This retrospective observational single-center study of patients with SAH by ruptured intracranial aneurysm was carried out at the University Hospital of Guadeloupe, the main and major hospital of the area, with 440 beds. As the most advanced center for the entire Guadeloupe archipelago, all patients with aSAH are managed there. Since 2010, all patients with suspected aSAH are transferred by airplane to one unique high-volume center in Paris, the Rothschild Foundation Hospital where digital subtractive angiography and subsequent coiling is performed. Although neurosurgical clipping is possible at the University Hospital of Guadeloupe, coiling is preferred [2, 11, 12]. Clipping in Guadeloupe is only performed in patients who are not transferred for coiling (e.g. lack of health insurance).

Airplane transfer includes hospital to airport transfer, installation of the patient in the airplane, flight time, exit from the airplane and transfer from airport to the reference center. Once aSAH has been diagnosed and airplane transfer to the reference center in Paris decided, patients are managed until the transfer based on transcranial doppler, clinical assessment and computed tomography, according to guidelines [13]. In addition, to secure and try to prevent complications during the 15-hour long transfer, including an 8-hour flight, additional specific guidelines for airplane transfer of aSAH patients were developed by both teams (University Hospital of Guadeloupe and Rothschild Foundation Hospital), based on our common experience, including a systematic control CT scan performed before flight. Figure 1 summarizes these guidelines (complete guidelines of the University Hospital of Guadeloupe are available in the Supplementary Material). University Hospital of Guadeloupe Guidelines for the management of aneurysmal subarachnoid hemorrhage).

Management of patients with aneurysmal subarachnoid hemorrhage prior to airplane transfer on a commercial flight according to our local guidelines. WFNS, World Federation of Neurological Surgeons

First, because external ventricular drainage (EVD) placement is impossible during the transfer and especially in-flight, EVD was performed under general anesthesia before the flight in patients with a Fisher scale [14] ≥ 3, since these patients are at high risk of developing hydrocephalus, defined by abnormal symptomatic accumulation of cerebrospinal fluid inside the cerebral ventricles and subsequent intracranial hypertension [15, 16]. Of notice, after EVD was performed, these patients remained intubated and mechanically ventilated for the whole transfer. Second, patients with a Fisher scale < 3 but a World Federation of Neurological Surgeons (WFNS) [17] grading scale ≥ 4 were preemptively intubated before the flight because these patients are at high risk of worsening level of consciousness and require endotracheal intubation, which can be difficult due to limited space on a commercial airplane. Our guidelines did not recommend the use of tranexamic acid [18]. Even if interventions before transatlantic transfers (e.g. EVD or intubation) were recommended by our guidelines, the final decision was under the supervision of the intensivist in charge of the patient. The entire transfer was managed by a team consisting of a physician and a nurse with a high expertise in airplane transfer of aSAH patients. The teams providing the transfer have not only received a theoretical training, but also a practical one. These teams have, in addition, carried out at least 50 transfers of ventilated intubated patients.

Continuous intracranial pressure measurement was performed in the ICU prior to the transfer, but measurement was not available during the transfer (monitors used for airplane transfer were not equipped with this function). Therefore, hourly ventricular drainage (about 10 mL) was performed if EVD was productive. If less than 10 mL was evacuated, the transfer team checked if EVD was permeable. If it was, EVD drainage was performed every couple of hours or hourly volume was minored. All these adjustments were done with close supervision of the patient.

On the ventilator, tidal volume was set to 6 mL/kg with plateau pressure < 30cmH2O and a respiratory rate so that end tidal CO2 remained between 38 and 42 mmHg. Blood pressure was managed in accordance with the last transcranial doppler realized before leaving the ICU, using either norepinephrine or antihypertensive therapies to fit the objectives. Oral nimodipine (60 mg) was administered each 4 h. Sedations were by either propofol (≈ 3 mg/kg/h) or midazolam (≈ 0.1 mg/kg/h) associated with sufentanil (20–30 µg/h). No tranexamic acid was used.

University Hospital of Guadeloupe is 8 km distant from the airport. Using a specific elevator positioned on the left back door of the plane, transferred patients are boarded three hours before the scheduled departure hour and other passengers. Six seats are required to install the patient supine on a flat stretcher, with the head at least 30° elevated, in the direction of the cockpit. Two medical oxygen kits (1,27 × 107 Pascals, 3250 L each) have previously been installed. All devices are then fixed for the flight (monitor, respirator, electric syringe pumps) (Fig. 2). Flight time is about 8 h (6751 km, Fig. 3). On arrival, patients are disembarked after passengers and transferred to the reference center (22 km from the airport). Total transfer time door to door from the University Hospital of Guadeloupe to the reference center is 14 to 16 h.

Steps for installation of the patient in the plane. Airplane silhouette is reprinted from [https://upload.wikimedia.org/wikipedia/commons/0/00/Avion_silhouette.svg] under a CC BY license, with permission from [Andreas 06, Public domain, via Wikimedia Commons], original copyright [2006] [19]. Other pictures are author’s personal properties

Transfer from University Hospital of Guadeloupe to Rothschild Foundation Hospital (reference center) in Paris. Maps are reprinted from [https://openstreetmap.org] under a CC BY license, with permission from [OpenStreetMapFoundation], original copyright [2018] [20]

Despite the fact the environment is less adapted to care, we opted for airplane transfers on commercial flights rather than ambulance flights because of the high number of daily commercial flights (three to five) and because ambulance flights should first come from mainland France, which would delay the transfer by at least 12 h. Of notice the cost of an airplane transfer in a commercial flight is about 14.000 euros, which is much lower than the cost of a medevac on an ambulance flight that is approximately 140.000 euros Two scheduled airlines operate sanitary airplane transfers. Minimal timing to organize an airplane transfer is 6 h (management of the patient, medical team for transport, availability on the flight or requisition by disembarking passengers). Therefore, if subarachnoid hemorrhage is diagnosed less than 6 h before the last flight, transfer is postponed for a day and so management in the reference center is also delayed, which increases the risk of complications (rebleeding, vasospasm, seizure…etc).

Consecutive patients admitted between January 1st 2010 and December 31st 2019 with non-traumatic subarachnoid hemorrhage confirmed by computed tomography or magnetic resonance imaging, were included. Patients with arteriovenous malformation, age < 18 years old and those who were not transferred to mainland France (no health insurance or treatment withholding) were excluded from the study. For patients with several admissions for non-traumatic subarachnoid hemorrhage, only the first stay was included in the analysis. Of note, since cerebral angiography was not available at the University Hospital of Guadeloupe, patients with non-traumatic subarachnoid hemorrhage in whom no arteriovenous malformation or aneurysm was seen on brain angio-computed tomography were assumed to be caused by an aneurysm and were subsequently airplane transferred for digital subtraction angiography and possible coiling.

We collected the following data: age, gender, previous arterial hypertension, day of bleeding, clinical characteristics at hospital admission (Glasgow coma scale, neurological deficit, WFNS grading scale and Fisher scale). We also collected the characteristics of the aneurysm (size, number, and location), the complications before securing (acute hydrocephalus based on computed tomography, intracranial hypertension based either on abnormal transcranial doppler defined by decrease of end-diastolic velocity < 20 cm/s and /or increase of pulsatility index > 1.20 or abnormal computed tomography suspected on disappearance of the cortical furrows or filling of the mesencephalic cisterns or disappearance of the ventricles, rebleeding) and the management before transatlantic airplane transfer (EVD, tracheal intubation and subsequent mechanical ventilation). We finally collected complications that occurred during the transatlantic airplane transfer and during the first day at the reference center in France mainland, 30-day and one-year mortality, modified Rankin Scale score [21] and Glasgow outcome scale [22] at discharge from the hospital.

The modified Rankin Scale is a tool used to classify the degree of handicap in stroke [21]. The scale is organized in six levels, with one more for patients who died: 0 is no symptoms; 1 is no significant disability, able to carry out all usual activities, despite some symptoms; 2 is slight disability, able to look after one’s own affairs without assistance, but unable to carry out all previous activities; 3 is moderate disability, requires some help, but able to walk unassisted; 4 is moderately severe disability, unable to attend to own bodily needs without assistance, and unable to walk unassisted; 5 is severe disability, requires constant nursing care and attention, bedridden, incontinent; 6 is when the patient has died.

The Glasgow outcome scale is used after brain damage to assess persisting disability [22]. The scale is organized in five levels.1 is when the patient has died. 2 is Persistent vegetative state: severe damage with prolonged state of unresponsiveness and a lack of higher mental functions. 3 is severe disability: severe injury with permanent need for help with daily living. 4 is moderate disability: no need for assistance in everyday life, employment is possible but may require special equipment. 5 is low disability: light damage with minor neurological and psychological deficits. According to previous publications [23] we considered a good neurological outcome at hospital discharge as values of modified Rankin Scale from 0 to 3 and as values of Glasgow Outcome Scale from 4 to 5.

Data collected were those of patients admitted between January 2010 and December 2019. Authors never had access to information that could identify individual participants during or after data collection.

Analyses were performed using Prism V9.4.1 (GraphPad software, Boston, MA) and R statistical software, version 4.0.4 (available online at http://www.r-project.org/). Quantitative variables were described as median [interquartile range] and were compared between groups using the non-parametric Wilcoxon rank-sum test. Qualitative variables were described as n (%) and were compared between groups using Fisher’s exact test. To evaluate a potential change in patient’s mortality and severity across the 10-year study period, we arbitrarily divided it into two 5-year study periods, 2010–2014 and 2015–2019. Potential changes in mortality rates over the study period were analyzed using a Fisher’s exact test. Potential changes of severity over time were analyzed using a Wilcoxon rank-sum test. Cox proportional-hazard regression analysis was used to identify the variables significantly associated with one-year mortality. Variables yielding p-values < 0.20 in the univariate analyses were considered. Variables used to calculate scores (e.g. WFNS) were not entered in the model. In addition, we did not enter variables related to interventions before airplane transfer or variables determined by the airplane transfer (e.g. intubation and external ventricular drain). Eventually, five variables were entered in the multivariate analysis. We performed backward selection on the model, stopping when the Akaike information criterion reached its minimum. Results were reported as Hazard ratios (HR), with their 95% confidence interval (95% CI).

Kaplan-Meier overall survival curves up to one year were separately computed for relevant variables and were compared using log-rank tests. No imputation was performed for missing values. A p-value less than 0.05 was considered significant.