DSE-Migraine-MA
Digital screen exposure and migraine burden: a systematic review and meta-analysis
Clinical Question
Is higher digital screen exposure associated with greater migraine or headache burden (risk, frequency, severity, disability) in adolescents and adults?
Bottom Line
In a random-effects meta-analysis of 10 studies (N=11,218), higher digital screen exposure was associated with a ~60% greater odds of migraine or headache (pooled OR 1.60, 95% CI 1.38–1.86, p<0.001), with directionally consistent adult and paediatric subgroup effects and robustness to leave-one-out. Overall certainty of evidence was low (GRADE), reflecting cross-sectional exposure ascertainment, residual confounding, and possible small-study effects; screen exposure should be regarded as a modifiable factor to assess in migraine care, pending prospective and interventional confirmation.
Major Points
- 22 studies (N=14,763) met eligibility criteria; 10 studies (N=11,218) with OR data were pooled by random-effects (DerSimonian–Laird) meta-analysis.
- Primary result: higher vs lower screen exposure — pooled OR 1.60 (95% CI 1.38–1.86, p<0.001); heterogeneity I2=60%, Cochran Q=22.3, df=9, p=0.008; 95% prediction interval 1.03–2.48.
- Adult subgroup pooled OR 1.87 (95% CI 1.43–2.44); paediatric/adolescent subgroup OR 1.46 (95% CI 1.24–1.72) — direction consistent across age groups.
- Leave-one-out sensitivity analysis: pooled OR ranged 1.52–1.67, indicating no single study drove the pooled estimate.
- Egger's regression suggested small-study effects (intercept 3.71, p=0.002); certainty of evidence rated LOW (GRADE).
- Two clinic-based comparative studies found no significant difference in headache frequency or severity by smartphone-use status — the only null narrative signals.
- Consistent secondary associations across the full 22-study set: sleep disturbance 18/22, reduced quality of life 20/22, visual/ocular discomfort 14/22, psychiatric comorbidity 13/22, musculoskeletal symptoms 11/22.
- Mechanistic convergence across included studies: blue light–mediated retinal ganglion cell activation, melatonin suppression / circadian disruption, and trigeminovascular / cortical hyperexcitability.
Design
Study Type: Systematic review and random-effects meta-analysis (DerSimonian–Laird) of observational (predominantly cross-sectional) and interventional studies
Randomization:
Blinding: Not applicable (dual independent screening, extraction, and risk-of-bias assessment)
Enrollment Period: No date restriction on included studies (published searches conducted for this review; database search dates as reported by authors)
Follow-up Duration: Not applicable (cross-sectional and observational data)
Centers: 0
Countries: multinational (studies from India, Egypt, Saudi Arabia, Bangladesh, Brazil, Turkey, France, Serbia, Denmark, Philippines, Germany, Italy, USA and others)
Sample Size: 14763
Analysis: PRISMA 2020–compliant; PROSPERO-registered (CRD420261404162). Databases searched: Embase, MEDLINE, PubMed, Scopus, Google Scholar. Two reviewers independently screened, extracted data, and assessed risk of bias using a modified Joanna Briggs Institute (JBI)/AHRQ checklist for cross-sectional and observational studies. Pooling threshold: ≥10 studies reporting OR with 95% CI for higher vs lower screen exposure — random-effects (DerSimonian–Laird) meta-analysis. Heterogeneity by I2 and Cochran's Q; small-study effects by Egger's regression; robustness by leave-one-out. Outcomes not amenable to pooling synthesised narratively. Certainty of evidence rated using GRADE.
Inclusion Criteria
- Observational (cross-sectional, case-control, cohort) or interventional studies
- Adolescent or adult participants with migraine or primary headache
- Quantifiable measure of digital screen exposure (screen time or device use)
- Reported migraine/headache outcomes (frequency, severity, disability) or secondary outcomes (sleep, quality of life, visual/ocular, psychiatric, musculoskeletal)
- English-language publication
- No date restriction on publication year
Exclusion Criteria
- Non-English publications
- Studies without a quantifiable screen-exposure measure
- Studies not reporting headache or migraine outcomes
- Case reports, narrative reviews, editorials, and conference abstracts without full data
- Duplicate publications
Arms
| Field | Higher screen exposure | Control |
|---|---|---|
| Intervention | Higher digital screen time / heavier device use (as dichotomised or top-category within each study) | Lower digital screen time / lighter device use (reference category within each study) |
| N | 0 | 0 |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Pooled odds of migraine or headache with higher vs lower digital screen exposure (random-effects DerSimonian–Laird meta-analysis of 10 studies, N=11,218) | Primary | Lower screen exposure (reference) | Higher screen exposure — pooled OR 1.60 | 1.6 | <0.001 |
| Adult subgroup pooled OR (higher vs lower screen exposure) | Secondary | Adults with lower screen exposure | Adults with higher screen exposure | 1.87 | |
| Paediatric / adolescent subgroup pooled OR (higher vs lower screen exposure) | Secondary | Paediatric/adolescent with lower screen exposure | Paediatric/adolescent with higher screen exposure | 1.46 | |
| Sleep disturbance associated with screen exposure (narrative synthesis) | Secondary | - | Reported in 18/22 studies (82%) | ||
| Reduced quality of life associated with screen exposure (narrative synthesis) | Secondary | - | Reported in 20/22 studies (91%) | ||
| Visual / ocular discomfort associated with screen exposure (narrative synthesis) | Secondary | - | Reported in 14/22 studies (64%) | ||
| Psychiatric comorbidity associated with screen exposure (narrative synthesis) | Secondary | - | Reported in 13/22 studies (59%) | ||
| Musculoskeletal symptoms associated with screen exposure (narrative synthesis) | Secondary | - | Reported in 11/22 studies (50%) | ||
| Two clinic-based comparative studies — headache frequency or severity by smartphone-use status | Secondary | No significant difference | No significant difference | ns | |
| Not applicable — observational meta-analysis of exposure/outcome associations; no interventional adverse events reported. | Safety | - | - |
Subgroup Analysis
Prespecified age-based subgroups showed directionally consistent effects: adults OR 1.87 (1.43–2.44); paediatric/adolescent OR 1.46 (1.24–1.72). Leave-one-out sensitivity analysis: pooled OR range 1.52–1.67 across sequential exclusion of each of the 10 pooled studies, confirming robustness.
Criticisms
- Almost all included studies were cross-sectional — cannot establish temporality or causality between screen exposure and migraine/headache.
- Screen exposure was self-reported and often heterogeneously categorised across studies, introducing exposure-measurement error.
- Moderate statistical heterogeneity (I2=60%) and a wide 95% prediction interval (1.03–2.48) indicate the underlying true effect could plausibly be small in some populations.
- Egger's test intercept 3.71 (p=0.002) suggests small-study effects and possible publication bias — smaller studies with larger reported effects are over-represented.
- Residual confounding (sleep quality, mental health, socioeconomic status, concurrent digital-media content) was inconsistently addressed across included studies.
- Two clinic-based comparative studies were null, contrasting with the pooled positive effect — highlighting setting-specific confounding.
- Language restriction to English may have excluded relevant regional evidence.
- GRADE overall certainty of evidence for the primary outcome was rated LOW, reflecting these limitations.
- Narrative outcomes (sleep, QOL, visual, psychiatric, musculoskeletal) were tallied qualitatively; the review did not pool these secondary outcomes quantitatively.
Funding
No funding was received for the conduct of this study.
Based on: DSE-Migraine-MA (Journal of Neurology, 2026)
Authors: Gowda N, Harish S, Tej G
Citation: J Neurol 2026;273:525. DOI: 10.1007/s00415-026-14062-y
Content summarized and formatted by NeuroTrials.ai.