Orestis Papatryfonos 1, Anna Largajolli 2, Li Qin 2, Michael Chappell 1, Amy Cheung 2
1 School of Engineering, University Of Warwick (Coventry, United Kingdom), 2 Certara Limited UK, Level 2-Acero, 1 Concourse Way (Sheffield, United Kingdom)
Objectives: Model-Based Meta-Analysis (MBMA) is a quantitative method that combines aggregated data from multiple randomised clinical trials (RCTs) to improve statistical power and characterise dose-response relationships, time courses, and treatment effects across trials. Evaluating the performance of MBMA in the context of paediatric extrapolation has the potential to support dose selection, particularly when paediatric data are sparse, clinical or mechanism-of-action understanding is limited, and RCTs are challenging to conduct. The objectives of this work were to 1) identify a postoperative pain indication that includes both paediatric and adult studies reporting common efficacy endpoints and treatments; 2) curate a paediatric analysis database; 3) develop an MBMA framework to characterise the selected postoperative pain indication in paediatric patients. This work was conducted within the Horizon EU-funded ERAMET project.
Methods: A systematic PubMed search of postoperative pain studies was performed according to a predefined protocol, following the relevant identification, screening, and eligibility assessment steps described in the Cochrane Handbook for Systemic Reviews of Interventions [1] and reported items in accordance with the PRISMA statement [2]. The pain indication that had the highest number of reported paediatric studies, was also reported in the adult population, and involved common treatments and efficacy endpoints across both populations was selected. A paediatric-only reference database was then constructed for the selected indication, and the analysis dataset was created by standardising variables across studies.
The modelling strategy proceeded in stages. First, a subset of studies including both placebo and active treatment arms was analysed separately to investigate potential time and dose relationships. Subsequently, the model structure was extended to include all eligible studies within the MBMA framework. All models were fitted in R using generalised least squares via the gnls function from the nlme package [3].
Results: The first systematic search identified 770 paediatric publications addressing postoperative pain, for several pain indications. Tonsillectomy was selected as the indication of interest due to its presence across both adult and paediatric populations, and the availability of common treatments and efficacy endpoints between populations. Specifically, 36 paediatric and 35 adult tonsillectomy studies were identified. The most frequently reported pain scale across both populations was Visual Analogue Scale (VAS), and paracetamol was among the most studied drugs. Eight eligible paediatric studies [4-11] were selected for curation in the MBMA dataset. Extracted variables included age, weight, dosing regimen, time-course VAS pain scores, and key trial design characteristics.
The initial model evaluation was performed using two studies that contained both placebo and active treatment arms to investigate potential time course and dose–response relationships. The placebo response was modelled non-parametrically using an unstructured time-specific approach for each study, implicitly capturing variability over time and across trials, while the drug effect was modelled parametrically. Neither of the evaluated time functions (exponential onset or time-fraction model) was supported by the data, consequently, the time effect was excluded from the final model structure. In contrast, a clear dose-response relationship was identified, parameterised as an Emax model, for the two studies, that resulted in stable parameter estimation results. The maximum paracetamol treatment effect (Emax) corresponded to a 0.97-unit reduction on the logit scale of the pain score, with the paracetamol dose leading to half of the maximum effect, (ED50) estimated at 7.72 mg/kg. The model described both placebo and paracetamol treatment arms, with good visual agreement between observed and predicted responses.
The same Emax model was then applied to all eight studies, including six studies without placebo arms, within the MBMA framework. For the full dataset, parameter estimates obtained were Emax = 0.73 logit units (132% RSE) and ED50 = 6.01 mg/kg (70% RSE). Parameter estimates remained stable and comparable. The Emax model consistently described the dose-response relationship across studies and provided robust fits across all treatment arms.
Conclusions: An MBMA framework was successfully developed to characterise the dose-response relationship of paracetamol in paediatric tonsillectomy pain. Future work will focus on expanding the curated tonsillectomy dataset, by including adult references to allow investigation of MBMA extrapolation capabilities, considering different drug classes, combining different pain scales definitions and potentially including different indications.
References:
[1] J. P. T. Higgins and S. Green, Eds., Cochrane Handbook for Systematic Reviews of Interventions, ver. 5.0.2, updated Sept. 2009. The Cochrane Collaboration, 2009.
[2] D. Moher, A. Liberati, J. Tetzlaff, D. G. Altman, and The PRISMA Group, “Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement,” PLoS Med., vol. 6, no. 6, p. e1000097, 2009, doi:10.1371/journal.pmed1000097.
[3] J. Pinheiro, D. Bates, and R Core Team, nlme: Linear and Nonlinear Mixed Effects Models, R package ver. 3.1-167, 2025, doi:10.32614/CRAN.package.nlme.
[4] A. Yenigun et al., “Intranasal diclofenac sodium, ibuprofen and paracetamol for pain relief after pediatric tonsillectomy,” Int. J. Pediatr. Otorhinolaryngol., vol. 198, p. 112552, 2025, doi:10.1016/j.ijporl.2025.112552.
[5] A. Jotić et al., “Pain management after surgical tonsillectomy: Is there a favorable analgesic?,” Ear Nose Throat J., vol. 98, no. 6, pp. 356–361, 2019, doi:10.1177/0145561319846065.
[6] M. Sener et al., “Paracetamol versus dipyrone by IV patient-controlled analgesia after tonsillectomy in children,” Braz. J. Anesthesiol., vol. 65, no. 6, pp. 476–482, 2015, doi:10.1016/j.bjane.2013.09.010.
[7] R. Soltani et al., “Lavender aromatherapy and post-tonsillectomy pain in children: A randomized trial,” Int. J. Pediatr. Otorhinolaryngol., vol. 77, no. 9, pp. 1579–1581, 2013, doi:10.1016/j.ijporl.2013.07.014.
[8] G. A. Dashti et al., “Rectal acetaminophen and postoperative pain after adenotonsillectomy,” Middle East J. Anaesthesiol., vol. 20, no. 2, pp. 245–249, 2009.
[9] J. P. H. Hamers and H. Huijer Abu-Saad, “Children’s pain at home following (adeno)tonsillectomy,” Eur. J. Pain, vol. 6, no. 3, pp. 213–219, 2002, doi:10.1053/eujp.2001.0326.
[10] H. Y. Uysal et al., “IV paracetamol versus tramadol after adenotonsillectomy in children,” J. Clin. Anesth., vol. 23, no. 1, pp. 53–57, 2011, doi:10.1016/j.jclinane.2010.07.001.
[11] S. Ozlugedik et al., “Honey for pain after tonsillectomy: A randomized placebo-controlled study,” Int. J. Pediatr. Otorhinolaryngol., vol. 70, no. 11, pp. 1929–1934, 2006, doi:10.1016/j.ijporl.2006.07.001.
Reference: PAGE 34 (2026) Abstr 11992 [www.page-meeting.org/?abstract=11992]
Poster: Drug/Disease Modelling - Paediatrics