Network design · Governance · Participation standards · Credentialing and audit · Coordination infrastructure
Designing a rural clinically integrated network
A rural county's hospital, primary care practices, EMS agencies, behavioral health provider and home health agency each sit below the scale at which a care coordinator, an analyst, a credentialing function or a payer contract can be afforded. A clinically integrated network holds those functions once, across all of them, and leaves every participant independent. The Paramedic Foundation designs that structure to a published federal reference standard, and designs it so that counsel can defend it.
A clinically integrated network is a layer of shared infrastructure sitting between independent rural health organizations that are each too small to afford it alone. The term appears in neither statute nor regulation. It comes from antitrust enforcement policy, and the operative test is whether the network changes how its participants practice or merely speaks on their behalf.
TPF designs that layer to a published federal reference standard and designs it to be defensible. It does not operate clinical services, and it does not clear a design legally.
Clinically integrated network
A private-market care management and contracting vehicle whose participants accept interdependent performance obligations. Its legitimacy rests on the ancillarity of its restraints, the former enforcement safety zones having been withdrawn in 2023.
Accountable care organization
A defined participant in a federal program, with prescriptive governance requirements and a payer-specific contract. A network may operate one. An accountable care organization is not thereby clinically integrated for any other purpose.
Participation standard
What membership obliges, in licensure, medical direction, credential, protocol adherence, insurance, data submission and responsiveness, each with a named evidence basis and a written consequence for a participant who falls short of it.
Unmatched request
A request for care that arrived, was specific, was located, and could not be filled. Recorded by county and service, the accumulated set maps where capability does not exist. Neither a survey nor a claims file produces it.
The problem is arithmetic before it is clinical
A rural county holds a hospital or none, a few primary care practices, one or two EMS agencies, a behavioral health provider covering several counties, home health, a pharmacy, and community organizations touching the same patients without seeing each other's records. Each of them is competent within its own scope. Each sits below the scale at which a care coordinator, a data analyst, a quality methodology, a payer contract, a credentialing function or a compliance officer can be carried on its own budget.
Each of those functions is affordable once across thirty organizations and unaffordable thirty times over.
So every participant performs a version of the same work separately, at a standard none of them would accept if they could see the others doing it, and at a cost none of them can carry. That is the argument for the network form, and it is an infrastructure argument rather than a clinical one. Infrastructure is what rural systems are least able to build for themselves.
Care coordination
Coordinators with a stated caseload, standardized referral pathways, and a defined loop-closure event and interval. Most rural referral failure is loop failure.
Measurement and analysis
An analyst, a measure set with baselines, and a quality improvement method applied consistently. Thirty organizations produce enough volume to measure. One rarely does.
Credentialing and audit
Organizational credentialing at entry and on re-assessment, practitioner verification against a nationally recognized certification, and a proportionate sampled audit.
Payer-facing contracting
The capability to negotiate as a body, which is also the capability that makes the antitrust analysis matter and that a design has to earn rather than assume.
Compliance and data governance
Data-sharing agreements and the technical means to honor them, business associate agreements, and a compliance function that no single participant could staff.
What clinical integration means in practice
The operative test from the enforcement record is short. A network is clinically integrated when it changes how its participants practice. Five elements carry that test, and a design is assessed against all five rather than against whichever one is easiest to evidence.
- Element 01Interdependence. Each participant holds a real stake in the others' performance. Where a participant can fail without consequence to anyone else, the arrangement is an association rather than an integration.
- Element 02Objective peer measurement. Performance is measured against peers by objective standards. Self-report satisfies the documentation and not the test.
- Element 03Actual facilitation of coordination. The structure does the coordinating. A governance document describing coordination that no one performs is the most common finding in a network design review.
- Element 04Enforced common clinical standards. Standards held through peer review, utilization management and outcome measurement, with consequences that reach a participant who does not meet them. The consequence sequence is written before it is needed.
- Element 05Health information technology as the hub. Technology carries the coordination the other four elements require of it.
Two cautions TPF repeats wherever they apply
Adopting information technology is not by itself clinical integration. It is a tool serving the other four elements. A design that leans the whole integration case on a platform purchase has not made the case.
The market-share safety zones are historical. The Federal Trade Commission withdrew the enforcement statements that contained them on July 14, 2023, and the Department of Justice withdrew the same statements. The analytic method and the vocabulary survive. The thresholds do not, and TPF says so in any document where the question could arise.
Precision
Four structures a network gets confused with
Each of these appears in solicitations as a synonym for a clinically integrated network. None of them is one, and scoping against the wrong definition produces a proposal that answers a question nobody asked.
Accountable care organization
A defined participant in a federal program, carrying prescriptive governance requirements and a payer-specific contract. It is a program participation status. Clinical integration is a characteristic of how an organization operates.
Grant-funded rural health network
A collaborative that sets no minimum membership standard, requires no legal entity and confers no authority to negotiate jointly. Valuable in its own right, and it obliges its members to nothing.
Health information exchange
An exchange moves data between organizations. A network commits organizations to interdependent performance. An exchange can serve a network and cannot substitute for one.
Messenger-model contracting
A vehicle in which the absence of joint pricing is precisely what keeps the arrangement lawful. Describing a messenger model as clinically integrated misstates the basis on which it is defensible.
Function
Three things a network does that its participants cannot
It sets the standard
What participation requires in capability, credentials, protocols and data submission, and what follows for a participant who does not meet it. This is the function most often written as an intention and least often built. A participation standard should be short enough to read, specific enough to audit, and explicit about the evidence that satisfies each element.
It matches need to capability
The binding problem in a rural system is rarely that nobody can do the work. It is that whoever holds the need cannot find out who can. Matching is a function with an intake, a published rule set, a record and an accountable operator. A network that performs it becomes the number that gets called.
It holds what no participant can hold alone
Data-sharing agreements and the technical means to honor them, the measurement method and the analyst who applies it, quality improvement discipline, payer-facing contracting capability, and compliance and audit. These are the functions the arithmetic section describes, assembled into one place.
A proposal that describes all three functions, names who performs each, and states what evidence will show each is working, is describing a network. A proposal that describes convening is describing a committee, which may still be worth funding under its own name.
What the evidence actually supports
This section is written to be uncomfortable. A sustainability argument that has not engaged with the findings below will not survive a technical reviewer who has read them.
Governance structure predicts savings more reliably than participation does
A difference-in-differences analysis published in the New England Journal of Medicine in 2018 found that physician-group organizations entering the Medicare Shared Savings Program in 2012 had reduced per-patient spending by $474, or 4.9 percent, by 2015. Hospital-integrated organizations in the same cohorts showed far smaller reductions, and those reductions were offset by bonus payments. A Milbank Quarterly analysis in 2020 found gross savings of $139 to $302 per patient for the earliest cohorts with no evidence of favorable risk selection. The Congressional Budget Office concluded in April 2024 that accountable care models produced small net savings, naming weak benchmark incentives, insufficient resources for participation in rural areas, and favorable selection through voluntary entry and exit.
The design implication is direct. Who governs, and on what terms, is a more consequential decision than how many organizations sign on.
Rural continuation is the hard number
A federal investment model that prepaid shared savings reached 45 organizations across 38 states, 36 of which had at least 65 percent of their delivery sites in rural areas. Rural Policy Research Institute analysis published in February 2026 found that of rural accountable care organizations starting in 2016, 60.4 percent participated in that model and 17.2 percent renewed afterward.
Grant-funded networks survive, in a qualified sense
A survey of federal network planning grant awardees from 2003 to 2018 found 88 percent had sustained at least some components, with engagement and alignment of divergent partner priorities the leading barrier at 42.9 percent, ahead of resources at 19 percent. The response rate was 18 percent, 42 of 377, so the finding carries survivorship bias in the optimistic direction and should be read with its denominator.
Information exchange does not substitute for integration
An Agency for Healthcare Research and Quality systematic review in 2015 covering 136 studies found no study reporting an impact on primary clinical outcomes, low-quality evidence for reduced duplicative testing and emergency department costs, and generally very low use of exchange capability inside the organizations that had bought it.
TPF's practice follows from this. It claims the coordination pattern, which a reviewer can verify against published material, and it does not claim that a particular application of the pattern is proven. Blurring those two claims is the fastest way to lose a technically sound proposal.
Scope
What TPF designs
Engagements are scoped to what a client actually lacks. Few need all nine, and a proposal offering all nine to everyone has not looked at the client.
Network design and governance
Legal entity, governing body composition and participant control, independent representation, conflict-of-interest policy, management structure, participation commitment, and the committee structure that holds clinical standards. TPF designs to the federal accountable care organization rule as a published reference standard rather than inventing structure, and states that this is what it is doing.
Participation standards
Licensure, medical direction, credential, protocol adherence, insurance, data submission and responsiveness, each with the evidence that satisfies it named alongside. Written to be audited rather than to be admired, and paired with a consequence sequence running from corrective action plan to suspension to removal.
Credentialing and audit
Organizational credentialing at entry and on re-assessment, practitioner verification against a nationally recognized certification rather than one the network invents for itself, and a sampled audit function proportionate to the risk it is checking.
Coordination and matching infrastructure
Intake, eligibility and triage rules, a maintained directory of what each participant is credentialed to do and where it will travel, a published matching rule set applied in fixed order, assignment with an accept-or-decline interval, and the record. The requester is owed an acknowledgement and a definitive matched-or-unmatched answer, and the unmatched answer has to arrive as fast as the matched one.
Medical direction that disturbs no statutory chain
Common practice across a network is produced by adoption rather than by direction. A network medical advisory committee composed of participating organizations' own medical directors authors and maintains the protocol set, and each participating organization's medical director adopts it by signature for the work that organization accepts. Every visit is then performed on the same protocols, under the direction of the practitioner's own medical director.
Agreement architecture
Participation agreement, protocol adoption instrument, business associate agreement, data use agreement, requester terms and governance charter, drafted to the point where counsel reviews rather than drafts. Counsel time is the scarce input in most of these engagements and the drafting stage is where it gets wasted.
Care coordination and transitions
Shared clinical priorities selected from the served population, standardized referral pathways with a specified loop-closure event and interval, care transition protocols each carrying an owner and a measure, and coordination designs that state a caseload. A care coordination plan without a caseload figure has not been costed.
Data and measurement
Instruments before technology: data obligations in the participation agreement, then data use and business associate agreements, then interoperability positioning described accurately, then the measure set. Four rules govern the measure set. Adopt the funder's own committed metrics. Report the denominator. Keep process counts separate from outcome measures. Take a baseline before the intervention, or record plainly that none exists.
Sustainability planning
Four routes exist and each carries a milestone that has to be hit inside the term: payer contracting, participant contribution, licensing what was built, and absorption by an existing institution. Absorption is a legitimate outcome and naming it as the intended one reads as more credible than promising perpetual independence. No route matures inside a first budget period.
Boundaries
What the structure does, and what it never does
The network does
Credential organizations and verify practitioners against an external standard.
Convey a request for care together with the clinical information necessary to act on it.
Maintain the protocol set that participating medical directors adopt by signature.
Hold the record of what was requested, what was filled, and what went unfilled.
The network does not
Receive emergency calls or affect any emergency response. It is not a dispatch center, and that is the first assumption an EMS audience makes.
License anyone. Credentialing and licensure are separate acts by separate bodies.
Direct care. A referral conveys a request rather than an order.
Displace any participant's own medical direction or governing board.
Legal opinions and filings rest with the client's counsel. The antitrust, privacy and liability positions in a network design are design positions, and they require confirmation before the function operates in any jurisdiction. Audited financial statements and tax returns rest with the client's accountants. TPF's design work is what makes a defensible answer available and puts the right question in front of counsel.
References
Who does this work
The bench behind network design engagements
Network design draws on rural and frontier EMS governance, financial sustainability and reimbursement, data and analytics, and program evaluation. Teams are assembled for the engagement rather than allocated from a fixed roster.
Regulatory architecture and entity formation. Regional authority and hub design. EMS finance, rate setting and cost analysis. Medicaid and alternative payment model strategy. Health economics and rural health research. Performance measure development and data governance. Healthcare data standards and systems integration. Nonprofit board development and governance training. Federal award compliance under 2 CFR Part 200.
Profiles, credentials and current roles are on the team page. TPF names the individuals proposed for an engagement in the proposal for that engagement, where the names can be matched to the scope.
About network design engagements
Common questions
Is a clinically integrated network defined in law?
No. The term appears in neither statute nor regulation. It came out of antitrust enforcement policy, and on July 14, 2023 the Federal Trade Commission withdrew the 1996 Statements of Antitrust Enforcement Policy in Health Care and the 2011 statement on accountable care organizations, saying they were outdated and that it would rely on general principles case by case. The Department of Justice withdrew the same statements. The analytic method and the vocabulary survive that withdrawal. The market-share safety zones do not, and a document that presents them as current guidance will be recognized as wrong by a reviewing attorney.
How is a clinically integrated network different from an accountable care organization?
An accountable care organization is a defined participant in a federal program, with prescriptive governance requirements and a payer-specific contract. A clinically integrated network is a private-market care management and contracting vehicle. A network may operate an accountable care organization. An accountable care organization is not thereby clinically integrated for any other purpose. Solicitations use the two terms interchangeably often enough that TPF checks which one a client means before scoping anything.
Does building a network mean consolidating organizations?
No. The case for a network is that it delivers coordination without consolidation. It holds the functions no single small organization can afford on its own and leaves each participant independent. The published evidence on vertical integration in health care is not favorable, which is part of why the network form is worth building, and language that treats consolidation as the goal undercuts the argument for the network.
What does the matching function actually produce?
Two things. It produces filled requests, which is what a participant notices. It also produces a dated, located record of the requests that could not be filled, which is what a state can act on. A survey reports what agencies intend to offer and a claims file reports what was billed. Neither shows demand that arrived, was specific, was located, and went unmet. That record is the most valuable output of the function and it is usually designed in as an afterthought.
Does TPF provide legal advice on antitrust?
No. TPF designs a network so that it can be defended and does not clear one. The antitrust, privacy and liability positions in any design are design positions requiring confirmation by the client's counsel before a function operates in any jurisdiction. TPF's contribution is that counsel is asked the right questions rather than handed a finished document to approve.
Building a network, or asked to show that one is more than a committee?
The decisions that determine whether a network holds together are made early: what participation requires, who governs what, how the matching function works, and what evidence will show it is working.