Mason Roberts

It’s not much different than the challenges of getting a VBC contract right, but it’s a completely different set of calculations and pricing.

#HealthActuary #ReinsurancePricing #ValueBasedCare #ActuarialScience #HealthcareFinance

And guess what the chances are the provider has similar enough contracts that you can pool the risk - it’s not common in my experience.

Each of the things I mentioned above have twists and turns like this. Nuenaces that if you don’t get right you’re out of business.

You’ll notice that this, in general, increases the variance by more than just the variance of the premium. It’s also the interaction of the premium and the claims, which means you need potentially a lot more patients to have a credible group.

That where you find this beauty of a formula, the Delta Method:

Var(R) ≈ R² ⋅ [CVx²+CVy²−2pxy ⋅ CVx ⋅ CVy]

Where R is the average MLR, CV are the coefficients of variation, and p is the correlation between claims (X) and premium (Y).

Let’s assume you have a MLR contract. How many patients do you need to credibly price, to trust the historic data?

There’s a formula for that, standard limited fluctuation credibility, but to use it you first have to calculate the variance of a ratio, the MLR.

To add to the fun, these features don't compound independently. They interact in ways that break standard reinsurance modeling assumptions.

I’m going to assume you’re ready to go fully down the rabbit hole if you’re still with me, so let me take you to math’town.

-- MLR credibility requires delta method variance adjustment for the ratio. The full-credibility sample size is materially larger than for straight claims.

Traditional medical reinsurance prices claims where VBC reinsurance prices a contract outcome, which is a non-linear derivative of claims.

-- Attribution and churn mean the population you underwrite at inception could be meaningfully different from the population measured at settlement.
-- Quality measure contingent triggers create a correlated two-trigger structure with almost no historical calibration data.

-- Trend guarantees (where a provider group sets a bar they have to beat) are effectively spread options. You're pricing the probability that one random variable exceeds another, not that it exceeds a fixed threshold.

Mathematical formula showing the Delta Method approximation for the variance of a ratio R = X/Y. The formula expresses Var(R) in two equivalent forms: first using means and variances of X and Y with a covariance term, then equivalently using coefficients of variation (CV) and the correlation ρ between claims X and premium Y. Used to illustrate how MLR variance exceeds claims variance alone when pricing VBC reinsurance.

VBC contracts are creating a new reinsurance frontier that existing actuarial tools weren't built for.

What we see everyday is that pricing reinsurance to match the risk of a VBC contract is a gift that just keeps giving in terms of challenges. Let me give you a sense of the complications:

I hope he knows they are not going it alone in the path to VBC. Many have tread this path before and they could lean on and learn from many in the industry.

#ValueBasedCare #HealthcareFinance #HospitalStrategy #RiskManagement #HealthPolicy

-- Incentive alignment within their organization and across partnerships.
-- Data infrastructure investments with real-time feedback loops.
-- A financial architecture redesign or true-up to include cost modeling, forecasting, etc.

This is a model that has been successfully implemented in many organizations. The items that he missed (maybe due to word count constraints) in his article that I think are crucial are:

-- Shift care to lower-cost settings (community, ambulatory, home, virtual).
-- Improve access timing (same-day, navigation, proactive outreach).
-- Build stronger community partnerships.
-- Focus on measurable outcomes, not facility utilization.

Ascension’s new CEO (leader of some 90 hospitals and 40 senior living facilities operating in 17 states) has said as much (see: buff.ly/ly05ADk).

He acknowledges that reimbursement is changing and his solution follows what I think of as a general VBC playbook:

Graphic titled “Hospital Systems Under Pressure” showing multiple interlocking gears around a large cracked central gear. Surrounding labels include: Incentive Misalignment (lack of motivation for VBC adoption), Reimbursement Changes (need to adapt to new payment models), Uninsured Populations (higher costs and reduced revenue), CMMI Programs (increased regulatory burden), Financial Architecture (inaccurate cost modeling and forecasting), and Data Infrastructure (inadequate data for informed decision-making). The cracked central gear symbolizes systemic strain within hospital systems.

This week seems to be revolving around hospital systems. My sense is that with the additional CMMI programs and the increases to our uninsured populations, hospital systems are feeling attacked from every angle (and thus the increase in news and posts).

Side note: The REIT owned facilities (about ~12 of SNFs) are going to be particularly interested since they have resources and there’s tenant risk for them. There’s maybe some larger facility groups that you shouldn’t sleep on, but thought I'd call this out.

#ValueBasedCare #PostAcuteCare #TEAM

-- Deploy actionable predictive analytics embedded in clinical workflows (not standalone dashboards).
-- Share the post-acute care risk using the TEAM portion of the post-acute care costs as the cap

So here’s how I would approach this:
-- Build preferred post-acute networks based on performance (cost and quality) data, not geography or convenience.
-- Establish shared metrics and routine joint performance reviews.
-- Use real-time data visibility across care settings.

I've seen clients drop their post-acute care costs by 20% just by being more thoughtful about where patients go post discharge.

Look, I know I'm a hammer looking for a nail. I’m biased towards a VBC solutions, but you can’t deny that it’s a good construct for shifting risk.

Most are going to need better integration with your data, which is non-trivial. But all this work is worth it. Basic transition programs (e.g., medication review, timely follow-ups, structured discharge planning) can reduce readmissions by ~30%.

So why not find your best post-acute care partners and approach these facilities to share in the risk?

Some post-acute care facilities are going to be better suited to take on the risk than others so you might have to bring them along with ramps to risk.

I think there’s a high likelihood that providers in hospital systems have a gut sense, but don’t really know, how much their preferred post-acute care facilities cost and what kind of outcomes they can ensure. A gut sense isn’t good enough in TEAM or AHEAD where there's real skin in the game.

The main risk I see: hospitals will treat post-acute care (SNFs, home health) as “external,” despite being financially responsible for outcomes and costs across the full episode.

I have no doubt that facilities can manage intake and ensure patients get the most appropriate care while with them. I’m a little more skeptical that they will be able to appropriately code patients on the front end and a lot more skeptical that they will control costs post-discharge.

Dark-background infographic titled “Strategies for Post-Acute Care Risk Sharing.” On the left is a series of concentric circles forming a target graphic, representing layered strategy. On the right, dotted arrows point to five labeled strategies with icons: Preferred Post-Acute Networks (globe icon), Shared Metrics and Reviews (handshake icon), Actionable Predictive Analytics (chart icon), Real-Time Data Visibility (document icon), and Share Post-Acute Care Risk (clock/medical cross icon).

CMS has been doubling down on pushing hospital systems into risk with programs like TEAM and AHEAD. Readmissions, SNF quality issues, and poor handoffs and post-acute care costs now directly hit hospital margins.

my understanding is this is different because it provides cost information. Let me know what you see in your systems or if you’ve built something similar.

#HealthcareFinance #ValueBasedCare #ClinicalVariation #HealthcareInnovation #HealthcareLeadership

Their method for calculating this seemed reasonable to me, but you may want to double check this.

This was applied in a facility setting but I could see something similar being applied within EHRs across VBC providers. I’m not an EHR expert, but

-- They had a few different levers of alerts based on the clinical context: “typical” EMR alters, passive nudges, and active alters. This part was the fuzziest for me.

They expected savings of $32.50 per admission and got $97.00.

-- Institution-specific cost integration (this provides who’s in your network and what do they cost)
-- They provided references to peer-reviewed literature + national guidelines