Discussion

This study demonstrates that this population of extracorporeal life support survivors accrued significant morbidity during the initial hospitalisation followed by improvement in Functional Status Scale throughout follow-up. In our population, younger age, genetic syndrome, and evidence of neurologic injury (electroencephalogram-determined seizures and central nervous system diffuse ischaemic changes) were each individually associated with an increase in the Functional Status Scale by ≥3 from admission to discharge. Children demonstrated improvement in the 12-month follow-up period, which was predominantly driven by improvement in the feeding domain. Functional Status Scale scores in our population increased from admission to discharge but then significantly improved from discharge to follow-up within 12 months.

Our finding of significant neurologic morbidity during hospitalisation was similar to findings previously reported. ^{Reference Kramer, Mommsen, Miera, Photiadis, Berger and Schmitt16,Reference Hervey-Jumper, Annich, Yancon, Garton, Muraszko and Maher17} These studies showed that 5.5–8.4% of patients experienced seizures during their hospitalisation, similar to the 5% of patients who experienced seizures in our population. ^{Reference Kramer, Mommsen, Miera, Photiadis, Berger and Schmitt16,Reference Hervey-Jumper, Annich, Yancon, Garton, Muraszko and Maher17} In contrast, Okochi and colleagues found a significantly greater proportion of patients experienced seizures detectable by continuous electroencephalogram, 23%. ^{Reference Okochi, Shakoor and Barton18} Okochi and colleagues specifically only included patients with at least 24 hours of continuous electroencephalogram monitoring while on extracorporeal life support, whereas in our population, a continuous electroencephalogram was only performed for a clinical concern. Thus, it is possible that a greater number of children experience seizures during their extracorporeal life support hospitalisation than are detected by routine clinical practice.

In contrast to our study that found that neurologic morbidity was the most common among survivors of extracorporeal life support, Beshish and colleagues found cardiovascular complications as the most common (59%) with acquired neurologic morbidity in just 14%. ^{Reference Beshish, Rodriguez and Hani Farhat14} Overall, Beshish and colleagues ^{Reference Beshish, Rodriguez and Hani Farhat14} also found a higher rate of new morbidity, with 40% of patients developing a new morbidity during their hospitalisation, whereas we found 24%. This may be secondary to our population being older. The median age in our study was 1.7 years, and theirs was just 3.1 months. Beshish and colleagues also found that longer duration of the extracorporeal life support run has previously been associated with worse functional outcomes. ^{Reference Beshish, Rodriguez and Hani Farhat14} We saw a trend in increased duration on extracorporeal life support being associated with an Functional Status Scale ≥3 from admission to discharge, but this did not reach statistical significance, which may be a consequence of our smaller sample size.

As an additional marker of morbidity, we investigated the readmission rate of our patient population within the first year of hospital discharge. A majority of our patients were readmitted within the first year, 73%. This is similar to previous studies, which have found that 36–62% of paediatric extracorporeal life support survivors required hospital readmission after their initial discharge. ^{Reference Jen and Shew3,Reference Lawrence, Sebastião, Deans and Minneci19} Specifically, we found that children who developed a new morbidity during their admission also had a longer cumulative hospital day burden in their first year post-discharge than children who did not acquire a new morbidity. Similarly, children with an increased number of complex chronic conditions have also been found to have a higher readmission rate. ^{Reference Lawrence, Sebastião, Deans and Minneci19}

While a substantial proportion of patients developed new morbidity during their hospitalisation, there was significant improvement in the Functional Status Scale score at one year. This is consistent with previous literature that has shown that most newborn infants treated with extracorporeal life support will have a normal neurodevelopmental assessment at 11–19 months of age. ^{Reference Khambekar, Nichani and Luyt20} However, this does not preclude children from developing neuropsychological impairments and school problems in longer-term follow-up, known as the growing into deficit phenomenon. ^{Reference Schiller, IJsselstijn and Hoskote21}

Perhaps the most meaningful impact of our findings is the improvement in the feeding Functional Status Scale score, which represents that a significant portion of children improved from requiring tube feeds to being able to complete age-appropriate feeding. Previous literature has found that a majority of children (87%) who are tube-fed have been found to experience at least one adverse effect from tube-feeding, such as nausea, vomiting, retching, and gagging. ^{Reference Pahsini, Marinschek, Khan, Dunitz-Scheer and Scheer22} The burden on caregivers is also significant. Though caregivers of children with a gastrostomy tube (G-tube) have not been found to be more depressed than caregivers of children without G-tubes, ^{Reference Heyman, Harmatz and Acree23} a substantial portion of parents (23%) describe a feeding as an anxious or intrusive experience. ^{Reference Pahsini, Marinschek, Khan, Dunitz-Scheer and Scheer22} Thus, this improvement during outpatient follow-up is impactful for both patients and their families.

The study has several limitations; its generalisability is limited given the inherent nature of the single-centre study. Some associations may be absent in our analysis given our sample size but may be exposed in larger, multi-centre studies. Additionally, Functional Status Scale scores have been previously shown to be easily reproducible with minimal interobserver variability ^{Reference Pollack, Holubkov and Glass12} ; however, when obtained retrospectively, they rely upon the documentation of clinical providers, and thus subtle findings may not have been detected. Our study used a standardised approach for transferred neonates, as reported in previous literature ^{Reference Beshish, Baginski, Johnson, Deatrick, Barbaro and Owens13,Reference Beshish, Rodriguez and Hani Farhat14} ; however, this approach does not account for pre-existing functional deficits, particularly in patients found to have a genetic syndrome.