Skip to main content

Bad CDC Data May Have Skewed Research

 |  By cclark@healthleadersmedia.com  
   October 29, 2012

They might be zombies.

A commonly used database run by the Centers for Disease Control and Prevention is fraught with serious errors, some of which indicate that patients were impossibly pronounced dead in the emergency department, but then intubated and admitted to the intensive care unit the same day, according to a Loma Linda University report.

In fact, the study, which looked at 10 years' worth of CDC data, found that one in four emergency room patients who were reported as being intubated—which in virtually all cases would be followed by death, admission to a critical care unit, transfer to another facility, or transfer to surgery or a catheterization lab—was instead reportedly discharged or referred to a non-critical care bed.

"A patient who is having a piece of plastic inserted down their windpipe is very, very sick," says Steven Green, MD, Loma Linda University School of Medicine professor in California and the author of an Annals of Emergency Medicinestudy documenting the inconsistencies. "These aren't the kind of patients, except in very rare situations, who would be discharged from the hospital."

Green categorically dismisses, with a chuckle in light of Halloween festivities this week, any suggestion the CDC anomaly could reflect the existence of zombies, or that it is possible for a patient to be alive and dead at the same time.

"I can't begin to speculate on what led to these errors, but one of the possibilities is the use of personnel who don't have a strong command of medical terminology," Green says.

The CDC database in question is the National Hospital Ambulatory Medical Care Survey or NHAMCS, which contains information on more than 350,000 emergency department visits between 2000 to 2009, the period used by Green for his analysis.

Among his study's other findings, Green says, of 875 intubated patients, 27% had incompatible dispositions. For example, 9% were reportedly discharged, 17% were admitted to a non-critical care unit.

Green explains that the problems he's discovered raise questions about the veracity of research projects that used NHAMCS to draw conclusions that large numbers of physicians don't deliver appropriate emergency care.

"For example, you might look at study that used the NHAMCS to see what percentage of patients with a broken bone got narcotics to treat their pain. They found numbers that look lower than what any of us would [expect, prompting] people to say ‘Whoa, that's lower than it should be,' and conclude that the doctors are doing a bad job of treating patients' pain."

Likewise, another study using NHAMCS data showed that physicians weren't always checking for pregnancy when teenage girls came to the emergency department complaining of abdominal pain.

"You'd see a low number and think there must be a lot of bad doctors, yet the alternative explanation that's suggested by the study that I've done is to say, wait a minute. Maybe there's a problem with the underlying data."

Green says he suspects that the personnel, who are hired by the CDC and assigned to go to hospitals and perform chart reviews to complete the survey and check all the boxes, may not "be trained to understand" the meaning of the term endotracheal intubation. "If you don't know what that means, the odds are you're not going to code it accurately, and that's my worry here."

Asked for a response, a CDC public information representative says the agency does not comment on journal articles that CDC personnel do not author.  Green says that after he informed a CDC official about his results, he was told that the inconsistencies he discovered are "under investigation."

Going forward, he says, researchers who want to use the NHAMCS data, or research projects stemming from it, "should have a little bit more concern about whether the underlying data is correct.  I just worry about the hundreds of papers already out there, or in press, and wonder how many of them have a result that is incorrect due to the underlying data."

Pages

Tagged Under:


Get the latest on healthcare leadership in your inbox.