Tuesday, November 7, 2017

2016 CMS Savings Under Chemistry Panel Pricing Rule; Strategic Impact in 2018

In 2016 and 2017, at the summer Clinical Laboratory Fee Schedule meetings, CMS discussed the impact of pending PAMA pricing law on its use of special "panel pricing policy" in clinical chemistry.  CMS has manualized policy (not statute or regulation) that caps the prices of analytes at no more than the price of a corresponding panel.   For example, if a 10-analyte panel is $20, and each analyte is individually $8, and you order 9 analytes, you might hope to be paid 9x$8 or $72, but CMS has special calculations that will cap your price at less the $20.   The CMS rules are complex, but that's enough to understand this article.

In August 2016, NILA and ACLA argued that CMS must discontinue panel pricing rules if and when PAMA sets the laboratory fee schedule in January 2018.  Rather, CMS must pay each code at the PAMA price.  While it may take courts and lawyers to decide, I think ACLA has a pretty good argument.

How much would this cost CMS?   I think there are two answers, one is calculable and one could only be guessed at.   We used 2016 utilization data, released recently by CMS.

In 2016, CMS spent $709,291,438 on clinical chemistry panels.  The vast majority of this was for codes 80053 (comprehensive metabolic panel) and 80061 (lipid panel), which totaled 85% of all panel spending at $604,854,656.


Exactly Calculable Savings from Panel Rule

CMS spent $43,026,852 on payments for the 24 analytes  in 2016.   However, if they had been paid at their fee schedule rates, payments would have been $87,262,430, about twice as much.  Cash savings appear to be $44,235,578 per year for the panel policy. [FN *]

click to enlarge

But Don't Miss The Impossible to Exactly Predict Impact

The above savings are in the existing panel pricing system, where labs have no incentive at all to submit claims for, say, 9 analytes on a panel of 10.   However, if the panel pricing rule vanished like a Penn & Teller trick (the PAMA Panel Vanishing Act), then labs would have a lot of incentive to submit analytes that don't match a panel definition and price.  For example, now, if you drop one code from a $20 panel, you get (say) $18.   Without the panel pricing rule, if you drop one code from a $20 panel, you might get $50 or more at the a la carte rate (e.g. 9*$6). 

This new perverse incentive could affect anywhere from 0% to 100% of the sixty million panels tested in 2016.   While it represents an absolute extreme, if every panel shown in the top figure was submitted as "N-1", the paid cost of the $709M of panels easily double to $1.5B or more, and this would be perfectly allowable since payments would either be on the PAMA code panel price, or the sum of all the individual analytes at their PAMA a la carte prices.[FN **]

Of course, despite the new incentive to submit "N-1" panels, labs in the US wouldn't submit all the panel tests at "N-1" a la carte prices.  So let's say this second effect is 10% of panels, or 10% of $700M, or about $70M.   The two effects together would tally about $110M.   Or more? [FN***]  All this offsets a substantial part of the several hundred million dollar savings caused by the rest of PAMA.

Point of Care Too

Another example is that today, if a point of care test company makes a kit that matches an existing ten test panel, a physician office lab will get about $20 from Medicare for it.  But if ACLA is correct that CMS must pay each analyte at its list price in 2018, if the device company makes an N-1 or nine-test panel, the physician might get, say, 9*$6 or $54 for that.

Excel for the above in the cloud, here.

Comparison to BRCA Panels

CMS performs roughly 25,000 BRCA tests per year.   If they are all coded at the 2018 rate for BRCA1-BRCA-2 as 81162, this would be 25,000 * $2253 or $56,325,000 payments.   If they were all paid at the BRCA Panel Rate (10 genes including the 2 BRCA genes), this would be 25,000 * ($838+$542) or just $34,500,000 payments, giving CMS a CY2018 savings of $21,825,000.   

(If BRCA were stack-coded as 81211+81213 that cost would be 25,000 * ($2396+$553) or $73,725,000 payments, so the savings under the panel code policy would rise to $39,225,000Recall from above that savings from the whole national panel chemistry pricing policy in 2016 were $44,235,578.  (If the new Secretary of Health would use either type of savings, he could pay for an extra $400,000 of charter jet flights every 3.6 days and it would be revenue-neutral.)

PAMA and MAC data for 2016 show that BRCA testing was: variably billed as 81211, or as 81211+81213, or as 81162, or sometimes 81432+81433, but other times 81432, the latter probably only under a MolDx policy that "prohibits payment for dup del analysis unless only it is dup del payment for 81213 on top of 81211").  CMS 2016 billing patterns used the full spectrum of somewhat ambiguous current CPT coding choices:




Policy Note

If ACLA is correct that lab tests must be paid at literal fee schedule prices under PAMA law in 2018, it could affect several MolDX local policies that variably act to upgrade and downgrade payments from fee schedule rates (here).

___

[FN *] Actual direct cash savings might be a bit less because some states pay a bit less than the fee schedule normal price.  However, I believe this is a small effect on a few codes in a few states, whereas the total affect is 50% over all codes and all states.

[FN **] Relative to 2016 fee schedule prices, in 2018, both panel and a la carte prices will typically drop 10%, but again, this is not a main driver of the large effect I'm discussing.  Note that 2018 utilization could probably be 10% higher than 2016 utilization.  I've used 2016 prices throughout the clinical chemistry section of this blog.

[FN ***] Hundred million dollar values aren't without precedent at all.  Recall that in 2014, CYP genetic testing shot up to $270M before being fought back down to only $25M in 2016.  CMS doesn't detect unprotected vulnerabilities quickly.   It has no current defenses against N-1 testing, because until December 31, they've paid 10% less, not 300% more.