Commentary: Does patient volume matter for low-risk deliveries?

That there is a volume-outcome relationship for deliveries, and that mortality is lower for births that occur in hospitals with higher level neonatal intensive care units (NICU) is well established. Numerous studies have found this relationship for all births, 1‐4 or selected groups of higher risk cases. 5‐8 Using all births in the German State of Hesse for a 10-year period, the study by Heller et al. 9 extends the volume-outcome, level of care analysis to low-risk births. The perinatal care system in Hesse is highly regionalized; over 90% of infants with a birthweight 1500 g (VLBW) are born in perinatal centres. Yet, even within this highly regionalized system the authors found that compared to births that occurred in high-volume delivery services, births occurring in hospitals with lower-volume delivery services had a 1.5- to 3-fold higher risk of mortality. The implications are a dramatic change from the past policy of regionalizing just high-risk deliveries. If replicated, these results would indicate that deliveries should be concentrated in high-volume facilities to the extent it is geographically possible. They enhance the case for increased regionalization, especially in countries such as the US with fragmented perinatal systems. 10 As the authors correctly note, their findings raise significant policy questions and more study is needed before there are any changes in policy. The Heller study was carefully done. They limited their analysis to infants with a birthweight 2500 g, excluding those infants with a documented congenital anomaly as a cause of death. They only counted deaths that occurred during delivery or within 7 days of birth. Since the regionalized system works so well for VLBW infants, it is likely that it works well for most previously identified high-risk deliveries. Thus, it is unlikely that the results are due to previously identified high-risk deliveries occurring at the low-volume hospitals. The study did not report the causes of death, but most of the deaths included in the analyses are likely related to acute obstetric emergencies. The authors erred on the side of caution in their analyses, with the likely effect being to underestimate the actual effect of patient volume. Importantly, they could not identify normal birthweight deliveries antenatally referred to a perinatal centre for a highrisk condition that was not a lethal congenital anomaly. The inclusion of time of day of delivery also probably contributes to an underestimate of the true effect of patient volume; some of the time of day effect is probably driven by the ability to respond to emergencies. The ability to identify high-risk cases is not perfect and some complications occur too late in the delivery process to subsequently move the delivery to a high-risk facility. Thus, some high-risk births will always occur at low-risk delivery services. The authors posit that the most likely cause of the results is the response time to medical emergencies at the lower-volume hospitals. While I am not familiar with the organizational details of care in Hesse, since these facilities focus on the care given to low-risk cases, it is likely that they are not as prepared to respond to medical emergencies as the perinatal centres. Most will not have 24-hour dedicated obstetric anaesthesia coverage and may lack continuous obstetric coverage. Further, many of these units will not be continuously staffed with the skilled personnel necessary for optimal neonatal resuscitation. Thus, it will take longer to perform emergency c-sections and to resuscitate newborns in circumstances where a few minutes makes a big difference. Since the availability of these personnel is likely to be inversely related to delivery volume, this also provides a plausible explanation for the volume gradient. While this study indicates that lives could be saved if all births were concentrated in high-volume tertiary centres, caution is needed before making such a radical change to the health care system. Geographical access to delivery services needs to be considered. In larger urban areas it is clearly possible to concentrate deliveries in a few large delivery services with minimal