ChatGPT is a remarkable tool, but it has its flaws. We decided to ask it to explain Medicare’s Inpatient Prospective Payment System. Here’s some of what it said:
The Medicare Inpatient Prospective Payment System (IPPS) is a method used by the Centers for Medicare & Medicaid Services (CMS) in the United States to reimburse acute care hospitals for inpatient services. IPPS was established as part of the Social Security Amendments of 1983 and went into effect in 1984. The main goal of IPPS is to control rising healthcare costs by setting predetermined rates for inpatient hospital services. Here is how IPPS works…
OK, but some quibbles. In theory, at least, IPPS is designed to reimburse hospitals for the national average cost of an admission. If a hospital can keep its cost under the national average, it makes money under IPPS. Roughly 15 percent of hospitals are profitable under IPPS. However, because IPPS excludes some costs in its calculation it can be difficult for some hospitals to beat the national average.
Given on-going increases in healthcare costs, based on ChatGPT’s description, it’s easy to conclude that IPPS has been a failure at controlling costs. IPPS comes with several goals; controlling cost is an important one, but there are other goals including utilization and ensuring hospital access in rural areas. ChatGPT ignores these goals.
The next paragraph is even worse.
Diagnosis-Related Groups (DRGs): IPPS uses a system of Diagnosis-Related Groups (DRGs) to classify patients into groups based on their diagnoses, procedures, age, sex, and other factors. Each DRG has a predetermined payment rate associated with it.
This is so inaccurate and incomplete as to be wrong. Medicare’s system is named Medicare Severity Diagnosis Related Groups. Perhaps not obvious from the name, Medicare policy says that more severely ill patients must also cost more to treat. Cost, then, becomes an important factor on how a claim is assigned to an MS-DRG.
Furthermore a single hospital might have a predetermined payment (or not) because there are multiple factors that go into determining payment. Those payments can also change through the year. The MS-DRG system does not function like a fee schedule as ChatGPT thinks, but is rather a much more complex formula for determining payment.
What MS-DRGs do have are predetermined relative weights which is a multiplier in determining payment. Relative weights are based on nationally indexed costs for each MS-DRG. So while an MS-DRG’s relative weight is the same for each hospital, the payment rate generally is not. Two hospitals close to each other can have significantly different payments for the same MS-DRG.
The ChatGPT answer sounds like it was swiped from a graduate student’s thesis someplace. It’s sort-of accurate, but fails to cover the significant nuances of IPPS. More importantly, it leaves out some vital facts which contribute to a misunderstanding of how the system actually works.