Efficacy of non-pharmacological interventions for pediatric primary headaches: a systematic review and network meta-analysis

Across a network of 16 trials, biofeedback was associated with fewer and less intense headaches than waiting-list control and ranked favorably for headache frequency. The evidence was very uncertain, so the review supports informed discussion rather than a promise that one approach is best.

Zhang FQ, Wu YN, Liu WD, Yang KH, Li XX, Wang SY.
World Journal of Pediatrics · Publication date: 2026-10-07
Added to site: 2026-10-09
DOI: 10.1007/s12519-026-01103-9

Published October 7, 2026. Added to Neurofeedback.com October 9, 2026.

Modalities

Biofeedback, Behavioral self-regulation

Clinical focus

Pediatric primary headache, Headache frequency, duration and intensity

Population

Children and adolescents with primary headache disorders

Sample

16 randomized controlled trials with 820 participants

Intervention / method

PubMed, Cochrane Library, Web of Science and Embase were searched from inception through August 1, 2026; risk of bias was assessed with the Cochrane tool and certainty with CINeMA; protocol registered in PROSPERO

Comparator

Multiple non-pharmacological interventions and control conditions, including waiting-list control

Clinician summary

Versus waiting-list control, biofeedback was associated with lower headache frequency (SMD -1.11, 95% CI -2.04 to -0.17) and intensity (SMD -0.85, 95% CI -1.59 to -0.11). Biofeedback ranked highest for frequency by SUCRA (81.4%), while biofeedback plus relaxation was associated with shorter duration. The authors explicitly state that rankings are relative probabilities, not proof of clinical superiority.

How to interpret the findings

Effect estimates were imprecise and the authors rated the evidence very low certainty. Network rankings are exploratory, included trials varied across interventions and outcomes, and a favorable SUCRA rank does not establish that an intervention is best for an individual child. The accessible publisher record did not provide trial-level adverse-event or disclosure detail beyond the review-level declarations.

Equipment and disclosures

Biofeedback methods and devices varied across included trials and were not enumerated in the accessible abstract; no product attribution is made.

Supported by National Natural Science Foundation of China grant 82004203. The authors declared no conflicts of interest and reported using ChatGPT for language improvement, with author review and responsibility for the manuscript.

Related evidence

Compare findings from the same clinical area, including mixed results.