Objective To determine whether improvements in school age outcomes had occurred between two cohorts of births at 22–25 weeks of gestation to women residents in England in 1995 and 2006.

Design Longitudinal national cohort studies.

Setting School-based or home-based assessments at 11 years of age.

Participants EPICure2 cohort of births at 22–26 weeks of gestation in England during 2006: a sample of 200 of 1031 survivors were evaluated; outcomes for 112 children born at 22–25 weeks of gestation were compared with those of 176 born in England during 1995 from the EPICure cohort. Classroom controls for each group acted as a reference population.

Main outcome measures Standardised measures of cognition and academic attainment were combined with parent report of other impairments to estimate overall neurodevelopmental status.

Results At 11 years in EPICure2, 18% had severe and 20% moderate impairments. Comparing births at 22–25 weeks in EPICure2 (n=112), 26% had severe and 21% moderate impairment compared with 18% and 32%, respectively, in EPICure. After adjustment, the OR of moderate or severe neurodevelopmental impairment in 2006 compared with 1995 was 0.76 (95% CI 0.45 to 1.31, p=0.32). IQ scores were similar in 1995 (mean 82.7, SD 18.4) and 2006 (81.4, SD 19.2), adjusted difference in mean z-scores 0.2 SD (95% CI −0.2 to 0.6), as were attainment test scores. The use of multiple imputation did not alter these findings.

Conclusion Improvements in care and survival between 1995 and 2006 are not paralleled by improved cognitive or educational outcomes or a reduced rate of neurodevelopmental impairment.