Distributing a Shared Savings Pool in a Clinically Integrated Network with Independent Physicians
Northshore Health Partners is a clinically integrated network of 220 primary care and specialty physicians: 96 employed by the two-hospital system and 124 independent. A commercial payer has just settled year one of a shared savings contract covering 9,000 attributed lives. The benchmark was $1,050 PMPM and actual spend came in at $1,012, a $38 PMPM gross saving. The contract sets a 3% minimum savings rate, a 50/50 split, and requires hitting 4 of 6 quality measures; the quality gate was met. Northshore spent $1.1M on care coordinators, a data platform, and pharmacist support. At the July network meeting the independent physicians demand an equal per-physician split of the check. The employed physicians argue their staff did the outreach, and the system wants its infrastructure costs recovered first. Two large independent groups have hinted they will not sign the year-two extension, which introduces 20% downside risk on a benchmark base of $113.4M.
1. Calculate the actual shared savings pool. Then critique the equal per-physician split and recommend an allocation method, showing the numbers.
Read the worked answer
Pool calculation.
- Minimum savings rate: 3% × $1,050 = $31.50 PMPM
- Savings above the MSR: $38.00 − $31.50 = $6.50 PMPM
- Annualized: $6.50 × 12 × 9,000 = $702,000
- Network share at 50%: $351,000
The $38 PMPM headline is not the check; the MSR absorbs 83% of the gross savings.
Why an equal split fails. $351,000 / 220 = $1,595 per physician. That is too small to retain a group and too large to be defensible to the physicians who actually did the work - it is a token that satisfies nobody and rewards non-participants identically to participants. Worse, it teaches the network that participation is optional.
Recommendation - a three-part formula, published before year two.
- Platform fee, capped and disclosed in advance: 20% of the pool ($70,200) to recover part of the $1.1M investment. Capping it matters: the honest message is that the infrastructure is an investment in future years, not a cost to be recovered from a $351,000 pool.
- Performance distribution (70%, $245,700): weighted 60% by attributed lives, 40% by a participation score (data sharing completeness, two quality measures, care-plan completion). This rewards the physicians whose panels generated the savings and keeps independents engaged.
- Reserve (10%, $35,100): held for year-two downside exposure and for practices that join mid-year.
Publish a worked example so every physician can compute their own number. The allocation is not really a math problem - it is a retention decision, and the transparency of the method matters more than the size of the check.
2. The year-two extension adds 20% downside risk. Quantify the exposure and state the conditions under which Northshore should accept it.
Read the worked answer
Size the exposure first. 1 PMPM across 9,000 lives = $108,000 per year. A 20% downside measured against the $1,050 benchmark is $210 PMPM, or $22.7M - roughly twenty times the network's cumulative investment and an uninsurable number for a two-hospital system. Even a 2% cap would be $2.27M. The first question to ask is not "do we accept risk" but "what is the cap, and is it measured against benchmark or against savings?"
Then test the cushion. To avoid a shared loss, year-two spend must stay at or below 97% of benchmark, or $1,018.50 PMPM. Year-one actual was $1,012 - a cushion of $6.50 PMPM, or $702,000. A single $600,000 readmission cluster equals $5.56 PMPM, which consumes 85% of that cushion. There is no actuarial margin here; the network is running at the edge of the MSR.
Conditions for acceptance.
- Cap the downside at 1.5–2% of benchmark ($1.9M–$2.5M), with a hard stop and no rollover.
- Rebase the benchmark for acuity and trend, and lock attribution rules in writing before the start date.
- Get claim-level data, not 90-day-lagged summaries; without it, the network cannot manage what it is being paid on.
- Build a minimum panel threshold so a handful of outlier patients cannot determine the result.
- Model one bad year against days cash on hand and the bond covenant before signing.
If the payer refuses a cap and timely data, decline the extension and keep the upside-only structure. Accepting uncapped risk to preserve a relationship is how a $351,000 check becomes a balance-sheet event.
3. The independent groups are threatening to leave. What governance and information-sharing changes would you make, and what compliance guardrails must the network respect?
Read the worked answer
Diagnose the real complaint. The independents are not primarily angry about $1,595; they are saying they contributed data and behavior change and received neither money nor information. Fix the information problem first, because it is cheap and it is the precondition for everything else.
Governance changes.
- Create a physician-led distribution committee with a written charter, seats proportional to attributed lives rather than employment status, and an appeals path with a 30-day decision standard.
- Give every practice a quarterly performance report: attributed lives, benchmark versus actual PMPM, where the variance came from, and their own contribution to the participation score.
- Publish the distribution formula and a worked example a year ahead, so no one is surprised by a check again.
- Define exit terms - vesting, tail liability for downside years, and data return - before year two, not after a group announces it is leaving.
Compliance guardrails.
- Independent physicians are competitors. All shared data must be aggregated or de-identified and routed through the CIN or a third-party convener; never let competing practices see each other's prices, payer rates, or patient-level panel data.
- Contracting must be conducted by the network, not by physicians jointly, and must rest on genuine clinical integration: shared protocols, joint quality measurement, real infrastructure investment, and a credible program to change utilization.
- Document the integration and keep the analysis current; the more the arrangement looks like a contracting vehicle for independent competitors, the more antitrust exposure it carries.
Framing for the July meeting. Show the $351,000 pool against the $1.1M investment and state plainly that year one was a build year. Then show the $702,000 cushion and the year-two economics. Physicians leave networks that give them neither dollars nor data; give them the data immediately and the dollars on a published formula.