June 2012, Vol 1, No 2
Rapid Changes in Reimbursement Protocols for Molecular TestsUncategorized
Molecular tests have proliferated over the past decade, bringing the precision of genomics into reality in the clinic, but the payment system for nucleic acid–based tests has been unchanging and primitive. This unfortunate state of affairs has limited the ability of payers to encourage (or control) the use of genomic tests and has seriously handicapped the ability of researchers to understand what genomic tests are being adopted into clinical medicine, how they are being used, and whether they are having an impact on health outcomes. In an era where there is increasing emphasis on the need for “practice-based evidence” (from observational studies and pragmatic trials that can supplement formal randomized clinical trials), our electronic medical records and our transaction systems for healthcare claims must be up to the task of telling us what genomic tests are being adopted.
Fortunately, an entirely new coding system is being introduced into the US transaction system for providers and payers in 2012 and 2013. Since the regulations and policies governing the use of medical procedure codes, especially in laboratory medicine, may be arcane to many stakeholders, we will provide an overview of how the current system works before describing the impending changes. While policies for coverage and payment may differ somewhat between private payers and federal payers like Medicare, we will focus on the standard-setting efforts of Medicare as the prototypic payment system for outpatient procedures. This article discusses why coding systems are really quite important for the smooth functioning of our healthcare system and how the coding system for genomic tests is being reengineered. We also sketch the shape of a policy battle among different stakeholders as to what payment system should be employed from 2013 forward for genomic tests. Finally, we discuss a few problems that the proposed coding system cannot yet address.
Codes as a Scaffold for Healthcare Procedures and the Flow of Funds
Despite calls for a more rational and holistic approach to payment for healthcare services in the United States, much of the reimbursement for outpatient healthcare flows through line item reimbursement – the longstanding “fee-for-service” system.1,2 Because thousands of individual services, drugs, and tests exist, the operation of a modern fee-for-service system requires a highly granular system of codes that allows providers to communicate with payers, informing the latter of what particular services have been performed on a given day. The US healthcare system has evolved a family of different, independent, legally recognized coding systems for different types of items and services: there are different code sets to categorize hospitalizations, physician services, laboratory tests, drugs, and supplies.
While the underlying coding systems for government and private payment systems are fundamentally the same because they are set by law, a particular payer can determine the means by which it negotiates or sets prices for the service represented by a certain code. For the largest payment system, the Medicare system, each set of codes (which may contain literally thousands of distinct codes) is crosswalked to a publicly available pricing table in which each service is assigned a price. And each Medicare pricing table reflects the result of the machinations and operations of one or another incredibly arcane set of Medicare policies – different sets of policies, often amazingly complex and granular, for hospital services, physician services, clinical laboratory tests, and drugs.
From the perspective of an administrator, Medicare’s prices are the result of a series of transformations from a clinical event to a coding system and finally to a pricing system, resulting in a line item dollar reimbursement that is sent back to the provider of the medical service. An economist’s view of the transaction would be different: it is a view where the dollar transaction is paramount, because a price encompasses crucial signals as to how the marketplace and recipient of a service value that service. Price tells the producer what to provide, when, where, and in what quantity. Prices are the invisible but powerful guiding hand of the market. Little of that wisdom is conveyed by Medicare’s prices, however, at least not intentionally, because Medicare’s prices blindly follow one or another set of dry accounting rules to distribute a limited pool of payments based on various tables of data and rules that manipulate those bits of data into a price.
This background leads to how Medicare reimburses for genomic tests. Up until 2012, Medicare has used an extremely simplistic system to reimburse for human germline and somatic genomic tests, consisting of a handful of mostly 20-year-old codes that represent a few basic laboratory bench processes like “DNA extraction,” “nucleic acid amplification,” and “sequencing.” Although there was a bit of state-to-state variation in pricing, these molecular steps generally paid about $20 each. A genetic test that required 5 such steps usually was reimbursed about $100, and a test that required 10 molecular steps was likely reimbursed about $200. Medicare (and most other governmental and private payers) reimbursed for genomic tests in this way because a small handful of process-specific codes were the only means provided in the federally designated codebook for laboratory tests, the Current Procedural Terminology (CPT) manual of the American Medical Association (AMA).3
We are at the edge of a truly unprecedented increase in the available codes for genomic tests, and they will soon be far more specific than in the past. By 2013, the AMA CPT manual will contain well over 100 specific Category I codes for named genetic tests, and another set of 9 codes representing tiers of increasing complexity. These tiers will collectively represent many dozens to eventually hundreds of additional genetic tests, with each named test assigned to a specific tier by text found in the AMA manual. During the 2012 calendar year, and perhaps into 2013, Medicare will create tables that match each new code to a price. Medicare policy staff will laboriously assign a new, administratively specific price to each of the AMA CPT genetic codes.
This article provides an orientation on how the laboratory test coding system for oncologic genetic tests worked up until now, how it is changing, and explains some of the policy options that Medicare administrators will face in 2012 as they apply one or another set of rules to the price-setting process. The outcome of these deliberations is important for the future of personalized medicine in oncology, because price signals carry enormous weight in encouraging or discouraging the supply of new medically necessary tests. In addition, the weight of these implications carries beyond Medicare beneficiaries alone, since most private payers closely follow Medicare’s tables for reimbursement, in lieu of the impossible task of setting thousands of prices de novo at each of hundreds of individual US insurers.
The Code Set: How It Was Through 2011 The AMA CPT code set in its modern form with 5-digit classifiers was initially established in 1970 and is updated annually.4-6 While the Health Insurance Portability and Accountability Act of 1996 is best known for its privacy and security rules, it also required the US Department of Health & Human Services to specify by regulation mandatory public code sets for transactions between healthcare providers and payers. In 2000, regulations finalized by the Centers for Medicare & Medicaid Services (CMS) endorsed and locked in the AMA CPT code set, the obligatory code set for most outpatient transactions such as physician and laboratory services.7
The CPT established the first molecular test process codes in 1993. By 1997, there were still only 6 molecular process codes (such as 83890, molecular diagnostics: isolation or extraction) to convey the provision of genetic tests to insurers.8 By 2010, despite enormous advances in genomic technologies and the widespread use of genetic and genomic tests, the AMA CPT handbook still provided only 20 codes, each reflecting one or another single generic process used at the laboratory bench (in some cases today, in electronic DNA test cartridges) during the workflow of genomic testing. There was increasing pressure from payers to provide a more comprehensive and rational code that would correspond to the actual test being performed – for example, a KRAS genetic test for a lung cancer patient, a BRCA genetic test for a woman at high risk of breast cancer.
The Code Set: How It Will Be in 2012/2013
The multidisciplinary AMA editorial panel that approves the entry of new codes into the CPT manual meets 3 times a year, approving codes to be published in the following year. Under the auspices of the editorial panel, a workgroup was convened by the College of American Pathologists and the Association of Molecular Pathologists. Over a 2-year period, the workgroup vetted the wording of over 100 new molecular CPT codes, passing them on in groups of several dozen codes to be approved by the AMA editorial committee.9
Early on, the workgroup determined that there would be approximately 100 gene-specific codes, representing genes that were either tests for germline mutations or for somatic mutations (eg, cancer).10 After those codes were established, the workgroup created another level of 9 CPT codes that represent levels of complexity for molecular tests and a list of multiple-named genes eligible to be represented by that code. In the future, highly validated and relatively popular tests may be assigned new specific CPT codes, and new but less common genetic tests will be assigned in groups to the 9 levels of the Tier 2 molecular codes. The new codes occupy 9 full pages of the 2012 edition of the AMA CPT handbook.3
A Pricing Policy Dilemma Blocks Medicare’s Use of New CPT Codes
For policy purposes, Medicare has long divided laboratory tests into 2 broad groups: laboratory tests that are viewed as physician pathology services, and clinical laboratory tests. Like MRI or PET scans, physician pathology services require a licensed physician to interpret them and issue a report. Medicare must distinguish between personal services of a physician (a Part B service) and services of an institution (eg, the technical component of a PET scan) in order to enforce longstanding distinctions between Part A (institutional) and Part B (physician) services. Usually the choice to categorize a laboratory test as a physician pathology test and a clinical laboratory test is unequivocal: for example, only a physician can sign out the intraoperative frozen section of a suspected tumor mass, while a thyroid hormone test in a blood sample is a benchtop laboratory service requiring the expertise of a technologist and a laboratory director (who may be a PhD or MD). However, as in most matters when a bureaucracy becomes involved, a few decades ago Medicare found it had to write specific regulations to enforce this seemingly commonsensical concept.
In 1980, during the Carter administration, Medicare issued a federal policy notice sharply limiting what services would be payable to a physician pathologist, resulting in a lawsuit against the US Department of Health & Human Services by the College of American Pathologists. In 1982, during the Reagan administration, Congress passed a new law that gave Medicare line item statutory authority to determine what was, and was not, payable as the services of a physician. The resulting regulations survived a court challenge in 1983 and are largely unchanged today. Medicare specified as a general rule that services payable to a physician (any physician) must be usually provided by physicians, followed by a more specific rule singular to pathologists: laboratory tests payable to a pathologist must be either surgical pathology services or specific cytology and hematology services determined to “require” a physician (such as the frozen section of a tumor mass noted above). Medicare also allowed payment for a consultation between a pathologist and a clinician in narrow circumstances when the consultation is requested in writing, not as a standing order, and required that such a consultation be based on a test report outside the expected range for the patient’s condition.
These 20-year-old regulations unexpectedly took on a new life in 2011 following activities at the AMA. Although the “stacking codes” for molecular tests had been classified since their initiation as clinical laboratory tests (ie, as a type of clinical chemistry test), the AMA’s new genetic codes were taken for review by the AMA Relative Value Update Committee, which exists to evaluate and value services paid on the Physician Fee Schedule, such as personal services of physician pathologists. Relative value units were assigned to the codes, as they were assigned to other new physician services, such as surgeries, and the results were forwarded for the review and approval of CMS.
Faced with a dilemma of whether to evaluate the new genomic codes on the Clinical Laboratory Fee Schedule or on the Physician Fee Schedule – 2 entirely different pricing processes with different rules – CMS announced on November 25, 2011, that it had assigned the new genetic CPT codes to neither fee schedule and would not use the codes at all in calendar year 2012.11 In a response letter, the College of American Pathologists criticized this decision (it was “astonished”) and urged CMS to use the codes, and to use them under the valuation processes and rules of the Physician Fee Schedule by classifying genetic tests as the personal service of a physician.12 In short, this request would have moved hundreds of millions of dollars worth of genetic tests from the ledgers of the Clinical Laboratory Fee Schedule (and its associated policies) to the Physician Fee Schedule.
What Will Happen in 2012?
CMS shunts new CPT codes for laboratory tests into 1 of 2 different pathways for pricing. Most new laboratory tests are of the clinical chemistry type, not pathologist services. These tests follow regulations that require announcement of the new clinical laboratory codes in July, solicitation of public comments as to how they should be priced, and publication of the final agency prices around December 1, effective for the upcoming calendar year. CMS may assign prices for clinical chemistry tests by “crosswalking” the codes to the price or sum of prices of existing (prior) codes, or may “gapfill” the prices by asking its claims processing contractors to assign prices in the first half of the coming year, from which it calculates median prices for national use.
Alternatively, CMS can treat the new laboratory test as a physician pathology service, in which case it publishes provisional values from the Relative Value Update Committee for the professional interpretation component of the test and for the technical component of the test in June. (The professional component is payable to the physician even in cases, like that of a hospital inpatient, where the technical component is bundled to a Diagnosis-Related Group and is not paid separately.) The assignment to a fee schedule has other policy implications. For example, as with other physician services, a 20% copayment is applied to physician pathology tests, whereas currently Medicare policy requires no copayment for clinical laboratory tests. Medicare takes comments on these Physician Fee Schedule payment values over the summer and publishes final prices for the coming fiscal year about November 1. In some cases, the prices appear for the first time on November 1.
While the outcome of Medicare’s policy decision will play out in the rest of 2012 and may carry into 2013, it seems likely that the new genetic codes will end up on Medicare’s Clinical Laboratory Fee Schedule. This is because of Medicare’s own regulations: Medicare wrote its own rules that determined that services payable to a pathologist must either be surgical pathology services or “require” a physician, and most genomic laboratory tests are signed out by PhD-holding lab professionals. Therefore, simply put, the tests do not meet the current legal bar of “requiring” a physician. However, the resolution of the fee schedule question depends on Medicare publicly announcing provisional decisions, taking pub – lic comments, and then coming to a final decision. Arguments for assignment to the Physician Fee Schedule include the position that it is a “cost-saving methodology”12 and will result in copayment requirements for patients, which could be a deterrent to test utilization.
Assessment of the New Genomic Codes There is no question that the 100-plus new genomic codes are a great improvement in precision over the process codes that bluntly described processes like DNA amplification and gave no clue to what gene was being tested. However, the CPT coding system is not designed for rapid change, and the advances in genomic testing are already rapidly accelerating. It is likely that at least for some specialized applications such as tumor exome analysis, techniques based on next-generation sequencing will become a practical clinical methodology within the next few years, but the coding system for 2013 is already set and makes no allowance for high-throughput sequencing approaches. To date, there have been no CPT codes for gene-expression algorithm–based tests, which are well accepted in US oncology practice for breast cancer prognosis and clinical decision making. The AMA CPT editorial panel has proposed a new group of codes, called Multi-Analyte Assays with Algorithms (MAAAs), which will appear in the AMA CPT manual for 2013. It may take a year or longer for the Medicare program to adopt these codes, but they will facilitate tracking the usage and utility of the tests in clinical practice.
Of some concern is the decision of AMA policy staff to include most future genomic tests in “bucket-like” categories of only a few codes, which are being called Tier 2 genomic codes. In this arm of the coding system, a dozen or potentially a hundred or more unique genomic tests could be represented by only 1 code. This would imply that it would be difficult for payers to price discriminate among tests represented by the same code and impossible to differentiate coverage policies among tests that are presented by Tier 2 codes.
An additional concern voiced by some stakeholders is that payers may want to know if a laboratory test is FDA-approved or not. The FDA has stated in guidance documents that it will, in general, only approve new indications of targeted drugs that require a genetic test if there is a corresponding and FDA-approved brand of the test. However, the FDA generally cannot enforce what test is used by clinicians and laboratorians. Payers will be concerned whether the test being used for the patient, and for which they are being billed, is appropriately validated. A fully validated and controlled test that is sensitive and accurate could certainly be a laboratory- developed test. However, there is no way for the laboratory to signal the quality of its test nor for the payer to cover a test it views as adequately validated. It may turn out that a coding system more granular than the CPT coding system can provide will be required. There are currently several efforts under way to develop such alternative systems, one of the most advanced being a coding system for molecular tests developed by McKesson, Inc and currently used by 1 Medicare contractor and being discussed in the trade press.13 How quickly this system, or a similar highly granular system, could take its place in the coding and payment process remains to be seen.
- Newcomer LN. Changing physician incentives for cancer care to reward better patient outcomes instead of use of more costly drugs. Health Aff (Millwood). 2012;31:780-785.
- Landrum MB, Keating NL, Lamont EB, et al. Survival of older patients with cancer in the Veterans Health Administration versus fee-for-service Medicare. J Clin Oncol. 2012;30:1072-1079.
- American Medical Association. Current Procedural Terminology – cpt. Chicago, IL: American Medical Association Press; 2012.
- Isetts BJ, Buffington DE; Pharmacist Services Technical Advisory Coalition. CPT code-change proposal: national data on pharmacists medication therapy management services. J Amer Pharm Assoc. 2007;47:491-495.
- CPTcodingbooks.com. CPT-4 code process – how a code becomes a code. www.cptcodingbooks.com/codes/information.html. Accessed April 13, 2012.
- Department of Health & Human Services. www.ncvhs.hhs.gov/97041614.htm. Accessed April 13, 2012.
- 45 CFR 160, 162.
- Root CB. Medicare coding and reimbursement for clinical laboratory services. Clin Chem. 1998;44(8 Pt 1):1713-1727.
- Paxton A. Molecular CPT codes topple old ‘stacking’ codes. CAP Today. April 2011.
- Association for Molecular Pathology. http://amp.org/committees/economics/AMPCPTReformProposal_Final.pdf. November 2009. Accessed April 13, 2012.
- Centers for Medicare & Medicaid Services. Medicare program; payment policies under the physician fee schedule, five-year review of work relative value units, clinical laboratory fee schedule: signature on requisition, and other revisions to part B for CY 2012. Final rule with comment period. Fed Regist. 2011;76:73026-73474.
- College of American Pathologists. www.cap.org/apps/docs/statline/pdf/cms_payment_policies.pdf. December 21, 2011. Accessed April 13, 2012.
- Ashford M. Harvard Pilgrim, Trustmark join other payors considering outsourcing MDx claims to specialty vendors. Genomeweb. Pharmacogenomics Reporter. www.genomeweb.com/mdx/harvard-pilgrim-trustmarkjoin-other-payors-considering-outsourcing-mdx-claims-s. April 18, 2012.
In a preliminary phase 1 study, the targeted drug crizotinib (Xalkori) stabilized disease progression and, in some children, eradicated cancer cells in relapsed, refractory, aggressive childhood cancers that are likely to harbor genetic abnormalities in the ALK gene: anaplastic large cell lymphoma (ALCL), inflammatory myofibroblastic tumors (IMT), and aggressive neuroblastoma. [ Read More ]
Advocating for biomarker-driven “smart” trials, making the case for patient selection for specific drugs, and the concept of rational combinations as potential targeted therapies in order to improve the efficacy of clinical trials were topics of an insightful presentation delivered by Dr Aleksandar Sekulic on behalf of Dr Pat LoRusso [ Read More ]