We thank Dr McCaffrey for his thoughtful letter and for giving us the opportunity to address his concerns. Dr McCaffrey raises two important issues: first, he suggests that, under our definition of ‘lethal’ congenital anomaly, neonates with Trisomy 13 and 18 should not have been included in our cohort. He points to two recent case series where 25 to 44% of infants with Trisomy 13 and 18 survived for more than a year when treated aggressively, far above the 15% cutoff we employed in our definition. However, one of those series included only nine patients,1 and the other consisted of neonates referred to a tertiary center for ongoing care after birth, indicating that their population was enriched for neonates who survived long enough to transfer.2 Without data on the intensity of medical treatment and survival of the neonates who were not transferred, the unusually high survival rate at 1 year was likely due to selection bias rather than contemporary medical intervention. Further population-based studies examining the relation between treatment intensity and survival of infants with Trisomy 13 and 18 are needed.
Second, Dr McCaffrey argues that the lack of correlation between intensity of treatment and length of life is best explained not by the failure of more aggressive medical intervention to improve survival, but by a self-fulfilling prophecy bias. We do not believe, however, that a self-fulfilling prophecy best fits our data. If the firm diagnosis of a presumably lethal anomaly led to the withdrawal of beneficial life-sustaining treatment, then there should be some evidence that, prior to the identification of that condition, intensive treatment correlated with survival. For postnatally diagnosed infants, however, we found the opposite—that increased treatment intensity predicted earlier death. Furthermore, for postnatally diagnosed infants who lived at least 3 days (n=21), excluding neonatal therapeutic intervention scoring system scores on the last 2 days of life—which, on McCaffrey's hypothesis, should be lower than those on the preceding days if confirming a lethal condition led to a transition to comfort measures—did not change the negative correlation (r=−0.45, P=0.004). It is most likely, therefore, that those infants who survive longest do not require intensive care, at least not in the newborn period. We reiterate here that our findings should not be generalized to older children with these diagnoses.
This is a preview of subscription content, access via your institution
