Wednesday, March 2, 2022

Reliving Medicare's Dumbest Reimbursement Decision Ever (2009)

Last week, at TRICON in San Diego, I gave a talk on CMS policy and incentives and disincentives to innovation (there are some of both).  It's being written up as a small publication.

I was reminded of an impossible, and quickly rescinded, CMS reimbursement decision in 2009.

Doctors were willingly using Product A, which was 10% the cost of Product B, and someone at CMS made a coding change that made Product A impossible to use.   It's also a case study for what happens when you are making strategic decisions but you lack the ability to think 1 or 2 steps ahead.

And there's a broader lesson for how CMS should incent innovation.   In the ACO world, CMS incents costs savings by "shared savings."  If the ACO's expected Medicare budget is $100M, and it comes in at $90M, it gets to share 50% or $5M of the savings.   But in the RVU system, for example, when a product comes in faster or cheaper, CMS immediately seizes 100% of the savings by immediately cutting the price of the innovation to the least possible level.   It would be smarter if CMS would actually incent providers to use the less expensive thing.

The Story


Before 2006, there were very limited treatments for wet age-related macular degeneration, which has an incidence of about 200,000, and a prevalence of about 3M, mostly age 55 and older.   The year 2006 saw the approval of LUCENTIS (ranibizumab), which is similar to AVASTIN (bevacizumab) but has a smaller molecular weight.

Even before Lucentis was FDA-approved, its success was known from Phase III trials, and there were some off-label clinical attempts to use ocular injections of Avastin.  You can still see online debates dating to 2006 (here) and 2009 (here and here) while an NIH-supported comparison trial found the clinical efficacy of the two agents was similar (here) in 2012.  In 2012, CMS held a MEDCAC (public policy meeting) on VEGF treatments in diabetic macular disease as well as macular degeneration, (here).


Leaving aside biosimilars codes (Q5108, Q5107), the Avastin code is J9035, 10 mg $67 and the Lucentis code is J2778, 0.1 mg $307.   Dosing of Lucentis is 0.5 mg per eye, or $1500 per eye per month.


As I understand the story, and as retold by the Washington Post in 2013 (here), Lucentis at that time cost about $2000 per injection and Avastin about $50 per injection (J9035 10 mg).  

So far so good.  But not for long.  

Someone at CMS decided to focus not on the apparent $1950 price saving for CMS ($2000-$50), but on the fact that Avastin was given in a much smaller dose (like 1 mg) but was being coded and paid for the smallest available code, "10 mg $50" or so. 

In a still-available document, dated August 28, 2009 (T1803, CR6626) CMS created new code Q2024, bevacizumab, 0.25 mg.   Thus, instead of the ophthalmologist being reimbursement in the $50, or $67, range, he would be reimbursed in the range of $1 per 0.25 unit, or $4 per mg meaning $4 per dose.

  • This coding for Avastin's tiny dose with Q2024 now put the ophthalmologist far below cost, including incorporated burdens like compounding fees.  
  • The brouhaha was immediate and loud.  

The August 28 CMS decision led to an October 2  article in Wall Street Journal, ridiculing CMS.  Here (archive here).  

See an October 6 article on the same topic at Healio here.

An OIG report on the topic attributes the recission of the code to letters from Congress to CMS or HHS in October 2009, and the decision to delete the code at the latest by November 2009.  See OIG here.   According to PolicyMed, the Congressional letter came from Senator Kohl (D-WI), here.  Kohl is said to have written, " a variety of medical authorities and advocates are complaining that the new coding system CMS implemented this month will reduce reimbursements to physicians for Avastin to a small fraction of the previous rate.”

Kohl praised CMS's decision to rescind the goofy coding change on October 28, 2009 (here).

Click to enlarge. From OIG report 2012.


I haven't identified the exact transmittal that deleted Q2024, but it was deleted by December 31, 2009, according to current HCPCS records. 

Regarding Q2024, see also T1805, CR6594, also dated August 28, 2009 here.


At least one MAC would code Avastin today by J7999, a generic compounded-drug code with local pricing (here).

At the time, in 2009, CMS didn't have an administrator, Mark McClellan having left in 2006 and Donald Berwick having arrived in 2010 (a job he held only 18 months).  Here.

There's CMS rulemaking around hospital outpatient drug pricing for Q2024/C9257 (citing Q2024 as deleted), November 20, 2009 74 FR 60491.

Kohl issued a press release October 28, 2009, praising CMS recission of the code, ad the press release contains a link to a PDF letter to CMS, but that link is now a dead link.  See press on the press release at MedPageToday from AAO here.

A note in Fierce Pharma October 14 basically quotes the WSJ October 2 article.