Would hospitals lie to you?
This question is essential for patients and policymakers, insurers, and regulators interested in controlling medical costs.
Mohsen Bayati of Stanford Graduate School of Business examined a recent version of this question in research on how Medicare patients are reported infections by hospitals.
Bayati is an associate professor at Stanford GSB. He says, “I read an article before starting this project that stated that data like these aren’t perfect.” A hospital’s infection report may not have been accurate because hospitals are motivated to record infections differently.
Bayati, Hamsa Basetani, and Joel Goh from Harvard University and the National University of Singapore were motivated to Study how Medicare reports hospital-acquired infection claims.
Researchers found that by examining the way hospitals reported on hundreds of thousands of Medicare patients’ infections, they were more likely, unintentionally or not, to code hospital-acquired infection (HAI) as an infection that was present at admission (POA). As many as 18.5% of diseases reported to be present on admission in states with less strict reporting requirements occurred at the hospital. This represents an estimated $200 million burden for Medicare each year.
Shifting Incentives
Infections acquired in hospitals are not for those who are cost-conscious or weak-hearted.
Bayati explains that patients may visit the hospital regularly and contract an infection. These infections are usually severe because the bacteria that survive in a hospital environment are bound to be wrong.
It’s not surprising that Medicare, a government-sponsored insurance program for seniors, passed a regulation in 2008 stating it would no longer pay for care related to HAIs and would instead place the financial burden on hospitals. Medicare will cover costs if the patient is admitted to the hospital with a severe infection.
Medicare, however, relies on hospitals self-reporting infection rates. This 2008 policy change could have changed the incentives for hospitals to identify and report precisely where infections originated. Bayati says, “If I am in a hospital, I can work hard to determine if someone has an infection. I will then be paid for that.” If I miss it, it will be considered hospital-acquired, and I won’t get paid. The hospital must have the infrastructure to detect such things up front. Even then, there may be mistakes and misclassifications.”
Reporting on Infections
Researchers looked at more than just raw hospital data to estimate the frequency of HAIs being misclassified. They also wanted to understand the various factors that affect infection reporting.
Bayati says that when you examine hospital data, it’s easy to see hospitals with a large number of POAs but a small number of HAIs. It would be tempting for people to assume they are misclassifying. This could be wrong. For instance, the facility might serve a population of patients with a higher risk of infection.
Researchers focused on differences in state reporting requirements to overcome this data challenge. Bayati says that some states require hospitals not to report infection rates for payment but to improve healthcare quality. For example, Massachusetts, New York, and New Jersey require registering all HAIs. Some states, like Arizona, Montana, and Texas, did not have reporting requirements.
Bayati, along with colleagues, compared the rates of infection reported by hospitals across states that had and did not have reporting requirements. This involved over 490,000 Medicare patients between 2009 and 2010. After controlling for patient risk, demographics, and billing practices, they found that hospitals with infection reporting requirements had higher rates of HAIs. This trend was evident even in hospitals with higher overall care.
Even the best hospitals reported higher HAIs in states requiring such reporting. Hospitals in states without such reporting requirements are more likely to classify HAIs under POAs for Medicare mistakenly. This study found that at least 10,000 HAIs are misreported each year. The estimated burden on Medicare is about $200 million per annum.
It’s a Question of Intent
What is the explanation for this pattern?
Researchers interviewed hospital staff and other sources to determine the cause of misclassification.
They hypothesize that hospitals that comply with state reporting requirements will be more concerned about auditing their records and, therefore, invest more in infrastructure and resources to classify the source of all infections accurately.
Bayati compares the intentionality argument with one used for income tax. He says that if you live in a state where taxes are audited closely, your federal and state returns will be more accurate.
He emphasizes, however, that the research does not provide any specific evidence of intentionality by the hospitals: “People may argue that hospitals misclassify intentionally but we did not test this.”
Promising policy interventions
Regardless of the motive or lack thereof, the findings point to possible policy interventions that could reduce hospitals’ misclassifications of infection types.
The most apparent deterrent is to require reporting of HAIs on a state-level basis. Bayati says that a stronger reporting regulation could encourage hospitals to invest in finding out the truth about these infections.
Another option would be to audit the entire federal infection reporting system. Bayati says the critical number is the ratio of HAIs to POAs in each hospital. Audits could be targeted at hospitals with high POA-to-HAI ratios.
Bayati concludes, “In all industries, we are moving towards using data to make decisions better.” It’s essential to know the integrity of data. According to our study, take steps to ensure that the data you use is accurate, as it may contain biases. “When humans input data, bias is always possible.”
