Impact of gestational age on risk of cerebral palsy: unravelling the role of neonatal morbidity

Abstract

Background: The contribution of adverse consequences of preterm birth to gestational-age-related risk of cerebral palsy (CP) has rarely been studied. We aimed to assess the potential mediating roles of neonatal morbidity on the association between gestational age and risk of CP.

Methods: In this Swedish population-based study, 1 402 240 singletons born at 22-40 gestational weeks during 1998-2016 were followed from day 28 after birth for a CP diagnosis until 2017. Potential mediators included asphyxia, respiratory-related, infection-/inflammatory-related and neurological-related diseases within 0-27 days of life. Cox regression was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Causal mediation analysis was performed to estimate the proportion of the association mediated through pathways involving the four sequential mediators.

Results: We found an inverse dose-response relationship between gestational age and risk of CP, where the strongest association was observed for 22-24 weeks (HR 47.26, 95% CI 34.09-65.53) vs 39-40 weeks. Compared with non-diseased peers, children with neonatal morbidity, particularly those with neurological-related diseases (HR 31.34, 95% CI 26.39-37.21), had a higher risk of CP. The increased risk of CP was, at 24 weeks, almost entirely explained by neonatal morbidity (91.7%); this proportion decreased to 46.1% and 16.4% at 32 and 36 weeks, respectively. Asphyxia was the main mediating pathway from 22 to 34 weeks, and neurological-related neonatal diseases led the mediating pathways from 34 weeks onwards.

Conclusion: Neonatal morbidity mediates a large proportion of the effect of preterm birth on CP, but the magnitude declines as gestational age increases.

Keywords: Preterm birth; cerebral palsy; mediation analysis; neonatal morbidity.

Figure 1

Directed acyclic graph to illustrate the possible structural relationship between gestational age, mediators, cerebral palsy (CP) and confounders. The arrow (0) represents the direct effect of A on Y, i.e. the effect that is not mediated through any of M1, M2, M3 or M4. All arrows (1)–(4) represent the natural indirect effect mediated through all pathways involving M1, M2, M3 and M4 jointly. All arrows (1) represent the natural indirect effect mediated through all pathways involving M1. All arrows (2) represent the partial indirect effect mediated through all pathways involving M2, but not through M1. All arrows (3) represent the partial indirect effect mediated through all pathways involving M3, but not through M1 or M2. The arrow (4) represent the partial indirect effect mediated through the pathway involving M4, but not through M1, M2 or M3. The dashed arrows represent the effect of C on A, M1, M2, M3, M4 and Y. A =gestational age; Y=CP; M1=asphyxia; M2=respiratory-related diseases; M3=infection-/inflammatory-related diseases; M4=neurological-related diseases; C=baseline confounders.

Figure 2–1

Mediation analysis for the association between gestational age and cerebral palsy (CP), Part I. (A) Total effect of gestational age on CP. (B) Direct effect of gestational age on CP. (C) Natural indirect effect mediated through all pathways involving M1, M2, M3 and M4 jointly. (D) Proportion mediated through all pathways involving M1, M2, M3 and M4 jointly. Solid lines represent risk difference in (A)–(C) and proportion mediated in (D), before 5 years of age. Dashed lines represent pointwise 95% CIs. M1=asphyxia; M2=respiratory-related diseases; M3=infection-/inflammatory-related diseases; M4=neurological-related diseases.

Figure 2–2

Mediation analysis for the association between gestational age and cerebral palsy (CP), Part II. (A) Natural indirect effect mediated through all pathways involving M1. (B) Proportion of the total effect mediated through all pathways involving M1. (C) Partial indirect effect mediated through all pathways involving M2, but not through M1. (D) Proportion of the total effect mediated through all pathways involving M2, but not through M1. (E) Partial indirect effect mediated through all pathways involving M3, but not through M1 or M2. (F) Proportion of the total effect mediated through all pathways involving M3, but not through M1 or M2. (G) Partial indirect effect mediated through the pathway involving M4, but not through M1, M2 or M3. (H) Proportion of the total effect mediated through the pathway involving M4, but not through M1, M2 or M3. Solid lines represent risk difference in (A), (C), (E) and (G), and proportion mediated in (B), (D), (F) and (H), before 5 years of age. Dashed lines represent pointwise 95% CIs. M1=asphyxia; M2=respiratory-related diseases; M3=infection-/inflammatory-related diseases; M4=neurological-related diseases.