January 2025 Update: Automated Patient Engagement, Claims Denials Costs, Healthcare Market Consolidation, the Economics of DeepSeek AI, and More
The January update covers contemporary topics and issues in healthcare, life sciences, AI, and technology. It begins with learnings from Change Healthcare fines from automated patient outreach, The restructuring of major Blues health plans, such as Blue Shield of California, reflects a trend toward improved care, affordability, and diversified investments, presenting opportunities for healthcare innovators. Meanwhile, disruptive shifts in AI (e.g., DeepSeek's open-source models) change the economics of AI, and persistent challenges like claims denials emphasize the need for smarter automation and integration, We conclude by uncovering the missing discipline that is at the root of transformation failure.
Joe Bastante
1/24/20256 min read
In this month’s update:
Why a major healthcare company was fined for member outreach and what can we learn
Why are Blues restructuring and why I should care?
Taking a bigger slice of the pie: insurers + providers + PBMs
Why businesses should care about DeepSeek's latest AI models
Care denials, latest trends and impacts on healthcare costs
88% of company transformations fail, what's the missing discipline?
Why a major healthcare company was fined for member outreach and what can we learn
In 2024, a settlement was reached with Change Healthcare, who made robocalls on behalf of an NC-based health insurer. The calls violated the Telephone Consumer Protection Act (TCPA), resulting in a class action dispute and a $1.67M settlement. TCPA is fairly complex, and highly relevant as many companies consider AI and automated outreach to improve engagement. The following points are key in safely engaging patients and members while avoiding TCPA-related fines and negative press. In general, TCPA disallows automated outreach (calls, faxes, SMS messages) to consumer without prior consent. Prior consent may not be required for non-profits, thought they must honor the do-not-call registry. Informational messages such as healthcare appointment reminders or prescription refill reminders are generally allowed so long as opt-out is provided. An established business relationship with a consumer may allow automated calling, but tread carefully as written consent is still needed for calling cell phones. Fines range from $500-$1,500 per call. Intentional violations carry the heavier fine. Consult a lawyer (since I'm not one) or the links below for more info.
Class action suit referenced above:
Nice resource summarizing TCPA:
Why are Blues restructuring and why I should care?
Blue Shield of California recently restructured under a new parent entity named Ascendiun. This has been a common move for many Blues plans. For example, in recent years, Blue Cross NC, Horizon Blue (NJ), and Louisianna Blue have similarly restructured. They are following earlier moves by Blues. For example, Florida Blue is under GuideWell and Elevance, the largest Blue, comprises a collection of companies. Stated motivations for restructuring center on improving care and affordability. To be more clear, some state-specific Blues are subject to greater state scrutiny over acquisitions and changes given they're viewed as resources critical to the state. Restructuring allows Blues to isolate the insurance business and become more competitive in diversifying, investing, and expanding offerings. This is good news for companies with impactful health offerings seeking investment. Blues already substantially invest together, for example, through the Blue Venture Fund, and they also invest independently and through smaller collections of Blues. These restructurings create opportunity for healthcare companies, especially those who can improve member health, save cost, and offer diversification and adjacency expansion. A favored investment would improve care quality and affordability for the investing Blue(s) and also offer expansion into other regions in non core insurance products.
See the Blue Shield of California announcement here:
Taking a bigger slice of the pie: insurers + providers + PBMs
This brief summary addresses a few misconceptions about healthcare costs, which have recently taken center stage. The horribly tragic shooting of UnitedHealthcare CEO Brian Thompson has raised concerns about health insurer profit seeking, which is understandable. UnitedHealthcare is one of the more profitable insurance companies. We must reign in healthcare costs, yet there's a common misconception that insurers keep whatever they don't pay in claims. In fact, for most lines of business, laws require the bulk of members' premiums to be spent on legitimate medical expenses, e.g., 85% for Medicare Advantage. UnitedHealthcare's operating margin was 5.8% in 2022 and 2023. United's Optum Rx had a 4.4%-4.5% operating margin 2021-2023. Of course, United is also a huge provider organization, achieving a 6.9%-8.5% operating margin during that same period. Some insurers are even investing in drug manufacturing (e.g., Blues investing in Civica Rx). Though United doesn't publish net profit margins by line of business, clearly profits do take a notable portion of healthcare spending, but it's cumulative across the complex U.S. health ecosystem, which would occur even if United didn't exist. They are just one of the many healthcare companies seeking a bigger slice of the healthcare pie. The reality is that healthcare cost will remain high so long as the value chain is so complex with each participant seeking favorable profits. Our challenge is to simplify healthcare.
Further info, UnitedHealth Group From 10-K, 2023:
https://www.unitedhealthgroup.com/content/dam/UHG/PDF/investors/2023/UNH-Q4-2023-Form-10-K.pdf
All United financial reports:
https://www.unitedhealthgroup.com/investors/financial-reports.html
Why businesses should care about DeepSeek's latest AI models
DeepSeek is a Chinese AI startup that released AI models creating quite a stir. They've produced state-of-the-art AI models and released them under MIT's generous open-source license model. DeepSeek offers an open-source alternative on par with OpenAI's best models. An open-source frontier model of this caliber is sufficiently disruptive, but possibly the bigger disruption will come from their cloud pricing. Their DeepSeek-V3 model, which is comparable to OpenAI's GPT-4o model, is much less costly. OpenAI is about 29.8 times more expensive than DeepSeek-V3 (that's right, 29.8 times not percent). DeepSeek-R1, which is a reasoning model similar to OpenAI's GPT-o1, is also similarly less costly than OpenAI's model (i.e., very roughly 30 times cheaper). While many will be hesitant to use Chinese AI services, no doubt they've dramatically changed the economics of AI for businesses. As in manufacturing, textiles, and automobiles, China is using economics to disrupt the AI industry. Only time will tell if they can persuade foreign business to embrace their services. Even if they don't, Chinese companies will have a distinct advantage in embedding AI into their products and processes. U.S. companies are already hosting and offering DeepSeek models, so the economics will undeniably change.
Summary of DeepSeek-V3 as compared to OpenAI GPT-4o
Summary of DeepSeek-R1 as compared to OpenAI GPT-o1
https://softreviewed.com/deepseek-r1-vs-claude-openai-o1-decoding-context-price-and-coding-prowess/
Care denials, latest trends and impacts on healthcare costs
Many of us share the interest in eliminating waste in healthcare, which often manifests as operational overhead between providers in payers. I had previously done a deeper dive on prior authorization, and in this snipped I summarize the issue of claims denials. Most research pins the in-network claim denial rate between 14%-20% broadly. Based on surveys of effort handling each denied claim, providers spend about $19.7bn each year handling claims denials. Of course, costs are incurred by payers too. Denial rates vary considerably by insurer. See the links below for detail. The American Hospital Associate stated that between 2022 and 2023 respectively. care denials increased an average of 20.2% and 55.7%. Automation can help, but keep in mind that CMS, state, etc., policies place limits on using AI to deny coverage. (I covered this in an earlier post.) Experian claims that the number of providers currently using some form of automation and/or AI to review claims has dropped from 62% in 2022 to 31% in 2024. As to why claims are denied, they're commonly denied for missing or bad data (patient info, claim info, billing codes) and for neglecting to submit prior authorizations, which tells us that many denials are avoidable. Those of us in healthcare can and must do better. We need better real-time integration between payers and providers to avoid submission of claims only afterword finding validation rules were tripped on the payer side.
The links below provide further data on the state of claims denials.
https://www.valuepenguin.com/health-insurance-claim-denials-and-appeals
88% of company transformations fail, what's the missing discipline
Most company transformations fail. Bain pegs the percentage at 88%, McKinsey says it's 70%. Whatever the proportion, it's concerningly large. Numerous causes have been implicated: unclear strategy, poor leadership team alignment, poor communication, cultural resistance, lack of short-term wins, poor oversight, poor planning, etc. After spending much time reflecting on my lessons learned and reviewing many articles and case studies, a clear fact emerged, that no individual causes are the ultimate root cause. Rather, the true problem is that there are many ways a transformational initiative can fail with risks crossing many departments, functions, and processes—and we expect the owning leader to understand them all, which is infeasible. A traditional risk management approach calls for collecting and assembling top risks from stakeholders. However, risks will be missed, especially those with unclear owners, and this won't be discovered until it's too late. What is needed is a unified method based on comprehensive lessons learned to manage risk and success factors end-to-end. I share this final topic since I worry many transformations will fail tripping on issues that are knowable in advance. It's a focus of mine to build out a holistic method, and I'm inviting anyone with insights and interest to reach out to me.
Countless resources are available on failure causes. Below are the two from which I obtained the 88% and 70% failure rates:
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.
