The actual rates were calculated from state and hospital birth and death records. The analysis showed that for every 1,000 births, the number of newborn deaths rose from 5.6 in 1996 to 6.6 between 1997 and 1999. From 2000 on, the rate fell back to about five deaths per 1,000 births.
The same trend was seen when possibly preventable fetal deaths were put in the rates. "As more and more hospitals closed, we expected things to get worse, but then improve," Lorch said. "We did not expect to see this, where the biggest effects were after two or three closures."
For comparison, Lorch and three coauthors from the University of Pennsylvania analyzed records from two geographic areas where maternity units were largely stable. One area comprised Philadelphia's surrounding suburban counties; the other comprised eight urban counties in Pennsylvania and California.
Neither comparison area saw a significant change in newborn mortality rates from 1995 through 2005. Their combined rates hovered around four deaths per 1,000 births.
While no one knows what caused the spike in Philadelphia's newborn-mortality rate, Lorch said the closures were the only major change in that period that dramatically impacted childbirth outcomes.
Why did the impact appear and then disappear so quickly? Anecdotal evidence suggests that the earliest maternity-unit shutdowns were more abrupt and disruptive than subsequent ones.
Local maternity-care experts affirmed that impression.
At Einstein Medical Center, deliveries rose from 2,002 in 1990 to more than 3,100 now. The hospital has spent $10 million to revamp obstetrics, but that didn't happen overnight.
In the beginning, "there were delays in patients getting prenatal care because they weren't used to traveling outside of their neighborhoods," said Arnold Cohen, Einstein's chair of obstetrics and gynecology. "When there were complications or emergencies, there were significant delays because of the increased volume that we weren't prepared for."
Bette Begleiter, deputy director of the Maternity Care Coalition, said: "There were heroic efforts made to make sure people stayed safe. But there was not much notice in the early closures. You could see the hospitals were unprepared."
Donald F. Schwarz, a Children's pediatrician who became commissioner of Philadelphia's health department in 2008, commended the study, published online in the journal Health Services Research. But he said factors besides the closures, such as changes in state Medicaid payments, may have contributed to the infant death rate.
"The good news is, there was a reaction in the early part of the last decade, either by the neonatal and obstetric community or the Health Department, that brought this back in line," he said.
Cohen noted another good outcome. The city's six remaining maternity units are at top-tier centers and communication between those units has improved.
Contact Marie McCullough at 215-854-2720 or email@example.com.