for Cardiovascular Angiography and Interventions statement (38) (endorsed by the American College of Cardiology) on SDD may be more conservative than what current “real-world” practice may be (39). Central to many discussions over the merits of SDD versus ON care is an assumption that more hospitalization, although clearly more expensive, is also by definition safer. No trials have attempted to look at the hazard of hospitalization in this patient population, but a recent systematic database review has suggested the risk from an overnight stay in the hospital includes 0.4% of skin ulcer formation, 3.4% suffering adverse drug reactions, and 11.1% of infection (40).
Complications after PCI include ischemic events, such as periprocedural MI and acute stent thrombosis, as well as bleeding and vascular complications. In the contemporary PCI era, there are several strategies to minimize both types of complications. Coronary stents have reduced the risk for abrupt closure, and potent antiplatelet and anticoagulant combinations reduce the risk for periprocedural MI. Appropriate dosing antithrombotic medications, or the preferred use of radial access, can reduce bleeding and vascular complications (41–43). Some observational studies suggest that the use of femoral closure devices in selected patients may also be associated with a reduction in vascular complications (44); however, the exact clinical benefit remains controversial. Given the myriad options to optimize PCI safety and efficacy, an SDD strategy may be a natural evolution of post-procedural care. Key to the success of this strategy, however, is judicious patient selection. Moreover, patient education regarding late complications that may occur after PCI is an integral part of an SDD program. A multidisciplinary team approach of all healthcare providers involved in patient pathways, as well as hospital administrators, is needed to build a successful SDD program.
Some investigators also reported patients' preference for and satisfaction with SDD strategy after PCI (13,45). It is clear that patients' education on the safety of outpatient PCI in selected patients is pivotal in ensuring the success of such a program. Because PCI is no longer considered a criterion for inpatient admission by default, the reimbursement scheme for elective PCI in the United States has shifted towards Ambulatory Payment Classification billing, instead of the Diagnosis-Related Group reimbursement system for inpatient procedures (38). Furthermore, in 2010, the U.S. Centers for Medicare & Medicaid Services extended its Recovery Audit Contractor program to all 50 states, with heightened scrutiny of hospital billing practices and inappropriate admissions related to procedures, which might have a noticeable impact on shifting current PCI practice toward shorter lengths of stay.
To the authors' knowledge, this systematic review and meta-analysis represents the most current and comprehensive analysis of available studies on SDD after PCI. Some of the observational studies included in this meta-analysis are of extremely small size, and the influence of the largest multicenter study to date (Rao et al. ) of >100,000 patients should be noted. An inherent limitation of any meta-analysis is that of publication bias. Although we included all randomized studies to date on this analysis, there was a large presence of single-center observational data (both retrospective and prospective). Other important issues such as patient satisfaction, quality of life, and costs were not addressed in those studies. They should be further studied in future comparative studies.
Considerable heterogeneity exists across published studies comparing SDD after PCI with ON, with significant variation in the populations studied, as well as the definition of outcomes. Despite a large sample size from pooling of randomized and observational studies, the point estimates for the OR between SDD and outcomes had wide CIs. With current heterogeneity between studies and wide CIs, an adequately powered multicenter randomized trial comparing SDD with ON would require a very large sample size. Until such a trial is completed, SDD after uncomplicated PCI seems a reasonable approach in selected patients.
The authors thank Dr. Lawrence Joseph, PhD, of McGill University, Montreal, Canada, for his statistical advice and assistance with the manuscript.
Dr. Abdelaal is supported by the Laval University Transradial Research and Education Fund. Dr. Bertrand is a research-scholar of the Quebec Foundation for Health Research. Dr. Rao has received honoraria as a consultant for Terumo Medical and The Medicines Company. Dr. Shroff has received honoraria from Cordis, Terumo Medical, and The Medicines Company. Dr. Caputo has received honoraria as a consultant for Boston Scientific, Medtronic, and Terumo. Dr. Pancholy has received speaker honoraria from The Medicines Company. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute coronary syndrome
- confidence interval
- glycoprotein IIb/IIIa inhibitor
- major adverse cardiovascular events
- myocardial infarction
- overnight hospitalization
- odds ratios
- percutaneous coronary intervention
- same-day discharge
- Received June 7, 2012.
- Revision received September 26, 2012.
- Accepted October 26, 2012.
- American College of Cardiology Foundation
- Singh M.,
- Rihal C.S.,
- Gersh B.J.,
- et al.
- Koch K.T.,
- Piek J.J.,
- Prins M.H.,
- et al.
- Bertrand O.F.,
- De Larochellière R.,
- Rodés-Cabau J.,
- et al.
- Heyde G.S.,
- Koch K.T.,
- de Winter R.J.,
- et al.
- Laarman G.J.,
- Kiemeneij F.,
- van der Wieken L.R.,
- Tijssen J.G.,
- Suwarganda J.S.,
- Slagboom T.
- Kiemeneij F.,
- Laarman G.J.,
- Slagboom T.,
- van der Wieken R.
- Banning A.P.,
- Ormerod O.J.,
- Channon K.,
- et al.
- Patel M.,
- Kim M.,
- Karajgikar R.,
- et al.
- Rinfret S.,
- Kennedy W.A.,
- Lachaine J.,
- et al.
- Popescu A.M.,
- Weintraub W.S.
- Liberati A.,
- Altman D.G.,
- Tetzlaff J.,
- et al.
- Agostoni P.,
- Biondi-Zoccai G.G.,
- de Benedictis M.L.,
- et al.