Sunday, April 30, 2023

Politico Says CMS on Brink of Releasing "TCET" Regulations

Last October, CMS leader Dr Lee Fleischer had a JAMA article on progress toward "TCET," Transitional Coverage for Emerging Technology - here.  Just as important may be the way CMS summarized his key points in CMS's own one-pager - here.  Since then, CMS held a two-day workshop on rebooting "CED," or coverage with evidence development ,presumably as a platform for TCET.

Now, on April 28, Politico has a short open access article that the TCET regulations should be imminent.  Read it here.  The rule has been noted on the OMB's website, putting it in final review.

cliok to enlarge

Saturday, April 29, 2023

AI Resources for PDFs: Still Buggy in May 2023

Update 5/24/2023.  I have often had poor results with ChatPDF, maybe it will get better.  For example, it would fail to find topics that were a major topic of the PDF.

In late May 2023, ChatGPT offered plugins to subscription members for GPT4.  There are two for PDFs, but both worked poorly.   

  • LINK READER APP gave great summaries but 3/3 times to the wrong paper, not what I provided a link for!   That's useless.
  • The other one, ASKYOURPDF, found the right documents, sometimes requiring a drag/drop upload if the link didn't end in PDF, but that's livable.  But ASKYOUR PDF seems to be designed only to retrieve a relevant snippet to your question, so it's pretty narrow or dumb AI.

I'm sure things will get better but ChatPDF website, and plugs ASKYOURPDF and LINK READER all had pretty big gaps for me so far.


A surprising entry in the world of AI add-ons is   Overcome the limits of ChatGPT text input (2000 words, a couple pages of text).  Article here.  It may not be the most brilliant AI, but it's free and worth checking out.


One of the limitations of ChatGPT is the it takes in text, not PDFs, and limited amounts (about 2000 words per instruction).   

Here comes  Even with a free account, you can upload 3 PDFs per day, and then see summaries and analysis of the PDF by AI.   You can ask questions of the PDF ("Summarize the main methods used in this paper.")  No more cutting text sections of a PDF into ChatGPT 2000 words at a time.  

Find it here:

(Free, 3 PDFs per day.  With an additional $5/mo, you can upload 50 PDFs per day).   

Some screenshots:

click to enlarge

click to enlarge

Click to enlarge

It's not perfect, for example, it couldn't identify authors of the paper, contact info, nor institutions.  Maybe it does not access footnotes.  I had to ask twice to get the "number of references," so the AI is not always brilliant.  But it's a novel model (uploading lengthy PDFs) and it's free.

See CHATPDF look at CMS's "TCET" plan for new technologies - here.

For websites that break very long texts into 2000-word chunks, see Andy Stapleton's video here.

Friday, April 28, 2023

Brief Blog: CMS Posts Preliminary Gapfill Rates for CY2024

On the public home page for the summer laboratory test pricing meetings - here - CMS has posted Prelminary Gapfill Prices for 2024, in a cloud zip file, here.   MAC pricing suggestions diverged, in 7 cases, by over $1000


To find the link yourself, on the CLFS webpage, click on the Meeting Notice, Agenda, Important Materials link, and scroll for CLFS Gapfill Determinations, where you will find 2023: Preliminary Determinations (Zip).

There are about 32 codes in the gapfill process.  All except four, are PLA codes (ending in "U.")  


The spreadsheet has 2 tabs, one for prices (from every MAC) and one for rationales.   The page for rationales has 7 columns, which appear to represent MACs, but are not labeled.  However, some columns refer to MOLDX and one columns refers in each case to "allowanced developed by First Coast and Novitas" (column D).  Related to MolDx, many of these cells simply list "price based on equitable pricing model" which is an internal rule set developed by MolDx (EPM).

Gapfill Statistics

There can be pretty wide divergence.  For example, 81418, a new CPT code for PGx panels, is priced at $917 by MolDx but $168 by Novitas.  (In many other cases, Novitas exceeded MolDx).

While the median gapfill prices always match the MolDx MACs, divergences among the three MAC systems (MolDx, Novitas/FCSO, and NGS MAC) were the rule.  

I found divergences - ranging from $11 to $7186 - in 31 of 32 codes.

Seven Divergences > $1000

7 of the codes had divergences (defined as max minus min) over $1000, ranging from $1400 to $7186.

NGS MAC often, but not always, matched MolDx.

Two of the highest divergences were:

0297U, Oncology, pan tumor, WGS (tumor, germline), MolDx $2919, and Novitas $7741. Delta, $4822.

0300U, also Oncology, pan tumor, WGS, MolDx $4183, Novitas $9713.  Delta, $7186.

click to enlarge but see also Cloud Google file

See My Cloud Spreadsheet with MAX, MIN Data


A Few Other Notes

0285U, cfDNA, response to radiation, ranged from $30 Novitas to $443 MolDx/NGS-MAC (Ratio over 10X).

0338U, rare disease WGS, ranged from $251 Novitas to $2435 MolDx (slightly different at NGS-MAC) (Ratio about 10X).

346U, Abeta ratio, amyloid, Alzheimer, ranged from $48 Novitas up to $93 MolDx and NGS-MAC.

Do Gapfill Prics Change, Draft to Final?

Last year, there were 35 codes in gapfill, and only 4 received summertime gapfill comments, per CMS's final spreadsheet.  All had upward changes (e.g. $263 to $640 and $1195 to $1266 and $2698 to $3033.

Molecular Diagnostics Investments: Some Bright Examples

There's a lot of news this year about pullback in valuations for biotechnology and digital health, and cutbacks in investments.  See. for example, potential Nasdaq delisting for Centogene, here, and a 33:1 reverse stock maneuver for GeneDx (formerly Sema4), whose valuation has plummeted, here.

But there are some bright spots for investments as well, especially the continuing activity in the minimal residual disease space.   

  • Haystack, an MRD company, to be acquired by Quest  for $300M-450M, here
  • Foresight Diagnostics closes a $60M investment round.  Foresight is a Stanford-spinout in the MRD space, garners $60M in Series B, here
  •  Earlier this month, Mercy BioAnalytics raised $40M in a Series A round, and picked up Myla Lai-Goldman as a board member; here, here.
And some labs are still cash-rich.  Fulgent Genetics had about $80M in cash in 2019; after the COVID emergency, it currently has about $800M cash on hand.

Changing Payor Dynamics

A current article in CAP TODAY on comprehensive genomic profiling of tumors devotes a large amount of space to discussion of ongoing reimbursement woes - entry point here.

A current deep dive article in Genomeweb (subscription; Adam Bonislawski; 2000 words) talks about how the lab industry will navigate a new era of "value based purchasing" witih payors.  Numerous interviews.  Here.   

A micro summary of Adam's article [GPT4]: Clinical labs face downward pricing pressure due to payors and industry consolidation, prompting exploration of alternatives to the traditional fee-for-service model. Clinical Lab 2.0 advocates propose leveraging testing and test data to impact patient health at the system and population level to better demonstrate value and capture a larger portion of that value.  A major payor now offers subscription services that provide lab data and insights to payors to help manage various patient populations.  Watch-out: Legal, administrative, and technical hurdles, such as anti-kickback laws, may hinder the implementation of alternative models between labs and payors.



Fun Post: Medicare Policy Talk with AI Pictures

 Fun post.  I am giving a Medicare hospital policy talk this week, and including some targeted illustrations that are generated through AI.  

E.g., "Elderly doctor reads Medicare regulations with magnifying glass," or "Runner races to finish line carrying stacks of books and regulations."

From (DALLE)

Thursday, April 27, 2023

Quest Acquires Haystack Oncology (Liquid Biopsy, MRD); Up to $450M

Summary: Quest Diagnostics is set to acquire Haystack Oncology to enhance its oncology portfolio with Haystack's advanced liquid biopsy technology. Haystack's LBx tech is for early detection of residual or recurring cancer. Haystack's MRD testing should be synergistic with Quest's existing portfolio of oncology, genomics, and pathology services.


In a press release, April 27, 2023, Quest announces it will acquire Haystack Oncology, a sensitive liquid biopsy technology suited for MRD - minimal residual disease diagnostics.

In particular, the technology was validated in a 2022 NEJM report, used in Stage II colon cancer, Tie et al. here. (Op ed by Mantagut & Vidal, here.)

The deal is $300M in cash now and up to $150M based on future milestones.

See also a subscription discussion at Genomeweb here. Open access note here.  See an earlier (subscription) review of Haystack, 11/2022, at Genomeweb here.

Overall, with other news, the stock fell today 5% ($147, $139).


Detailed summary of press release (GPT4):

Quest Diagnostics, a leading provider of diagnostic information services, has announced a definitive agreement to acquire Haystack Oncology, an early-stage oncology company focused on minimal residual disease (MRD) testing. The all-cash equity transaction is set to enhance Quest's oncology portfolio by adding Haystack's cutting-edge liquid biopsy technology.

Haystack's technology is designed to aid in the early, accurate detection of residual or recurring cancer, thereby informing therapy decisions more effectively. Minimal Residual Disease (MRD) testing is a rapidly growing category of liquid biopsy tests that detect circulating tumor DNA (ctDNA) in the bloodstream of patients post-surgery and treatment for cancer. Haystack's ctDNA-based technology, developed over 20 years of research and development by esteemed experts from Johns Hopkins University, is specifically designed for MRD detection.

The merger aims to significantly improve personalized cancer care. By combining Haystack's highly sensitive MRD cancer-detection technology with Quest's broad expertise and scale in oncology, genomics, and pathology, the merged entity is expected to lead in the fast-growing MRD testing category. This includes solid tumor sequencing, specimen collection, and EHR connectivity. The development efforts post-acquisition will initially focus on MRD tests for colorectal, breast, and lung cancers.

After the acquisition closes, Quest plans to adapt the MRD test developed at Haystack as the foundation for new clinical lab services, which are expected to be available beginning in 2024. The combined strengths of Haystack's liquid biopsy technology with Quest's capabilities in screening, pathology, and sequencing aim to broaden access to MRD insights, thereby potentially improving long-term outcomes for cancer patients.

In financial terms, Quest will pay $300 million in cash at closing, net of cash acquired, and up to an additional $150 million on achieving future performance milestones. The acquisition is expected to be modestly dilutive to earnings over the next three years but accretive by 2026. The transaction is anticipated to be completed in the second quarter, subject to customary regulatory review.

Journal Club: NEJM Reviews AI in Infectious Disease Surveillance

Today's NEJM contains a detailed review article on applications of AI in infectious disease surveillance.  The Boston-based group is reported as Brownstein et al. NEJM 388:1597.  It's part of a new series of topics for AI in medicine.

The article does not have an abstract and is 3300 words long.  

Below, I asked GPT4 to provide a 75-word summary as well as a detailed summary (which comes out at 600 words).

If this topic is of interest, see also a very recent NEJM review on "mainstreaming wastewater surveillance" with molecular methods, Mello et al.


GPT4 / Short

The articles discuss the role of AI in infectious-disease surveillance, with examples from Covid-19 tracking. AI methods, such as reinforcement learning, can optimize testing resources and aid early detection. However, overfitting, underrepresentation of certain populations, and privacy concerns present challenges. The future of surveillance involves emerging technologies like large language models, yet success relies on understanding the limitations of AI, ensuring representativeness in datasets, and prioritizing international cooperation.


GPT4 / Long

Part One discusses the application of artificial intelligence (AI) in infectious disease surveillance, particularly in the context of the Covid-19 pandemic. It emphasizes the use of AI and machine learning in analyzing vast amounts of data in real-time, which is beyond human capacity. This includes data derived from various sources such as social media, mobile devices, internet search trends, electronic health records, and genomics databases. These data sources can be classified as syndromic surveillance, digital surveillance, and genomic surveillance.

Syndromic surveillance involves collecting health-related data before official diagnoses are made, which can provide early indications of an outbreak. Digital surveillance uses data from sources like social media and internet searches, tracking disease-related search terms or symptom mentions to detect and monitor outbreaks. Genomic surveillance is based on sequencing pathogen genomes, allowing for tracking of disease spread, identification of new variants, and informing vaccine design.

AI-enhanced surveillance can help identify unusual disease patterns and predict future trends. For instance, during the Covid-19 pandemic, AI models predicted the trend of the disease, identified disease hotspots, and facilitated public health responses. The Part One also emphasizes the importance of developing AI models that can adapt to new challenges and identify signals amidst noise.

Part Two focuses on the risk assessment of infectious diseases, and the application of AI in managing and reducing disease transmission. It provides examples of how AI has been used in real-world scenarios, like health QR codes in China for real-time risk assessment and AI-driven chatbots for health-related queries. It also describes the use of AI in Greece for border surveillance, where an AI algorithm called Eva was used to screen travelers for Covid-19, outperforming tests based on epidemiologic metrics.

The application of reinforcement learning in Eva is discussed, where the algorithm sorts travelers according to various factors like origin country, age, sex, and time of entry. Test results are used to update the system and assign travelers for Covid-19 testing based on recent prevalence estimates for their type. The system continuously learns and optimizes the allocation of limited testing resources.

The text then moves onto exploring extended applications, including wearable technology for early infection detection, pinpointing outbreak hotspots, and leveraging AI for passive surveillance of adherence to non-pharmaceutical interventions. These AI-driven approaches can complement traditional public health surveillance, which is accurate but slower, and participatory surveillance, which provides real-time insights but lacks the confirmatory nature of traditional reporting.

However, the document also acknowledges several challenges and roadblocks for AI in surveillance. Data volume and quality are crucial for surveillance, but there can be issues with overfitting (tailoring predictions too specifically to existing data) and difficulties in distinguishing similar clinical manifestations without molecular testing. Surveillance systems need to be recalibrated regularly to adapt to new pathogen variants and changing external factors like vaccination.

There is also a concern about underrepresentation of certain populations in databases, which can lead to skewed and inequitable outcomes. This is both a moral and a methodological issue that needs addressing. The document provides an example of an AI trained on a database of chest radiographs in children which ended up distinguishing adults from children rather than diagnosing Covid-19.

Privacy issues arise as surveillance models incorporate more data streams from digital interactions and connected devices. The document suggests the use of federated learning, a technology that allows for local calculations on personal data without centrally storing it.

The document ends by emphasizing that AI cannot replace the cross-jurisdictional and cross-functional coordination required for combating emerging diseases, despite its power in improving surveillance infrastructure. The future of infectious-disease surveillance will rely on emerging technologies like biosensors, quantum computing, and augmented intelligence, including large language models like GPT-4. However, the success of these next-generation AI-driven surveillance tools will depend heavily on our ability to unravel their shortcomings and identify which achievements can be generalized.


For a humorous interaction with GPT4 regarding a hypothetical clinical psychology patient, here.

CAP Today: Update on Tumor Gene Profiling, Reimbursement (CGP)

CAP Today's April cover article focuses on tumor gene panel testing, clinical standards, and continuing reimbursement woes. Also discussed: Standing orders and pathologist orders for genomics.


Molecular pathology is always a moving front - NIPS, hereditary gene panels, tumor gene panels, tumor gene panels by-liquid-biopsy, minimal residual disease testing, transcriptomics, multi-cancer early detection - each has its own collection of evidence, guidelines, FDA labeling, its interactions with NCDs or laws, and other factors.

On one of the oldest fronts, hereditary gene panels, recent news again that legacy guidelies and coverage rules miss far too many cases (a "policy-based" false negative, rather than a lab-based one) - here.  Meanwhile, in the past year, MolDx has taken the position that it won't pay for BRCA testing alone, it ust be part of a panel (testing alone is both more costly and per MolDx, below a stardard of care that Medicare requireres.)  See open access news here.  Subscription coverage at Genomeweb here.

On another of the older fronts, tumor gene panels, like the Foundation Medicine test, see a deep-dive article by Karen Titus of CAP TODAY in the April issue.  Here.

A wide range of topics are discussed in this article, including the history of the field, the switch to next-gen sequencing, the Medicare NCD in this space, the lack of equity in test availability, and contrasts between what happens major cancer centers and community care.

Also catching my attention, a discussion of standing orders (so no patient gets missed) and pathologist-based ordering.  NOT mentioned, but I think relevant, a Medicare law that diagnostic test orders must be made by the patient's treating physician who is directly provider her/his medical care.  This is found at 42 CFR 410.32, and I discuss it in an April 2022 blog with some history and further references.  Here. Medicare policy manual 15:80.6.4 gives very limited zones for pathologist test orders, such as seeing TB-like specks on a H&E and ordering a TB stain "to complete the case."   The CAP TODAY article also brings up direct physician payment to an MD, which occurs for "physician pathology tests" like a lymph node biopsy, but not for genetics.   This is related to a regulation that a physician service must be performed by a physician >50% of the time (see 42 CFR 415.130, 415.102(a)).   I discussion those links in my April 2022 blog also.  

(I also link to an even nerdier 2013 white paper here.  The topic also turns out to be tied into historical legal definitions of what services are paid to a hospital (TC) and paid to a physician (PC)).

With regard to lack of payment for genomic interpretation services, Medicare would argue that they pay $3500 for, say, a Foundation Medicine test, out of which the company can pay whatever it chooses to, to either physician lab directors or PhD professionals who are involved signing out in the test.

I haven't been a practicing pathologist since 2000, but reading the article I had half-an-idea there may be two uses of the term Standing Orders.  One is per-patient (an order to take the particular patient's glucose every day at 8 am and 8 pm x 7 days), the other is per-protocol (order a Foundation Medicine test on every metastatic cancer this year).   Medicare has some rules for the former.


Regarding turn around time for cancer sequencing, see a March 24 premium article by Andrew Han at Genomeweb here.

Regarding uninterpretable 30-page genomics reports, see an article by Beauchamp et al. on "integrative diagnostics," here.   (There is also a forthcoming paper by MGH on this topic).

The NCCN Flow Chart for Lung Cancer:  We Should Be Appalled, Not Take as Status Quo

Regarding starting chemotherapy in lung cancer before molecular testing is done, there's a silent testimony to this problem baked right into NCCN guidelines.   They have flow charts when genomic testing is quirkcly available, and separate flow charts for other patients, specifically when genomic testing is delayed.   Having to enshrine that dysfunction in major protocols is sad.

See guideline 2.2023, NSCL-20, with one flow chart if a critical EGFR mutation is discoveried in time for treatment, and another flow chart of the same patient if all too often the discovery of the EGFR mutation is delayed.

click to enlarge


Monday, April 24, 2023

ChatGPT4 Takes On Legal Hypotheticals: Conflict of Courts vs FDA Injunctions

Readers will know I have enjoyed experiments with how ChatGPT4 approaches medical policy decisions and discussion.

I was wondering what happens if more and more courts take on cases that review FDA decisions, some courts asking that Drug ABC be taken off the market, and other courts ruling that Drug ABC is properly approved and shall stay on the market.  Now, I am not an attorney, and GPT4 is not an attorney, so this essay is just an experiment in learning the style and approach of AI answers to legalistic questions. here we go.  My human text is in blue.

You are a legal expert, experienced in jurisprudence and judicial procedure, and especially in the federal court system. The issue is discrepant rulings (such as injunctions), each of which is styled to have national impact immediately, from different judges among the 700 federal court judges. 

The case study in this hypothetical is a drug, ABC, approved by FDA in 2000. 

·        Plaintiff A in District A asks for an injunction against its sale, due to mistakes in the FDA approval process. She wins her case with Judge A. Sales of ABC are injoined nationwide.

·        Plaintiff B in District B asks for an injunction against the sale for its indications 2 and 3, while expliciting rule that indication 1 is a proper agency action and shall stand. Judge B grants this national injunction.

·        Plaintiff C in District C also asks for an injunction against all indications due to irregularities in the FDA approval, but this plaintiff loses and the judge rules that, as FDA requests, all indications can stay on the market.

·        Clearly, these back-and-forth cases and stakeholders could multiply further.  Turning to the appellate courts, their rulings are also contradictory (!), some staying the lower court pending hearings, others allow the lower court national injunction to remain in effect. 

      The Supreme Court does not want to have an activist "shadow docket," yet doing nothing seems to be a road to chaos with chaotic lower court and appellate contradictory rulings - are sales banned by injunctions, if so, which ones and in what degree, and what is the role of a particular judge's "anti-injunction" ruling affirmatively that the FDA drug is properly and legally approved and shall stay on the market.

Finally, as one piece of context, in 2023, FDA-related cases of the type in this hypothetical are actually happening. Discuss the situation from a jurisprudence perspective and discuss possible options and outcomes.


As an AI language model, I am not a legal expert, but I can provide some insights into this hypothetical situation involving discrepant rulings in federal courts. This scenario presents a challenging and complex issue that highlights the potential for conflicting decisions among lower federal courts and the implications for the federal court system, stakeholders, and the broader public.

1.            Jurisprudence perspective: The situation reflects the fact that federal judges can have differing interpretations of the law, which can result in conflicting decisions. While judges strive to apply the law consistently, they may disagree on the application of legal principles or the weight of the evidence. In this case, the main issue centers on the validity of the FDA's approval process for the drug ABC and whether any mistakes or irregularities warrant an injunction against its sale.

2.            Role of appellate courts: The appellate courts serve to review the decisions of lower courts and provide more uniformity in the interpretation and application of the law. However, appellate courts may also disagree with one another, leading to further confusion and potential chaos. In this situation, it is essential for the appellate courts to act swiftly and provide clear guidance to the lower courts, stakeholders, and the public.

3.            Role of the Supreme Court: The Supreme Court has the ultimate authority to resolve conflicts among the lower courts and provide a uniform interpretation of the law. The Court may be reluctant to engage in "shadow docket" decisions, which are rulings without full briefing or oral argument. However, the Court may be compelled to intervene to provide clarity and prevent chaos resulting from conflicting lower court decisions.

4.            Possible options and outcomes:    a. Supreme Court intervention: The Supreme Court could grant certiorari to one or more of the cases and issue a decision that resolves the conflict among the lower courts. This would provide a definitive resolution to the issue and guidance for future cases involving similar disputes. b. En banc review: If appellate courts issue conflicting rulings, one or more of them may choose to rehear the case en banc, meaning before all of the judges on the court. This can help to resolve inconsistencies and provide a more consistent interpretation of the law. c. Congressional action: In response to the chaos resulting from conflicting judicial decisions, Congress may choose to pass legislation that clarifies the FDA's authority or establishes a more explicit framework for the approval and regulation of drugs like ABC. This could help to minimize future disputes and provide clear guidance to courts, the FDA, and stakeholders.

 In conclusion, the hypothetical situation involving discrepant rulings on the FDA-approved drug ABC highlights the potential for conflicts among federal courts and the challenges of achieving uniformity in the interpretation and application of the law. Possible solutions to this issue include intervention by the Supreme Court, en banc review by appellate courts, or congressional action to clarify the legal framework surrounding drug approval and regulation.


You have focused on middle to long term events, such as a definitive SCOTUS ruling, or legislation. 

What happens in the shorter term? For example, one district rules that due to problems at the FDA, sales are nationally banned, and another district rules the opposite, that the drug is properly approved and shall remain on sale. What would happen if the Supreme Court declined to act (other than to accept cases for next year's docket)?

In the short term, when conflicting district court rulings exist, and the Supreme Court declines to act, the legal landscape can be quite uncertain and chaotic. Here are some possible short-term outcomes:

1.            Status quo and conflicting injunctions: If the Supreme Court does not intervene, the conflicting district court rulings will remain in place, causing confusion and uncertainty for the parties involved. This could lead to a patchwork of injunctions across the country, with sales of the drug being allowed in some jurisdictions while being banned in others.

2.            Stay of injunctions: The appellate courts may issue stays of the district court injunctions, effectively halting their enforcement until the appellate courts reach a decision. This would maintain the status quo while the cases are under review, but it may still create uncertainty for stakeholders and the public.

3.            Consolidation of cases: One possible solution to the conflicting rulings in the short term could be the consolidation of the cases at the appellate level. The appellate courts may decide to hear the cases together, either by consolidating the appeals or coordinating their review in some manner. This could help streamline the legal process and promote a more consistent interpretation of the law among the appellate courts.

4.            Voluntary compliance by stakeholders: In the face of legal uncertainty, the manufacturer of the drug or other stakeholders may choose to voluntarily suspend sales of the drug until the legal issues are resolved. This would be a strategic decision by the parties involved to minimize their exposure to liability and mitigate the potential consequences of the conflicting injunctions.

5.            State-level action: State governments or regulatory agencies may also take action in response to the uncertainty created by the conflicting rulings. They might impose their own restrictions or guidelines on the sale of the drug, which could help provide some level of clarity for stakeholders and the public within their jurisdictions.

It is essential to note that in the short term, the situation would likely remain uncertain and potentially chaotic, as various stakeholders navigate the conflicting rulings and legal landscape. The ideal resolution would come from a higher authority, such as the Supreme Court or legislative action, to provide a more definitive and uniform interpretation of the law.


I have one last hypothetical. In District A, stakeholder A has won a ruling from Judge A that ABC should be banned by his injunction against the FDA.

 Another stakeholder, in District E, is distressed by this ruling. She actually enters the same case and request in District E, but intentionally defective and poorly argued. Judge E in District E therefore quickly rules that the FDA has acted properly and ABC is properly on the market (creating a conflict of rulings). How does this sound as a legal strategy for opponents of the District A ruling?

 While this hypothetical scenario presents an interesting legal strategy, it is unlikely to be an effective or advisable approach for several reasons:

 1.            Forum shopping: Intentionally filing a poorly argued case in a different district to obtain a conflicting ruling could be seen as an attempt to manipulate the legal system, which is frowned upon. Judges and the court system generally disapprove of forum shopping, where a party seeks to gain an advantage by choosing a specific court or judge likely to provide a favorable ruling.

2.            Judicial discretion: Judges have discretion in handling cases and may not necessarily issue a ruling based on the intentional weaknesses in a party's arguments. If a judge suspects that the case has been filed with the intention of manipulating the outcome, they may dismiss the case, abstain from issuing a ruling, or decide the case on other grounds.

3.            Ethical considerations: Attorneys have an ethical duty to represent their clients zealously and in good faith. Intentionally presenting a defective or poorly argued case could be seen as a violation of an attorney's professional responsibility and could potentially result in sanctions or other disciplinary actions.

4.            Precedential value: Even if the strategy were to result in a conflicting ruling, the poorly argued case may carry less precedential weight in subsequent litigation due to the weaknesses in the legal arguments. This could limit the impact of the ruling and make it less persuasive in future cases.

5.            Inefficiency and costs: Pursuing this strategy would add to the workload of the courts and contribute to an inefficient use of judicial resources. It would also likely result in increased legal costs for the parties involved.

 Instead of intentionally filing a poorly argued case, parties opposed to the District A ruling should consider more legitimate legal avenues to challenge the decision. These could include appealing the District A decision, pursuing a separate case with well-reasoned arguments in a different jurisdiction, or advocating for legislative action to address the underlying issue. These approaches would be more likely to result in a meaningful and lasting resolution of the conflict while adhering to the principles of the legal system and professional ethics. for "conflicting court rulings"

CMS and Coverage with Evidence Development: Severe Underperformance for Minorities

CMS hopes to release a program of coverage updates under the term, "Transitional Coverage for Emerging Technologies," or TCET.  So far, most CMS discussions have focused on Coverage with Evidence Development, including a workshop on the topic in February 2023.

However, access to care under CED has been severely biased against minority patients, as noted in a new article at Reuters and other sources.


The Reuters article, by Deena Beasley and Julie Steenhuysen, focuses on the pending full approval of Alzheimer drugs.   Find it here.   Under current policy, new Alzheimer drugs will be available to Medicare patients only in a registry, even if the drugs have full approval (as opposed to accelerated approval).  Under the CED paradigm, access to Alzheimer-related services for Black patients has been severely curtailed.   The Reuters article describes the PET-amyloid registry like this:  "A first registry failed to recruit enough minorities. In the second, started two years ago, there are still 18 states without a single site enrolled."   

Original Data

We can quickly confirm the original data from the 2019 JAMA publication on the CMS registry - here.   Of 11,000 Medicare patients receiving services, less than 3% - only a couple hundred nationwide over five years - were Black.   

In addition, the patients in the Medicare CED study were biased to certain socioeconomic groups, with 25% of patients having graduate degrees (and this, in a population that averaged over 70 years old).  

2023 Data

A second Medicare patient registry, IDEAS-2, continues the problem into 2023.  For example, there are zero participating facilities in two states with among the highest Black populations, North Carolina and South Carolina (22% and 27% Black, respectively.)  Data here.  

New Mexico, with an 11% Native American population, and 50% Latino, also has zero participating facilities.

For a recent review of bias against minorities in clinical trials, Turner 2022, here.

Sunday, April 23, 2023

Journal Club: Beauchamp on "Integrative Diagnostics" = Path & Radiol Together

 Several times I've referred to the cutting-edge policy articles in J Amer Coll Radiol as being worthwhile even for those who work in the other diagnostics field - lab medicine.   (See my large article from 2022 here and a short follow up here).

A great example is a new article at JACR by Beauchamp et al. that lays our a roadmap for "Integerative Diagnostics" meaning bringing together imaging with lab medicine & pathology.   Find the article open access here:

The authorship is attributed to the "International Society for Strategic Studies in Radiology."  This article also include's an editor's Visual Abstract, a trick I've seen used recently in several journals.   The visual abstract is here:

While laying out a roadmap and a range of different value arguments, the authors conclude:

Integrative diagnostics has the potential to identify correct therapies more quickly, modify treatment when appropriate, and terminate treatment when not effective, ultimately decreasing morbidity, improving outcomes, and avoiding unnecessary costs. Radiology, laboratory medicine, and pathology already play major roles in medical diagnostics. Our specialties can increase the value of our examinations by taking a holistic approach to their selection, interpretation, and application to the patient’s care pathway. We have the means and rationale to incorporate integrative diagnostics into our specialties and guide its implementation in clinical practice.

See the authors' own visual comparison of the status quo - different reports from different departments of the hospital - and their vision for the future, where all those divisions become a "back office" (their term) to a comprehenive diagnostic analysis and conclusion.

Since the article is open-access, have-at-it if it sounds worthwhile to you.  If you'd like one last perspective, here's a GPT4 AI summary of the article's dicussion section.

  • [GPT4] Medical diagnostics play a crucial role in healthcare decision-making and spending. Integrative diagnostics (ID) has the potential to improve cost-effectiveness, streamline care pathways, and address disparities in healthcare access, particularly for minority and underserved populations.
  •  However, barriers to ID implementation include local reimbursement systems, lack of financial support for informatics infrastructure, and cultural resistance within healthcare organizations. 
  • ID can also enhance clinical trials and medical education by fostering interdisciplinary collaboration and efficient use of resources. 
  • Initial efforts at implementing ID in academic medical centers have faced challenges due to IT infrastructure limitations, but progress is being made. 
    • Recommendations for promoting ID include expanding multidisciplinary teams, integrating medical education across disciplines, enhancing IT infrastructure, and developing financial models that demonstrate ID's value. 
  • Ultimately, clear vision and effective leadership are needed to catalyze change in the healthcare system to fully adopt ID.

Tuesday, April 18, 2023

Hey, I'm On a Podcast !

I'm proud to be on Episode 12 of healthcare expert Rick Gannota's podcast, HEALTHCARE NATION.

I discuss the Medicare agency and innovation.

Find the website here:

Find the YouTube channel, which includes the video feed:


Or you should be able to access your on your favorite Podcast app under "Healthcare Nation."  (On my own podcast app, there were two Healthcare Nation channels, one more up to date than the other, FYI).

This is an AI summary from GPT4...
  • Dr. Bruce Quinn is a healthcare consultant with expertise in Medicare, business innovation, and federal health policy.
  • Healthcare innovation faces challenges in navigating Medicare and payment systems, with new technologies sometimes struggling to gain recognition and appropriate reimbursement.
  • The impact of technology on healthcare is significant, including advancements in telemedicine, remote care, genomics, and precision medicine, but startups must be mindful of the regulatory and payment landscape.
  • Artificial intelligence and machine learning have the potential to revolutionize healthcare, but their true impact remains uncertain and faces regulatory hurdles.
  • Genomic testing, particularly for cancer screening and detection, is promising but faces payment and pricing challenges within the Medicare system.
A longer-format summary is here.

An Insight into Application of ACA Preventive Services Law: House Report on Birth Control Coverage

The Affordable Care Act mandates coverage of preventive services, including contraceptives, by commercial health insurance, as endorsed by the USPSTF and two other bodies. A US House report from October 2022 examines "inappropriate" non-coverage or copay requirements for some contraceptive methods, raising questions about similar issues in other healthcare verticals like preventive diagnostic tests.


The Affordable Care Act requires commercial health insurance to cover preventive services endorsed by the USPSTF, or by two other bodies.  The other bodies are a vaccines committee at CDC, and a women's health committee at HHS.    It is the women's health body (not USPSTF) that makes contraceptive services an ACA preventive health benefit.

While there are plenty of legal cases and rulemakings navigating around contraception and religious freedoms, in general, the HHS interpretation is that health plans must cover all contraceptives or contraceptive methods endorsed by the HHS group.   (Similarly, on the USPSTF side, I believe plans are required in principal to cover all colorectal screening methods endorsed by USPSTF).   

A report from the US House of Representatives was released in October 2022 (that is, under Democratic leadership at that time in the House), and looks at "inappropriate" non coverage or requirements for copays for some contraeptive methods.   I mention this, on this blog, because similar issues may apply in other healthcare verticals, such as preventive services with diagnostic tests.

Beyond that, I'm not an attorney, and I'll leave interpretations of correct, or incorrect, non coverage to readers.   Here's the House PDF document on noncoverage and improper copays:

See a May 2022 WaPo article on the topic:

See a July 2022 HHS press release on the same topic:

See an October 2022 news article based on the then-new House report:

See a 2023 policy note at KFF here.  See a 2021 report on the same topic (same news) in 2021 here.

Summary at top generated by GPT4.


Monday, April 17, 2023

MolDx Updates Pharmacogenomics Tech Assessment Rules

Labs working with MOLDX must follow multiple levels of instructions:  An LCD, a billing article with the LCD, in some cases an FAQ document, and special and regularly updated Technical Assessment Forms, often in Excel format.  Sometimes the public comment Q&A that appeared with the original LCD can be insightful as to what they mean by various statements.

MolDx updates its Pharmacogenomics form, PGX PF 007, as of April 3, 2023.  They updated their general assessment form GEN CQM 003 Checklist, on February 3.

Note that the GEN Checklist doesn't have a change log, but the PGX document does.

Here are changes in the April 2023 PGX document, click to enlarge.

They've included a number of examples such as filling out the top row of a spreadsheet with sample data to give you the look-and-feel with sample data of what the column headers really mean.  In another case, they clarified that a certain column name expects a Y/N answer.

Click to enlarge.


Sunday, April 16, 2023

PLA Codes: This is a Busy Quarter

AMA has released applications for PLA codes, submitted April 6, 2023.  They'll be discussed in committee around APril 25 and finalized at the May 2023 AMA CPT meeting.   In the past, codes from this April cycle have been included quickly in the June/July CMS new code pricing meetings.  Interested parties can request the chance to comment up to April 21; see PDF.

Find the five-page PDF here;

There are circa 23 new codes, and some interesting deletions, such as the very recently created 0397U, Agilent Resolution ctDx First (109 genes) for ctDNA.  Envisage deletes its similarly recent Barrett's esophagus classifer.  

PreciseDx applies for an AI-based breast biopsy test with 8 digital/morphologic features.  See also an AI test from LungOI which seems to include whole slide imaging analysis with gene analysis (eg EGFR).  (Separately, see an article in Precision Oncology about intraoperative diagnosis with AI for brain tumors; here.)

Two centers apply for optical genome imaging.  I believe there's also an application in the regular CPT process for optical genome mapping for a non proprietary (Category I) code.

There's a plasma amyloid Alzheimer test, Precivity AD.

Thursday, April 13, 2023

Brief Blog: CMS Summer Lab Pricing Meetings for 2023: June 22, and July 19/20

 CMS has released announcement documents for the summer new lab test pricing meetings (aka "crosswalk, gapfill meetings").   

The public comment meeting will be June 22, a Thursday.    The expert advisory panel meetings will be July 19/20, a Wednesday/Thursday.

CMS should release an agenda, including all code names, at least 30 days ahead of June 22.

Find the preliminary announcements online at Federal Register here:

The June 22 meeting permanent announcement is here.  Presentations due by June 1.  Registration will be via email with a specific subject line (for commenters only).  Public streaming won't require registration.

The July 19/20 permanent announcement is here.  Stakeholders with codes are "standby speakers" who will answer questions if asked, and must pre-register (by email).  More details will appear about 2 weeks before the meeting.  


Last year there was a bit of a fuss over whether a code can be crosswalked to a code in the active gapfill process - answer, NO.   And whether a code can be crosswalked to an ADLT code.   For the latter, CMS has no rule against it, but you can't force CMS to pick an ADLT code as a crosswalk and if they are disinclined to do so, they simply won't.   Codes are stated to be crosswalked to, "existing test, multiple existing test codes, or a portion of an existing test code" which doesn't rule-out an ADLT code choice.  

Big Win for Novel Diagnostics: Clinical Synuclein Assay for Parkinson's Disease

The California diagnostics firm AMPRION scored a major win this week for its novel proteomic diagnostic test SYNTap(r) to accurately diagnose early Parkinson's disease.   The test has a unique in vitro synthesis step that amplifies the signal for synuclein (a Parkinson's protein) in CSF. The formal name for the technique is "seed amplification."

THe new news is an 1100-patient study published in Lancet Neurology this week, with a strongly favorabe Op Ed.  The Michael J Fox Parkinson Foundation also is in the news with a strongly favorable assessment (the foundation also supports a large biobank and this research).   Favorable press is also appearing at Endpoints, STAT, and Washington Post.

Read more at the sources below.

  • Amprion home page.
  • Lancet: Siderowf et al. here.
  • Lancet, Op Ed, Berg & Klein here.
  • MedPage Today here.
  • StatNews here.
  • Endpoints here.
  • WaPo here.  Forbes here.  Guardian here.
  • MFJ Foundation press release here.
  • Genomeweb covered on April 26 here.

A documentary about Michael J Fox will release on AppleTv on May 12.

New drugs have remained scarce; one approach led by Abbvie failed at FDA this spring.

(Note, disclosure, I am a policy consultant for Amprion).

Wednesday, April 12, 2023

Very Brief Blog: AMP Publicizes September Date for Economics Conference

Association for Molecular Pathology (AMP) has an annual sumit in DC on health economics & molecular pathology.

This year, look for it on Wednesday September 13, 2023.  It's at the Omni Shoreham, the huge grand old hotel (built 1930; 800 rooms) about two miles north of the White House (Woodley Park Metro/Zoo).

More info here:

Tuesday, April 11, 2023

AI Actually Helps: Quickly Educates Me on the Doctor-Patient '3 Talk Model'

Summary: While reading an MCED article, I found GPT4 was helpful in quickly getting some information on a medical decision making model that was new to me.


This week, Annals of Internal Medicine publishes an article by Selby et al. on Multi-cancer early detection tests, MCED.  The goal is to prepare doctors to have discussions about this testing with patients who have questions right now.  The article is here.

In the short article, the second citation is to decision making with high uncertainty, and the third citation is to something called the "3 Talk Model."  

I had never heard of the 3-talk model; it is by Elwyn et al. and dates to BMJ 2017 (359;J4891; PMID 29109079). It's now open access.  By the way, PMID 29109079 shows both related articles and articles that cite Elwyn directly.

A large table in Selby is built around the 3-Talk Model.  I asked GPT4 "What is the 3 Talk Model?" and got a decent looking answer (as much as I need to know.)  I then asked, "Are there any problems with the 3 Talk Model?" and got what seems like a reasonable answer again.


In medical decision making what is the "3 talk" model?

GPT4. The "3 talk" model, also known as the Three-Talk Model, is a structured approach to medical decision-making and patient communication. The model emphasizes the importance of clear and effective communication between healthcare providers and patients, particularly when it comes to discussing sensitive issues, treatment options, and decisions related to patient care. The model is comprised of three distinct stages:

Team Talk: This stage involves healthcare providers discussing and reviewing the patient's medical history, current condition, and available treatment options among themselves. This helps to ensure that all team members are on the same page and that they have a clear understanding of the patient's medical situation before talking with the patient.

Option Talk: In this stage, the healthcare providers present the patient with the various treatment options available, including the potential risks, benefits, and outcomes associated with each option. This stage is crucial for enabling patients to make informed decisions about their care. It is essential for healthcare providers to present the information in an unbiased manner and to be transparent about the uncertainties surrounding each option.

Decision Talk: The final stage involves discussing the patient's preferences, values, and concerns, and helping them make a decision that aligns with their personal goals and values. Healthcare providers should support and guide the patient in this process, ensuring that they understand the implications of their choice and feel comfortable with the decision they make.

Overall, the Three-Talk Model is designed to facilitate effective and empathetic communication between healthcare providers and patients, with the ultimate goal of helping patients make well-informed decisions about their care.

Have their been any criticisms of this model?  It seems commonsensical.

GPT4. While the Three-Talk Model has been widely accepted and utilized for its structured approach to medical decision-making and patient communication, there are some potential criticisms and limitations:

  • Time constraints: In a busy clinical setting, healthcare providers may not always have the time to engage in in-depth discussions with their patients at each stage of the model. This can lead to rushed conversations or inadequate information sharing, which could impact the effectiveness of the model.
  • Variability in communication skills: The success of the Three-Talk Model relies heavily on the communication skills of the healthcare provider. Not all providers may have the same level of expertise in communicating complex medical information or addressing sensitive issues with patients. This variability could lead to inconsistencies in the quality of patient-provider interactions.
  • Patient comprehension and health literacy: For the model to be effective, patients need to understand the information provided to them and actively engage in the decision-making process. However, some patients may have difficulty comprehending complex medical information, particularly if they have low health literacy. This can limit the effectiveness of the model in supporting informed decision-making.
  • Cultural sensitivity: The Three-Talk Model does not specifically address cultural differences and how they might affect patient-provider communication. In some cases, cultural beliefs or language barriers might hinder effective communication, making it more challenging for healthcare providers to engage patients in the decision-making process.

Despite these potential criticisms and limitations, the Three-Talk Model does provide a valuable framework for healthcare providers to follow when discussing medical decisions with patients. It emphasizes the importance of clear communication and patient-centered care, which are essential for supporting informed decision-making and improving patient outcomes.  AI pics

I asked GPT4 directly how to have a structured conversation about MCED with a patient.  That is, it's sort of the poor-man's, AI-only version of the Selby article.  I've posted here.

Monday, April 10, 2023

Brief Blog: Medicare Advantage Must Cover "Traditional Medicare's" Benefits; Enhanced Rules

You may have seen articles, like this one at STAT, that based on brand-new federal rulemaking, Medicare Advantage plans must cover fee-for-service ("traditional Medicare") benefits.  As defined in NCDs and LCDs.

You may have wondered, what's new?  That has been the rule for years, and even enshrined in a regulation as well as the policy manuals (e.g. managed care Chapter 4 section 10.16).  However, CMS states that denials still occur, such as repetitive pre-auth denials for covered services, or denials not appealed when physicians don't understand their rights or the rules.

The short answer, CMS seems to be beefing up its regulations (such as 42 CFR 422.101).  See the early release typescript of the Medicare Advantage rule; from page 205 forward (205-251, but then followed by prior auth rules which are also of interest).   

Find the new rule here (p 205 fwd):

See April 12  Fed Reg for the final typeset rule.

The rulemaking goes on many pages, but includes a comment that MA plans must use current evidence in "widely used treatment guidelines."  Much of the new text (inserted at 42 CFR 422) has to do with processes for MA plans making coverage documents that apply outside of NCD and LCD rules.

There is a discussion of NCD/LCD processes (page 239).  There is a remark that an existing regulatory statement about complying with Medicare manuals is deleted.   This is important, because some types of coverage are found only in manuals.  CMS states, somewhat vaguely, that Medicare manual compliance remains required - so it is pretty fuzzy why go to the trouble to delete this phrase from regulations.  

CMS multiple times cites an OIG report in April 2022 on MA plans and denials -