A total of 768 CS patients were included in 49 centers, among whom 359 (46.7%) did not require ventilation, 118 (15.4%) required NIV, and 291 (37.9%) MV. Clinical characteristics of these patients are presented in Table 1. The mean age, gender, and risk factors were similar in the three groups. There was no difference regarding medical history except less history of previous heart disease in the MV group (47.8 vs 50 in NIV and 64.9% in NV groups, p < 0.001). MV patients were less frequently under long-term cardiological treatments than those under NIV or NV (Beta-blockers, ACE inhibitors or ARB2, Furosemide, and anti-aldosterone).

The most frequent CS triggers were ischemia, ventricular, and supraventricular arrhythmia without between groups difference except for ischemia, which was less frequent in NV group (29% vs 45.8 and 45%, p < 0.001).

The clinical, echography, and biologic presentations are presented in Table 2. Patients in the MV group were more often hospitalized in the ICU versus ICCU than patients under NIV or NV (58.8% vs 15.5 and 10.9% respectively, p < 0.001). MV patients presented with more previous cardiac arrest (21.3 vs 6.8 and 2.2%, p < 0.001), more skin mottling (51.4 vs 34.2 and 29.9%, p < 0.001), and higher lactate at admission than NIV or NV group (3.8 vs 2.7 vs 2.3 mmol/L, p < 0.001). Renal and hepatic functions were similar between groups. There were no between groups difference regarding echocardiography at admission besides tricuspid annular peak systolic velocity tissue doppler imaging which was higher in the MV group than NIV or NV (10 vs 8 vs 7 cm/s, p < 0.001).

In-hospital management and parameters at discharge are presented in Table 3. The MV group received more volume expansion during the first 24 h of management than the NIV or NV groups (56.7 vs 49.2 vs 27.5%, p < 0.01). During hospitalization, they benefited from higher doses of inotropes and vasopressors (dobutamine, norepinephrine, and epinephrine). Moreover, organs support was more frequently used in this group with higher use of acute mechanical circulatory support (MCS) (36.8 vs 8.5 vs 6.4%, p < 0.01) mainly by VA-ECMO support (71% in the MV group), and higher use of RRT in the VM group (27.2 vs 8.5 vs 8.4% p < 0.01). On the contrary, there was a more common use of diuretics (89.4% vs 74.6% in the MV group, p < 0.01) and less acute MCS use (6.4%) from different types with a predominance of IABP (65.2%) in the NV group.

Half of CS patients had undergone coronary angiography. A 3-vessel disease was found in about 20% of the cases but only the culprit lesion was revascularized in about 80% without significant differences between groups. There were also no differences regarding right-heart catheterization, pacemaker or defibrillator implantation, or radiofrequency ablation.

At discharge, LVEF was significantly higher in patients in the MV group (41 vs 32.1 vs 31.7%, p < 0.01). Patients in the MV group were hospitalized longer (20 vs 16 days, p < 0.01) but without difference in the discharge mode (home, rehabilitation, or care center).

At 30 days, MV group presented higher mortality (crude HR (MV vs No Ventilation) 1.41 [1.05–1.90], p = 0.022 and Fig. 1, log-rank p = 0.012) and a higher rate of major adverse events (death, heart transplantation or ventricular assistance) as compared with others groups (crude HR (MV vs No Ventilation) 1.52 [1.16–1.99], p = 0.003 and Fig. 2, log-rank p = 0.002). At 1 year, the between-groups difference subsists (Supplementary Fig. 1, log-rank p = 0.052), especially with a higher mortality for MV patients (crude HR (MV vs No Ventilation) 1.28 [1.02–1.61], p = 0.032).

30 day survival according to the maximal level of ventilatory support used during hospitalization

30 day survival free from heart transplantation or LVAD/BiVAD support according to the maximal level of ventilatory support used during hospitalization. BiVAD, biventricular assist device or total artificial heart; LVAD, left ventricular assist device

Interestingly after adjustment for known independent predictors of 30-day mortality [5] (age, LVEF < 30%, mechanical circulatory support, RRT, use of norepinephrine and use of diuretics), the between groups difference in 30-day all-cause mortality disappears (Supplementary Table 1).

No difference in all-cause mortality (Crude HR (NIV vs No Ventilation) 0.79 [0.49–1.26], p = 0.315) or MAE (Crude HR (NIV vs No Ventilation) 0.83 [0.54–1.27], p = 0.399) was found between NIV and NV groups either at 30-day or 1-year (Crude HR (NIV vs No Ventilation) 0.95 [0.69–1.31], p = 0.752 for all-cause mortality).

Among MV group, no difference in terms of 30-day mortality was found between patients intubated directly (n = 237, 84%) and patients first ventilated by NIV and then intubated (n = 44, 16%) (respectively 31.7 and 31.8%, p = 0.899) (Supplementary Fig. 2). No difference was found neither between patients intubated directly, patients first ventilated by NIV and then intubated within 24 h (n = 28, 10%), or patients first ventilated by NIV and then intubated after 24 h (n = 16, 6%) (respectively 31.7, 28.6 and 37.5%, p = 0.899) (Supplemental Fig. 3).

Factors associated with increased use for MV (Table 4) are previous cardiac arrest (OR 5.48, p < 0.001), infectious CS trigger (OR 2.55, p 0.001), presence of mottling (OR 2.25, p < 0.01), and higher lactate at admission (OR 2.13 for the third tertile of lactate, p 0.003). On the other hand, older patients (OR 0.97 for 1 year more, p < 0.001), non-observant patients (OR 0.17; p = 0.001), and those on long-term furosemide were less managed by MV (OR 0.47; p = 0.006).