Neonatal Death and Best Gestational Age for Delivery at Term
It is understood that during experimental observation the 'dropout' of patients from a group may affect, either positively or negatively, the assessment of results for the remaining subjects. As a result, examining only the remaining subjects fails to take into account the possible effects of those who dropout (attrition bias). There is considerable attrition due to delivery between 37 and 39 weeks gestation. Many of these are from spontaneous labor without sequelae. However, the interval between 37 and 39 weeks is not a zero-risk period. Variation in susceptibility to outcomes associated with obstetric complications occur, which result in fetal death, maternal morbidity (and rare mortality) and newborn morbidity and mortality. National data indicate that over 50% of patients entering the 37th week of pregnancy deliver by 39 weeks. Failure to account for the attrition of these delivered mothers and babies results in a selection bias. This may result in overestimating the reduction in neonatal morbidities by prolonging pregnancy while underestimating the risks. Attrition bias has been discussed by Dumville et al., who suggested that when the number of patients lost is <5%, there is little chance of significant bias. Conversely, when a loss ≥20% occurs, 'readers should be concerned about the possibility of bias.' This raises significant questions about relying on 39-week morbidity data without taking into account the effects that earlier delivery has on outcome (attrition bias).
Similar to the examination of mortality, there seems to be an assumption that any increased morbidity, which occurs before 39 weeks, must be due to gestational age. Admission to the neonatal intensive care unit (NICU) is often used as a surrogate for a gestational age-related morbidity, yet NICU admission alone should not be assumed to be without important confounders. Reporting on only respiratory morbidity, Hibbard et al. observed an 11.8% NICU admission rate at 37 weeks (compared with 6% at 39 weeks). Yet, 90% of the 37-week admissions did not require ventilator support. Overall, no difference in the incidence of hyaline membrane disease was observed comparing 38 with 39 weeks, with only a 1% incidence at 37 weeks (with 90% of those reported as mild). Similarly, Tita and co-workers reported neonatal outcomes following elective repeat cesarean and found NICU admission rates of 12.8% at 37 weeks compared with 5.9% at 39 weeks. As in the prior example, rates of ventilator support were only 1.9 and 0.4% at 37 and 39 weeks, respectively. A substantial proportion of these NICU admission in both of these series likely represent precautionary efforts rather than an actual need for ventilator support. These two studies also report the rates of infectious morbidities to be higher at 37 weeks vs 39 weeks. Believing that such conditions are the result of early delivery is a seductive conclusion; however, many of these morbidities may not be the result of, but rather contributed to, the earlier delivery. Starting antibiotics for presumed infection is sound newborn practice, but it should not be taken as an adverse event in and of itself when infection is excluded. Although publications have attempted to report detailed morbidities by gestational age, the current data are insufficient to determine what amount of observed neonatal morbidity is truly related to gestational age. Without data that control for potential confounding issues (susceptibility, selection, attrition bias), and which clearly identify gestational age-related morbidities, the potential 0.5% decrease in respiratory morbidities between 37 and 38 weeks may be importantly overestimated.
Neonatal Morbidities and Gestational Age
It is understood that during experimental observation the 'dropout' of patients from a group may affect, either positively or negatively, the assessment of results for the remaining subjects. As a result, examining only the remaining subjects fails to take into account the possible effects of those who dropout (attrition bias). There is considerable attrition due to delivery between 37 and 39 weeks gestation. Many of these are from spontaneous labor without sequelae. However, the interval between 37 and 39 weeks is not a zero-risk period. Variation in susceptibility to outcomes associated with obstetric complications occur, which result in fetal death, maternal morbidity (and rare mortality) and newborn morbidity and mortality. National data indicate that over 50% of patients entering the 37th week of pregnancy deliver by 39 weeks. Failure to account for the attrition of these delivered mothers and babies results in a selection bias. This may result in overestimating the reduction in neonatal morbidities by prolonging pregnancy while underestimating the risks. Attrition bias has been discussed by Dumville et al., who suggested that when the number of patients lost is <5%, there is little chance of significant bias. Conversely, when a loss ≥20% occurs, 'readers should be concerned about the possibility of bias.' This raises significant questions about relying on 39-week morbidity data without taking into account the effects that earlier delivery has on outcome (attrition bias).
Similar to the examination of mortality, there seems to be an assumption that any increased morbidity, which occurs before 39 weeks, must be due to gestational age. Admission to the neonatal intensive care unit (NICU) is often used as a surrogate for a gestational age-related morbidity, yet NICU admission alone should not be assumed to be without important confounders. Reporting on only respiratory morbidity, Hibbard et al. observed an 11.8% NICU admission rate at 37 weeks (compared with 6% at 39 weeks). Yet, 90% of the 37-week admissions did not require ventilator support. Overall, no difference in the incidence of hyaline membrane disease was observed comparing 38 with 39 weeks, with only a 1% incidence at 37 weeks (with 90% of those reported as mild). Similarly, Tita and co-workers reported neonatal outcomes following elective repeat cesarean and found NICU admission rates of 12.8% at 37 weeks compared with 5.9% at 39 weeks. As in the prior example, rates of ventilator support were only 1.9 and 0.4% at 37 and 39 weeks, respectively. A substantial proportion of these NICU admission in both of these series likely represent precautionary efforts rather than an actual need for ventilator support. These two studies also report the rates of infectious morbidities to be higher at 37 weeks vs 39 weeks. Believing that such conditions are the result of early delivery is a seductive conclusion; however, many of these morbidities may not be the result of, but rather contributed to, the earlier delivery. Starting antibiotics for presumed infection is sound newborn practice, but it should not be taken as an adverse event in and of itself when infection is excluded. Although publications have attempted to report detailed morbidities by gestational age, the current data are insufficient to determine what amount of observed neonatal morbidity is truly related to gestational age. Without data that control for potential confounding issues (susceptibility, selection, attrition bias), and which clearly identify gestational age-related morbidities, the potential 0.5% decrease in respiratory morbidities between 37 and 38 weeks may be importantly overestimated.
SHARE
