Doctors and medical researchers often worry if they can trust the Social Security Administration to accurately tell them when Americans die.
Federal death records have long been used to determine if critically ill patients are still alive as they wait for organ transplants. Researchers also need reliable information for health studies that follow people over many years, even a lifetime, after they've had a particular medical procedure or been identified as having certain traits.
Researchers typically turn to one or both of two databases to determine if someone is dead or alive -- the National Death Index, which contains virtually all death certificates filed in the U.S. since 1979, and the Social Security Death Master File, which contains records of nearly 90 million benefit enrollees reported to the government as deceased since 1937.
"Knowing when a subject died gives us some finality," said Dr. Alexander Turchin, a researcher at Harvard Medical School and Partners Healthcare System in Boston. "It's the one invariable. We can argue whether someone had a heart attack or kidney disease, but you can't argue whether someone has died."
Except when the Social Security system occasionally mislabels as dead someone who is still alive.
"Mistakes are a concern, because death is a relatively rare outcome, so if just a few deaths are erroneous, the conclusions of the study may be in error as well," Turchin said.
Turchin and several colleagues published a study online last November that matched electronic medical records against Social Security death reports for nearly 160,000 patients going back to 2000.
Their report, titled "I Am Not Dead Yet: Identification of False-Positive Matches to the Master Death File," found that more than 24,000 of the patients had at least one medical record element, such as new bills, medicine disbursement, lab reports or vital signs, still being recorded a month or more after the patient supposedly died.
Further sifting of the records and other data found there were about 800 patients reported dead through the DMF who were actually alive.
"It's not an enormous number, but the more errors we can take out up front, the better the work will be," Turchin said.
The DMF is also used by the United Network for Organ Sharing -- a private, nonprofit organization that manages the U.S. organ transplant system under a contract with the U.S. Department of Health and Human Services -- to determine eligibility of organ transplant candidates.
"In order to maintain an efficient organ allocation system, UNOS ... provides a periodic report to transplant centers of their waiting list candidates who match with a record in the Social Security Death Master File," said UNOS spokesman Joel Newman.
He said UNOS is unaware of any case in which an organ transplant candidate was adversely affected by a false death report on the DMF. "We can't say it has never happened, but we don't know of such an instance," Newman said.
Harvard epidemiologist Frank Sesso, said researchers recognize "there is really no perfect system for tracking survival or mortality. Both the NDI (National Death Index) and the Death Master File have some gaps."
A decade ago, Sesso and colleagues ran a comparison of death records for a group of male college alumni between the two databases and found that the Social Security file agreed on death status 94.7 percent of the time; and matched the Death Index for subjects being alive 97.5 percent of the time. The research also showed that the Social Security records were less useful in identifying deaths among older women, who were often not included in records because they didn't work and pay into the Social Security Trust Fund.
Researchers know there are other gaps in the DMF, since not all deaths are recorded by the agency. If, for instance, they're not a citizen or are a federal worker outside the main retirement system.
However, the DMF is updated monthly, while it takes the Centers for Disease Control and Prevention, which runs the NDI, about two years to incorporate death certificates in the record. This makes the Social Security files a much more "real time" tool for identifying most deaths, including deaths that may occur among Americans living overseas. That's something the NDI does not capture since it is based on state vital records.
On the other hand, the NDI offers much more detail, including a cause and manner of death for most subjects.