Wednesday, October 14, 2020

Medicare's Unorthodox Spending on Code 81408 in CY2019: 80% in Red States

81408 Isn't Just a Code, It's the HIGHEST PAID Genetic Code

81408 isn't just any CPT code, it's the highest-paid genomic code in 2019 (here).  Growth exploded, in round numbers, from $5M in 2017, to $100M in 2018, to $300M in 2019.   Everything about this story is bizarre.

80% of the 2019 payments for 81408 go to 2016's red states, and 100% of the 2019 payments go to red states plus NJ and DC, which are served by a MAC in an adjacent 2016 red state.

_______

The Bizarre Explosion of 81408 Billing

I've previously reported that Medicare's overall molecular pathology spending has been rising dynamically, from about $500M in 2017 to about $1B in 2018 to about $1.7B in 2019.   See previous blogs, one focused on 81408 spending as a whole (here), one focused on the spending being 99% from Novitas and FCSO MACs (here), including data on who in particular billed that code in the most recent year with NPI-level data, which is 2017 (here).   

Early billing on 81408 in CY017 is associated with DOJ genetics fraud press releases and news stories in 2019.  


81408 in Medicare is ALWAYS Highly Anomalous

81408 is what AMA calls a "tier 2 level 9" code which holds a grab-bag of genes studied by full sequencing whose use is by definition rare, and often for obscure fatal pediatric disorders.

As I've discussed earlier, payments for 81408 are de facto extremely anomalous, because nationwide Medicare payments were almost zero in 2017 and earlier, and because later payments occur only in two MACs, which have policies that explicitly have no edits for 81408.  

In the other five MACs, 81408 is nonpayable as not medically necessary.  Said differently, the highest-paid CPT code in 2019, 81408, is medically nonpayable in 5 of 7 MACs.  


81408 Paid Mostly in RED States in 2019

A new data analysis today shows that 81408 in CY2019 was paid 78.5%, almost 80%, in RED states.




A few more findings.  

In Blue states, payment, if any went to NJ & DC.

First, while around 20% of 81408 payments were in blue states, it was highly focused, and never occurred in either NGS MAC or MOLDX MACs.   

Said differently, 19/21 blue states got ZERO payments for 81408.

Nearly 100% of the spending on 81408 in BLUE states was in either D.C. or in New Jersey, two (2016) blue states whose payment policies are governed by the Novitas MAC, which is in Pennsylvania, a 2016 red state.   

CMS MUE Edits for 81408 - OMG

Second, CMS medically unlikely edits are in place to help stem fraud, and provide no more than units of 1 for most genetic tests, but allow TWO UNITS PER DAY of 81408.   Again, absolutely unbelievable data. (I almost wrote, un-effing-believable.)


Why hasn't CMS released missing May 2020 date on CY2018 payments?

Third, and finally, in a departure from official policy, CMS has declined to release CY2018 payment data by NPI and CPT code, which it should have released in May 2020.  See blog on this topic here and see background here including my correspondence with CMS this fall.   

Since CMS isn't releasing the 2019 NPI/CPT data on the standard CMS website for this data, I've separately put in my own FOIA request this week, here.

Note that this data (NPI/CPT) should be released annually by CMS in May.  It's not ad hoc.  It reflects official policy positions of CMS (see here and see Federal Register, 79FR3205, January 14, 2014 here.)

Helping Voters Decide 

If 2018 data were released by CMS, as would normally happen last May, stakeholders could see ahead of the election whether recipients of the anomalous payments were also big political donors.


___

I note that 2019 81408 payments were 78% in red states.  

Obviously, in 2016 there were more red states, with an electoral count, roughly reflecting population, of 57%.  

However, as to the circa 44% population in blue states, who got 20% of 81408 billings, 100% of that went to only two places, NJ and DC, under the Novitas MAC.  19 of 21 Blue states, including New York, Illinois, and California,  had exactly 0% of the 81408 gravy train.  Which, as far as we know, may be as fast or faster in 2020 as it was in 2019.