PBM Trends, Insurer Profitability Stats & Headwinds, Quantum Computing in Healthcare, and More
In February we examine PBM trends, scrutiny by regulators, and legislation impacting the prescription benefit industry. We review medical cost trends and the impact on insurer profitability given the spike in care costs in Q4 of last year. An update on prior authorization regulations is included as is information on which insurers deny the most claims in Medicare Advantage. We wrap sharing perspectives on quantum computing’s relevance for healthcare and how DeepSeek’s AI models have led to U.S. innovation and cost cutting.
Joe Bastante
2/26/20257 min read
In this month’s update:
The latest PBMs trends and why the growing concerns by regulators
Q4 was bad for most insurers, but just how bad was it?
Update on state prior authorization restrictions
Update on claims denials, who denies the most?
Google and Microsoft are the latest to produce quantum chips, should healthcare care?
You can ignore they hype on DeepSeek now
The latest PBMs trends and why the growing concerns by regulators
In last month's update, I shared revenue figures explaining why healthcare companies are increasingly owning insurer, provider, and PBM businesses. Drilling deeper into the PBM market, the market size was about $573 billion in 2024 and is projected to grow to $897 billion by 2032 (5.8% CAGR). Nearly all large insurers own PBMs, which together dominate the market: CVS/Aetna Caremark, Cigna Express Scripts, United OptumRx, Humana Pharmacy Solutions, and Prime Therapeutics/Magellan Rx (Blue Cross). In 2023, PBMs owned by large insurers held 89% of the PBM market. PBMs are facing increasing scrutiny, especially for alleged lack of financial transparency. PBMs are making moves to address concerns and more directly benefit patients (e.g., Cigna's Accredo Specialty Pharmacy offering counseling, training, and 24x7 access to pharmacists & nurses, Prime Therapeutics performance-based reimbursement where payment is tied to drug efficacy). Even so, scrutiny has intensified as government officials accuse PBMs of inflating prices, disadvantaging independent pharmacies, and offering little transparency. The FTC issued a scathing report last summer, and a bipartisan bill was introduced in December, which would require health insurers to sell their PBMs. Most U.S. states already have legislation targeting PBMs, for example, banning PBM gag clauses (i.e., limiting what pharmacists can tell customers about drug costs and options), requiring PBM reporting on rebates, and limiting patient cost sharing. PBMs are seeking to change the narrative, for example, OptumRx announced they will pass 100% rebates to consumers by 2028. Yet, scrutiny will continue as government agencies seek to reign in the meteoric rise in healthcare costs. Just this past week, the FTC was green lighted to pursue a suit against Caremark, Express Scripts, and Optum Rx for allegedly inflating insulin prices.
News on bipartisan legislation that would require insurers to sell PBMs:
Thorough resource covering PBM legislation by state:
https://nashp.org/state-tracker/state-pharmacy-benefit-manager-legislation/
FTC's 2024 interim report on PBMs:
https://www.ftc.gov/reports/pharmacy-benefit-managers-report
FTC's suit against PBMs for allegedly inflating insulin prices
Q4 was bad for most insurers, but just how bad was it?
2024 was a difficult financial year for most insurers. Nearly all major insurers saw increases in MLR (i.e., medical loss ratio, which is the proportion of premiums spent on patient care and costs). For example, from 2023 to 2024, MLRs increased as follows: CVS went from 86.2% to 92.5%, Humana from 88% to 89.8%, Elevance from 87% to 88.5%, Centene from 87.7% to 88.3%, and UnitedHealthcare from 83.2% to 85.5%. The fourth quarter of 2024 was particularly difficult. For example, Humana's Q4 MLR was 91.5% resulting in a loss for the quarter of $693 million. CVS/Aetna had an even greater MLR in Q4 at 94.8%. Common reasons cited include: continuing post-COVID high utilization, increases in specialty pharmaceutical costs, GLP-1 impacts, increase in behavioral health utilization, and lower Medicare STARS ratings. Drilling in a little deeper, we see many changes collectively create profitability headwinds such as the Inflation Reduction Act's limiting Medicare beneficiaries to $2,000 max out of pocket per year, risk adjustment changes reducing risk payments, changes in the STARS methodology reducing scores and payments on average, etc. Given the headwinds, Moody downgraded the health insurance industry credit outlook to negative. That's not to say all health insurers lost money in 2024, but there's been downward pressure on profits even as the number of enrolled members has increased. The Trump administration's focus on reductions in federal spending for Medicaid and ACA will add uncertainty and volatility in 2025.
Moody's downgrade to the credit outlook for health insurers:
2024 Medical loss ratios for major insurers:
https://www.beckerspayer.com/payer/payers-ranked-by-2024-medical-loss-ratios.html
Update on state prior authorization restrictions
While we've covered prior authorization stats in depth in past posts, this snipped includes updates on notable state regulations and recent data released on prior authorizations in Medicare Advantage. We begin with a Rhode Island bill introduced last month, which would significantly constrain prior auths in that state. According to a Rhode Island's press release, "The legislation would prohibit insurers from imposing prior authorization requirements for any admission, item, service, treatment, or procedure ordered by an in-network primary care provider, including general internists, family physicians, pediatricians, geriatricians, OB-GYNs, nurse practitioners, physician assistants and other health care providers". This would have a significant impact on allowable prior authorizations if approved. An Indiana bill under consideration would eliminate prior auths for drugs under $100, cap authorization rates to 1% of claims, and prohibit insurers from mandating step therapy protocols (i.e., can't pursue a treatment till other options were tried).
Regarding prior authorization rates in Medicare Advantage, The data from 2019 through 2023 showed minimal variation between years. For example, prior authorizations per member per year have hovered between 1.4 and 1.8. Likewise, denial rates have ranged between 5.6% and 7.4%. The statistic I'm always surprised by is the rate of successfully overturned denials (i.e., were denied but then approved when appealed). The rate has ranged from 81.3% to 83.2%. Check out the KFF research for further details.
Press release for Rhode Island's prior authorization bill:
Indiana prior authorization bill:
https://indianacapitalchronicle.com/2025/01/23/senators-advance-several-meaty-health-bills/
KFF Medicare Advantage prior authorization research released January 2025:
Update on claims denials, who denies the most?
Last month's update included details and statistics on claims denials. Just a few days after I sent out the update, KFF released the latest research on ACA claims denials by insurer. In-network claims denial rates varied drastically between insurers ranging from 1% to 54% of claims. Since I have friends in many insurance companies, I won't call out any insurers, but I'd recommend you take a look at what they found.
So long as we're again discussing denials, Blue Cross Blue Shield of Michigan will begin enforcing a policy which charges providers for repeated appeals for denials. If a denial is re-appealed and the provider seeks an external review, the provider will be charged the cost of the review if the denial decision is upheld.
Claims denial rates by insurer:
BCBS Michigan enforcement of charges for repeated appeals:
Google and Microsoft are the latest to produce quantum chips, should healthcare care?
Google and Microsoft have recently announced new quantum computing chips, so I thought it would be a good time to review what they are (briefly), their applicability to health and life sciences, and their readiness for use. Since my intent is to keep these summaries brief, I'll skip many technical details fascinating though they are. The main point to be understood is that quantum computers are fundamentally different from and complementary to traditional computers. Traditional computers are deterministic, that is, given inputs, you can determine the output and it will always be the same. This is just what's needed to run your Windows or Mac computer, complete a bank transfer, or calculate the patient's share of a medical bill. Quantum computers are probabilistic, meaning they produce answers, often more than one, based on probabilities. Quantum computers are particularly effective at modeling real world systems, and for such uses they can be orders of magnitude faster than traditional computers. On a quick side note, many leaders are concerned having heard that quantum computers can crack modern encryption algorithms. While this may be true, new algorithms are being developed, including by the NIST, which are not subject to quantum cracking.
Quantum computers will have applicability in health and life sciences applications, for example, in drug discovery through simulating complex molecular-level interactions, in modeling how an individual will react to a therapeutic compound, and in simulating and optimizing complex global supply chains. Even so, should healthcare leaders care today about quantum computing? In most cases, no. Those in highly scientific areas should track progress, maybe even explore some early uses. For most others, the technology will need to develop further to be of practical use and therefore doesn't represent an immediate priority.
If you'd like to know more about quantum computing, IBM offers an informative introduction:
You can ignore the hype on DeepSeek now
In last month's update I explained how DeepSeek changed the economics of AI for companies and consumers. Their initial pricing was 30 times less than OpenAI's, though prices are now being lowered by major AI companies to be more competitive with DeepSeek. While DeepSeek should be credited with improved training techniques, their claim of training their model for $6 million was misleading. The amount only covered the last round of training, not the complete development cost. So, they were more economical in training their model, but not to the extent claimed. Researchers from Stanford and the University of Washington recently trained an AI “reasoning” model for under $50 similarly using distillation, a technique using existing models to tune new models. DeepSeek's innovations were one of many that have been unfolding and will continue to unfold as AI develops. Regarding DeepSeek, their achievements were noteworthy, but it's time to move on. No doubt the next generation of reasoning models will be out before summer and with them will come new disruptions.
Article on Standford and the University of Washington's $50 model:
As always, feedback, suggested topics, or questions are welcomed. I’m here to help. Contact me anytime.
Contact us
Whether you have a request, a query, or want to work with us, use the form below to get in touch with our team.
