Procedural volume and outcomes in patients undergoing VA-ECMO support

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is increasingly used for treatment of patients with critical cardiopulmonary failure and as a rescue therapy to stabilize critically ill patients with circulatory compromise [1, 2]. In the last decade, VA-ECMO therapy has seen a rapid increase in the western world, although in-hospital mortality rates remained at a high level [3, 4]. We have recently reported that utilization of VA-ECMO for cardiopulmonary support increased by more than 30-fold in Germany between 2007 and 2015, while in-hospital mortality remained unchanged around 60% [5].

As a result of the rapid increase in VA-ECMO implantations, the number of implanting centers has also increased. To what extent an increase of the VA-ECMO implanting hospitals and their procedure volume of VA-ECMO implantations has a role in terms of outcome remains unclear. Several studies have shown that higher hospital procedure volume is associated with improved outcomes in interventional procedures such as transcatheter aortic-valve replacement [6], percutaneous mitral valve repair [7], percutaneous coronary intervention [8], and coronary artery bypass grafting [9]. It is speculated that this observation is the expression of a learning curve for interventional procedures. Whether this might be translated to VA-ECMO implantation is currently unknown.

Importantly, not only mortality, but also incident complication would be an outcome of interest, as 50% of all VA-ECMO patients experience therapy-limiting complications such as major bleeding, stroke, limb and abdominal ischemia, thrombosis, and infection [10‐12]. Nevertheless, whether treatment of these complications in higher volume centers affects outcome is unknown.

The aim of this study was to investigate the association between hospital VA-ECMO procedure volume and outcomes in a large, nationwide registry. Findings regarding volume-mortality association could be of importance to inform the complex decision-making processes when considering initiation of VA-ECMO therapy for cardiogenic shock.

In this large, nationwide study, most VA-ECMO cases were treated at hospitals with a lower annual procedure volume. Treatment at a hospital with a lower annual procedure volume was associated with a higher mortality risk and likelihood for complications compared to higher volume centers.

Previous studies have shown that higher procedural volumes of cardiovascular interventions were associated with lower mortality rates [6, 8, 9]. Selected medical conditions including VA-ECMO in cardiac failure and critical care conditions such as mechanical ventilation are shown to have a significant volume-outcome association in which higher hospital case numbers are associated with survival benefit [15‐17]. Particularly, an international study using hospital ECMO volumes from 290 international centers found strong associations of higher hospital-level ECMO volume and lower mortality for neonates and adults, but not for children [18]. A similar observation has been made in Japan; for cases involving ECMO for respiratory failure, a higher hospital procedure volume of ECMO treatment for any indications was associated with lower in-hospital mortality [19]. Contrary to the traditional volume-outcome relationship, other studies on VA-ECMO reported increased mortality associated with high-volume hospitals and in patients transferred to tertiary centers [20, 21]. However, additional information is needed about patient characteristics and selection, management strategies for VA-ECMO, and organization of care within high- and low-volume centers to improve treatment and outcomes among these critically ill patients. In the present study, the majority of patients with VA-ECMO support were treated at hospitals with a low procedural volume (≤ 50 VA-ECMO implantations per year). However, 30-day in-hospital mortality across the categories of hospital procedural volume showed a significant difference between the lowest and highest annualized hospital procedural volumes. Lowest 30-day in-hospital mortality was observed at hospitals with an annualized hospital procedural volume > 100 VA-ECMO cases, especially. However, at hospitals in the lowest procedural volume category, a greater proportion of patients with ACS, prior CPR, CAD, CKD, and advanced age underwent therapy. In contrast, patients treated in the highest procedural volume categories showed a higher proportion of AHF and CHF, whereas sex differences between the hospital procedure volume categories were not observed. As expected across all categorical divisions and compared to the overall cohort, patients with prior CPR had a higher 30-day in-hospital mortality on VA-ECMO in general. Again, differences in baseline criteria were observed between the groups. The proportion of patients suffering from ACS decreased with a higher procedural volume. At hospitals with ≥ 50 annual procedural volume, the leading indication for VA-ECMO implantation slightly shifts from ACS to AHF. Significant differences in patient characteristics indicated by a higher proportion of patients with ACS, prior CPR, and comorbidities such as CAD and CKD in lowest procedural volume category could explain the higher 30-day in-hospital mortality compared to the highest volume categories with > 50 VA-ECMO implantations per year. However, after adjusting for potential confounders, the significant association with an increased risk in patients treated at lower volume centers persisted.

VA-ECMO is a potentially life-saving therapy for reversible circulatory failure not responsive to conventional therapies [3, 22]. However, complications on VA-ECMO are very common and are associated with a significant increase in morbidity and mortality [4, 23].

In the present study, we observed high complication rates of bleeding with 20.0%, limb ischemia with 7.4%, abdominal ischemia with 7.0%, and stroke with 1.0% in the overall cohort.

Stroke occurs in approximately 4% of VA-ECMO patients [11, 24]. This is similar to our results. In addition, the rate of stroke varies by indication and cannulation technique. It has been reported that femoral artery cannulation had a noticeably lower risk compared to central access [25]. Furthermore, the cause of stroke is multifactorial and can be associated with hemodynamic instability and thromboembolic events [26].

In a meta-analysis of 1866 patients by Cheng et al., lower extremity ischemia occurred in 16.9% of patients undergoing VA-ECMO support [11]. However, our analysis showed a lower complication rate. Although this might be explained by underreporting in the used registry, the use of prophylactic antegrade perfusion catheters might have reduced the frequency of limb ischemia over time [27].

Abdominal ischemia during VA-ECMO support is a known complication. To date, incidence rate of bowel and intestinal ischemia on VA-ECMO therapy is largely unknown. Among the total cohort, the incidence rate of abdominal ischemia was 6.7%. Hence, it can be deduced that abdominal ischemia is a relevant complication on VA-ECMO therapy and should be carefully considered before initiation of this invasive treatment.

We found a strong association between hospital procedure volume and complications. Although the crude incidence of complications was higher in hospital with a higher procedural volume, after adjusting for confounders, a higher likelihood for complications was observed at hospitals with lower vs. higher annual VA-ECMO volume. This finding is in line with previous cardiovascular and cardiac surgical studies that have demonstrated a relationship between higher hospital procedure or operator volume and fewer complication rates [28‐30].

The observation that higher procedural experience for VA-ECMO was associated with better in-hospital outcomes could be explained by differences in patient characteristics, technical factors, and more experienced handling of patients with complications at higher volume centers.

In a large nationwide registry with more than 10,000 patients, a significant volume-mortality association was shown for VA-ECMO procedures in cardiogenic shock. The majority of patients with VA-ECMO support were treated at hospitals with a low procedural volume in Germany. Thirty-day in-hospital mortality risk and likelihood of complications were higher at hospitals with the lowest annual procedure volume.