Many centers delivering infants at 22 to 25 weeks' gestation have limited data regarding their outcomes. A meta-analysis of the 4- to 8-year neurodevelopmental outcomes and exploration of the limitations of meta-analysis would aid physicians and parents to plan care for these infants.
To determine the rate of moderate to severe and severe neurodevelopmental impairment by gestational age in extremely preterm survivors followed up between ages 4 and 8 years, as well as to determine whether there is a significant difference in impairment rates between the successive weeks of gestation of survivors.
A peer-reviewed search strategy obtained English-language publications from MEDLINE In-Process & Other Non-Indexed Citations, MEDLINE, and EMBASE. Personal files and reference lists from identified articles were searched. Contemporary cohorts were obtained by restriction to those published after 2004. Inclusion criteria were prospective cohort studies, follow-up rate of 65% or more, use of standardized testing or classification for impairment, reporting by gestation, and meeting prespecified definitions of impairment. We excluded randomized clinical trials, highly selective cohorts, consensus statements, and reviews. Of 1771 identified records, 89 full-text publications were assessed for eligibility. Using the full text of each publication, 2 authors independently followed a 2-step procedure. First, they determined that 9 studies met inclusion criteria. Next, they extracted data using a structured data collection form. Investigators were contacted for data clarification.
All extremely preterm infant survivors have a substantial likelihood of developing moderate to severe impairment. Wide confidence intervals at the lower gestations (eg, at 22 weeks, 43% [95% CI, 21%-69%]; heterogeneity I2, 0%) and high heterogeneity at the higher gestations (eg, at 25 weeks, 24% [95% CI, 17%-32%]; I2, 66%) limit the results. There was a statistically significant absolute decrease in moderate to severe impairment between each week of gestation (6.5% [95% CI, 2%-11%]).
CONCLUSIONS AND RELEVANCE:
Knowledge of these data, including the limitations, should facilitate discussion during the shared decision-making process about care plans for these infants, particularly in centers without their own data. More prospective, high-quality, complete cohorts are needed.