External and construct validity of a standardised measure of neonatal noxious-evoked brain activity.

2523 

Abstract Submission 

Roshni Mansfield^1,2, Marianne Aspbury¹, Luke Baxter¹, Aomesh Bhatt¹, Maria Cobo^1,3, Caroline Hartley¹, Annalisa Hauck¹, Simon Marchant¹, Vaneesha Monk¹, Kirubin Pillay¹, Ravi Poorun^4,5, Marianne van der Vaart¹, Rebeccah Slater^1,6

¹Department of Paediatrics, University of Oxford, Oxford, United Kingdom, ²Newborn Care Unit, John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom, ³Universidad San Francisco de Quito USFQ, Colegio de Ciencias Biologicas y Ambientales, Quito, Ecuador, ⁴College of Medicine & Health, University of Exeter, Exeter, United Kingdom, ⁵Children’s Services, Royal Devon & Exeter NHS Foundation Trust, Exeter, United Kingdom, ⁶Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom

Roshni Mansfield, BA, BM BCh, MRCPCH  
Department of Paediatrics, University of Oxford|Newborn Care Unit, John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust
Oxford, United Kingdom|Oxford, United Kingdom

Marianne Aspbury, DPhil  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Luke Baxter, DPhil  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Aomesh Bhatt, MRCGP  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Maria Cobo, DPhil  
Department of Paediatrics, University of Oxford|Universidad San Francisco de Quito USFQ, Colegio de Ciencias Biologicas y Ambientales
Oxford, United Kingdom|Quito, Ecuador

Caroline Hartley, PhD  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Annalisa Hauck, MD  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Simon Marchant, MSc  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Vaneesha Monk, PhD  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Kirubin Pillay, DPhil  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Ravi Poorun, MBPhD  
College of Medicine & Health, University of Exeter|Children’s Services, Royal Devon & Exeter NHS Foundation Trust
Exeter, United Kingdom|Exeter, United Kingdom

Marianne van der Vaart, DPhil  
Department of Paediatrics, University of Oxford
Oxford, United Kingdom

Rebeccah Slater, PhD  
Department of Paediatrics, University of Oxford|Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford
Oxford, United Kingdom|Oxford, United Kingdom

Electroencephalography (EEG) can be used to measure neonatal noxious-evoked brain activity. Brain-derived measures of acute nociception are more objective than neonatal clinical pain scores, and clinical trials of neonatal analgesics would benefit from the improved reliability of these objective metrics. The external and construct validity of brain-derived measures must be demonstrated to support their use in clinical trials. The noxious neurodynamic response function (n-NRF) is an EEG-derived measure of neonatal noxious-evoked brain activity, developed in term neonates in Oxford, UK (Hartley et al. 2017). This study aimed to test the external and construct validity of the n-NRF in two independent datasets from University College London Hospital (UCL) and Royal Devon & Exeter Hospital.

The UCL dataset is publicly available (Jones et al. 2018). The Exeter dataset was collected from 2021 to 2023 (NHS ethics: 12/SC/0447, 21/LO/0523). Both datasets include term and preterm infants. Heel lances were clinically required. For control heel lances, the heel lance device was rotated 90 degrees so that the blade pierced the air. Stimuli were time-locked to the EEG as described previously (Worley et al. 2012). Data were acquired using Compumedics Grael V2 EEG System and recorded using CURRYscan8 neuroimaging suite, with eight electrodes on the scalp (Cz, CPz, C3, C4, FCz, T3, T4, Oz), a reference (Fz), ground (forehead) and ECG (left clavicle) electrode. EEG data from Cz electrode were pre-processed using an EEGLAB-based pipeline, including filtering (1Hz to 30Hz, 50Hz notch filter), epoching (0.5s pre- to 1s post-stimulus), and baseline correction to the pre-stimulus mean. Cz EEG data were regressed onto the n-NRF to calculate n-NRF magnitudes. We first aimed to reproduce the finding that the n-NRF magnitude is significantly larger following a clinical heel lance compared to a control heel lance (Hartley et al. 2017). We then tested whether the mean n-NRF magnitude (scaled to an average heel lance magnitude of 1.0 in the original Oxford dataset (Hartley et al. 2017)) is equivalent to 1.0±0.2 in independent cohorts. Lastly, we assessed whether the n-NRF magnitude following a heel lance increases with postmenstrual age (PMA) in preterm infants (Schmidt Mellado et al. 2022). The study protocol was peer-reviewed and pre-registered with the journal Cortex (https://doi.org/10.17605/OSF.IO/ZY9MS).

The n-NRF magnitude was significantly greater following a heel lance than a control heel lance in both the UCL (n=60; mean difference (MD) 0.88; 95% confidence interval (CI) [0.64, 1.13]; one-tailed paired t-test p<0.0001) and Exeter datasets (n=31; MD=0.31; 95% CI=[0.02, 0.61]; p=0.02) (Figure 1). Using the TOST (two one-sided t-tests) procedure, we did not confirm equivalence of the n-NRF heel lance magnitude to 1.0 in either dataset (UCL: n=72, mean=1.33, 90% CI=[1.18, 1.52]; Exeter: n=35, mean=0.92, 90% CI=[0.74, 1.22]). An exploratory meta-analysis (Figure 2) including Oxford pilot data, UCL and Exeter data, showed significant heterogeneity between sites (random effects model mean n-NRF magnitude=1.10, 95% CI=[0.57, 1.62], X2=7.86, p-value=0.02). There was significant positive correlation between PMA up to 37 weeks and heel lance n-NRF magnitudes (n=65; one-sided Pearson's R, adjusted for site: 0.24; 95% CI=[0.06, 1.00]; p=0.03).

We reproduced two findings: the n-NRF magnitude was larger following a noxious than a non-noxious stimulus; the n-NRF magnitude increased with PMA up to 37 weeks. We did not reproduce equivalence of the heel lance n-NRF magnitude to 1.0 at other sites; we showed significant inter-site heterogeneity. This may relate to site differences in equipment, research staff, and study populations. Replicating these findings provides evidence supporting the external and construct validity of the n-NRF. However, if used for multisite trials, adjusting for site effects is required to account for inter-site differences in mean magnitude.

EEG ²

Perception: Pain and Visceral ¹

Electroencephaolography (EEG)

Pain

PEDIATRIC

Pre-registration

Other - Replication study; infants; neonates