Considerations in paediatric sample selection
Sample selection and sample size determination in paediatric trials require carefully considering the developmental heterogeneity within the paediatric population and the ethical imperative to include children who are likely to benefit while protecting those at greatest risk. The following concepts are important to consider when selecting the paediatric sample for a clinical trial.
Age groups
The ICH E11 guideline defines the following paediatric age groups, which are widely used in regulatory submissions and trial design: preterm neonates; term neonates (0–27 days); infants and toddlers (28 days–23 months); children (2–11 years); and adolescents (12–17 years). These groupings reflect key developmental transitions that influence drug disposition, disease presentation and the ability of children to participate in clinical procedures. Trials should be designed to enrol appropriate age groups based on the intended clinical use of the product and separate analysis strata or substudies may be needed for different age groups within a single trial. All age groups are to some extent arbitrary and need to be justified according to the product.
Age de-escalation
Age de-escalation means evaluating drugs in sequential age cohorts of children, starting with adolescents and then working down in age into children, infants and neonates. The rationale for de-escalation is to move incrementally from the most physiologically mature children, enabling caution regarding drug exposure and safety. Age de-escalation has historically been the primary approach, aiming to balance the scientific need to study drugs across all distinct paediatric age groups against the ethical imperative to protect younger, more vulnerable children from unknown risks. However, age de-escalation effectively amounts to performing multiple trials in series and thus adds to delays in establishing the dosing of new agents for young children. For this reason, over the past decade, there has been a push to study drugs simultaneously across wider age groups or weight bands. Paediatric HIV research has led the way in promoting the inclusion of children of all ages as soon as an appropriate dose is known, with trials designed to start across age groups while enabling delays for the youngest children when age-appropriate formulations or dosing are not yet available. Increasing the accuracy of first dose and advances in physiologically based pharmacokinetic modelling have increased both the efficiency and safety of studying drugs for children.
Weight bands
Weight bands are predefined weight ranges (such as 3–6 kg and 6–10 kg), each assigned a fixed dose designed to translate precise pharmacokinetic dosing (typically mg/kg or allometric scaling) into a system that is practical to deliver in real-world settings in which individualized calculations are not feasible. Rather than calculating an exact dose for every child, weight bands group children into ranges that receive a standardized dose validated to achieve adequate exposure across that range, reducing dosing errors and improving adherence, especially in high-volume or resource-limited settings such as HIV, tuberculosis and malaria programmes for children.
Because a poorly designed band can under-dose children at the lower end or over-dose those at the upper end, weight bands are developed through population pharmacokinetic modelling and simulation to ensure that safe, adequate exposure is maintained across the full range. WHO and GAP-f have also worked to harmonize weight bands across therapeutic areas, since children treated for multiple conditions previously faced different, disease-specific dosing charts.
A harmonized approach to weight-band dosing across disease areas could substantially streamline prescribing for young children, although implementation would require coordinated revision of existing dosing recommendations across multiple therapeutic guidelines. This has direct implications for designing paediatric clinical trials, which should, when possible, adopt weight-band dosing structures that are consistent with the emerging harmonized framework to facilitate real-world implementation of trial findings. WHO issued updated guidance on harmonized, weight band–based dosing schedules for antiretroviral therapy for neonates, infants and children living with HIV and is working with partners to scale up implementation on the ground, given its relevance across therapeutic areas.
Special paediatric populations
Certain paediatric populations require additional methodological and regulatory considerations beyond those applied to the general paediatric trial population, given their smaller numbers, altered physiology or increased vulnerability. Table 3 summarizes the key challenges and design considerations for children with rare diseases, comorbidities and chronic disease.
Table 3. Key challenges and design considerations for special paediatric populations
| Population | Key challenge | Design and regulatory considerations | Key consideration |
|---|---|---|---|
| Children with rare diseases | Total eligible patient population may number only in the hundreds or thousands worldwide or even less than 100 for ultra-rare diseases | Multinational, multicentre designs; adaptive and innovative trial designs; regulatory frameworks specific to rare disease drug development (EMA orphan designation, United States Food and Drug Administration rare paediatric disease designation) | Engagement with patient registries and natural history studies is essential for understanding the available population and informing trial design |
| Children with comorbidities | Comorbidities may affect drug metabolism, treatment response and adverse event risk | Standard adult-derived exclusion criteria may need to be adapted to avoid excluding children representative of the eventual target population | Concomitant medications – which may interact with the investigational product – should be carefully documented and analysed |
| Children with chronic disease | Conditions such as epilepsy, HIV, tuberculosis or malnutrition may alter pharmacokinetics and risk profiles compared with healthy children | Long-term trial design must account for altered pharmacokinetic (across age groups) and risk profiles specific to the chronic disease population, which may differ by ethnicity | Must also consider the trial’s impact on normal childhood development, schooling and family functioning |
Approaches to calculating sample size
Calculating sample size in later-stage paediatric trials follows the same statistical principles as in adult trials – specifying the primary endpoint, the expected effect size, the desired power and the significance level – but is often constrained by the limited size of the paediatric population, especially for rare diseases or young age groups. When standard frequentist sample size calculations would require an infeasibly large number of participants, alternative approaches include:
- Bayesian adaptive designs that use accumulating data to update sample size estimates (see Table 3);
- extrapolation from adult data to reduce the sample size required for the confirmatory paediatric study; and
- use of external controls (such as natural history data or historical controls) as part of a hybrid analysis.
All sample size assumptions should be clearly justified in the protocol, and sensitivity analyses should explore the impact of departures from key assumptions.