Flawed Measures, Appearance of Bias in 340B Study

by Admin | March 25, 2020 2:45 pm

March 25, 2020 – A new research study[1] published in Health Services Research finds hospitals newly enrolled in the 340B program provide more charity care, by offering more patients discounted care. Specifically, for newly enrolled 340B hospitals, the study finds:

• The amount of charity care provided increased by nearly 28.9 percent after 340B enrollment.
• The probability of offering discounted care increased.
• The income level below which patients are eligible for discounted care increased 19 percentage points (from 294% to 313%).

These three findings met their strict test for statistical significance. Methodological flaws raise questions about other study findings.

First, the authors assert that the increase in charity care at the hospital level was offset by decreases in other types of community benefit. However, their measure of community benefit is flawed, and the finding does not meet their own measure of statistical significance.

Instead of creating a hospital level measure of community benefit, the authors measured community benefit at the level of the health care system using data from IRS Form 990.  They made no attempt to account for the size of the 340B hospital relative to the entire system or the fact that many systems have a mix of 340B and non-340B hospitals with hospitals both entering and leaving the program at different times. For example, Ascension Health has over 100 hospitals; less than half are 340B and those that joined at various times. Inova Health System, a single state system in Virginia, has only one 340B hospital among its five members. The researchers’ approach thus dilutes the value of hospital-specific community benefit activities making the results highly inaccurate.

Using this flawed measure, they claim that the increase in charity care was offset by reductions in other community benefits. However, they note that when they applied a stringent standard of statistical significance, “we find that 340B participation was only associated with an increase in charity care spending and that estimated decreases in other categories of community benefit spending were no longer statistically significant.” Yet the authors continued to emphasize negative findings that did not meet their own tests for statistical significance, giving the appearance of bias.

The study also looked at the provision of low-profit services. These included care for substance abuse, burn and trauma care, dental care, AIDS-related care, obstetrical care, and psychiatric care. They leave out other services, like neonatal intensive care and primary care that have been deemed “essential services” by the Medicaid and CHIP Payment and Access Commission (MACPAC). Also, they fail to recognize that profitability for any service depends on the payer mix of the service area, which can be very different for safety-net versus non-safety-net hospitals.

They did not find an increase in the service lines they chose to review, but adding service lines takes time and needs to be looked at in relation to community need. Earlier research shows that 340B hospitals are more likely to offer these services over time.[i][2] Furthermore, some highly specialized services, like burn and trauma care, are not appropriate service offerings for smaller hospitals, again failing to account for differences in hospital size. They also did not measure the many other ways that 340B hospitals contribute to care for patients with low incomes or living in rural communities.

The report also has factual errors which detract from its overall credibility. For example, they claim the deep cut in Medicare Part B payments to many 340B hospitals has been “blocked” by hospitals. In fact, the policy, which began in 2018, remains in effect while the courts consider an appeal of a lawsuit challenging the legality of the policy. A federal judge has already found the policy to be “unlawful” and the federal government is appealing that verdict.

Research into the impact of the 340B program is an essential element of the national debate over the program. Overall, this study finds that hospitals do in fact increase services to low-income populations when they enroll in the 340B program. The authors try but fail to negate these findings by applying a poor measure of overall community benefit, ignoring the role of hospital size, and emphasizing findings that don’t meet their own stringent standards for statistical significance.  This kind of shoddy work only manages to further confuse policymakers and health care providers working to address the health care needs of vulnerable patients.

[i][3] L&M Policy Research. Analysis of 340B Disproportionate Share Hospital Services to Low-income Patients. 2018 Mar 12.

Source URL: https://340binformed.org/2020/03/flawed-measures-appearance-of-bias-in-340b-study/

---