Application and Comparison of Different Prognostic Scoring Systems in Patients Who Underwent Cardiologist-Managed Percutaneous Cardiopulmonary Support

Abstract

Background

Temporary mechanical support, including percutaneous cardiopulmonary support (PCPS), is crucial for reversing patients’ compromised hemodynamic function. Knowledge about whether cardiologists can directly manage patients receiving PCPS and about the predictive values of different prognostic scores is insufficient.

Methods

We examined the data and in-hospital mortality of 45 eligible patients receiving cardiologist-managed PCPS from July 2012 to January 2019 in our institute. We compared different prognostic scores [namely Survival After Veno-arterial ECMO (SAVE), modified SAVE, prEdictioN of Cardiogenic shock OUtcome foR acute myocardial infarction patients salvaGed by VA-ECMO (ENCOURAGE), and Sequential Organ Failure Assessment (SOFA) scores] through area under the receiver operating characteristic curve (AUC) analysis.

Results

The patients’ mean age was 64.3 ± 11.3 years, and 71.1% were men. The overall in-hospital survival rate was 35.6%. Compared to survivors, nonsurvivors were more likely to have an ischemic etiology, cardiopulmonary resuscitation, and higher lactate levels. Survivors had higher SAVE (-5.9 vs. -11.4) and modified SAVE (4.2 vs. -7.1) scores than nonsurvivors (both p = 0.001), but SOFA (9.7 vs. 10.3) and ENCOURAGE (24.8 vs. 26.8) scores were similar (both p > 0.1). In multivariate models, only modified SAVE score remained statistically significant (hazard ratio: 0.96, 95% confidence interval: 0.93-1.00; p = 0.047). Modified SAVE score showed the best risk discrimination (AUC = 0.78).

Conclusions

Establishing regular and continual training protocols can enable cardiologists to perform emergency PCPS (without on-site surgery) and daily care for patients with refractory cardiogenic shock. The modified SAVE score facilitates risk stratification and future decision-making processes.

INTRODUCTION

Cardiogenic shock (CS) is a state of systemic hypoperfusion caused by a primary cardiac disorder. CS management remains a clinical challenge owing to its diverse etiologies, rapid progression, and high in-hospital mortality rates.¹ Incorporating medical, surgical, and mechanical circulatory support (MCS) is key to rescuing a failing heart and reversing CS.² Several MCS devices have been introduced in clinical practice, such as intraaortic balloon pump systems, extracorporeal membrane oxygenation (ECMO), and ventricular assistance devices.¹ Peripheral veno-arterial ECMO, also called percutaneous cardiopulmonary support (PCPS), has the advantages of miniaturization, percutaneous cannulation, and rapid priming (autopriming) designs, and it can enable cardiologists to initiate extracorporeal life support in eligible patients without additional assistance from surgeons or perfusionists, which is required in traditional ECMO systems.^3-6

The volume of patients who are at risk when undergoing a high-risk percutaneous coronary intervention (PCI) or new percutaneous cardiovascular intervention is increasing.^6-12 Although current evidence and guidelines support performing PCI without on-site cardiac surgery, they recommend that interventional centers should have access to MCS systems, such as PCPS, for emergency management.^8,13

The temporal trends of PCI and ECMO volumes have continually increased in Taiwan.^9,14 Accordingly, developing a reliable prognostic scoring system for facilitating patient selection and risk stratification is imperative. Several scoring systems have been proposed for predicting survival in patients treated with veno-arterial ECMO, such as the Survival After Veno-arterial ECMO (SAVE) score, prEdictioN of Cardiogenic shock OUtcome foR acute myocardial infarction (AMI) patients salvaGed by VA-ECMO (ENCOURAGE) score, and modified SAVE score.^15-17 However, these scoring systems have been established for traditional ECMO systems that involve teams composed of cardiovascular surgeons and perfusionists, and whether these scoring systems can be applied to PCPS systems that involve teams composed of only cardiologists and nurses is unclear.

This study investigated the in-hospital mortality of patients who underwent cardiologist-managed PCPS and explored the discrimination ability of the aforementioned prognostic scoring systems in the study population.

METHODS

PCPS versus ECMO teams

Our institute — a tertiary medical center in Taipei, Taiwan — has established a well-organized and functional ECMO team, with cardiovascular surgeons and perfusionists constituting the team’s core members. We subsequently realized the importance of the initial timing of ECMO use in patients with refractory CS and the increasing volume of patients undergoing high-risk PCI or other percutaneous interventions in our cardiac catheterization laboratory. Accordingly, since July 2012, we have used an alternative MCS system — PCPS system — to shorten delays in MCS initiation in eligible patients with refractory CS. The operating protocol of this system is briefly described as follows. The PCPS system involves a team comprised of interventional cardiologists who are trained to perform percutaneous femoral cannulation (with a 16.5-Fr. arterial cannula and 21-Fr. venous cannula) and initiate PCPS (Terumo Inc, Tokyo, Japan) in the cardiac catheterization laboratory for a clinically irreversible patient. The entire cannulation process is performed under fluoroscopic guidance, with the tip of the arterial cannula being placed at the iliac artery and the tip of the venous cannula being placed at the junction of the inferior vena cava and the right atrium. Trained, on-duty nurses stationed in the nearby intensive cardiac care unit (CCU) perform PCPS through the autopriming method. Three trained cardiologists, designated as PCPS specialists, receive an emergency call simultaneously, and one of them responds to provide support in post-PCPS care and troubleshooting. When a patient’s hemodynamic and cardiac function improve, a surgeon weans the patient from PCPS through the surgical repair of vessels. All involved cardiologists and nurses must undergo regular PCPS training courses and attend case-based discussions for each patient treated with PCPS.

Study population

In this study, we enrolled 45 consecutive patients aged > 20 years from an observational PCPS cohort from July 2012 to January 2019. All enrolled patients underwent PCPS administered by cardiologists in the cardiac catheterization laboratory and subsequently received daily care from cardiologists; cardiovascular surgeons were involved only for vascular repair during PCPS weaning. Patients were excluded if (1) PCPS was implemented by a cardiovascular surgeon with or without a perfusionist and not in the cardiac catheterization laboratory (n = 4), (2) surgeons or perfusionists participated in subsequent post-PCPS care (n = 2), or (3) PCPS was used for pulmonary support in a veno-venous configuration (n = 1). PCPS was initiated either under active cardiopulmonary resuscitation (CPR) or indication of profound shock, defined as systolic blood pressure less than 75 mmHg — despite receiving an intravenous inotropic agent — that was associated with altered mental status and respiratory failure. The choice between the PCPS and ECMO was made by the cardiologist in charge along with a PCPS specialist or cardiovascular surgeon who was included in the process of evaluating screening criteria.

Data collection and prognostic scoring systems

Clinical details were retrospectively extracted from prospective records of the PCPS database. Baseline characteristics included demographic characteristics, medical history, laboratory data, and reasons for CS. The reason for ischemic shock was identified for those who developed refractory CS due to disturbances in coronary macro-flow and myocardial ischemia, including the following conditions: (1) acute myocardial infarction, (2) complications related to myocardial ischemia/infarction, and (3) complications related to coronary angiographic procedures with a rise in cardiac biomarkers (creatine kinase-MB or troponin) above the upper limit of normal. The others were categorized as being nonischemic reasons, including acute myocarditis, pulmonary embolism, dilated cardiomyopathy, and valvular heart disease. We carefully recorded the details of patients who had a circulatory collapse that required CPR. Patients who required chest compression in the hospital were defined as having an in-hospital cardiac arrest (IHCA), whereas patients who required chest compression before hospital arrival were defined as having out-of-hospital cardiac arrest (OHCA). PCPS under CPR was defined as active CPR during the initiation of PCPS. Intervals from the earliest recorded time of cardiac arrest and activating the PCPS system to initiating extracorporeal circulation were recorded as arrest to PCPS time and activation to PCPS time, respectively. The procedural time was estimated from the arrival time at the catheterization laboratory if the patient underwent PCPS as the first procedure, or from the request of PCPS preparation if the patient was already on the table in the catheterization laboratory to the complete establishment of extracorporeal circulation. The outcome recorded in this study was in-hospital mortality. This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Mackay Memorial Hospital (19MMHIS198e).

We explored the discrimination ability of four scoring systems: SAVE, ENCOURAGE, modified SAVE, and sequential organ failure assessment (SOFA). We selected these scoring systems because they have been implemented and applied to study patients with CS who underwent veno-arterial ECMO, similar to our study population.^15-18 Records of prothrombin activity, an ENCOURAGE variable, were not available in our hospital database. Therefore, we derived prothrombin activity by calculating the international normalized ratio (INR) according to a previously proposed equation: prothrombin activity (%) = 1/(0.028 INR - 0.018).¹⁹

The primary endpoint was defined as in-hospital mortality, and this endpoint was analyzed on a time-to-event basis. If death did not occur, the date of censoring was considered the date of final follow-up.

Statistics

Baseline characteristics and prognostic scores were stratified into two groups according to in-hospital survival status, and the two groups were compared. Continuous variables are expressed as mean ± standard deviation, and categorical variables are presented as percentages. For group comparisons, we used the Student’s t test for continuous variables and the chi-square test for categorical variables. Survival probability curves were plotted using the Kaplan-Meier method and compared using the log rank test. Cox proportional hazards regression analysis was performed to evaluate the association of variables with in-hospital mortality. Independent baseline variables with a p value of < 0.05 in univariate analysis were included in multivariate analysis. The modified SAVE score was calculated as follows: SAVE score + 15 (if lactate level < 75 mg/dL) or SAVE score + 0 (if lactate level ≥ 75 mg/dL). The rationale behind this calculation is that in a previous study, the effects of a lactate level of < 75 mg/dL appeared to outweigh the effects of the SAVE score in terms of survival prediction, with the corresponding odds ratios being 8.74 and 1.17, respectively.¹⁷ Accordingly, because of the inevitably strong interaction between lactate levels and the modified SAVE score, we did not include data regarding lactate levels in multivariate adjustments for the modified SAVE score. The accuracy of the various prognostic scoring systems in predicting in-hospital mortality was assessed using area under the receiver operating characteristic (ROC) curve (AUC) analysis. We used SPSS (version 22.0; SPSS, Chicago, IL, USA) and MedCalc (version 19.0.5; MedCalc Software, Mariakerke, Belgium) for all statistical analyses. We considered a p value of < 0.05 to be statistically significant.

RESULTS

Baseline characteristics and survival

The 45 patients enrolled in this study had a mean (range) age of 64.3 (53-76) years. Among these patients, 71.1% were men, all of whom underwent cardiologist-managed PCPS with a mean support time of 95.5 (± 69.9) hours. In addition, 16 (35.6%) patients survived to hospital discharge (the survivors group), and 29 (64.4%) patients died (the nonsurvivors group). The nonsurvivors were more likely to have had an ischemic etiology for CS, received CPR before PCPS implementation, and had higher lactate levels (all p < 0.05) compared with the survivors. No significant differences were observed between the two groups in terms of demographics, history of comorbidities, refractory cardiac arrest resuscitated by salvage PCPS, total CPR time, activation time, procedural time, hematocrit levels, or estimated glomerular filtration rates. All of the patients with ischemic reasons received coronary revascularization, including PCI (n = 32/35, 91.4%) and coronary bypass surgery (n = 3/35, 8.6%). PCPS in those with non-ischemic reason was most used as a bridge to cardiac recovery except in two cases, one of whom underwent aortic valve replacement surgery for etiology of severe aortic valve stenosis, and the other received thrombolytic therapy for acute pulmonary embolism. In univariate analysis, the hazard ratios [HRs; 95% confidence intervals (CIs)] for pre-PCPS cardiac arrest, ischemic etiology, and lactate levels were 4.05 (1.21-13.6; p = 0.02), 3.72 (0.88-15.7; p = 0.07), and 1.01 (1.01-1.02; p = 0.001), respectively. Notably, of all patients, 75.6% experienced at least one cardiac arrest (IHCA or OHCA) and 60% were resuscitated by salvage PCPS (ECPR), indicating the clinical severity of our study population (Table 1).

Table 1. Baseline characteristics of patients who underwent PCPS according to in-hospital survival status.

	All patients	Survivors	Non-survivors	p value	Univariate
	n = 45	n = 16	n = 29		Hazard ratio (95% CI)	p value
Demographics
Age, year	64.3 ± 11.3	63.6 ± 12.7	64.7 ± 10.6	0.75
Sex, male (%)	32 (71.1)	12 (75.0)	20 (69.0)	0.74
Body mass index, kg/m2	25.5 ± 3.9	26.4 ± 3.9	25.1 ± 3.9	0.29
Pre-PCPS SBP, mmHg	65.2 ± 22.0	71.7 ± 10.6	58.8 ± 28.5	0.16
Medical history, n (%)
Previous MI	8 (17.8)	2 (12.5)	6 (20.7)	0.69
Previous stroke	2 (4.4)	1 (6.3)	1 (3.4)	1.00
Pre-PCPS cardiac arrest, n (%)	34 (75.6)	8 (50.0)	26 (89.7)	0.009	4.05 (1.21-13.6)	0.02
IHCA	29 (64.4)	7 (43.8)	22 (75.9)	0.05
OHCA	5 (11.1)	1 (6.3)	4 (13.8)	0.64
PCPS under CPR, n (%)	27 (60.0)	8 (50.0)	19 (69.5)	0.35
CPR time, (min)	34.1 ± 23.0	32.6 ± 22.0	34.6 ± 23.7	0.84
Activation to PCPS time, (min)	38.0 ± 14.7	37.5 ± 13.6	38.3 ± 15.6	0.87
Arrest to PCPS time, (min)	36.6 ± 24.0	24.1 ± 14.7	40.7 ± 25.3	0.12
Procedural time, (min)	17.9 ± 4.8	16.1 ± 3.7	18.9 ± 5.2	0.07
Duration of PCPS support, (hr)	95.5 ± 69.9	115.8 ± 48.9	84.3 ± 77.6	0.15
Shock reasons, n (%)
Ischemia	35 (77.8)	8 (50.0)	27 (93.1)	0.002	3.72 (0.88-15.70)	0.07
Laboratory
Hematocrit, (%)	36.8 ± 9.3	38.9 ± 6.7	35.9 ± 10.5	0.40
eGFR, ml/min/1.73 m2	41.4 ± 24.6	53.9 ± 27.4	36.3 ± 21.9	0.08
Lactate, mg/dL	95.8 ± 57.5	70.8 ± 61.4	110.3 ± 50.8	0.04	1.01 (1.01-1.02)	0.001
Prognostic scores
SOFA	10.1 ± 4.1	9.7 ± 4.8	10.3 ± 3.9	0.65
ENCOURAGE	26.1 ± 7.3	24.8 ± 8.2	26.8 ± 6.8	0.40
SAVE	-9.5 ± 5.9	-5.9 ± 4.6	-11.4 ± 5.5	0.001	0.92 (0.86-0.98)	0.01
Modified SAVE	-3.1 ± 10.7	4.2 ± 9.8	-7.1 ± 9.1	0.001	0.95 (0.92-0.99)	0.005

PCPS, percutaneous cardiopulmonary support; 95% CI, 95% confidence interval; SBP, systolic blood pressure; MI, myocardial infarction; IHCA, in-hospital cardiac arrest; OHCA, out-of-hospital cardiac arrest; CPR, cardiopulmonary resuscitation; eGFR, estimated glomerular filtration rate; SOFA score, Sequential Organ Failure Assessment score; ENCOURAGE score, prEdictioN of Cardiogenic shock Outcome foR AMI patients salvaGed by VA-ECMO score; SAVE score, Survival After Veno-arterial ECMO score.

Prognostic scores

Compared with the nonsurvivors, the survivors had higher SAVE (-5.9 vs. -11.4, p = 0.001) and modified SAVE (4.2 vs. -7.1, p = 0.001) scores. However, the two groups did not exhibit statistical differences in SOFA (9.7 vs. 10.3, p = 0.65) or ENCOURAGE (24.8 vs. 26.8, p = 0.4) scores. As presented in Table 1, the HR (95% CI) for the SAVE score was 0.92 (0.86-0.98; p = 0.01), and that for the modified SAVE score was 0.95 (0.92-0.99; p = 0.005).

In multivariate analysis, the predictive power of the SAVE score for mortality was lost (HR: 0.95, 95% CI: 0.88-1.02; p = 0.15), as presented in Table 2 (model 1). By contrast, a relatively high modified SAVE score was independently associated with hospital survival (HR: 0.96, 95% CI: 0.93-1.00; p = 0.047), as shown in Table 2 (model 2). We conducted assessments to determine the optimal cutoff value of the modified SAVE score for creating a dichotomous variable, and a mean value of the score (-3) appeared to be the optimal cutoff. Accordingly, we applied this cutoff (-3) to stratify the entire cohort into binary groups: one comprised of patients with a modified SAVE score of > -3, and the other with a modified SAVE score of ≤ -3. As indicated in Table 2 (model 3), a modified SAVE score of > -3 was more significantly associated with hospital survival compared to a score of ≤ -3 (HR: 0.37, 95% CI: 0.15-0.94; p = 0.04). We also plotted a Kaplan-Meier survival curve (Figure 1), which showed that a modified SAVE score of ≤ -3 was associated with a lower survival probability than a modified SAVE score of > -3 throughout the observation period (hospital survival rate: 9% vs. 29%; log rank p = 0.005).

Table 2. Predictive values of SAVE score and modified SAVE score in multivariate analyses.

	Hazard ratio (95 % confidence interval)	p value
Model 1
SAVE score	0.95 (0.88-1.02)	0.15
Pre-PCPS cardiac arrest (yes vs. no)	1.72 (0.66-4.45)	0.27
Ischemic reason (yes vs. no)	2.33 (0.52-10.50)	0.27
Model 2
Modified SAVE score	0.96 (0.93-1.00)	0.047
Pre-PCPS cardiac arrest (yes vs. no)	2.63 (0.62-7.98)	0.22
Ischemic reason (yes vs. no)	1.60 (0.35-7.62)	0.55
Model 3
Modified SAVE score (> -3 vs. ≤ -3)	0.37 (0.15-0.94)	0.04
Pre-PCPS cardiac arrest (yes vs. no)	2.47 (0.71-8.67)	0.16
Ischemic reason (yes vs. no)	1.50 (0.33-6.84)	0.60

PCPS, percutaneous cardiopulmonary support; 95% CI, 95% confidence interval; SAVE score, Survival After Veno-arterial ECMO score.

Figure 1.

Hospital survival percentage in enrolled cohort according to modified SAVE score. Survival probability for patients with a modified SAVE score of -3 or less was lower than that of those with a modified SAVE score of more than -3 throughout the study period (hospital survival rate: 9% vs. 29%; log rank p = 0.005). SAVE score, Survival After Veno-arterial ECMO score.

We plotted ROC curves (Figure 2) to evaluate the accuracy of the various scoring systems in predicting in-hospital mortality. According to the results, the SOFA and ENCOURAGE scores exhibited limited predictive accuracy (AUC observed for SOFA score = 0.54; AUC observed for ENCOURAGE score = 0.62). The SAVE score exhibited a relatively high-risk discrimination ability (AUC = 0.74), but it remained inferior to that of lactate level (AUC = 0.76). The modified SAVE score exhibited the best risk discrimination ability (AUC = 0.78).

Figure 2.

ROC curves for in-hospital mortality in patients who underwent emergency PCPS according to different variables. Various AUC values were observed during the prediction of in-hospital mortality: SOFA = 0.54; lactate = 0.76; SAVE score = 0.74; modified SAVE score = 0.78; and ENCOURAGE score = 0.62. AUC, area under the ROC curve; ROC, receiver operating characteristic curve; SOFA score, Sequential Organ Failure Assessment score; ENCOURAGE score, prEdictioN of Cardiogenic shock Outcome foR AMI patients salvaGed by VA-ECMO score; SAVE score, Survival After Veno-arterial ECMO score.

DISCUSSION

This study demonstrated the feasibility of establishing a PCPS team without on-site surgeons or perfusionists to respond to and treat hemodynamically irreversible patients in a cardiac catheterization laboratory. The overall rate of survival to hospital discharge in this study was 35.6%, which is similar to rates reported in previous studies and registries.^14,17,20,21 Compared with existing scoring systems for veno-arterial ECMO, the modified SAVE score provided the best risk discrimination (AUC = 0.78). A modified SAVE score of ≤ -3 was independently associated with an increased risk of in-hospital mortality.

Mechanical support in CS

The rate of mortality from CS has been reported to be high, with the rate approaching 60%-80% in patients without advanced therapies.^22,23 A comprehensive heart-lung support system such as ECMO is sometimes pivotal to stabilizing patients’ hemodynamic status and affords sufficient time for recovery or to receive subsequent therapeutic modalities regardless of the etiology.^7,24-27 Incorporating ECMO into conventional CPR (ECPR) can be also effective in saving patients with cardiac arrest (either IHCA or OHCA).^28,29

Intraaortic balloon pumping only provides partial hemodynamic support, and might have limited clinical benefits.^30,31 Peripheral ventricular assist devices have also been developed for use in emergency situations, but they are still unavailable in Taiwan.³² Moreover, they have other limitations, including the requirement of additional procedures and devices for biventricular support, lack of extra-oxygenation in some devices, and high cost. Therefore, peripheral ventricular assist devices should be only applied to selected patients in emergency situations.³³

Role of PCPS in the interventional era

Implementing traditional ECMO systems is a time-consuming process and requires backup by surgeons or perfusionists. PCPS has attracted attention because of its simple and timely implementation, particularly in the current interventional era.^3,4,26,29 To the best of our knowledge, the present study is the first to introduce the implementation of a PCPS team composed of cardiologists and CCU nurses and to report the corresponding in-hospital outcomes. Specifically, in our hospital, the team responsible for percutaneous interventions or daily care can be transformed into a PCPS team without on-site surgeons upon encountering a patient with refractory CS. This conceptis in line with current recommendations, which state that interventional centers can effectively perform emergency MCS procedures such as percutaneous interventions without on-site surgery and should implement effective quality control policies. However, consolidated training programs and relevant case-based discussion should be particularly emphasized for all teammates, because most will be inexperienced owing to the relatively low volume of PCPS. Insufficient training or unfamiliarity with procedural processes will prolong the implementation time and increase the risk of operating errors. Our study also reflected the real-world condition that there was still a time lag between activation and procedural time because some cardiologists still needed additional waiting time for the assistance of PCPS specialists to implement PCPS. In this study, the rate of survival to hospital discharge was 35.6%, which is similar to the survival rate (41.4%) of patients in the global registry of the Extracorporeal Life Support Organization (ELSO).²¹ The slightly lower survival rate in our cohort might be because the patients in our cohort exhibited a higher prevalence of cardiac arrest before or during PCPS and had a major etiology of myocardial ischemia/infarction (77.8%) compared with the patients in the ELSO registry. Previous studies on ECMO have demonstrated that cardiac arrest and an etiology of myocardial ischemia/infarction are poor prognostic factors.^15,28,29,34

Prognostic scores for cardiologist-managed PCPS

Precise risk prediction for patients undergoing PCPS can help physicians to stratify the risk of candidate patients, help them evaluate the quality of care retrospectively, and in particular facilitate their future decision-making processes.³⁵ The findings of this study are consistent with those of Chen et al., who reported that the modified SAVE score had the best risk discrimination ability (AUC = 0.78), and that stratifying patients by using a cutoff value of -3 was independently associated with in-hospital mortality.¹⁷ The possible reasons for these observations are provided as follows. First, the population enrolled in the study by Chen et al. and that enrolled in our study comprised Taiwanese patients who had acute presentations, high prevalence of pre-ECMO cardiac arrest (74.7% vs. 75.6%), and high lactate levels (97.4 vs. 95.8 mg/dL). Second, the patterns and quality of clinical care in Taiwan are similar. Czobor et al. initiated PCPS in a cardiac catheterization laboratory and proposed the initial SOFA score to be a major prognostic factor (odds ratio = 0.577); nevertheless, their results might not be generalizable to our patients because they enrolled a limited number of patients (n = 25) with a relatively low degree of clinical severity.²⁰ The SAVE and ENCOURAGE scores have been validated in studies involving relatively large samples of patients treated with veno-arterial ECMO.^15,16 However, the application of these two scores to our population was primarily limited by differences in baseline characteristics between the patients. Patients enrolled in the previous study conducted on the SAVE score had less severe conditions and fewer ischemic etiologies,¹⁵ and all of the patients enrolled in the previous study using the ENCOURAGE score had acute myocardial infarction.¹⁶

Study limitations

This study has several limitations. First, this was a retrospective single-center study and some missing data (such as neurologic function, initial rhythms of arrest) and potential bias were inevitable. The generalizability of our findings to other hospitals could be restricted. Because of this study’s single-center design and small cohort, some unknown confounding factors may have affected the study results, despite the statistical adjustments. Second, all patients were cannulated under fluoroscopic guidance in our cardiac catheterization laboratory to prevent complications or unsuccessful cannulation. We could not determine whether this could have resulted in potential patient selection bias. Third, the small cohort may have introduced bias and limited the predictive value of any baseline factors, regardless of factor adjustments. Therefore, we could not provide a new scoring system. Fourth, eliminating the disease heterogeneity in our study population was difficult. Finally, a surgeon participated in our PCPS system in the final step of surgical repair of femoral vessels. The effectiveness and safety of the applied percutaneous closure device was not tested.

CONCLUSIONS

In conclusion, patients with refractory CS were effectively treated by our PCPS team composed of interventional and critical care personnel. However, all PCPS team members should receive regular and continual training to prevent operating errors. The modified SAVE score is useful for risk stratification in patients who have received PCPS and can help in making future therapeutic decisions.

CONFLICTS OF INTEREST

All authors declare no conflicts of interest.