TUESDAY, April 3, 2012 (HealthDay News) -- Recent reports showing a big dip in U.S. pneumonia cases and related deaths may be the result of a glitch in the way hospitals code for pneumonia and associated illnesses, rather than a treatment breakthrough, a new study suggests.
"We had observed that over a very brief period of time, between 2003 and 2009, the number of admissions to hospitals for pneumonia was decreasing rather dramatically and, at the same time, the mortality rate for patients hospitalized for pneumonia was also decreasing rather dramatically," said study author Dr. Peter Lindenauer, director of the Center for Quality of Care Research at Baystate Medical Center in Springfield, Mass.
"We don't have a new drug that might suggest the drop, so one of the things we thought we needed to investigate were changes in diagnostic coding choices," said Lindenauer, also an associate professor of medicine at Tufts University School of Medicine in Boston.
Pneumonia, one of the nation's top killers, causes more than a million hospitalizations a year in the United States.
For the study, published April 4 in the Journal of the American Medical Association, Lindenauer and colleagues analyzed trends in hospital admissions and survival in patients with pneumonia, and in patients diagnosed with sepsis or respiratory failure, two related illnesses, combined with pneumonia. Sepsis is a life-threatening bacterial infection.
The researchers calculated how many cases were coded as a principal diagnosis of pneumonia, and how many cases were given a principal diagnosis of sepsis or respiratory failure with a secondary diagnosis code of pneumonia.
"A physician's primary responsibility is to document a clinical diagnosis, impressions, and create a patient treatment plan," said Lindenauer. He said it's ultimately the job of the hospital coders (large staffs engaged in coding) to go through a patient's chart and use software that helps them assign codes that will enable insurance companies to render payments.
"So the choice of codes has tremendous implications for hospital payment," added Lindenauer.
Using a large, national hospital database, the researchers also evaluated changes in hospitalization and death rates in patients who had conditions they thought might be less susceptible to changes in coding.
The investigators found that from 2003 to 2009, the hospitalization rate of patients with a principal diagnosis of pneumonia decreased from 5.5 to 4.0 per 1,000 patients, a drop of 27 percent. And pneumonia deaths declined from about 6 percent in 2003 to about 4 percent in 2009.
Over the same time period, the hospitalization rate for patients with a principal diagnosis of sepsis and a secondary diagnosis of pneumonia increased 178 percent, from 0.4 to 1.1 per 1,000 patients.
"What we came to hypothesize was that a shift was taking place whereby patients who'd previously been the sickest pneumonia patients were increasingly being coded as having sepsis," said Lindenauer.
He said hospitals get reimbursed more for patients with sepsis, so there's a strong financial incentive for coding based on sepsis versus pneumonia.
During the same time period, a national campaign was launched to raise awareness about sepsis and its treatment among physicians, which may also have influenced coding.
"There was probably a double-whammy of increased reimbursement and also an increase in physician awareness, so more were writing it in the chart and that allowed more hospitals to get reimbursed," said Lindenauer.
The researchers looked at the data in another way, too. They combined patients with sepsis, pneumonia and respiratory failure, and analyzed results. While they found hospitalizations were still declining for this combined group, the figures didn't drop nearly as much as when the pneumonia group was analyzed separately from the sepsis and respiratory failure group.
"Our study would suggest if you really want to know what's happening with pneumonia, it's not enough to look at coded groups for pneumonia. You have to also look at these other groups with a secondary diagnosis of pneumonia," said Lindenauer.
An accompanying editorial published in the journal suggests coding nuances may also affect the way other diseases, such as heart disease, are evaluated.
Dr. Rohit Bhalla, chief quality officer at Montefiore Medical Center in New York City, who was not involved in the study, concurred that coding is complex and needs further analysis. He said new coding methods will soon be introduced that could complicate matters further.
"You have to understand administrative data's strengths and limitations," Bhalla said. "You have to be careful and circumspect about implying that small changes in data relate to care." He suggested it might be more accurate to use both administrative and clinical information from a patient's chart when studying disease trends, treatments and survival.
For more on pneumonia, see the U.S. National Library of Medicine.