This study is a large multi-centre collaboration between a busy regional paediatric intensive care transport service (Children's Acute Transport Service, CATS), four large paediatric intensive care units (PICUs at Great Ormond Street Hospital, St Mary's Hospital and Royal London Hospital in London, and Addenbrookes Hospital in Cambridge) and the Department of Paediatrics at Imperial College, London. CATS transports over 800 sick children on life support to the three PICUs each year.

We aim to improve our understanding of the genetic basis and biological pathways by which children with acute infection or injury become critically ill and develop failure of vital organs, and how these factors influence outcome. We will establish well-characterised cohorts of sick children with diverse pathologies, in whom blood, urine and other samples will be collected at an early stage of critical illness. Samples will be analysed using genomic, transcriptomic, proteomic and metabolomic approaches. Advanced bioinformatics techniques will be used to identify biomarkers for early diagnosis and robust risk stratification. We will focus on biomarkers to help distinguish between serious bacterial infections, viral infections and other causes of critical illness; diagnose incipient organ failure; and accurately identify, early on, children at high risk of developing a poor outcome.

We will recruit critically ill children at first contact with the CATS team, during emergency transport to PICU. Due to the emergency nature of the research, and minimal risk associated with the study procedure, we will seek deferred, written informed consent from parents/guardians once their child has been stabilised, within 24-48 hours following PICU admission.

By studying these important questions, we aim to better understand how we can diagnose and provide early life-saving treatments to critically ill children. The research team have an established track record of successful completion of several large clinical studies in critical care as well as validation of biomarkers in other diseases.

This is a prospective cohort study involving the collection of clinical data and biological samples. No interventions will be performed on patients.

Sample collection Blood and urine samples will be collected at two time points. Stool and throat swabs will also be collected if possible.

Timepoint 1: During stabilisation by the CATS team (first contact with an intensive care team) The CATS team will collect samples during routine sampling, prioritised in the following order: DNA (2 ml); RNA (2 ml); Serum (2 ml); Plasma (2 ml) and Urine (10 ml).

Timepoint 2: Within 24-48 hours of PICU admission The research team in the participating PICU will collect samples during routine sampling, prioritised in the following order: DNA (2 ml); RNA (2 ml); Serum (2 ml); Plasma (2 ml) and Urine (10 ml). In addition, stool and throat swab samples will be collected on the PICU.

In infants less than 10 kg in weight, we will restrict total blood volume collected at each time point to 0.8 ml/kg body weight, in line with MCRN Clinical Trials Guidelines, 25 July 2008.

In addition, faecal samples will be collected on admission to PICU (day 1) and 24-48 hours after PICU admission. If faeces are not available, an anal swab will be taken for microbiological analysis. SCFA will be measured on the first available faecal sample.

Sample transport, processing and storage Samples will be transported by the CATS team to the admitting PICUs in cool boxes (4°C). Sample processing and storage will occur at the hospital where the patient is admitted. Processing will include anonymisation of the sample by generating a unique study number, which will be used for all references to the samples thereafter.

Clinical data collection A standardised case report form and data collection manual with definitions and rules will be produced. Data will be collected by research staff at each unit. Clinical data will cover the key stages of the patient's pathway - intensive care referral and transport, PICU course and outcome. We will perform detailed clinical phenotyping of the patient, including reason for PICU referral/admission (infection, trauma, neurological, cardiac, respiratory and other), severity of illness (PIM-2 score), severity of organ failure (PELOD) and outcome at discharge from PICU (death, severe disability). Recognising that short-term outcome at PICU discharge may not be the most meaningful outcome for parents/carers, or provide an accurate picture of long-term outcome in children, we will also seek parents/carers' consent to be contacted 1 year after PICU admission to complete short questionnaires to assess the child's functional status and quality of life. Questionnaires will be either administered by telephone, email or web-based means, depending on parental preference.

Where possible, in order to avoid duplication of effort, clinical data from CATS retrieval and PICU admission will be obtained by electronic linkage from the paediatric intensive care national clinical audit (PICANet, University of Leeds and Leicester). Data linkage will be performed using the CATS retrieval number, at the end of the study period. Collection of these data will follow a standard procedure - any transfer of data (requests for data and the return of the full dataset) will be performed securely (with full encryption).

Data analysis Initially, each type of information, i.e. genetic, transcriptomic, proteomic, epigenetic, clinical will be analyzed separately to identify associations and correlations at a first stage. Differential behaviour and association patterns will be derived by various statistical and bioinformatic analyses, both univariate and multivariate, as described above. At a second stage, information will be combined across datatypes into predictive models. A systems biology approach will be employed in which networks of co-regulated genetic, transcriptomic, proteomic and epigenetic modules will be constructed in both case and control cohorts. Network connections which will be different between cases and controls will be identified as "predictors" of the outcome.

Graphical modelling will be used to visualize those complex networks. Heavy statistical analysis will be required to identify the most influential connections that once removed they would have the greatest impact on the network topology. Models will be refined, re-evaluated, validated and replicated across the different disease cohorts until the best possible out of sample classification is achieved. The expertise of the mathematical and statistical group will provide the ability to link the multiple layers of biological information from DNA to phenotype into a network systems biology approach.

Sample size No formal sample size calculations are possible but we will recruit a minimum number of patients in each group of interest. We aim to recruit adequate numbers of patients with different reasons for intensive care admission into the study (50-80 patients per group of interest).

Study duration Patient recruitment will take place over a 24 month period, with a 6 month lead in time for staff training and 6 months lead out time for patient follow up and completion of databases.

Patient recruitment Eligible patients will be recruited to the study during transport by medical and nursing staff from the CATS intensive care transport team.

Consenting procedure We will seek deferred written informed consent from parents/guardians once their child's condition stabilises, usually within 24-48 hours following PICU admission. The consent process will be supported by Patient Information Leaflets. There are several precedents for this approach in emergency research, including in paediatric intensive care. Support for deferred consent in urgent life-threatening situations is also provided through the latest EU Regulation on Clinical Trials (2013).

The main reasons for seeking deferred consent in this study are:

1. Clinical: Emergency intensive care transport of children represents an urgent situation where life-saving treatment cannot be delayed and transport to a PICU needs to be undertaken rapidly. In addition, parents/guardians may not be present at the time of stabilisation to discuss consent.
2. Practical: Taking an additional staff member for the transport purely for research consent is not feasible due to the limited space in the ambulance. The 24/7 nature of the CATS service and the fact that multiple transports occur concurrently also makes having dedicated research staff impractical.
3. Methodological: Difficulties in obtaining full informed consent may result in selection bias being introduced. In addition, if study samples are not taken at an early stage of critical illness, before treatments (such as fluids or inotropes) are initiated, deriving robust biomarkers of critical illness may be flawed. While methodological issues cannot provide an argument for including data when research subjects expressly deny consent, it does make an ethically valid case for including samples or data collected before consent has been taken, and where such explicit denial of consent does not exist.
4. Ethical: Due to the difficulties of obtaining informed consent in emergency situation, the sickest patients are often subjected to anecdotal medical practice. It may be unethical to deny sick children the benefits of new research and therapeutic strategies based on the most robust scientific understanding of critical illness, which is what our study aims to advance.

If parents/guardians do not wish to consent or are unable to consent, or staff are unable to make contact with parents/guardians, the samples will be discarded and no clinical data will be collected.

Inclusion Criteria:

• Critically ill children aged 0-16 years requiring emergency transfer by the Children's Acute Transport Service (CATS) to participating paediatric intensive care units
• Presence of an indwelling catheter (arterial and/or venous) for blood sampling
• Presence of a urinary catheter for urine sample collection.

Exclusion Criteria:

• CATS transfers to other intensive care units or non-intensive care destination units
• Premature newborns (under 36 weeks corrected gestational age)
• Children with a known do-not-resuscitate (DNR) order in place.

Critically ill children transported to an intensive care unit in an emergency