A previously healthy and physically fit 48-year-old male kayaker of Norwegian ethnic origin sets out to paddle alone around an island (Varalds?y) in the Norwegian Hardanger fjord - an estimated trip of about 4-5 hours.

Gore-Tex? shell with thin wool underneath. Flotation vest without collar. Waterproof GPS-receiver (Global Positioning System) and cellular phone in a waterproof container. Polar? heart rate monitor consisting of an ECG chest band and a wristwatch recording heart rate data.

The helicopter was airborne 7 minutes after alarm, and arrived 20 minutes later over Bondesundet (22 NM from home base). The civilian high speed boat was already there searching. After 8 minutes search from the air the kayaker was found floating prone with his head completely submerged in seawater. There were no signs of life.

The kayaker was taken ashore in the high speed boat, and attempted single-helper CPR was started as it followed the helicopter to a nearby ferry dock. Once ashore quality CPR and endotracheal intubation was performed. Helicopter transport to our tertiary hospital was then undertaken while performing manual chest compressions. Ventilation was by an Oxylog? ventilator. No drugs were given, and defibrillation was not attempted. The only monitoring was capnography.

The accident happened on a Saturday; hence the thoracic surgery team was summoned from home. Because of this, CPR by the pre-hospital team was continued into the cardiothoracic OR (Operating Room). On arrival the patient was asystolic with a bladder temperature of 20.6?C.

In the OR interposed abdominal compression-CPR [³] was applied. Monitoring invasive arterial blood pressure, this technique clearly produced higher arterial pressures than conventional chest compressions (Figure 1 and Figure 2).

During surgery to establish femoral vascular access for the CPB, short coordinated pauses in chest compressions were needed (shown in video).

pH 6,67, pO2 8,4 kPa, pCO2 17,0 kPa, BE -20 mmol/l, Hb 17,1 g/dl, Na⁺ 151 mmol/l, K⁺ 3,1 mmol/l, Cl^- 113 mmol/l, glucose 20,5 mmol/l and lactate 19 mmol/l.

Full CPB circulation was established 2 hours after cardiac arrest occurred.

After core temperature reached 30?C, the patient was in VF (ventricular fibrillation). Cardioversion was attempted 5 times, but VF returned until at 34?C, the VF became more course. A final shock was delivered and stable ROSC (Return of spontaneous circulation) was achieved.

Following rewarming and ROSC the patient was intentionally kept hypothermic at 34?C the first 24 hours. Initially ha was treated for low cardiac output (cardiac index) < 2.0 l/m² and pulmonary oedema. He was kept sedated and an ICP (intracranial pressure) transducer was placed to monitor for expected oedema-induced rise in ICP. A rise in ICP to 35 mmHg on day 3 led to treatment with Na-pentothal after which ICP fell to normal values. S-Na was intentionally kept around 150 mmol/l. Sedation was discontinued on day 5, and cardiac index had normalized at this time. He gradually regained consciousness the following days. Further complications included critical illness neuropathy, delirium and pneumonia. After 21 days of treatment in the ICU he was transferred to a conventional neurological ward. After a total of 33 days in hospital he was discharged to a rehabilitation unit.

(From author?s private handheld PDA/GPS).

GPS based estimation of time of submersion/ drowning:

Assuming linear and constant speed movement of the objects in the water and by plotting lines through the points where we located the kayak and the kayaker at two times with known time-interval, we could estimate the time and position when and where these two objects parted in the water. This is likely the time and place of loss of consciousness. Cardiac arrest either caused this loss of consciousness of followed shortly thereafter. With an assumed error margin of +/-5minutes this yields 11-21 min submerged time before he was found.

This GPS based estimate correlates fairly well with the HRM data (21 minutes) which we used as the best estimate as GPS position data was not logged with great precision.

We used the software provided with the Polar? HRM for visualisation and analysis of his heart rate data. His telephone log and our air-ambulance log provided us with supplemental information.

Note: The Polar? HRM is not intended for recording arrhythmias. Sadly, according to the manufacturer, a real-time ?error-correcting? algorithm in the device potentially changes abnormal data before it is stored. Thus asystole and very rapid ventricular rhythms are probably lost. Noise artefacts from extreme muscle activity may also disturb the recordings. The HR data is recorded and stored at 5 sec intervals. The data can be seen visualized in figures 3 and 4. Prolonged straight lines are assumed to indicate signal fault or no signal from the ECG electrodes. We have assumed there is real heart rate data with interpolated straight lines during periods of noise, missing or faulty data. Key times are indicated with green arrows and text (Figure 3 and Figure 4).

There is, as expected, a brief increase before a sharp drop in HR following cold water immersion. Heart rate first rises from approximately 130 min^-1 to 137 min^-1 and then drops to 43 min^-1 after approximately 11 minutes. From there on the HR seems to vary between 78 min^-1 and 35 min^-1.

The variations in HR with repeated patterns of accelerations and decelerations are assumed to be caused by physical struggling e.g. trying to get back into the kayak.

38 minutes after immersion these variations stop, probably due to final incapacitation. Cardiac arrest and head submersion is likely to have occurred simultaneously or shortly thereafter. The next recorded HR is 35 minutes later at close to 100 min^-1. Manual chest compressions at a rate of about 100 min^-1 had commenced at this time and may be the reason for this recording. The following spike in HR to 210 min^-1 coincides with helicopter start-up with patient onboard and may have been caused by disturbances from the helicopter.

It is of course possible but unlikely that all data after immersion is corrupt, but we have no way of knowing if this is the case. Completely straight lines are obvious errors, but we have assumed the points between these lines to be genuine data.

Disappointingly, the recorded heart rate data is clearly sporadic limiting further analysis (Table 1).

On the time of discharge from ICU the patient was fully alert and achieved a Glasgow Coma Score of 15.

The neurological examination was conducted at the end of the ICU stay and revealed no signs of focal damage to the brain.

EEG: Three months after discharge from the ICU the patient?s EEG showed non-epileptoform pattern without clinical significance.

Normal at three months.

Reduced conduction velocity in fingers on both hands at three months.

Normal at three months.

CANTAB (Cambridge Neuropsychological Test Automated Battery) [⁴] is a semi-automated neuropsychological test battery of memory, attention and executive functions using a touch screen ?tablet? PC. Tests: motor screening test (MOT), short-term memory and forced decision-making (DMS), spatial planning and motor control (SOC), and visual memory and learning (PAL).

The patient?s results are compared to an integrated age- and sex-matched normal population giving us his results as a z-score. A z-score of 0 represents a result equivalent to the mean result in the normal population. [⁵]. A change in |z-score| > 1.0 indicates significant improvement or deterioration (Table 1).

Cold cardiac arrest patients are not dead until both warm and dead.

Written informed consent was obtained from the patient for publication of this case report, accompanying images and video. A copy of the written consent is available for review by the Editor-in-Chief of this journal.

The authors declare that they have no competing interests.

FKL was the attending anesthesiologist on call in the air ambulance helicopter, treated the patient pre-hospitally, recorded GPS data, analyzed and interpreted the data regarding the heart-rate, GPS, telephone log and medical logs. JGRT performed the cantab cognitive examinations, analyzed and described the results of these. HKF was senior intensivist on call in the ICU. He video-filmed the initial treatment in the OR, and treated the patient in the ICU. All authors read and approved the final manuscript.

(The patient did not fulfil the PAL tasks and these results are excluded from the analysis).