Continuous positive airway pressure plus low flow oxygen versus usual care of severe acute cardiogenic pulmonary edema in the pre-hospital setting: A randomised controlled trial

Study design and setting

This randomised, controlled, parallel-group trial, which was non-blinded but with good concealment of allocation, was undertaken in patients with a diagnosis of severe ACPE from July 2009 to July 2010. The Joint Tasmanian Ethics Committee (Tasmania Health & Medical Human Research Ethics Committee; EC00337) approved the study with patient consent waived (Ethics Reference Number: H0010364), as in such an emergent therapeutic area both treatments were accepted practices, time was of the essence and neither the patients nor relatives were in a position to make informed choices. The study was registered with the Australia New Zealand Clinical Trials Registry (Trial ID, ACTRN12609000410257).

The RCT consisted of two treatment arms: a control arm using standard positive pressure ventilation with high flow oxygen with a bag-valve-mask assisting patients’ ventilations (bagging), versus an active arm using a continuous positive airway pressure (CPAP) device (Whisperflow) and oxygen supplied at 28–33%. All participants were transported by ambulance to the Emergency Department of the Royal Hobart Hospital (RHH). The RHH is the tertiary hospital for the state of Tasmania and is the only major acute public general hospital serving a population of 250,000 in Southern Tasmania, Australia. No changes to the trial were made after commencement.

Selection of participants

Patients were included if they were ≥18 years of age and diagnosed by paramedic ambulance staff with severe ACPE requiring ventilator assistance. Patients were excluded if they were in respiratory or cardiac arrest, had a Glasgow Coma Score (GCS) <12 or had a systolic blood pressure <90 mmHg.

Interventions

Initially, all subjects received standard therapy according to Ambulance Tasmania (AT) guidelines, namely initial high-flow oxygen, sublingual nitroglycerin as a primary therapy in increasing incremental doses from 400–1600 µg, increasing every 5 min as long as systolic blood pressure was >100 mmHg, with 1–2 mg IV morphine as an anxiolytic and 40 mg IV frusemide for severe respiratory distress in patients with a transport time greater than 20 min.

We used computerised random number generation to allocate patients to treatment arms. This procedure ensured that treatment allocation was concealed before randomisation. Neither paramedics nor the research team were blinded to treatment after randomisation. Subjects entered the study and were randomised at the point where the ambulance personnel considered that they required non-invasive ventilatory assistance. This could be immediately on initial contact or after some standard therapy but subsequent deterioration. Patients in the active arm received 10 cm of H₂O CPAP delivered by Whisperflow® (flow created by a mix of oxygen and air delivering a fixed oxygen flow rate providing 28–33% oxygen), with continued concurrent therapy in the AT Protocol for ACPE patients. In the control arm, positive pressure ventilation was provided by a conventional bag-valve-mask system with high flow oxygen at a rate of 8–15 l/min (bagging), again along with the same concurrent per-protocol therapy. Pulse oximeters were used to measure oxygen saturations in both groups. Baseline and follow-up measurements of oxygen saturation and vital signs, including GCS scores were recorded on the computerized Ambulance Report Form (Victorian Ambulance Clinical Information System). Vital signs recorded at the time of initial assessment were used as baseline measures.

Measurements

In total, 98% (62/63) of paramedics from AT agreed to participate and were trained in the use of the CPAP device and the study protocol. Randomization and allocation concealment were achieved using computer-generated numbers printed on cards indicating the assigned device, which were sealed in sequentially numbered opaque envelopes placed with the ventilation/CPAP equipment on every ambulance.

Having identified a suitable study patient, the paramedic opened the next envelope and entered the randomization number into the computerized ambulance reporting system. A list of these patients was sent to the study coordinator each week. As is standard with such severe patients, they were pre-notified to the Emergency Department (ED) at RHH as an impending Category 1 (high priority) patient, but also that the patient should be recorded as part of the CPAP trial. At arrival at the ED, paramedics were asked to request ED personnel to follow the study protocol and to draw arterial blood gas (ABG) within 30 min; the study was then complete and conventional patient care continued under the treating emergency physician.

The randomisation cards were collected and accounted for by the research team (MA and MG) shortly after each use. The team also followed up each set of ambulance case notes to ensure that the correct device had indeed been used, since neither paramedics nor the research team were blinded to treatment after allocation. In addition, a full list of patients during the study period who had been transported by ambulance with an ED and hospital discharge diagnosis of ACPE was collected, and case notes were assessed to determine if any patients who either required ventilatory assistance or who received it were missed during this time.

Two independent physicians blinded to group allocation retrospectively reviewed randomised study patient computerised ambulance and hospital records to confirm that they had met the set diagnostic criteria for severe ACPE (crackles on auscultation, acute onset shortness of breath and hypoxia requiring assisted ventilation); where appropriate, the cause of mortality was also confirmed from the notes and/or death certification. Discrepancies were reviewed by a third physician (EHW) who was blinded to treatment allocation for resolution. Hospital admission data, including arterial blood gas results were obtained from the RHH Patient Information Medical System. Dispatch, arrival and transport data were obtained from dispatch records from AT to calculate length of pre-hospital treatment times. The third physician (EHW) also reviewed the hospital clinical records of those patients who died, while blinded to the treatment arm, to assess at what point critical deteriorations had occurred and crucial management decisions had been made. AT provided in-kind support for consumables, paramedic training and IT, but unfortunately pulled out at the 12-month time point for budgetary reasons, owing to the expense of CPAP consumables.

Outcomes and statistical analysis

The primary outcome measure was pre-hospital or in-hospital mortality. The secondary outcomes were: requirement for invasive ventilation, arterial blood gas values upon arrival in the ED (pH, PO₂, PCO₂, bicarbonate), length of hospital stay (days) for patients who survived their admission to hospital, and vital signs (oxygen saturation, heart rate, blood pressure, respiratory rate and GCS score).

Where only venous blood samples were available for blood gas assessments we used published formulations to estimate corresponding arterial values for pH, carbon dioxide, and bicarbonate¹⁵. Arterial and converted venous blood gases were compared using Student’s t-tests, and if no difference was detected they were combined for subsequent analyses. It was not possible to convert venous paO₂ since respective arterial paO₂ values are not directly comparable and so not provided in the conversion equations, so only measured arterial PaO₂ values were analysed.

Power calculations for the primary outcome measure were based on the findings of Hubble’s work¹⁶, who report a mortality rate, although in patients not as severely ill as ours, of 5.4% for ACPE patients receiving CPAP compared with 23.2% for patients receiving standard therapy. A total sample size of 74 participants (37 in each group) would thus provide 80% power to detect this 17.8% absolute reduction in mortality in the CPAP arm. It was estimated from a retrospective review that at least 50 patients per year with a diagnosis of ACPE requiring ventilatory assistance were transported by AT to the RHH; therefore, the trial was intended to run for 18 months (see below). Unlike more conventional randomized controlled trials, it was not envisaged that there would be drop-outs after recruitment, between randomization and arrival, dead or alive, at the ED.

Baseline characteristics for the patient cohort were described using frequencies for categorical variables and mean (±SD) for continuous variables. We used log binomial regression to compare the risk of death for patients in the two treatment arms. For the secondary outcomes, linear regression was used to compare post-treatment vital signs after adjustment for pre-treatment measurement, and to compare blood gas measurements. Non-normally distributed data were transformed as required. Variables that could not adequately be transformed were analysed using the Mann–Whitney test. Negative binomial regression was used to compare length of hospital stay for patients who survived, with the effect estimate presented as a ratio of mean length of stay for the intervention arm relative to the control arm. A ratio less than one would indicate a decrease in length of hospital stay associated with the intervention, whilst a ratio greater than one would indicate an increase. We summarized pre-hospital drug treatments as frequencies and percentages of patients treated with nitroglycerin, morphine or frusemide, and the number of doses received. Only intention-to-treat analyses were performed. Stata/IC version 12.1¹⁷ was used for all analyses and a P-value of 0.05 was considered statistically significant.

Study sample

A total of 377 patients with varying severity of ACPE were treated and transported to the RHH during the first 12 months, before the study was prematurely terminated as a result of budgetary concerns from AT. Of these, 50 (13%) met the inclusion criteria and were considered eligible for inclusion in the study according to the attending paramedics: 26 were assigned to the conventional bagging arm, and 24 to the CPAP arm (Figure 1). Of the remaining 327 patients transported, none received CPAP or bagging, so no missing patients were identified (Figure 1). All patients received treatment with the allocated device. Although the intended recruitment numbers were not achieved, the CIs decided to continue with the analysis as planned to avoid publication bias and to ensure that the data are available for meta-analysis. Thus, the analysis was completed on an intention-to-treat basis with the numbers obtained. Baseline characteristics in the two groups were similar for age, initial oxygen saturation, respiratory rate, blood pressure, heart rate, GCS score and pre-hospital treatment time; however, there were more females in the CPAP arm (Table 1). ED diagnosis confirmed the opinion of the ambulance personnel for all patients enrolled as ACPE, and this was also endorsed by a later review of investigator notes. The median treatment time (scene arrival to ED) was 35 mins.

Figure 1. Participant flow diagram.

Comparison of CPAP and bagging

We found modest evidence for a reduction in overall mortality in the treatment arm (RR, 0.24; 95% CI, 0.06–1.00; p=0.051; Table 2) from 34.6% (9 deaths) in the bagging arm to 8.3% (2 deaths) in the CPAP arm. This represents a 76% relative reduction in mortality. There were 11 deaths, all in hospital, with 8 occurring in the first 24 hours. Of these deaths, nine were due to a cardiac cause (essentially heart failure), one patient died of a stroke, and the final death at 9 days was due to an unexpected abdominal catastrophe (volvulus and secondary necrosis).

The mean (SD) length of hospital stay was 7.2 (5.11) days for surviving patients in the control arm (n=17) and 3.23 (2.37) days for patients in the CPAP arm (n=22). From the regression analysis (Table 2), it is estimated that the expected length of hospital stay would decrease by 55% with CPAP compared with usual care (ratio of means, 0.45; 95% CI, 0.29–0.70; p≤0.001). Blood gas analysis was undertaken within 30 minutes in 78% (39/50) patients; 48% (24/50) were arterial and 30% (15/50) venous. Patients receiving CPAP were significantly less likely to have acidosis (pH <7.35) and hypercarbia (PaCO₂ >45 mmHg) (Table 2); the hospital arrival post-ambulance-treatment oxygen saturation of patients was significantly lower in those who received CPAP (Table 3). Only two patients were intubated in the ED, both having been bagged, and both died in the ED. There were no significant differences in ED arrival heart rate, blood pressure and GCS between the two study arms (Table 3).

Use of out-of-hospital medications was similar between the two groups: sublingual nitroglycerin was used in 73% (19/26) of patients in the control arm and all patients (24/24) in the CPAP arm; frusemide was used in 50% (13/26) of patients in the control arm, and in 75% (18/24) of patients in the CPAP arm; low-dose morphine was used in 23% (6/26) of patients in the control arm and 25% (6/24) of patients in the CPAP arm. Of those patients receiving drug treatment, the median (IQR) total doses for sublingual nitroglycerin sprays (400 ug/spray) was 2.8 sprays (2.8) for the control arm and 2 sprays (2.8) for the CPAP arm. For frusemide, most patients (92% in the control arm and 89% in the CPAP arm) received a total dosage of 40 mg, one in each arm received 20 mg and one in the CPAP arm received 80 mg. Dosage data were recorded for 9/12 patients receiving morphine: the range of total dosage in the control arm was 1–3 mg and was 1–2 mg in the CPAP arm.