Everyone on the planet knows what a coronary angioplasty is, since it makes the news every time someone in public office gets their coronaries reamed out. Angioplasties happen—a lot—and when they do happen, because of significant differences in the language of procedure coding between ICD-9 and ICD-10, even the best map cannot predict actual financial impact. You may have heard me once or twice say, “of course the differences between ICD-9 and ICD-10 are significant, otherwise there would be no point in switching.”* Here is a classic case of that sentence at work in the real world.
The crux of the difference is in how the coronary arteries themselves are classified. In ICD-9, the axis of classification is the number of distinct coronary vessels treated—the left anterior descending, the right main, the obtuse marginal, and so forth. In ICD-10, the axis of classification for coronary artery procedures is the total number of sites treated, regardless which coronary artery it is.
Words matter—you may have heard me harp on this once or twice as well. In this case, the ICD-10 method of describing the total amount of work involved in an angioplasty procedure is more generous than in ICD-9. One site treated would be described in ICD-10 as 02703ZZ Dilation of Coronary Artery, One Site, Percutaneous Approach. Two, three, even four or more separate sites angioplastied—same vessel, different vessel, ICD-10 doesn’t care—are recorded in the code as total number of sites in ICD-10. For example: 02733ZZ Dilation of Coronary Artery, Four or More Sites, Percutaneous Approach.
In ICD-9, however, if all of the sites of coronary artery plaque occur on the same coronary artery, even though they required separate angioplasty treatment, they are all described in ICD-9 like this:
00.66 Percutaneous transluminal coronary angioplasty [PTCA]
00.40 Procedure on single vessel
In ICD-9 and ICD-10 MS-DRGs, all the procedures described so far are currently assigned to the same MS-DRG, 251 Percutaneous cardiovascular procedure without coronary artery stent.
That first example is the lowest MS-DRG in the angioplasty hierarchy. At the top of the hierarchy is when four distinct arteries are angioplastied or four stents are placed. This number of vessels/stents is the cut-off for a different DRG. When a stent is placed for every site treated, the difference between ICD-9 and ICD-10 does not matter because the ICD-10 codes contain the total number of stents placed per site, for example, 027134Z Dilation of Coronary Artery, Two Sites with Drug-eluting Intraluminal Device, Percutaneous Approach. ICD-9 tracks stents and number of vessels separately so there is no correlation between number of stents and number of vessels. (A cardiologist with an interest in a stent manufacturer could place four stents for every PTCA performed, and boost both the stent stock price and the MS-DRG. Not that anyone would do that.)
The difference between ICD-9 and ICD-10 surfaces when four distinct vessels are treated and fewer than four stents are placed. Now the more generous ICD-10 definition of total work, based on sites rather than distinct coronary vessels, bites it for the short term. Since we are stuck with using MS-DRG definitions designed to make the most of ICD-9, ICD-10 MS-DRGs cannot take account of the new terminology using total sites until sufficient ICD-10 data has been collected to design new MS-DRGs that make the most of ICD-10.
The ICD-10 MS-DRGs does its level best to assign the same medical record coded in ICD-9 and ICD-10 to the same DRG. In this situation, ICD-10–coded records with a total of four or more sites but fewer than four stents are consigned to the lower-paying MS-DRG. This is a classic example of real work being done in the real world and clearly described in the ICD-10 that cannot be reimbursed using a payment system designed around the terminology of ICD-9.
A map will not and cannot expose this difference. The GEMs equate sites to vessels in what are necessarily simplified translation alternatives. See the GEMs documentation for technical users, part 3 for details. The translation disconnect between the two systems is hairy enough—if the GEMs were to attempt to give every possible PTCA alternative based only on the code, it would take a whole separate GEM to list all the possibilities.
I’ve shot my blog wad on the general principle, so I have to split this in two. Part two will show how this difference plays out in real life and in the sometimes bizarro world of financial impact forecasting.
*This sentiment borrowed from the principal inventor of ICD-10-PCS, Dr. Bob Mullin, sorely missed in recent years as a mentor and commiserator.
Rhonda Butler is a Senior Clinical Research Analyst with 3M Health Information Systems.