Careers  |   Log In  |   Register  |   Welcome Center  |   Follow Us  Facebook  Twitter Google Plus

Diagnostic coding shifts behind drop in pneumonia deaths

NEW YORK (Reuters Health) – Pneumonia mortality didn’t fall between 2003 and 2009 because care was getting better, but because the sickest pneumonia patients were getting a different primary diagnosis code, according to a new study published yesterday in the Journal of the American Medical Association.

“These results suggest that secular trends in documenting and coding, rather than improvements in actual outcomes, may explain much of the observed change in this and other studies,” said Dr. Peter K. Lindenauer, director of the Center for Quality of Care Research at Baystate Medical Center in Springfield, Massachusetts, and his colleagues in their report.

While substantial improvements in mortality among pneumonia patients have occurred in recent years, Dr. Lindenauer and his team write, “in the absence of care-transforming technologies, other explanations must also be considered.”

To investigate their hypothesis that changes in documentation explained drops in mortality, the researchers performed a temporal trends study using data from the 2003-2009 National Inpatient Sample for pneumonia patients with a principal diagnosis of pneumonia, sepsis, or respiratory failure. The sample included around eight million cases per year.

For patients with a primary diagnosis of pneumonia, the hospitalization rate decreased from 5.5 to 4.0 per 1,000, a 27.4% decline.

But hospitalization rates per 1,000 rose from 0.4 to 1.1 for patients with a primary diagnosis of sepsis and a secondary diagnosis of pneumonia (i.e., a 177.6% increase), and from 0.44 to 0.48 for patients with pneumonia and a primary diagnosis of respiratory failure (a 9.3% increase).

When all of the patients were combined, hospitalization rate decreased from 6.3 to 5.6 per 1,000.

Inpatient mortality rates among all three patient groups fell during the study period, from 5.8% to 4.2% for pneumonia patients (a 28.2% decline); from 25.1% to 22.2% for patients with a primary diagnosis of sepsis (a 12% reduction); and from 25.1% to 19.2% for patients with a primary diagnosis of respiratory failure (a 23.7% drop).

But inpatient mortality in the three groups combined actually increased from 8.3% to 8.8%, once the researchers took age and gender into account. When the model was also adjusted for comorbidities, overall mortality declined from 8.3% to 7.8%.

The researchers found declines in hospitalization and mortality for control conditions including ischemic stroke, ST-segment elevation myocardial infarction, and rupture of the thoracic or abdominal aorta, but all changes were significantly smaller than the 28% seen for pneumonia patients.

They estimate that nearly 150,000 fewer people received a principal diagnosis of pneumonia in 2008 compared to 2003.

“There’s always been and continue to be financial incentives associated with documenting and coding sepsis because the reimbursement rates for hospitals are in fact higher than if they just have pneumonia,” Dr. Lindenauer told Reuters Health. He pointed out that the “Surviving Sepsis” campaign was underway during the study period, which may also have led to more people receiving sepsis as a primary diagnosis.

“Whatever the cause, it’s really hard to imagine that over seven years the rate of sepsis could have actually tripled, which is what we observed. Clinically that just doesn’t make any sense,” Dr. Lindenauer said.

Based on the findings, he added, performing analyses of trends in outcomes for pneumonia or comparing quality of care from hospital to hospital may be biased if the analysis only includes patients with a primary diagnosis of pneumonia.

“Administrative data have played a very important role in health services research and I think they will continue to play an important role,” Dr. Mary Vaughan Sarrazin of the University of Iowa Hospitals and Clinics in Iowa City, who co-authored an editorial accompanying the study, told Reuters Health. “But their usefulness depends on our ability to interpret the data correctly and use them correctly. There’s all sorts of nuances that you need to be mindful of.”

While the current study specifically investigated pneumonia, Dr. Sarrazin added, it’s likely that similar trends may be going on in other conditions.


JAMA 2012;307:1405-1413.