Saturday, May 31, 2014

Judges Reject Decades-Old Medicare Ban on Transgender Surgery

On May 30 and 31st, the media buzzed with reports that judges at Medicare had suddenly thrown out the longstanding federal ban on payment for transgender surgery.  For examples, see the CNN report, here, and USAToday's coverage, here.

I won't talk about the important social and societal issues, but if you would like more information on how the wheels turned, how the rules and procedures played out within the Medicare agency, and links to hard-to-find key documents, see more after the break.


What Happened?

As most of the news reports captured accurately, a federal board of three judges overturned Medicare's ban against transgender surgery, while leaving some implementation judgments to the agency.    As CNN stated:
     The decision on whether Medicare pays for sex-change operations is now in the hands of regional administrators, thanks to a decision Friday by a federal board that overturned a national policy of not covering such surgeries.
     The ruling by the Departmental Appeals Board of the Department of Health and Human Services gives authority to 15 [sic - 10] regional contractors who administer the Medicare program to decide whether sex-change surgeries will be covered in their territories.
     "The national policy barring Medicare from covering gender transition surgery has been invalidated," said Centers for Medicare and Medicaid Services spokesman Aaron Albright in an e-mail to CNN.
     "As with all such determinations, CMS will carry out this independent board's ruling through Medicare Administrative Contractors, who manage Medicare claims payment systems. These contractors may cover this care case-by-case or under a local coverage determination based on clinical evidence to determine medical appropriateness," added Albright, whose agency administers Medicare and Medicaid.

What was the NCD?

The National Coverage Determination (NCD) dates to the 1980s, and was found in a pre-2005 manual called the "Coverage Issues Manual."   These old decisions, often only a paragraph long, were issued a generation ago by CMS administrators (then, the agency was called HCFA - the Healthcare Financing Administration).   Some of these old decisions are short and arcane - for example, an NCD from that era says that Medicare may not cover a white cane for blind people, because the cane does not "treat" or cure blindness.  (See a current essay on the white cane, at NYT).

In sharp contrast, a modern National Coverage Determination is now frequently 20-50 pages long, containing an elaborate analysis of the published literature on the topic, a review of credible guidelines, and are subject to public comment and review.  It now takes the agency six to twelve months of review, draft announcements, and public comment on the draft documents before a new NCD can be finalized.

The old transgender NCD was very short, simply stating that transgender surgery was not validated by randomized trials, had risks, and was therefore "experimental" and not covered.    The full text is here:

Definition    Transsexual surgery, also known as sex reassignment surgery or intersex surgery, is the culmination of a series of procedures designed to change the anatomy of transsexuals to conform to their gender identity. Transsexuals are persons with an overwhelming desire to change anatomic sex because of their fixed conviction that they are members of the opposite sex. For the male-to-female, transsexual surgery entails castration, penectomy and vulva-vaginal construction. Surgery for the female-to-male transsexual consists of bilateral mammectomy, hysterectomy and salpingo-oophorectomy, which may be followed by phalloplasty and the insertion of testicular prostheses.
Indications and Limitations of Coverage 
     Transsexual surgery for sex reassignment of transsexuals is controversial. Because of the lack of well controlled, long term studies of the safety and effectiveness of the surgical procedures and attendant therapies for transsexualism, the treatment is considered experimental. Moreover, there is a high rate of serious complications for these surgical procedures.
For these reasons, transsexual surgery is not covered.
The statements appear online at CMS, as cited, and the statements were also published in the Federal Register, August 21, 1989; 54 Fed Reg 34555, 34572.  [As cited by the judges.]

Appeal of an NCD Has Two Confusing Phases

The Medicare Modernization Act of 2003 - which created the Part D prescription drug program - put the definition of National Coverage Determinations and Local Coverage Determinations in statute, concisely, to create appeal processes for each.    See 42 CFR 426.  The appeals must come from a beneficiary who needs the service that is non-covered by the NCD.   However, once such a patient is nominated, they can be helped by pro bono law firms, stakeholder groups, and associations.

It is crucial to understand that the NCD appeal follows two phases.

  • In the first, the judges determined whether the written record for the NCD is adequate.   This is what happened on December 2, 2013 - judges ruled that the written record was not adequate to support the NCD as it had been written and implemented (here).   And this December ruling marked the first time an NCD appeal had made it to first base since the 2003 appeal procedures were created.   
  • A second phase of these cases then commences, which in this case would aim to determine the NCD fails under the statutory standard of making agency policy that is "reasonable" in the view of an objective individual. Since almost all NCD appeals have been dismissed, historically, the next phase of this case trod new territory.  It is this second phase that was released on May 30,2014 (on the CMS website now, here).   
It helps to put this the opposite way: under the applicable rules for Phase 2, the judges would have found the NCD could stand, as long as it was not clearly "unreasonable."  In other words, the judges would defer to the position of the agency as long as it could be propped up as representing at least one reasonable position, perhaps among several alternatives.  If an agency could choose one of four policy positions, A,B,C,D, and A,B,C are at least reasonable, and D is simply objectively unreasonable, the only position that an outside party could have thrown out under a reasonableness standard would be "D."

  • It is somewhat confusing that this is a "reasonableness" standard for a policy position that could apply to any federal agency - the Department of Energy, etc.  (Has the Department of Energy taken a position that is "reasonable" - at least, not overtly unreasonable, even if not the outcome preferred by the litigants).  Medicare also frequently uses the same word "reasonable" in a different sense and context, "reasonable and necessary" evidence for a medical service to be paid.
  • Re: "reasonable" standard for striking or upholding agency decisions, see Pojanowski, NW Law Rev 2014, 104.799, online here.   This echoes the Chevron deference to agency decisios (here) - an agency's implementation of statute need not be the only choice allowed, nor even the best one (depending on who judges "best"), but only a "permissible" one.  If permissible implementations are A,B,C, and the agency chooses A, the litigant wants B, and the judge prefers C, the agency's choice "A" will stand in deference to a permissible interpretation.   Chevron cases may refer to textual analysis of the law (where is the common, what does this adjective refer to.)  The Hayes case that struck down the Least Costly Alternative drug pricing found that the agency's interpretation of statute failed to meet a Chevron test.

I'm not sure why the two stage process is used - first, whether the record is adequate to support the NCD, and second, whether the position is frankly unreasonable.   It seems if the record is adequate, then the case can stop and no further judicial resources are required.  Perhaps, hypothetically, if the record is inadequate, the agency could somehow at the second stage still justify the policy as hanging on by a thread... under the low legal standard for phase two, of at least not being completely unreasonable.

National Coverage Determination appeals have their own home page at the agency, here.  All complaints registered to date have their own table on a separate page, here.  Transgender surgery was the twelfth acceptable complaint logged in twelve years.   It appears that all the other complaints listed below, although formally acceptable, were dismissed without an action favorable to the moving party.

This complaint, the twelfth, received the holding in December 2013 that: Ruling issued finding that the NCD record was not complete and adequate to support the validity of the NCD. [Styled as NCD Ruling #2].  And then, in May 2014, that furthermore: the NCD failed to meet the standard of "reasonableness" as an agency policy position. [Styled as DAB Ruling #2576].

What Happened in This Case

One of my first questions was, since the paperwork for this complaint was filed in March 2013, and most subsequent paperwork seems to have been filed by July 2013, why is this suddenly news on May 30, 2014?    Did the judges take that long for a short ruling that seems to reflect an open and shut opinion?  

As described above, the first part of the case was decided on December 2, 2013 (the record was inadequate to support the NCD - the record being a box of yellowed 1970s documents, I suppose). this was the only document online on May 30, 2014, when the news story broke.

The case was filed on March 26, 2013, supplemented on April 18, 2013, and ruled as acceptable on April 29, 2013.  CMS submitted the "NCD record" (e.g. its historical files) on May 15, 2013, and the moving party submitted a report on June 14, 2013 that the case files were not adequate to support the NCD.   Six advocacy organizations supported the case with filings from June 20 to July 10, 2013.  Even before the final filing, on June 26, CMS notified the Department Appeals Board that it would not contest the case.

The Technology Assessment.  The NCD was based on a 1981 report by the National Center for Healthcare Technology (NCHCT).   While unfortunately I could not find this report online, if you google the words "NCHCT evaluation of transsexual surgery 1981" you will find a variety of secondary sources that discuss it (e.g. an online essay here which attributes its authorship to Janice Redmond) and this may be a optical character reader copy of the NCHCT document here. [NOTE: Prior link showed the NCHCT document in 3/2014; it doesn't work as of 6/2016].  The National Center for Healthcare Technology is usually considered a predecessor to the current AHRQ, e.g. here and here.   By 2012, a California State Department of Insurance document specifically calls out nonpayment for transgender surgery as discriminatory (here).  This 1980 NCHCT report was harshly criticized as dated and prejudicial by the current advocacy groups, and the three judges appeared to agree.

CMS Could Have Cut the Process Short

There is another side to the back story as well.   In March 2013, CMS apparently opened, for one day, comment on a motion to withdraw the NCD.   (2013 online article links, though no longer correct, point to an NCD page).  This was mentioned in Buzzfeed's 2014 article, here.  The online website Advocate has an archive that still shows two articles from March 2013, one announcing the opening of the NCD change (here) and one about twelve hours later announcing that CMS had withdrawn the webpage, here.  At the time, CMS said, it should not reopen and review the NCD because a legal action had been filed.  This seems to be an option that CMS chose -- not one they were forced to choose.   CMS would have been within its abilities to revise (e.g. delete) the NCD.  In 2006, CMS changed an NCD in response to an NCD complaint (Case A-04-89).  While this case was working its way through the appeals process from March 2013 to May 2014, CMS  created a public policy to rapidly withdraw dated and obsolete NCDs (August 2013), and posted ten NCDs for withdrawal after a short 30 day comment process on November 27, 2013, but that list of ten obsolete NCDs did not include the obsolete transgender one (see CMS here).   Even in March 2013, before the expedited renewal process was formalized, CMS could have accomplished removal of the NCD in 90 days or so.  For comparison, this would be like the government withdrawing the "don't ask don't tell" policy when it was challenged in court, rather than letting the court process continue for a year or two.

The current story appears to have appeared in Friday's news (May 30) - for many businesses and organizations, Friday news releases are often considered stories that will get less visibility than weekday stories.   CMS did not issue a website proactive press release on this rather convoluted policy matter, although it did field questions through a press officer.  The press seems to be favorable, for example, including the Fox News story, here.

May 30, 2014: The Release of Phase 2 of the Case

Three detailed reports by Chris Johnson at Washington Blade do piece together the administrative process and match up to the CMS website.  On April 1, 2013, Johnson noted that a beneficiary had filed an appeal in the case, here.   On December 12, 2013 (here), Johnson reported the 12-page judicial ruling I quickly found on the CMS website - that judges had found by December that the evidentiary record was inadequate to support the NCD.  The December 2013 ruling made only the tiniest blip in the media.

On May 30, 2014 Johnson reports his reading of a new 28-page document finding further that transgender surgery is medically reasonable and necessary (his story here and the 28 page document here.).

Note that the official online record of the case was then brought up to date.  Whereas the initial public posting found that the "the NCD record was not complete and adequate" [NCD Ruling #2] the finalized public posting shows that the NCD holding is now in fact, by legal standards, not a "reasonable" one [DAB Ruling 2576].  Again, this does not mean the NCD was not "reasonable and necessary" but rather that the NCD was simply not "reasonable" at all.

An interesting hypothetical - since the NCD was found "not reasonable" - is whether the NCD was ever reasonable, but today, in 2014, it no longer is.   This would be like holding that a law demanding school segregation and no mixing of races was once reasonable (say, in 1840 or 1940) but no longer is (say, as found under Brown v Board of Education in 1954).

Bringing CMS into the Present

A brief online review of commercial payer policies show that they often cover transgender surgery.  For example, see policies at Anthem Blue Cross, here, which was representative of some other policies I found online.  Several other major national insurers had similar policies covering transgender surgery.   Even the Medicaid program in California - MediCal - covers transgender surgery, see here.

The Transgender NCD as an Interesting Case Study in Risk-Benefit Decisions

A few months ago I wrote a blog on a few aspects on the evaluation of risk-benefit in medical decisions - here.  Of course, there are hundreds of articles and even full length books written on this subject.   A recent MedCAC meeting on the evaluation of CT screening for smokers to reduce deaths from lung cancer also hinged on hours of discussion by panelists and stakeholders about risk-benefit decisions (here).

Taken on its own terms, the old Medicare decision said that there were three reasons for non coverage of transgender surgery:

  1. It was controversial.
  2. There were no randomized controlled trials.
  3. There were surgical risks, that could not be balanced against "benefits."
What would a "randomized controlled trial" look like?  There is hardly placebo transgender surgery, so critics could say the trial (even if conducted) was too flawed to interpret.   The "risks" of transgender surgery could certainly be established in one arm trials (e.g. a 2% death rate or other risks).   The "randomized" arm would obviously have no risks of surgery.   But even with more data of this type, it would be hard to balance risks and benefits at the level of a white-coated panel in an ivory tower, since the difference in kind between the "risks" and "benefits" is so enormous.   Obviously, for someone with mild and variable gender dysphoria, the risks would be too large, and for someone with debilitating and badly life-impairing gender dysphoria, the risks would not be too large (we know many such people have chosen the surgery).   So there isn't a simple answer if the risks exceed the benefit in the abstract, when the frame of reference is untied from the condition and judgment of a particular patient.   Interestingly, this idea - that risk/benefit decisions may be unique to patients - is currently much under discussion in regulatory circles in the US and Europe. For example, such-and-such a multiple sclerosis drug may be worth the risks for certain patients with certain histories and needs, and not appropriate for others.

While the future Medicare surgical cases will now be up to the discretion of its ten local contractors, it would be surprising indeed for any of them to try to hold the surgery is always not reasonable and necessary, since the judges have determined that holding that position is not within the domain of "reasonable" decisions that one can hold.   In theory, they could determine that a particular individual patient did not have significant enough gender dysphoria to make the surgery "necessary."  

Coding Corner

It is an issue far below the social, policy, and medical ones, but I wondered if there were CPT codes for transgender surgery.  I haven't made an exhaustive search, but there don't seem to be, although there are scattered (and unpleasant sounding) codes such as "amputation of the penis."   For hospital procedures, a different procedure code set called ICD-9-PC and ICD-10-PC would be used - I couldn't google my way to find any ICD-PC codes for transgender surgery, either.  ICD-9 and ICD-10 disease codes for gender dysphoria exist.  I found one list here.

A final bit of coding trivia

I found this:

A final note

On August 1, 2014, another case, the 13th NCD appeal, was opened, regarding the NCD for vagus nerve stimulation, but no further public docket is available, at least through 9/2014.