Same-day discharge after percutaneous coronary procedures—Structured review and comprehensive meta-analysis

We conducted a literature search on EMBASE, PubMed and CINAHL following the general recommendations from the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) [13] for randomized trials, and the Cochrane Collaboration and Meta-Analyses of Observational Studies in Epidemiology (MOOSE) group for observational studies [14]. For the search we used the MeSH search terms (“same-day discharge” OR “outpatient” OR “ambulatory” OR “daycare”) AND (“coronary artery” OR “PCI” OR “intervention” OR “angiography”) NOT (“psychiatric” OR “orthopedic” OR “vascular” OR “peripheral”). As detailed in Fig. 1, from the initial 1512 potentially relevant titles 1360 mismatches were removed by further filtering. From the remaining 152 references, 23 studies with exclusively femoral access or without exact numbers on access sites for both SDD and OS patients were excluded. Further 25 editorials/comments or letters and 16 studies with a cost-analysis or nursing focus were removed leaving 44 studies with either a prospective (Table 1) or retrospective design (Table 2), with specific data on radial access in SDD PCI and reporting outcome data for 30-day and/or 24‑h follow-up. For the meta-analysis, 9 prospective trials (three RCTs and six 2-cohort studies) reporting separate outcomes for SDD and OS were extracted.

This meta-analysis was performed following a standardized protocol by collecting information on study design, sample size, population demographics, coronary angiographic characteristics, access site for PCI, procedural adjuvant pharmacotherapy, procedural success rate, outcome (MACE), and follow-up data. Studies reporting only a single cohort, retrospective studies, and sources not reporting dedicated outcome for 30 days with absolute numbers were excluded from the analysis. Study level data were analyzed. The dichotomous MACE data for the SDD and OS groups were entered into a statistical software program (the Cochrane Collaboration’s Review Manager [RevMan], version 5.4.1, Nordic Cochrane Centre, Copenhagen, Denmark). Heterogeneity across studies was assessed using the Cochran’s Q statistic and the I² statistical test, with values of 25% or less considered low heterogeneity, 25–50% considered moderate, and values over 50% considered substantial. We calculated pooled ORs using a fixed effect model with the Mantel-Haenszel method in cases of low statistical heterogeneity and a random effect model in cases of moderate and substantial heterogeneity (see Fig. 2). The dichotomous outcomes were reported as odds ratios (OR) with their 95% confidence intervals (CI). The weight of each trial on the overall results was calculated as a percentage of the total number of patients included in each outcome analysis. Additionally, a sensitivity analysis was performed based on study design (randomized versus randomized and observational studies).

As shown in Fig. 1 the literature search provided 44 clinical trials on SDD PCI with a focus on radial access. Outcome data were analyzed either within a single SDD cohort or in comparison with OS patients. The 22 prospective trials and RCTs containing a total of 5804 patients treated as SDD are shown in Table 1, 22 retrospective trials with 206,517 SDD patients in Table 2. Of the SDD procedures in the 22 prospective trials listed 91% were performed via radial access, in the group of 22 retrospective trials the percentage was lower (39%), mostly caused by 2 large registries with low radial access rates [4, 15]. Calculating from the remaining 20 trials the percentage was 60%. The 44 trials listed are consistent in demonstrating a favorable safety profile of radial access SDD procedures, despite the fact that some studies were performed with routine administration of glycoprotein GPIIb/IIIa antagonists and comprised a wide range of interventional techniques including complex PCIs like multivessel (MV), multistent, bifurcation, CTO or left main coronary artery (LM) interventions. The largest group of interventions in the majority of studies were elective and ad hoc PCIs. In prospective trials, 30-day MACE ranged from 0 to 3.2%, whereas the majority of trials reported MACE rates of 0 or < 1% (see Table 1). In retrospective trials MACE ranged from 0 to 1.5% (see Table 2). In one trial reporting SDD outcomes after NSTE-ACS, 30-day readmission was substantially higher for both SDD (7%) and OS (11%) without statistical significance between the groups (p = 0.06) [16]. In those studies where SDD adverse events can be attributed to the time interval from the end of the clinical observation period (6 h postprocedure) to 24 h postprocedure, these are listed in a separate column. Remarkably, 17 out of 19 prospective trials and 12 out of 12 retrospective studies/registries either did not report any major adverse events from 6–24 h or none at all, consistently highlighting the solid safety profile of SDD protocols. Consequently, the general safety precaution of one overnight stay would not have provided additional postprocedural safety in this selected patient group [17]. Those 2 studies reporting very low MACE rates of 0.26% [18] and 0.8% [19] within 24 h post-PCI (both due to stent thrombosis) are among the oldest in this review (2004 and 2005; Table 1 #2 and #4) and may not represent current PCI tools and P2Y12 inhibitor treatment.

With the retrieved trial records we performed a meta-analysis representing current clinical practice focusing on radial access SDD. As shown in Fig. 1, we selected 9 clinical trials per standardized protocol from the initial 44 references. These trials (see Fig. 2) prospectively reported separate outcomes of 30-day MACE for both SDD and OS, and included data on radial access, the majority being performed exclusively with radial or ulnar access. Single-cohort studies and trials without specific data on access site were excluded. One study was based on the prospective Australian VCOR registry. The analysis itself, however, was done retrospectively [20], so the study was excluded. One of the RCTs was a substudy of the EXCEL trial focusing on left main interventions [21], therefore more high-risk procedures. In terms of frequency of MACE, there was no relevant difference to the other studies selected. Our meta-analysis contains a total of 23,017 patients, 2584 within RCTs and 20,433 within prospective 2‑cohort studies, and a total of 2405 SDD procedures. For the purpose of a brief analysis, only MACE data were extracted from the different sources.

The meta-analysis of all 9 prospective trials shows an odds ratio (OR) for MACE of 1.00 (95% CI 0.68–1.48; p = 0.21; I² 8% Fig. 2a). The incidence of MACE was low in both investigated groups: 3% (49 of 1817) vs. 3% (87 of 3099; Fig. 2a). In one study no MACE occurred within 30 days, only a number of minor adverse events were detected [22], therefore, odds ratios were not estimable.

When extracting data only from the three RCTs, the meta-analysis showed an OR of 0.66 (95% CI 0.35–01.24; p = 0.19; I² 0% Fig. 2b), indicating non-inferiority of SDD versus OS in carefully selected patients, which was consistent with our meta-analysis from all 9 prospective trials and the previously published meta-analysis data [9‐12]. Again, the incidence of MACE was low in both investigated groups and numerically higher in the OS group: 2% (15 of 979) vs. 4% (58 of 1605), respectively.

Consistent with the findings of Small et al. ([27]; see 5.1), data on the feasibility of SDD following complex interventions taken from a recent Spanish multicenter registry have been published in 2019 [28]. A complex intervention was defined as either left main PCI, bifurcation with a 2‑stent technique, multivessel PCI with ≥ 2 vessels, rotational atherectomy, antegrade CTO or graft PCI. The SDD was feasible in 64% of patients with complex lesions versus 81% in patients with simple angioplasty [28]. The most common reasons for switching from SDD to OS were clinical symptoms (65%; chest discomfort, arrhythmias or at decision of physician in charge), suboptimal angiographic result (20%), vascular access complications (4%), crossover to femoral access (1.3%), rejection of SDD by patient or relatives (0.5%) and excessive contrast use (1%) [28]. In 791 patients out of 1047 discharged as intended, only a single MACE event occurred in the simple PCI group (n = 592), while no adverse events at all occurred in the complex PCI group (n = 199) at 24 h or 1 month [28]. A total of 20 of the 22 prospective studies in Table 1 and 19 of the 22 retrospective studies in Table 2 either specifically mentioned complex PCI procedures or PCI subgroups with features indicating more complex PCI (MV, bifurcation, LM, CTO, SVG), therefore showing an adequate representation of complex PCI procedures within the SDD cohorts studied.

Rao et al. analyzed data from 107,018 patients aged > 65 years from the CathPCI Registry in 2011. Among 1339 older patients discharged on the day of procedure, the rates of death or rehospitalization at 2 or 30 days postprocedure were not different from 105,679 patients admitted overnight [29]. Remarkably, consistent with the clinical standard at the time of publication, 96% of procedures in the SDD group and 98% in the OS group were performed with femoral access. Vascular closure devices were used in 65% and 50%, respectively [29].

An earlier retrospective cohort study examined the safety of SDD PCI performed via radial access between 1998 and 2001 in patients below and above the age of 75 years [30], including single vessel, multivessel and bypass graft interventions featuring lesions from type A to type C. In a self-reported outcome questionnaire detailing adverse events either from 0–24h or 1–30 days from discharge, there were no significant differences in major adverse events in 797 patients < 75 years and 146 patients > 75 years of age [30].

Older patients usually have more underlying chronic conditions, more advanced coronary and peripheral atherosclerosis, rendering radial/ulnar access more challenging. The recent expert consensus statements from the American College of Cardiology (ACC) [17] and the Society for Cardiovascular Angiography and Inteventions (SCAI) [31, 32] do not mention a strict age limit for SDD patients. In our experience, older patients aged > 75 years should be carefully screened considering their suitability for SDD treatment. Frequently not only adverse anatomical/vascular challenges are compromising SDD PCI, but also cognitive factors like the ability to comply with medication prescriptions and to organize reliable transport to and from the clinic on the day of the procedure.

The feasibility of SDD rotational atherectomy (RA) has been considered critical by some authors [33]; however, in high-volume centers with experience in RA these interventions are increasingly being included in the array of SDD procedures. Of the prospective studies four ([28, 34‐36]; Table 1) and two of the retrospective studies ([15, 37]; Table 2) in our review included small numbers of RA PCI procedures. This is consistent with a retrospective analysis of 4591 RA procedures from 2007 to 2014 in England and Wales published by Taxiarchi et al. [38]. The authors show an increase of SDD following rotablation from 6.7% in 2007 to 35.5% in 2014, representing 2.8% of the total uncomplicated elective PCIs in the registry. In terms of 30-day mortality, there was no superiority of OS compared to SDD [38]. Patients with MV PCI (14.0% SDD vs. 18.8% OS; p > 0.001) and LM PCI (8.0% SDD vs. 13.1% OS; p < 0.001) were found less frequently in the SDD cohort. On the other hand, SDD patients had PCI via radial access more frequently (48.3% SDD vs. 30.1% OS; p < 0.001), highlighting some of the selection criteria used by the majority of centers. The outcome with a 30-day MACE of 0.5% for SDD and 0.35% for OS (p = 0.409) showed an excellent safety profile with respect to this selection practice [38]. In the registry of SDD in complex lesions by Cordoba-Soriano et al., only 21 out of 1047 patients were treated with rotational atherectomy, and only 9 out of these (43%) were discharged on the day of the procedure [28]. The compatibility of highly aggressive PCI techniques with SDD programs should be critically evaluated by each interventional center depending on interventional experience and scope of procedures.

The PCI of chronic total occlusions (CTO) generally involves a more aggressive setup with larger sheath diameter, high-tipload wires, microcatheters and dual access at least for the retrograde approach. Most CTO procedures also require multiple stents to cover longer segments of vessel occlusion. Therefore, the eligibility of CTO procedures for SDD has been viewed critically [33]. On the other hand, 10 of the 44 trials reviewed in Tables 1 and 2 included smaller numbers of CTO procedures, two studies [39, 40] (#17 and #20 in Table 2) specifically targeted CTO procedures on a SDD basis. A retrospective study from 2021 compared the 30-day outcome of 51 CTO interventions performed as SDD procedures with 122 performed with conventional OS. There were no demographic differences between both groups but non-SDD patients were more likely to have diabetes mellitus (non-SDD 51% vs. SDD 31%; p = 0.015) and arterial hypertension (non-SDD 89% vs. SDD 67%, p < 0.001), while SDD patients had a higher BMI and were more frequently smokers [40]. In SDD procedures, radial access was used as the single vascular access, while 17% of non-SDD patients had at least 1 femoral access. Outcome analysis showed an in-hospital MACE of 0% for SDD vs. 1.6% for non-SDD patients and a 30-day MACE of 0% and 1.6%, respectively. Antegrade wire escalation was the dominant crossing strategy for lesions in the SDD group; however, a multivariate logistic regression model showed that only diabetes mellitus and procedure time were independently associated with the decision to maintain a SDD strategy [40].

Similar results were found by Taxiarchi et al. in a retrospective longitudinal study from the UK, covering 7576 SDD cases versus 13,763 OS cases from 2007 to 2014 [39]. The percentage of SDD management among CTO procedures increased from 21.7% in 2007 to 44.7% in 2014. Patients in the OS cohort were more likely to have relevant medical histories, such as prior MI (OS 43.6% vs. SDD 39.2%; p < 0.001), prior coronary artery bypass graft (CABG) surgery (OS 16.3% vs. SDD 12.8%; p < 0.001), prior PCI (OS 35.6% vs SDD 31.9%; p < 0.001) and significant comorbidities as multivessel disease (OS 32.0% vs. SDD 29.7; p < 0.01) and arterial hypertension (OS 61.6% vs. SDD 59.4%; p < 0.01). Unadjusted 30-day mortality was lower for SDD patients (0.12% vs. 0.31% for OS; p < 0.01). After adjustment for clinical severity of cases, SDD was no longer independently associated with 30-day mortality (OR 0.54; 95% CI 0.25–1.15) [39]. Interestingly, while observed mortality rates for OS cases were well within expected mortality rates calculated from the British Cardiovascular Intervention Society (BCIS) risk score model [41], they were lower than expected for SDD patients throughout 7 of the 8 years of observation, although this trend did not reach the level of statistical significance [39]. The presence of ≥ 1 enabling strategies (dual access site, rotational atherectomy, intravascular ultrasound, and use of penetration catheters or microcatheters), was independently associated with overnight stay; however, the authors showed that in high-volume centers surgeons were more likely to treat more challenging cases than SDD [39].

Therefore, while the decision to include CTO interventions in the array of SDD procedures is consistent with the 2018 SCAI recommendations [31], it should certainly be critically appraised depending on the individual center’s expertise and caseload of CTO interventions.

Left main coronary artery (LM) PCI was listed as an exclusion criterion for SDD by the 2009 SCAI position paper [33] and by the more recent 2020 SCAI position statement for PCI in ambulatory surgical centers [32] (as opposed to hospitals featuring in-house cardiac surgery departments). The 2021 ACC expert consensus document does not explicitly mention LM PCI as an exclusion criterion for SDD PCI [17]. Currently in Austria LM PCI is not routinely performed with SDD patient management; however, substantial experience does exist with SDD even in this high-risk setting. Out of 44 trials listed in Tables 1 and 2 14 included LM PCI procedures, 3 trials [21, 24, 42] specifically targeted LM PCIs in a SDD setting. A Canadian study compared the outcome of elective LM PCI in 267 patients treated as SDD to 194 patients with OS [42]. Patients in the SDD group were younger (70.9 ± 10.1 years versus 73.4 ± 10.8 years), more frequently had prior cardiac catheterization and showed a larger percentage of protected LM stenoses (51.7% versus 35.1%). The composite primary endpoint of 30-day mortality, myocardial infarction and rehospitalization was significantly lower in the SDD group (OR 4.3; 95% CI 1.1–6.0, p = 0.002) [42]. This remarkable outcome depends, at least in part, on the inherent selection bias of SDD suitable patients following current recommendations. Conversely, however, it does underline the robustness of the SDD concept with correct patient selection even in complex interventions. One of the most remarkable findings regarding the standard OS as a safety precaution was, once again, the timing of complications: All MACEs occurred beyond 48 h postprocedure, therefore OS would have provided no additional safety benefit [42].

A large retrospective multicenter analysis evaluated the outcome of 6452 LM PCI interventions in England and Wales from 2007 to 2014 [24]. The authors found that SDD treatment after LM PCI had almost doubled during the study period (all LM PCI 19.9–39.8%; unprotected LM 20.7–41.4%) in parallel with an increase in procedural complexity including rotational atherectomy and multistent strategies. Nevertheless, SDD was not associated with an increase in 30-day mortality in general LM PCI procedures (OR 0.70, 95% CI 0.30–1.65) and in unprotected LM PCI (OR 0.48, 95% CI 0.17–1.41) [24]. Consistently, a subgroup of 100 LM interventions within the prospectively randomized EXCEL trial were performed on a SDD basis ([21]; see Table 1 #21). Compared to 835 OS LM procedures, there were no significant differences in MACE at 30 days (4.0% SDD vs. 5.0% OS, adjusted OR 0.52, 95% CI 0.12–2.22; p = 0.38) or 5 years (20.6% SDD vs. 22.1% OS, adjusted OR 0.72, 95% CI 0.40–1.29; p = 0.27).

In summary, LM PCI in carefully selected clinically stable patients without excessive calcifications appears to be feasible and safe in a SDD setting; however, it must be taken into account that the multicenter study cited above reported substantial heterogeneity of LM PCI frequencies among different clinics [24] and that the favorable outcome is largely driven by high-volume centers with great expertise in complex and LM PCI. The individual decision to offer SDD for a LM PCI patient should be made individually involving the interventional team’s level of expertise, the operator’s preference and the clinical course during and after the intervention.