Main findings
In this population based study, we found that the risk for CP increased with increasing intervals between PROM and delivery of the child. An interval of more than 24 h was associated with a 60% excess risk for CP compared with the reference group, independent of other antenatal risk factors for CP. However, the absolute risk for CP was low and there was no excess risk for death during delivery or in the neonatal period.
Among children with CP, there was no association between increasing intervals between PROM and delivery and CP subtypes or severity of impairments. However, among those who had been examined with cerebral MRI, the proportion with findings consistent with hypoxic-ischemic injuries (i.e. diffuse cortical injury and basal ganglia pathology) increased with increasing intervals, while the proportion consistent with white matter lesions, usually a sequelae of injuries in mid-pregnancy, decreased.
Strengths and limitations
Strengths of the present study are the large number of births and the prospective recording of data in the MBRN and the CPRN. The percentage of children with PROM (11.6%) is comparable to what has been reported in the literature (6–19%) [1]. The main findings are unlikely to be caused by chance as indicated by the low P-values. The correctness of the CP diagnosis in the CPRN at five years of age, and the unbiased selection of cases as documented in a validation study are further strengths [23]. According to updated information from the register, 83% of all children with CP born during in Norway during 1999–2009 were registered in the CPRN [24]. The corresponding misclassification of approximately 100 children with CP as controls is negligible compared with the 500,000 children comprising the latter group.
Although the exact timing of the membrane rupture, the onset on contractions, and the delivery is recorded in medical records of obstetric departments, this information is categorized by the MBRN as described in the methods section. This categorization implies that an interval of less than 12 h between PROM and delivery is included in the reference group. This lack of detailed information leads to some misclassification. In addition, some children in the reference group may have been born more than 24 h after rupture of membranes, even if the rupture occurred at or after onset of contractions. However, these limitations have most likely diluted the differences between the groups.
In the study period, according to Norwegian guidelines, intrapartum antibiotic prophylaxis was recommended to reduce the risk of GBS disease in the newborn if “there were signs of infection or maternal fever intrapartum (prolonged rupture of membranes, foul smelling amniotic fluid, fetal tachycardia, increasing C-reactive protein).” [25] Thus, intrapartum antibiotic prophylaxis was most likely not administered in deliveries where a prolonged interval from PROM to delivery was the only risk factor. The lack of information on maternal antibiotic treatment and suspected maternal infection (i.e. fever, or clinical chorioamnionitis) around delivery, in particular maternal and infant GBS-status (routine screening for maternal GBS colonization is not performed in Norway), is another limitation of the study. However, although the data prevents us from speculating on infectious etiology, we consider it most unlikely that the increased risk for CP associated with a prolonged interval is explained solely by a more restricted use of intrapartum antibiotic prophylaxis in Norway than in some other countries.
The cerebral MRI images were interpreted by different radiologists at different hospitals, which is another limitation. The reliability of these descriptions is therefore uncertain. Moreover, MRI examinations were completed at different ages of the children. Thus, caution is needed in the interpretation of these results, although they may be considered plausible.
The study was restricted to term born singletons without congenital malformations to exclude confounding by prematurity, multiple births and malformations. In particular, the management of PROM in preterm deliveries is likely to be significantly different from the management of PROM in the otherwise uncomplicated pregnancy at term. Furthermore, multivariable analyses did not suggest confounding by the available antenatal, known risk factors for CP.
Comparison to literature and interpretation of findings
Few studies have addressed the potential association between increasing intervals from PROM to delivery and CP. In a systematic review on risk factors for CP, McIntyre et al. identified one study in term born children [10]. That study found a non-significant increased risk for CP following an interval between PROM and delivery of more than 24 h (OR 1.54, CI 0.37–6.27) [11]. Albeit non-significant, the size effect was similar to the one we found, and the lack of statistical difference was probably due to a limited sample size (101 cases and 308 controls).
We managed to identify two more studies on the association between prolonged intervals from PROM until delivery, and CP. However, both studies included preterm children, and are therefore not comparable to our study [26, 27].
Our findings may be compared with three studies on the association between a long interval from rupture of membrane to delivery and CP, although these studies did not study PROM. The sample sizes of these studies were smaller than in our study, but the reported effect sizes were comparable to our findings (i.e. OR 1.4–2.4) [28,29,30], however the results were not statistically significant.
The association between a prolonged interval from PROM to delivery and CP could be caused by a perinatal infection. While our study does not provide data supporting a causal relationship between a prolonged interval, infections and CP, other studies have found that a prolonged interval increases the risk for perinatal infections [8], and that perinatal infections increase the risk of CP [9].
However, it may be noteworthy that in the group of children with CP, the proportion with diffuse cortical injury and basal ganglia pathology on cerebral MRI increased with increasing intervals from PROM to delivery. These findings are characteristic of hypoxic-ischemic injuries at term [31,32,33], and they are also highly predictive of CP [34]. We therefore speculate that a long interval between PROM and delivery may increase the vulnerability of the neonatal brain, lowering the threshold at which a reduction in blood supply to the fetus may lead to perinatal hypoxic-ischemic injury. This vulnerability might be caused by an infection and/or inflammation. Moreover, the decreasing proportion of white matter lesions with increasing intervals from PROM to delivery may lend further support to this speculation, since such findings are likely to suggest prolonged partial asphyxia [35], and are characteristic of injuries occurring earlier in pregnancy. On the other hand, 10 children with CP in the prolonged interval-group had white matter injury, the typical lesion in children with CP born preterm. In these children it is possible that the brain injury predated PROM.
Another possible explanation of our findings could be reverse causality, i.e. that the birth process is slower in fetuses with an antenatal brain injury. However, we consider this less likely, based upon the cerebral MRI findings, and the exclusion of congenital malformations and plurals.
The decreasing prevalence of stillbirths with increasing intervals between PROM and delivery can most likely be explained by reverse causality as antenatal stillbirths are likely to be induced early.