Objective To investigate differences and calculate pooled incidence of any intraventricular haemorrhage (IVH), severe IVH (Grade III/IV, sIVH) and ventriculoperitoneal shunt (VPS) placement in preterm infants across geographical, health and economic regions stratified by gestational age (GA).

Design MEDLINE, Embase, CINAHL and Web of Science were searched between 2010 and 2020. Studies reporting rates of preterm infants with any IVH, sIVH and VPS by GA subgroup were included. Meta-regression was performed to determine subgroup differences between study designs and across United Nations geographical regions, WHO mortality strata and World Bank lending regions. Incidence of any IVH, sIVH and VPS by GA subgroups<25, <28, 28–31, 32–33 and 34–36 weeks were calculated using random-effects meta-analysis.

Results Of 6273 publications, 97 met inclusion criteria. Incidence of any IVH (37 studies 87 993 patients) was: 44.7% (95% CI 40.9% to 48.5%) for GA <25 weeks, 34.3% (95% CI 31.2% to 37.6%) for GA <28 weeks, 17.4% (95% CI 13.8% to 21.6%) for GA 28–31 weeks, 11.3% (95% CI 7.3% to 17.0%) for GA32–33 weeks and 4.9% (95% CI 1.4% to 15.2%) for GA 34–36 weeks. Incidence of sIVH (49 studies 328 562 patients) was 23.7% (95% CI 20.9% to 26.7%) for GA <25 weeks, 15.0% (95% CI 13.1% to 17.2%) for GA <28 weeks, 4.6% (95% CI 3.5% to 6.1%) for GA 28–31 weeks, 3.3% (95% CI 2.1% to 5.1%) for GA 32–33 weeks and 1.8% (95% CI 1.2% to 2.8%) for GA 34–36 weeks. Europe had lower reported incidence of any IVH and sIVH relative to North America (p<0.05). Proportion of VPS across all GA groups was 8.4% (95% CI 4.7% to 14.7%) for any IVH and 17.2% (95% CI 12.2% to 26.2%) for sIVH. Heterogeneity was high (I² >90%) but 64%–85% of the variance was explained by GA and study inclusion criteria.

Conclusions We report the first pooled estimates of IVH of prematurity by GA subgroup. There was high heterogeneity across studies suggesting a need for standardised incidence reporting guidelines.