Mechanical chest compression devices in the helicopter emergency medical service in Switzerland

We conducted a retrospective study examining the period lasting from December 31, 2013, to June 30, 2016 (912 days). The institutional review board of the Bern cantonal ethics committee (KEK) reviewed the study design and granted permission for use of patient data without individual consent, according to the federal act on research involving human beings, and the ordinance on human research with the exception of clinical trials. The permission covered the use of patient data collected during the HEMS operation and related hospital stays (KEK Bern, September 13, 2016, reference number 2016–01473).

Swiss Air-Rescue (Rega) conducts around 11,000 HEMS missions annually. In 2013 Rega operated 13 helicopter bases, which are distributed throughout the country in such a way that they can fly to any location at any time within its operational area within 15 min. The helicopter fleet comprises 7 Eurocopter EC145 machines, stationed at the lowland bases in Zurich, Basel, Bern, Lausanne and St. Gallen, and 11 AgustaWestland Da Vinci located at the mountain bases in Untervaz, Locarno, Erstfeld, Samedan, Wilderswil, Mollis and Zweisimmen. All 13 helicopters are equipped for winch operations and night-time flying.

In Switzerland, the HEMS crew includes a pilot, a paramedic and an emergency physician. The paramedic also serves as winch operator and is therefore not available to treat the patient on-site in case of a winch rescue. In 2013, the AutoPulse® device was added to the standard equipment of the Rega HEMS. Primary missions take place in a large variety of environments and situations, ranging from providing support for ground ambulances in urban areas to autonomous operations in extreme mountainous terrain. We included every operation (primary and secondary mission) during the investigation period.

After every HEMS operation, a variety of information is recorded in the Rega operations database (SAP Software), including whether the MCD was used or not. In a first step, we filtered the database for all operations where the use of the MCD had been reported, and we excluded other operations. For the included operations, we then gathered the physicians’ medical operations report forms from the Rega archive. In a second step, we examined these reports and entered key information in a database (Labkey). For the remaining operations, we identified the ones in which the patient was transported to a hospital, and we asked the respective hospitals to provide us with the discharge letters of these patients. In a further step, we analyzed the hospitals’ discharge letters and also integrated their key information into our database. The fourth and final step was to analyze the collected data.

The characteristics of all patients were compared using descriptive statistics.

The binary variable diagnosis (trauma versus medical diagnosis) was excluded due to a separation problem (i.e. only one survivor in the trauma group). To further assess the impact of the diagnosis variable, a Firth’s bias reduced logistic regression model was performed to evade separation [14]. The resulting estimate and standard error for the diagnosis variable remained very large. Hence, the diagnosis variable was excluded for the analysis to avoid bias.

Due to the relevant number of missing data in the variables CPR by layperson (n = 70), shockable rhythm (n = 62), and survival (n = 70), a sensitivity analysis using multiple imputations was performed. All variables of the model were used for imputation in the target variable using the pmm method, generating 10 imputed data sets. Multiple imputation was performed using the mice R-Package [15].

In 158 cases (33%), the death of the patient was documented on the scene. In the remaining 320 missions (67%), the patient was transported to a hospital. The transported patients could be divided into a first group of 49 cases (15%) where CPR was ongoing when the patient was loaded into the HEMS and a second group of 271 cases (85%) where the patient had regained spontaneous circulation before being loaded into the HEMS (Fig. 1). In the first group, ROSC could be achieved during transport in five patients (10%). However, only one patient (2%) survival to hospital discharge; 46 patients (94%) died in the hospital; In two cases (4%) the endpoint remains unknown because of missing hospital reports. Of the patients in the second group, 20 (7%) experienced a second CA during the flight, two (1%) could be resuscitated with a single defibrillation attempt, and the remaining 18 (7%) needed full CPR efforts, including the use of the MCD. Those 18 patients died in hospital. The remaining 251 patients (93%) from this group with ROSC on the scene remained haemodynamically stable during the flight.

In two cases (0.4% of all primary operations) a malfunction of the MCD was documented. One malfunction was due to rough terrain hindering the correct positioning of the device; the second was an unexplained inability of the device to start chest compressions when cardiac arrest occurred in flight.

Univariate logistic regression models showed that male sex and layperson CPR were associated with better survival as well (Table 3; Fig. 2). However, these effects were no longer statistically significant when adjusted for the type of cardiac arrhythmia (shockable versus non-shockable) in the multivariate analysis (Tables 4 and 5; Fig. 2).

Amongst the secondary operations, 102 (91%) did not require use of the MCD, which was preinstalled and on standby. Of the rest, five (5%) patients were loaded into the helicopter with ongoing CPR; ultimately, all of these patients died in the hospital. The remaining five (5%) patients went into cardiac arrest during the flight and CPR was initiated, with one of them achieving ROSC before landing (Fig. 1).