'Glaring Flaws' Found in Data Affecting Hospital Reimbursement

Researchers find serious underreporting of body mass, alcohol abuse, and tobacco use in hospital billing data and recommend two solutions.

The Nationwide Inpatient Sample (NIS), which is derived from billing data, is used for everything from calculating hospitals' risk for readmission or surgical complications to researching things like the effect of health policy changes and access to care.

But there are "glaring flaws" and gaps in the data, thanks to dramatic underreporting of patients' alcohol and tobacco use, as well as their weight and body mass, according to researchers at Johns Hopkins University School of Medicine. The study was published in the journal PLOS ONE.

"We sort of knew going into it [that] it would be bad. We didn't know how bad," says the study's last and corresponding author Susan Hutfless, PhD.

Just how bad was it? Here's an example: According to the NIS, the United States prevalence of overweight is just 0.21%, and the prevalence of obesity is 9.6%. The researchers compared NIS data to information that's in the Behavioral Risk Factor Surveillance System (BRFSS), a federally sponsored telephone-administered survey where more than 500,000 American adults answer questions about their health. According to BRFSS data, 35.8% of Americans are overweight and 27.4% are obese.

"Why is no one coding overweight at all when it's 33% of the population?" Hutfless says.

Body mass isn't the only place where there's significant underreporting in the NIS. It reports alcohol abuse in just 4.6% of the population, compared to 18.3% in the BRFSS, and tobacco use in 12.2% of the population, versus 20.1% in the BRFSS.

This underreporting can have significant implications for a hospital's bottom line, Hutfless says, since missing data can result in inaccurate risk adjustments, and therefore, unfair reimbursement.

"A very high number of patients are having their risk coded as lower than it actually is," she says.

Yet the information is being collected: Questions about patients' weight, alcohol, and tobacco usage are standard. It's "impossible to find a provider who doesn't ask that," says Elie S. Al Kazzi, MD, MPH, the study's first author.

But critical information about weight, alcohol abuse, and tobacco use that's included in the recorded medical history doesn't usually make it onto a hospital bill, and the NIS is based on billing data.

'The Data is There'
"When people are coding, they are coding the big-picture conditions," Hutfless says. They're coding myocardial infarction, but not necessarily that the patient is also an overweight smoker. That information would certainly affect the patient's risk of readmission. And "missing information in the database leads to inaccurate health quality assessments and could have a devastating economic impact on hospitals that see the sickest patients," according to a release about the study.

In an accompanying editorial, Hutfless and Al Kazzi write about several possible solutions to the problem. "The data is there. It's just an issue of actually downloading it and transferring it," Al Kazzi says.

One short-term solution would be to merge the databases, data is from HHS's more than "1,800 publicly available databases could be used to complete missing information," the editorial says.

The authors also suggest that "incorporating meaningful use measurements into datasets that comprise billing codes could enhance the completeness without adding additional coding burden." As for possible long-term strategies, they write that "financial incentives in coding, publicly available information on items used to adjust for risk in the existing Medicare products, and random audits" could increase accuracy.

"In my opinion, it's easy" to fix, says Hutfless. "It's administrative fixes, but I don't think that the amount of manpower and brain power is large… especially for the return on investment."