It is known from the previous dispatch study [6] that all paediatric drowning cases with an Injury Severity Score (ISS) > 15 recorded in the NSW Trauma Registry occurring during the operational hours of the HEMS service were identified by the parallel tasking system that operated till March 2011. All such cases were intubated by the P-HEMS team and transported directly to a paediatric specialist centre. In the one year period after March 2011 when only the centralised dispatch system was in operation there were seven drowning cases with ISS >15 during the HEMS operational hours, of which only four (57 %) were identified for P-HEMS response. Two of the remaining three cases were transported by road paramedics to adult trauma centres then secondarily transported to a paediatric specialist centre where they subsequently died, arriving at the specialist centre more than four hours after the incident in both cases. As these three missed cases were not treated by the reporting P-HEMS service they are not included in the present study. Additionally one of the four patients identified for P-HEMS response was managed by a NSW Ambulance physician team and is also not reported.

Neurological outcome

No child with a GCS greater or equal to eight on arrival of the P-HEMS had neurological impairment at the time of follow up. Outcomes of all children with GCS less than eight are shown in Table 4. All twenty two children with GCS below eight were intubated and ventilated at the accident scene by the P-HEMS team, ten (45 %) requiring adjuvant anaesthesia. Of four children who survived with neurological impairment, three had a GCS of three on arrival of the P-HEMS team. One child however, who suffered an asystolic cardiac arrest, survived and had a normal neurological examination as assessed by a paediatric rehabilitation specialist 18 months after the incident (PCPC =1).

Age	Asystole first reported rhythm	Initial GCS	Cardiac output on ED arrival	Outcome	PCPC score	Outcome Comments
5.3	Y	3	N	Died ED	6
1.1	Y	3	N	Died ED	6
1.5	Y	3	N	Died ED	6
3.5	Y	3	N	Died ED	6
2.0	Y	3	N	Died ED	6
2.5	Y	3	N	Died day 4	6	MRI- multiple areas of infarction involving brain stem, deep nuclei and cortex
2.8	Y	3	Y	Died day 2	6	Bradycardic on ED arrival, MRI- diffusely swollen brain, transtentorial and foramen magnum herniation, no arterial or venous flow
6.4	Y	3	Y	Died day 2	6	MRI extensive cerebral and cerebellar oedema with BG and brainstem involvement. Tonsillar herniation. No flow on MRA
7.7	Y	3	Y	Died day 2	6	CT complete loss grey- white matter differentiation, herniation of cerebellar tonsils
1.2	Y	3	Y	Died day 9	6	MR bilateral globus pallidus, hippocampal and bilateral cortical areas of ischaemic injury. No purposeful movements, occasional breaths, intensive care withdrawn.
3.5	Y	3	Y	Died 17 months	6	At 3 months - severe spastic quadriplegia with bulbar palsy, incomprehensible sounds. Died at 17 months
1.3	Y	3	Y	Neurological impairment	4	At 6 months - significant dystonia, trunk hypotonia, epilepsy
1.0	Y	3	Y	Normal	1	CT - early cerebral oedema, MRI no diffusion restriction. Early seizures on PICU. At 18 months -normal neurological and developmental exam
1.7	N	3	Y	Neurological impairment	3	Pre accident diagnosed Aspergers and ADHD. At 1 month – right sided weakness, loss of language. At 3 ½ years - distractible, special help at school
2.2	N	5	Y	Normal	1	At 12 months -ahead in development
2.8	N	5	Y	Normal	1	At 11 months -Normal function
1.2	N	5	Y	Normal	1	Tonic clonic seizures in PICU. At 1 yr 5 months development normal.
2.1	N	6	Y	Normal	1	No neurological deficit at discharge
3.9	N	6	Y	Normal	1	At 7 months normal behavior and skills
9.9	N	6	Y	Baseline*		*Had seizure disorder, autism, ADHD, moderate –severe disability before accident. Returned to baseline state.
1.3	N	7	Y	Neurological impairment	3	At 2 yrs - behaviour problems, aggression, repetitive movements/routines
1.9	N	7	Y	Normal	1	No neurological deficit at discharge

*Patient not neurologically normal at follow up but had returned to pre-incident baseline

Survival and disability rates are summarized in Fig. 1 according to the Utstein reporting template.

There is little previously published data on the role of P-HEMS in paediatric drowning cases and their utility in this context has yet to be established. There was a very high intervention rate in severely injured (ISS > 15) drowning cases in this series treated by the P-HEMS. All of these patients were intubated with 45 % requiring adjuvant anaesthesia. Consensus guidelines on management of drowning [5] emphasise early restoration of ventilation and circulation. Activation of teams able to provide anaesthesia assisted intubation and direct transport to specialist paediatric centres is therefore consistent with current best practice. P-HEMS are ideally constituted to deliver both advanced airway interventions and direct transport [6]. Definitive evidence that these interventions improve outcome is however not yet available. A large multicenter trial would be required as individual EMS systems are highly unlikely to encounter adequate patient numbers to make such a study feasible.

The previous study of the paediatric trauma case identification system used in Sydney indicated that direct identification by the P-HEMS service significantly outperformed paramedic identification from a central control room, and had profound effects on the overall performance of the Sydney prehospital trauma system [6]. Similarly whilst the P-HEMS case identification system was in operation in this study, all paediatric drowning cases with an ISS > 15 were identified for P-HEMS response and were transported directly to a paediatric specialty centre. After the NSW Ambulance withdrew access by the P-HEMS to the CAD system only four of seven cases were identified. Two of the three non-identified cases were transported by road paramedics to adult trauma centres with delays of several hours before transfer to a paediatric specialist centre. This indicates that further investigation of the effects of cessation of the P-HEMS case identification system should be undertaken and reinstatement of the parallel identification model may be warranted. Beyond the case identification process the system worked well, with all severe drowning cases identified for P-HEMS response receiving intubation and direct transport to a paediatric specialist centre.

The P-HEMS was also very accurate in identifying high risk patients for respiratory complications in this small series. No child that was not intubated during the prehospital phase of care subsequently required intubation in the hospital.

All children with a GCS over 3 on arrival of the P-HEMS survived. Twenty five percent of children who suffered a cardiac arrest survived which is a relatively high survival rate for out of hospital cardiac arrest. For all causes of reported paediatric out of hospital cardiac arrest survival rates range from 6.4 % to 12 % [9–12] with neurologically intact survival in half of these patients [9, 12, 13]. Arrested drowning victims are reported to have better survival rates than the all-cause arrest group. In a pooled review of 41 studies drowning associated arrests had a 22.7 % survival [9]. This is similar to the current series and emphasises the need to aggressively resuscitate paediatric drowning victims who have suffered cardiorespiratory arrest in the pre hospital setting as individual outcomes cannot be predicted.

No child survived who did not have a return of spontaneous circulation before admission to the emergency department. In all causes of paediatric cardiac arrest in a series of over 200 children in Melbourne, Australia less than 1 % survived if spontaneous circulation remained absent on arrival at hospital [10]. In a large US study of 599 out of hospital paediatric cardiac arrests the longest duration of CPR in a survivor with a good neurological outcome was 42 min [14] and a series of children who drowned reports no children receiving more than 25 min CPR having a good outcome [15]. A recent nationwide study of paediatric drowning outcome in the Netherlands found that no child who received CPR for more than 30mins without return of spontaneous survived with a good neurological outcome [16]. Prehospital management is therefore critical and again would suggest a role for P-HEMS in this population.

The decision surrounding cessation of cardiopulmonary resuscitation after drowning has been complicated by case reports of survival after long periods of both submersion [15] and CPR [17, 18, 19] in hypothermic patients. It is likely that hypothermia is only protective if it occurs before irreversible hypoxic-ischaemic cerebral injury which is less likely to occur in Australian water conditions compared to cooler areas of Europe or North America.

The rates of bystander CPR in this series were high, suggesting community education programs are having effect. All but one child with a GCS less than 8 on arrival of the P-HEMS had received bystander CPR, and in all asystolic children CPR was underway. Bystander CPR is associated with improved neurological outcome in children admitted to ED after drowning [20]. In cases of cardiopulmonary arrest secondary to drowning, positive pressure ventilation via mouth or mask is required in addition to chest compressions for resuscitation to be effective. In all causes of paediatric out of hospital arrest rates of favorable outcome increase if breaths are added [13].

An initial GCS above 8 at the scene was associated with full recovery. However a GCS below 8 did not predict a poor outcome, seven out of eight children with initial GCS between four and seven had good neurological outcomes. This is in accordance with other studies which also show initial neurological parameters such as low GCS, lack of response to pain and lack of pupillary reaction do not necessarily predict poor neurological outcomes [21, 22].

One child in this series survived neurologically intact after a period of asystolic cardiopulmonary arrest. Normal survival after cardiopulmonary arrest due to drowning has been well described [23] and in a meta-analysis of 442 drowning associated arrests the intact survival rate was 6 %, putting it well above rates for all other causes of cardiopulmonary arrest [9].

P-HEMS played a significant role in the management of severe paediatric drowning in this case series. Requirement for interventions that are only provided by P-HEMS in the Sydney EMS system, specifically anaesthesia facilitated intubation, were high and all cases identified for P-HEMS response were transferred directly to a paediatric specialist centre. Discontinuation of the P-HEMS direct case identification system that operated during the majority of the study period resulted in deterioration in system performance with some paediatric drowning patients not identified for P-HEMS response being transported to adult hospitals.