Acute Respiratory Failure After Drowning: a Retrospective

Original article 295

Acute respiratory failure after drowning: a retrospective multicenter survey Pierre Micheleta, Fouad Bouzanaa, Olivia Charmensata, Fabrice Tigerb, Jacques Durand-Gasselinc, Sami Hraiechd, Samir Jabere, Jean Dellamonicaf and Carole Ichaig

Objectives Despite the extensive literature on drowning, improvement of oxygenation and short ICU length of stay clinical data are still lacking on the best medical strategy to [3 (1–14) and 2 (1–7), respectively]. use. Acute respiratory failure (ARF) is the main component Conclusion Despite the absence of recommendation for of drowning pathophysiology. The objectives of this NIV use in case of drowning-related ARF, this technique was multicenter study were to analyze the clinical course of often used with safety and efficacy. The decision for NIV use drowning-related ARF patients and to describe the efficacy was mainly based on the preserved or improved of the ventilatory strategies used. neurological status. European Journal of Emergency Methods Medical records of drowned adult patients Medicine 24:295–300 Copyright © 2017 Wolters Kluwer admitted in seven ICUs after prehospital emergency Health, Inc. All rights reserved. medical care during three consecutive summer periods European Journal of Emergency Medicine 2017, 24:295–300 were retrospectively analyzed. Keywords: acute respiratory failure, continuous positive airway pressure, Results Among the 126 patients (58 ± 21 years) admitted, drowning, mechanical ventilation, noninvasive ventilation

38 patients with cardiac arrest at the scene were not aEmergency Department, Timone 2 Hospital, Aix-Marseille University, Marseille, analyzed, 26 received mechanical ventilation (MV), and 48 bIntensive Care Unit, Antibe General Hospital, Antibe, cIntensive Care Unit, Toulon-La Seyne General Hospital, Toulon-La Seyne, dRespiratory Distress and patients received noninvasive ventilation (NIV). Compared Infectious Intensive Care Unit, North Hospital, Aix-Marseille University, Marseille, with patients placed under MV, the NIV patients presented a eIntensive Care and Anesthesia Department, Saint-Eloi Hospital, Montpellier f ’ better initial neurological (Glasgow Coma Scale of 7 ± 4 vs. University, Montpellier, Medical Intensive Care Unit, l Archet Hospital, Nice University and gIntensive Care Unit, Pasteur 2 Hospital, Nice University, Nice, 12 ± 3, P < 0.05) and hemodynamic status from the France prehospital stage (mean arterial pressure of 77 ± 18 vs. Correspondence to Pierre Michelet, MD, PhD, Emergency Department, Timone 2 96 ± 18, P < 0.001). With comparable ARF-related Hospital, UMR MD2 P2COE, Aix-Marseille University, 13005 Marseille, France hypoxemia to MV, the NIV was maintained with success in Tel: + 33 638 741 313; fax: + 33 491 386 943; e-mail: [email protected] 92% (44/48). Both MV and NIV were associated with rapid Received 9 July 2015 Accepted 2 December 2015

Introduction ARF related to drowning remains unknown. Nevertheless, Drowning is still a relevant medical challenge and a leading this technique could prove to be interesting on the basis of cause of accidental death worldwide. According to the WHO, the use of positive end expiratory pressure (PEEP) over more than 500 000 deaths occur each year after unintentional the oxygen supply early in the clinical course. Moreover, drowning [1]. Although age less than 14 years appears to be a the French health organization includes emergency medi- key risk factor notably in low-income countries, repetitive cal service (EMS), with physicians on scene allowing the surveys clearly indicate a trend toward a reduction of early use of NIV in the prehospital setting. drowning among children in high-income countries [2]. The aim of our study was to analyze the clinical course of Conversely, considering a French survey, more than half of drowning-related ARF patients both in the prehospital unintentional drownings occur near the shoreline, in the first and the ICU settings and to describe the efficacy of NIV 300 m, and concerns mainly adults. The current definition of when used. Related to the absence of clear indication for drowning highlights the role of acute respiratory failure (ARF) NIV use in this case, we have retrospectively analyzed in the pathophysiology: ‘drowning is the process of experi- medical reports of patients admitted in several ICUs encing respiratory impairment from submersion/immersion in located in the south French coast for ARF following liquid’ [3]. Furthermore, a uniform way to report data after a drowning. drowning event has been suggested with the adoption of Utstein template [4]. Despite these efforts, the optimal Methods strategy to support the respiratory function is still imprecise. Study design The possible applications of noninvasive ventilation (NIV) This survey was conducted during three consecutive sum- have increased in the past decades, both in the hospital and mer periods between June and September (2010–2013). We extrahospital settings [5], but safety and efficacy of NIV in selected all adult patients (older than 18 years) admitted for

ARF following drowning in the Mediterranean Sea in seven an epidemiological register. Nevertheless, these surveys do ICUs. These ICUs were located along the Mediterranean not include medical parameters except for mortality. coast in the south of France. The study design was approved Considering the surveys of InVS during the study period by the ethics committee of the Marseille University Hospital and more specifically the data extracted for the ICU (Comite de Protection des Personnes Sud Méditerranée I), areas, it appears that ∼ 730 drownings occurred in the which waived the need for informed consent according to studied area. Among these drownings, 170 deaths French legislation, as that study was only observational. occurred on scene and more than 500 surviving patients were not admitted in participating ICUs (most of them Methods and measurements went to emergency departments without available We conducted a retrospective evaluation of prospectively estimation). included EMS and ICU records. Two ICU physicians independently reviewed these records with care to col- lect the Utstein Style variables. In this way, both Acute respiratory failure demographic data and scene information were collected As some of our patients were not intubated, the criteria by EMS physicians including neurological, respiratory, used for ARF definition were the association of the and cardio-circulatory initial vital signs and EMS inter- following: ventions such as cardiopulmonary resuscitation, intubation, and NIV use. (1) Pulse oximetry less than 92% saturation. (2) Hypoxemia (100 mmHg < PaO2/FiO2 ≤ 200 mmHg). After ICU admission, collected data included respiratory, (3) Extensive infiltrate on chest radiograph. hemodynamic, biologic parameters at admission and during the first days, complications, and outcomes. Severity of illness was assessed using the Simplified Ventilatory strategy Acute Physiology Score II at ICU admission [6] and the The choice of ventilatory strategy was determined by the Sequential Organ Failure Assessment Score [7]. The physician in charge of the patient both in the field and in neurological status of survivors was assessed using the the ICU. Both continuous positive airway pressure Glasgow Outcome Scale (GOS) [8]. (CPAP) and bilevel positive airway pressure (BiPAP) The proportions of patients admitted to the ICU were were recorded as NIV strategy. No systematic protocol assessed by comparison with the whole population of was applied by the different teams, as there is no drowned patients during the same period and the same recommendation yet regarding the respiratory manage- area using the results of InVS (Institut National de Veille ment of ARF following drowning. During the ICU stay of Sanitaire – National Institute of Health Watch) surveys [2]. patients submitted to MV, teams used protective ventilatory strategies associating reduced tidal volume, PEEP, and limited plateau pressure. French setting This survey was conducted in the French Mediterranean coast for three consecutive summers. All the ICUs par- Definition of patient groups ticipating to this survey were located less than 10 km To analyze the results with clinical accuracy, we have sepa- from the sea coast, except for the Montpellier Unit. The rated the patients into three groups: MV, NIV, or oxygen water temperatures for the studied coastal area ranged supply. These groups were defined after patient admissions from 19 to 25°C during the summer period. in respective ICUs based on respiratory parameters. French sea cost supervision benefits from the association between lifeguard and EMS organizations. Each beach Statistical analysis area is under the observation of trained first-aid workers Statistical analyses were conducted using statistical soft- who are able to take victims out of the sea, provide ware (SPSS Inc., Chicago, Illinois, USA). A descriptive cardiopulmonary resuscitation, oxygen supply, and rapid analysis was performed on the whole population. call to EMS. French EMS teams are composed of an Categorical variables were expressed as percentages. emergency physician and a nurse with the skills to pro- Continuous variables were expressed in the form of vide specialized care including NIV and mechanical mean ± SD or median (25th–75th percentiles), according ventilation (MV) according to the ‘stay and play’ model to the data distribution (Kolmogorov–Smirnov test). All P used in France. values were two-tailed, and a P value less than 0.05 was Every 2 or 3 years, the national institute of sanitation watch considered significant. For continuous variables, com- (InVS) conducts a national survey on drowning during the parisons were performed using the Student’s t-test or the summer period (between June and September). All EMS, Wilcoxon’s test when the distribution was non-Gaussian. emergency departments, and ICUs in France have the obli- The χ2-test was used for comparison of discrete variables gation to declare the drowning cases with implementation of or Fisher’s exact test, as appropriate.

Results intervention on scene was 17 ± 16 min, and the time During the study period (three consecutive summers), between the first call and hospital admission was about 730 cases of drowning occurred in the area of 78 ± 32 min. recruitment, with a total mortality of 23%. The number of At the scene, 26 patients received MV, 25 were placed patients admitted in the ICUs for drowning represented under NIV, and 37 received only oxygen supply. No 3% of the total ICU admissions. Knowing the epide- cardiac arrest occurred in these three groups of patients. miological reports of InVS [2] and the statistics of each Among the 37 patients receiving only oxygen supply, 23 ICU, about 17% of patients were transferred to ICUs were placed under NIV at arrival in the ICU related to (Fig. 1). To reduce the risk of missing drowned patients oxygen desaturation (Fig. 1). during the study period, all electronic files of patients admitted to the seven ICUs for ARF or CA were analyzed. Telephone contact with private ICUs was also Respiratory status and ventilatory strategy performed to screen for potential cases directly admitted The ventilatory strategies used are reported in Fig. 1. to these units. The main characteristics and parameters related to the initial ventilatory strategy used (oxygen supply, NIV or Characteristics of study participants MV) are displayed in Table 1. For this first assessment, Among the 126 patients admitted for ARF after drown- the oxygenation was better in the oxygen group com- ing, 38 presented cardiac arrest at the scene and were pared with the NIV and MV groups (P < 0.01 for each). excluded from further analysis. Most Ustein Style parameters were available for the 88 other patients (Table 1). They included victim identifier, sex, age, pre-existing Oxygen group illness, precipitating event (alcohol intoxication = 20; sui- A total of 37 patients received oxygen supply only at the cide attempt with drug use = 5; sport accident = 1), body of scene. Among these patients, 23 were placed under NIV water (100% salt water), EMS called (100%), initial vital after ICU admission because of an aggravation of signs assessed by the EMS (100%), and average body respiratory failure (oxygen group, n = 14). temperature at the scene (36 ± 2°C). Loss of consciousness wasrecordedin41patients. MV group Prehospital conditions This group included the 26 patients receiving MV at the All the drowning events were witnessed, and 72% life- scene and four additional patients intubated after failure guards were present in the area. The delay for EMS of NIV before ICU arrival (MV group, n = 30).

Fig. 1

Patients with ARF following drowning N = 126 Cardiac arrest resuscitated N = 38

Oxygenotherapy Early NIV N = 37 N = 25 Initially intubated N = 26

N = 21 NIV failure N = 4 ICU admission NIV from ICU N = 23 N = 30 = N 44 MV group NIV group

Discharge alive Discharge alive Discharge alive N = 14 N = 44 N = 30

Ventilatory strategies for the 126 drowned patients admitted for acute respiratory failure (ARF) during the study period. The different groups in regard to the ventilatory strategy used were defined from the ICU admission, as represented by the dotted line. MV, mechanical ventilation; NIV, noninvasive ventilation.

Table 1 Utstein style for drowning parameters BiPAP or CPAP first. Four patients (two under BiPAP and Oxygen group NIV group MV group two under CPAP) required intubation and MV because of (N = 14) (N = 44) (N = 30) an aggravation of respiratory failure (in three patients) and Age (years) 56 ± 21 65 ± 14 58 ± 23 neurological deterioration in one. After ICU admission, 21 Sex (male/female) 9/5 23/21 7/23 still received NIV and 23 additional patients who initially Cardiovascular disease 2 (14) 20 (45) 8 (25) (%) received only oxygen supply were placed under NIV (NIV Respiratory disease (%) 1(7) 3 (6.5) 5 (18) group, n = 44). Neurological disease (%) 1 (7) 6 (14) 8 (27) Loss of consciousness 2 (10) 18 (36) 21 (26) The main explanation reported by physicians to use NIV [n (%)] was a better neurological [Glasgow Coma Scale(GCS) of Glasgow Coma Scale 13 ± 2a 14 ± 1a 7 ± 2 Corporeal temperature 36.2 ± 136.7± 1.4 36.2 ± 1.2 12 ± 3 vs. 7 ± 4, P < 0.05] status compared with those (H0) (°C) placed on MV. The oxygenation improved in the NIV Mean arterial pressure 90 ± 22 96 ± 18 77 ± 18 (H0) (mmHg)a group and in the MV group from the 6th hour of ICU Heart rate (H0) 91 ± 25 92 ± 24 87 ± 26 without difference between groups (Table 2). (beats/min)a pH (H0)a 7. 3 3 ± 0.06 7.31 ± 0.08 7.23 ± 0.09 On first radiographic assessment, 49 (55%) patients pre- a b PaFiO2 (H0) (mmHg) 243 ± 154 156 ± 92 149 ± 95 a sented radiographic aspects of alveolar infiltrates pre- PaCO2 (H0) (mmHg) 43 ± 844± 852± 11 − a HCO3 (mmol/l) 22 ± 222± 322± 4 dominantly in the peri-hilum area with Kerley B-lines, Lactate (H0) (mmol/l)a 2.0 ± 0.6 2.9 ± 1.9 3.4 ± 2 but 20 (22%) presented additional lobar consolidations. SAPS 2 score 24 ± 828± 850± 19 SOFA score 1.7 ± 12.4± 26.5± 4 Incidence of infectious 0 1 (2) 6 (20) Neurological status pneumonia [n (%)] Length of ventilator – 1.4 ± 0.7 3 ± 2 For the first 25 drowned patients receiving NIV at the support scene, the first neurological assessment performed by ICU length of stay 1.4 ± 0.5 2 (1–7) 3 (1–14) EMS reported a GCS of 12 ± 3. The second neurological MV, mechanical ventilation; NIV, noninvasive ventilation; SAPS, Simplified Acute evaluation performed once the clinical evaluation was Physiology Score; SOFA, Sequential Organ Failure Assessment; H0, intensive completed and oxygen supplied reported a GCS of 14 ± 1. care first assessment. This second neurological assessment took place during aData recorded at the ICU admission. b For PaFiO2 ratio calculation, the measured PaO2 on blood gas analysis was the EMS transport or upon hospital arrival. In the MV divided by 80% FiO2 as oxygen concentration in the heavy oxygen supply mask. group, cardiac arrest patients excluded, the average GCS was of 7 ± 2(P < 0.05 compared with the NIV group). Table 2 Respiratory parameters between noninvasive ventilation group and mechanical ventilation group at ICU admission (H0); Outcome after 6 and 12 h following ICU admission (respectively, H6, H12) During the ICU stay, six patients in the MV group (20%) pH (H0) pH (H6) pH (H12) and one (2%) in the NIV group developed pneumonia. In NIV 7.31 ± 0.08* 7.36 ± 0.07§ 7. 3 8 ± 0.06* the MV groups, these pneumonias were complicated by MV 7.23 ± 0.09 7.31 ± 0.09§ 7. 3 3 ± 0.08 septic shock in five patients. In the MV group, another PAFI (H0) PAFI (H6) PAFI (H12) three patients were in a vegetative state (GOS 2) and two

§ had moderate disabilities (GOS 4). All the patients of the NIV 156± 92 207 ± 92 204 ± 115 MV 149 ± 95 215 ± 107§ 281 ± 109 NIV group made a good neurological recovery (GOS 5) (Table 1). PaCO2 (H0) PaCO2 (H6) PaCO2 (H12) NIV 44± 8* 38 ± 6§ 36 ± 4* MV 52 ± 11 42 ± 10§ 42 ± 8 Discussion − − − Facing a leading cause of accidental death worldwide, our HCO3 (H0) HCO3 (H6) HCO3 (H12) study reports the effectiveness of NIV in ARF related to NIV 22± 321± 422± 4 MV 22 ± 422± 422± 4 drowning. Moreover, our results demonstrate that the

− choice for ventilator strategy is mainly based on the PAFI: PaO2/FiO2 ratio; PaCO2 are expressed in mmHg; HCO3 are expressed in mmol/l. neurological status. MV, mechanical ventilation; NIV, noninvasive ventilation; H0, H6 and H12: measurements at ICU admission, 6 and 12 hours after respectively. Some countries have conducted repetitive epidemiolo- *P < 0.05 between NIV and MV groups for the same time of assessment. gical surveys to obtain a more accurate picture of § P < 0.05 by paired samples t-test between H0 and H6 for the same group. drowning to adapt public policies. In France, these surveys conducted by the InVS have reported that 1500 NIV group cases of drowning are responsible each summer period for Initially, 25 patients received NIV at the scene with more than 350 accidental deaths [2]. These results 8 ± 2cmH2O average level of PEEP. Among these patients, highlight the need for improving lifeguard coverage and 14 patients initially received CPAP during the prehospital training to ensure rapid extraction from water to dry land, phase, which was secondarily replaced by BiPAP in the but they raised the question of advanced EMS strategy ICU. There was no difference between patients receiving once the patient is on land. In this way, our results

Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. Acute respiratory failure after drowning Michelet et al. 299 support the validity of the stay and play model with predominantly hypoxemic nature of the ARF observed in regard to potential rapid improvement of clinical status our study rather than hypercapnic. Therefore, the CPAP once specialized care is provided by EMS. application that is often more easily applicable than BiPAP in the prehospital setting should be considered. Considering drowning-related ARF, our results demon- After ICU admission, the choice of BiPAP could be strate that ARF is characterized by a profound but explained by the prevention of respiratory exhaustion and reversal hypoxemia without relevant hypercapnia. This hypercapnia. hypoxemia explains the superiority of NIV either initially at the scene or after ICU admission for several patients The positive effect of NIV on hypoxemia was associated who initially received oxygen supply only. As no suffi- with a limited period of NIV application. This correlates cient data were available for further analysis of the oxy- well with clear indications of NIV in emergency condi- gen supply patient group, we have only analyzed the tions such as COPD or cardiogenic pulmonary edema. respective course of patients placed under NIV and MV. This rapid clinical improvement was associated with a These results demonstrated that both NIV and MV were reduced length of ICU stay and a favorable outcome. associated with a rapid improvement in oxygenation. As One could argue that these positive results could be previously reported [9], pneumonia occurred more fre- explained by the absence of ‘real and profound ARF’.To quently in the MV group, although NIV and MV groups answer this concern, we have defined ARF and we did should arguably not be compared as the MV group was not analyze medical records including moderate respira- more severe. tory dysfunctions. Moreover, the results of respiratory Within the six-grade system of Szpilman’s classification, our assessment by repetitive blood gas measurements did not patients were mainly classified as grades 3 to 5 with stable support this hypothesis. Another explanation could be hemodynamic parameters but impaired respiratory function based on the pathophysiology of drowning. Patients of associated with a variable neurological evaluation [10]. The the NIV group presented a moderate impairment of Szpilman’s classification is first based on the evaluation of neurological status but without complete loss of con- neurological status and second on the respiratory and hemo- sciousness. Thus, the amount of water inhaled was dynamic status. Although this classification is not always probably limited. As the pulmonary lesions are caused by known and/or used by the EMS teams, the initial neurolo- a temporary insult, these patients could be responsive to gical evaluation factually appears to be of paramount impor- NIV application [14]. tance to initially choose a ventilatory strategy. Indeed, the With regard to the neurological status, the improvement existence of neurological distress is an indication for intuba- observed for many patients in our study was not clearly tion and MV [10]. Whereas several patients received MV reported in the literature previously. Although specific related to the persistence of neurological impairment at the management of neurological disorders after drowning scene, others presented a rapid improvement of their neuro- remains unclear [16], this improvement could be dis- – logical status (assessed by GCS) in the first 10 15 min, with cussed in regard to the particular settings of our study. oxygen supply and/or NIV applied by the EMS team. Indeed, a rapid intervention of lifeguards and EMS team Interestingly, NIV was efficient for more than 80% of the has been associated with basic life support maneuvers, patients, with only four patients requiring intubation. In prevention of stomach content aspiration, airway protec- regard to the retrospective methodology of our study, the tion, and a high oxygenation supply [17]. Moreover, it is decision to use NIV was not based on recommendations well accepted that loss of consciousness is mainly related but rather on the confidence in the reversible nature of to hypoxemia rather than hypercapnia, as observed in our drowning-related ARF and the absence of neurological patients [18]. In addition, during the studied summer distress for eligible patients. Whereas CPAP was used as periods, the seawater temperature was between 18 and frequently as BiPAP in prehospital settings, all the 26°C. This was associated with a normal body tempera- patients received BiPAP from ICU admission. We are not ture in most cases, and it limited the alteration of con- able to explain the choice of BiPAP rather than CPAP by sciousness related to hypothermia [19]. Therefore, a ICU physicians, whereas the interest of PEEP has been second clinical assessment by the EMS physician has repeatedly suggested [11–13]. Gregorakos et al. [14] have frequently excluded the need for intubation, MV, and analyzed a retrospective cohort of patients admitted in an finally allowed the use of NIV. emergency department for ARF and reported the possibility to avoid intubation in noncomatose patients. Nevertheless, the ventilatory strategy used was not clearly Study limitations described. Dottorini et al. [15] have reported two cases of Although we have payed attention to reduce the risk of successful treatment by NIV in patients without loss of missed drowned patients, the possibility that EMS teams consciousness, highlighting the importance of neurologi- have transported ARF patients to emergency depart- cal status with regard to the choice of ventilatory strategy. ments with no information to referent ICU could induce The interest of PEEP application is also supported by the a bias.

Furthermore, EMS electronic data systems were not 4 Idris AH, Berg RA, Bierens J, Bossaert L, Branche CM, Gabrielli A, et al. available in all studied areas. Only notes of successfully American Heart Association. Recommended guidelines for uniform reporting of data from drowning: the ‘Utstein style’. Circulation 2003; resuscitated submersion patients who sustained cardiac 108:2565–2574. arrest were recorded, and the exact number of patients 5 Boldrini R, Fasano L, Nava S. Noninvasive mechanical ventilation. Curr Opin who died at the scene was based on the InVS survey. Crit Care 2012; 18:48–53. 6 Gall Le Jr, Lemeshow S, Saulnier F. A new Simplified Acute Physiology Finally, the patient groups were not comparable in terms Score (SAPS II) based on a European/North American multicenter study. JAMA 1993; 270:2957–2963. of severity and prognosis. Nevertheless, one objective of 7 Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, et al. this study was to look into factors associated with the The SOFA (Sepsis-related Organ Failure Assessment) score to describe choice of NIV rather than compare patients related to the organ dysfunction/failure. On behalf of the Working Group on Sepsis- Related Problems of the European Society of Intensive Care Medicine. strategy used. Intensive Care Med 1996; 22:707–710. 8 Jennett B, Bond M. Assessment of outcome after severe brain damage. – Conclusion Lancet 1975; 1:480 484. 9 van Berkel M, Bierens JJ, Lie RL, de Rooy TP, Kool LJ, van de Velde EA, Our study is the first to report, in a large cohort of drowning- Meinders AE. Pulmonary oedema, pneumonia and mortality in submersion related ARF, the safety and effectiveness of NIV. The victims; a retrospective study in 125 patients. Intensive Care Med 1996; positive results of NIV in these cases were associated with a 22:101–107. 10 Szpilman D. Near-drowning and drowning classification: a proposal to stratify stable and/or a rapid improvement of neurological status. mortality based on the analysis of 1831 cases. Chest 1997; 112:660–665. The neurological status should be considered as a variable 11 Layon AJ, Modell JH. Drowning: update 2009. Anesthesiology 2009; – clinical parameter that could influence the strategy of care. 110:1390 1401. 12 Ruiz B, Calderwood H, Modell J, Brogdon J. Effect of ventilator patterns on Therefore, our results support the need for further pro- arterial oxygenation after near-drowning with freshwater – A comparative spective evaluation on ventilator supports (NIV, high-flow study in dogs. Anesth Analg 1973; 52:570–576. 13 Bergquist RE, Vogelhut MM, Modell JH, Sloan SJ, Ruiz BC. Comparison of oxygen device), as well as on classification system and stay ventilatory patterns in the treatment of freshwater near-drowning in dogs. and play model. Anesthesiology 1980; 52:142–148. 14 Gregorakos L, Markou N, Psalida V, Kanakaki M, Alexopoulou A, Sotiriou E, et al. Near-drowning: clinical course of lung injury in adults. Lung 2009; Acknowledgements 187:93–97. Conflicts of interest 15 Dottorini M, Eslami A, Baglioni S, Fiorenzano G, Todisco T. Nasal-continuous There are no conflicts of interest. positive airway pressure in the treatment of near-drowning in freshwater. Chest 1996; 110:1122–1124. 16 Topjian AA, Berg RA, Bierens JJ, Branche CM, Clark RS, Friberg H, et al. References Brain resuscitation in the drowning victim. Neurocrit Care 2012; 1 Szpilman D, Bierens J, Handley A, Orlowski J. Drowning. N Engl J Med 2012; 17:441–467. 366:2102–2110. 17 Venema A, Groothoff J, Bierens J. The role of bystanders during rescue and 2 Thélot B, Lasbeur L. Drowning Epidemiological Survey. National Institute of rescuscitation of drowning victims. Resuscitation 2010; 81:434–439. Health Watch 2010. 1 June – 30 September 2009, Saint-Maurice, France. 18 Kristoffersen MB, Rattenborg CC, Holaday DA. Asphyxial death: the roles of 3 van Beeck EF, Branche CM, Szpilman D, Modell JH, Bierens JJ. A new acute anoxia, hypercarbia and acidosis. Anesthesiology 1967; 28:488–497. definition of drowning: towards documentation and prevention of a global 19 Wood SC. Interactions between hypoxia and hypothermia. Annu Rev Physiol public health problem. Bull World Health Organ 2005; 83:853–856. 1991; 53:71–85.