Abstract
Population health has failed not from lack of ambition but from lack of specificity. The official definition—improving health status of entire populations through upstream interventions and social determinant screening—has consistently asked delivery organizations to take responsibility for outcomes they cannot control: too broad to execute within a single health system’s competency, too focused on upstream determinants that health care systems cannot change at scale. 1 Most organizations spend resources on universal interventions they cannot scale while ignoring the specific populations they could actually serve well.
This commentary presents a conceptual and operational framework—not empirical research—for health systems and provider organizations assuming defined financial risk for clinical populations. The central argument is that health systems should focus on core clinical services within their competency, avoid taking financial risk for outcomes they cannot control, refrain from building services that qualified community providers already deliver, and pursue strategic partnerships rather than defaulting to direct ownership. It is not intended as a critique of public health or social policy but as a pragmatic execution model for organizations with clinical accountability.
Targeted social determinant interventions can materially impact outcomes when integrated into segment-specific care models through strategic partnerships. Food insecurity screening and medically tailored meals for diabetes patients with complications, housing stabilization for high-risk elderly populations, and transportation coordination for dialysis patients demonstrate measurable clinical impact when health systems partner with community organizations whose core competency is delivering these services. The distinction is operational: social determinant interventions succeed when health systems refer to or subsidize organizations with established expertise (Tiers 1 and 2) rather than building parallel social service capacity that duplicates existing community resources.
The parallel mistake is full capitation across diverse populations. Assuming one financial arrangement works for young adults with diabetes, frail nursing home-eligible elderly, and complex multiple sclerosis patients reflects fundamental confusion about clinical segments. Each population has different cost structures, different intervention intensity requirements, and different outcome expectations. One-size-fits-all risk arrangements generate predictable failures: risk selection, undertreatment, cost-shifting, or system losses.
Organizations attempting global capitation have repeatedly failed despite substantial resources and infrastructure. Geisinger lost millions on their health plan while Kaiser’s capitated model works only within Kaiser’s closed system and has not been replicated successfully elsewhere despite decades of attempts. Even PACE, the most targeted capitated model for a specific population (frail elderly), has struggled to achieve meaningful enrollment and expansion beyond a handful of successful programs. Most health systems attempting capitation have lost money systematically, gamed risk selection rather than improving care, or both. The Alternative Quality Contract of Blue Cross Blue Shield of Massachusetts represents a notable exception—an 8-year two-sided global budget model that achieved sustained savings and quality improvements—but its success demonstrates precisely that focused accountability with genuine financial risk requires infrastructure and delivery competency that most organizations have not built. 2
This does not mean segment-specific accountability cannot work—it means successful execution requires far more operational discipline and delivery system competency than most health care organizations possess. Organizations pursuing focused financial risk for defined populations must have genuine integrated delivery capability across the required care continuum, clinical leadership committed to outcomes accountability rather than volume production, longitudinal data infrastructure enabling real outcome measurement, and financial arrangements creating consequences for poor performance. Few organizations have built this competency foundation before assuming risk. Even those that have faced enrollment, expansion, and sustainability challenges.
A third failure stems from measuring what is easy rather than what matters. 3 Process metrics (did you deliver the intervention?), access proxies (30-day readmissions), and compliance counts (percentage screened) obscure whether patients actually improved. These metrics correlate more with social factors, system gaming, and luck than actual care quality. Organizations celebrate referring all diabetes patients to nutrition but never track whether A1C improved or complications were prevented. Academics publish research assessing quality based on referral completion and screening rates rather than documented patient outcomes, further embedding measurement theater into the literature.
The core problem: health care organizations lack the 3 executable elements required for sustainable population health. They attempt universal interventions for undefined populations, they pursue capitation without the competency to execute it, and they measure compliance rather than improvement. This produces predictable underperformance while generating the appearance of systematic effort—organizations overreach into services and risk they cannot manage while neglecting the core clinical populations where their competency could produce genuine improvement.
Sustainable population health requires starting with specificity: (1) Define the specific population precisely, (2) Establish what success actually looks like for that population with realistic targets, (3) Build a targeted delivery system with the operational competency and financial alignment to achieve it.
Pillar 1: Defined Patient Population
Population health begins with specificity. 4 Not “our diabetes patients” but “patients with type 2 diabetes requiring medication intensification with documented complications or multiple comorbidities.” Not “frail elderly” but “community-dwelling adults age 55+ with 3 or more chronic conditions, current ADL assistance requirement, and high-cost health care utilization.” Not “COPD patients” but “stages III and IV COPD with recent exacerbation or multiple emergency visits.”
This specificity serves two purposes. First, it defines the clinical segment requiring common interventions. Patients with early diabetes managed entirely in primary care require entirely different capabilities than diabetic patients with active nephropathy and neuropathy. Mixing them in one program design ensures neither gets appropriate care. Second, it bounds financial risk. When you define “our frail elderly program serves community-dwelling adults 55+ meeting these specific criteria with these specific service needs,” financial partners can accurately price risk. Vague population definitions ensure capitation pricing includes hidden risk subsidies.
Defined populations typically cluster around 5 characteristics: age range, clinical diagnostic criteria, acuity/complexity level, functional status requirements, and comorbidity patterns. 5 Consider these real examples:
Defined population examples
High-risk diabetes with complications
Type 2 diabetics age 50+ with documented albuminuria, eGFR decline, or retinopathy requiring active coordination across endocrinology, nephrology, ophthalmology, and primary care. These patients account for disproportionate cost and benefit substantially from integrated management. They are clinically and financially distinct from type 2 diabetics with controlled disease in primary care alone.
COPD stages III and IV
Patients with FEV1 <50% predicted, recent exacerbation or emergency department (ED) utilization, or oxygen requirements. They require pulmonary oversight, coordinated primary care, rapid exacerbation response, home monitoring capability, and psychiatric/substance use disorder (SUD) screening. Mixing them with milder COPD obscures what coordination is actually needed.
Frail elderly (PACE-eligible)
Community-dwelling age 75+ with 3 or more chronic conditions, current activities of daily living (ADL) assistance requirement, and health care utilization indicating institutional risk without intervention. This population uniquely benefits from integrated primary care, geriatric assessment, care coordination, and preventive social services.
Defined populations also enable accurate financial modeling and appropriate quality metrics. When you know exactly who qualifies, you can calculate current spending patterns, identify high-cost drivers, estimate intervention costs, and project realistic outcomes. Vague populations guarantee financial surprises.
Pillar 2: Achievable Meaningful Outcomes
Once population is defined, specify what success looks like. 6 This requires discipline: achievable means realistic for this population with available resources. Meaningful means it actually reflects whether patients improved. This eliminates most current metrics.
Stop measuring: referral completion rates, office visit counts, screening completion percentages, process metric achievement, 30-day readmission proxies. These reflect administrative compliance, not patient improvement. Outcomes matter because they reflect whether patients improved. 7 They are also measurable: clinical data systems can track A1C, eGFR, exacerbation events, functional assessments, and falls. Real improvement requires tracking outcomes longitudinally—not just recording that a service was delivered but documenting whether the patient’s condition improved afterward.
These metrics require infrastructure: longitudinal data systems, outcome analytics capability, risk adjustment methodology, and systematic capture at clinical encounters. You cannot manage outcomes without measuring them accurately and consistently over time.
The operational challenges of longitudinal outcome capture should not be understated. Data interoperability across care settings remains fragmented, patient attribution in multi-provider environments is imprecise, and risk adjustment methodologies for segment-specific populations are still maturing. Consistent with this framework’s principle of focusing on what is within reach, organizations should not pursue measurement infrastructure beyond their current operational capacity. A practical threshold for accountability includes: consistent longitudinal tracking of 3–5 condition-specific clinical outcomes within the segment, patient-reported outcome collection at standardized intervals, attribution models accounting for plurality of care rather than requiring exclusivity, and risk adjustment using validated condition-specific severity indices rather than generic demographic adjustments. Organizations should measure what they can control within their core service and build measurement capability iteratively alongside program development—not delay implementation while pursuing enterprise-wide data infrastructure they cannot yet operate.
Pillar 3: Efficient Delivery System Organization
Once population and outcomes are defined, organize delivery efficiently. This does not mean owning all services. It means structuring referrals, partnerships, and selective ownership to achieve outcomes. Three tiers:
Tier 1: Optimize existing referral relationships
Most communities already have qualified specialists, therapists, and service providers. The primary failure is not capacity but coordination. Organizations default to building in-house services without first attempting to optimize existing external relationships.
Tier 1 execution requires: active identification of all qualified community providers for required services, establishment of warm handoff processes ensuring patients actually connect with providers, closed-loop feedback confirming patients received services and documenting outcomes, and navigation support helping patients overcome practical barriers (transportation, scheduling, insurance).
The IMPaCT community health worker program demonstrates this approach: health systems partnering with community health workers—rather than building in-house social service capacity—to address social complexity within defined clinical segments, producing improvements in chronic disease control and reduced hospitalizations through structured referral optimization. 8
Example: For high-risk diabetes requiring retinal screening, first establish active referral relationships with existing ophthalmologists, negotiate outcome-reporting mechanisms, implement patient navigation ensuring appointments are kept, and track whether screening actually occurred and what interventions resulted. Do not build an in-house ophthalmology service until you have exhausted referral optimization and documented genuine capacity shortfall.
Tier 2: Subsidize existing provider capacity
After optimizing referrals, genuine capacity gaps may exist: qualified providers exist but have insufficient capacity or financial incentive to serve your population. Tier 2 addresses this through targeted capacity-building subsidies.
Instead of building in-house services, negotiate specific capacity expansion with existing providers. Offer practice subsidies, outcome-based bonuses, guaranteed patient volume, or infrastructure support (data systems, care coordinators embedded in their practice) in exchange for expanded capacity and outcome accountability.
Example: A pulmonology group serves COPD patients well but lacks capacity for rapid exacerbation response. Rather than hiring your own pulmonologists, negotiate subsidy supporting expanded office hours, dedicated urgent slots for exacerbation management, and care coordinator embedded in their practice. Share savings from reduced ED visits and hospitalizations. Both parties benefit: pulmonologists gain predictable revenue and infrastructure support, health system achieves outcomes without building redundant capacity.
Published evidence supports this strategic partnership approach. The Johns Hopkins Community Health Partnership (J-CHiP) targeted high-risk Medicaid and Medicare beneficiaries through partnerships with community organizations rather than building parallel services, achieving $3,000 per member per year lower costs and 9%–26% reductions in hospital admissions and readmissions. 9 The IMPaCT program used existing community infrastructure and community health workers to improve chronic disease control for socially complex patients, demonstrating that health systems achieve better outcomes through strengthening community partners than through duplicating their capacity. 8
This approach succeeds when subsidies are: specific (defined services, defined capacity expansion), outcome-accountable (financial arrangements tied to outcome achievement), and sustainable (revenue-sharing or permanent capacity payment, not one-time grants).
Tier 3: Selective in-house service development
Only after rigorously executing Tiers 1 and 2 should organizations build in-house services. Tier 3 is justified when: no qualified external providers exist despite market search, existing providers refuse capacity expansion even with reasonable subsidies, service is so integral to the clinical model that direct employment is strategically necessary, or you have genuine competitive advantage in delivering this service.
Most organizations skip directly to Tier 3 without attempting Tiers 1 or 2. This represents the most common form of overreach—duplicating community capacity, building services without competitive advantage, creating fixed costs that cannot flex with patient volume, and competing with rather than partnering with existing providers.
Successful Tier 3 execution requires: documented failure of Tiers 1 and 2 (proving genuine gap exists), competitive analysis demonstrating you can deliver service at competitive cost and quality, financial modeling showing service achieves breakeven within a realistic timeframe, and integration plan ensuring service connects to the overall population management model.
Example: PACE programs typically own primary care directly (Tier 3) because geriatric primary care is so integral to their model, and community capacity is genuinely insufficient. But they contract with external specialists, therapists, and facilities (Tiers 1 and 2) where qualified providers exist. Selective ownership plus strategic partnerships.
Financial alignment: Outcomes require accountability
Delivery system organization fails without financial alignment. If outcomes matter but dollars reward volume, clinical teams face impossible conflicts. Financial arrangements must create genuine accountability for defined outcomes.
This does not require full capitation across entire populations. It requires focused accountability arrangements aligned to defined populations and specified outcomes:
For high-risk diabetes: bundled payment covering endocrinology, nephrology, ophthalmology, and care coordination with bonuses for A1C control, nephropathy prevention, and complication reduction. Financial risk is limited to a defined patient segment (not all diabetes patients, only those meeting complexity criteria). Duration sufficient to achieve outcomes (annual or multiyear arrangements, not quarterly). For frail elderly: per-patient-per-month capitated or risk-adjusted payment creating accountability for total cost of care. Bonuses for functional status preservation, delayed institutionalization, and outcome metrics achieved. Penalties for preventable ED visits and hospitalizations. For COPD: bundled or episodic payment for exacerbation management with 30/60/90-day readmission responsibility. Shared savings if exacerbation rates decline, revisit rates for stable patients decline, and oral steroid exposure decreases.
These arrangements are not full capitation across entire populations or health systems. They are focused on financial accountability for specific clinical segments with bounded populations and defined outcomes. They create incentives for integrated care without requiring complete vertical integration. They align dollars with outcomes.
Critical requirement: Financial arrangements must match population definition and outcome specification. If you defined outcomes as A1C control, retinopathy management, and nephropathy prevention, your bundled payment must reward these metrics, with less value placed on reducing ED visits (which may be orthogonal to diabetes management quality). Financial misalignment creates perverse incentives.
Contract design is central to the bounded risk principle. Open-ended or unlimited downside exposure represents overreach—organizations taking financial accountability beyond what their delivery competency can manage. Risk arrangements should incorporate bounded downside exposure (typically 5%–15% of total contract value in early years, escalating as operational maturity develops), multiyear contract terms (minimum 3-year commitments allowing investment recovery and outcome trajectory establishment), and graduated risk corridors protecting against catastrophic losses during implementation phases. The end-stage renal disease (ESRD) Care Model and Alternative Quality Contract both employed multiyear designs with escalating risk matched to organizational capability, contributing to their demonstrated sustainability.2,10 Short-term or open-ended risk exposure discourages the upfront investment in infrastructure, workforce, and care redesign that focused population health requires.
Centers for medicare & medicaid services (CMS’s) own evolution from the Comprehensive ESRD Care Model to the Oncology Care Model to current disease-specific total cost of care models reflects a regulatory trajectory consistent with this framework—moving from undifferentiated global capitation toward bounded, segment-specific risk arrangements where accountability is matched to organizational delivery competency and tailored quality measures replace generic process metrics.
Execution: The Practical Pathway
Secure funding sources: Identify payers (typically insurers) willing to fund population-specific outcomes. They benefit financially from improved management of high-risk segments. Select strategic populations: Choose clinical segments where you have existing capability or a clear development pathway. High-complexity, high-cost segments (frail elderly, complicated diabetes, severe COPD) offer the best ROI.
Workforce constraints—particularly in geriatrics, pulmonology, and cardiology—reinforce this framework’s core principle: stay within reach. Organizations should begin with one segment where existing clinical leadership and specialty capacity are strongest, not launch multiple segment programs simultaneously. Attempting to staff risk-bearing programs across multiple clinical segments without adequate specialty depth is itself a form of overreach—taking accountability for populations the organization cannot yet serve with the required competency. A phased approach builds operational infrastructure and outcome measurement capability through a single initial program, then expands to additional segments as workforce capacity and institutional competency permit. Virtual specialty care, advanced practice provider-led teams, and shared specialist models can extend limited capacity. Where the speciality workforce is genuinely insufficient, Tier 2 partnerships—subsidizing existing community specialists rather than recruiting scarce clinicians away from them—preserve the strategic partnership principle central to this framework.
Define population precisely: Specify age range, diagnostic criteria, acuity level, functional status, and comorbidity burden. Document clear inclusion/exclusion criteria. Establish measurable outcomes: Define what success means for this population, what’s realistically achievable, and how you’ll measure it. Build measurement infrastructure first. Map existing resources: Catalog community services, qualified providers, current referral success rates, and genuine capacity gaps. Require data, not assumptions. Execute 3-tier delivery approach sequentially: Tier 1: Optimize referrals to existing providers through navigation, warm handoffs, closed-loop feedback. Measure outcomes for 6–12 months. Tier 2: Subsidize existing providers’ capacity expansion with outcome-based contracts where qualified partners exist but lack capacity. Tier 3: Build in-house services only where genuine gaps persist and you have a competitive advantage. Align financial accountability: Negotiate contracts creating real risk for outcomes. Ensure payment and clinical responsibility match. Build outcome infrastructure: Deploy longitudinal data systems, outcome analytics, risk adjustment, and systematic capture. Data platforms integrating electronic health records, claims data, and patient-reported outcomes enable the longitudinal tracking essential to segment-based accountability. Virtual specialty care extends limited specialist capacity without requiring organizations to recruit scarce specialists away from community partners—consistent with the Tier 2 partnership approach. Remote monitoring—continuous glucose monitoring for diabetes, home spirometry for COPD, wearable activity tracking for frail elderly—generates real-time clinical data supporting proactive intervention within the organization’s core service competency. These technologies enable focused execution at scale rather than requiring organizational expansion into new service lines. Install clinical leadership: Physician and advanced practitioner leadership focused on outcomes accountability, not volume production.
Evidence from Focused Segment-Based Models
Published evidence demonstrates that segment-specific accountability within an organization’s competency outperforms expansive risk-taking. Medicare’s Comprehensive ESRD Care Model—the first specialty-oriented accountable care organization—achieved $126 per beneficiary per month reductions in Medicare payments, 5% decreases in hospitalizations, and 8% reductions in readmission likelihood, outcomes that primary care-based accountable care organization (ACOs) failed to achieve for the same population. 10 A nephrology-led organization focusing on its core service for a defined segment outperformed generalized organizations managing the same patients within broader risk arrangements.
The Oncology Care Model enrolled over 3200 oncologists across approximately 200 practices managing 6-month chemotherapy episodes. While Medicare results were modest ($297 per episode savings outweighed by incentive payments), commercial populations demonstrated $6287 reductions in episode spending.11,12 The mixed results illustrate the importance of bounded risk design—the model worked when financial exposure was proportionate to the participating organization’s delivery competency.
COPD bundled payment programs demonstrate that precise population definition prevents overreach. A single-center bundled payments for care improvement (BPCI) program showed differential impact by clinical complexity: the most complicated patients (DRG 190–191) achieved savings of $1,753 to $1,897 per episode, while less complicated patients (DRG 192) had additional costs of $4,184. 13 Organizations should focus bundled risk on the complex patients where their specialized capacity creates genuine value, not extend risk to populations manageable through standard care. A separate COPD care bundle reduced 30-day readmissions from 21.7% to 11.8% (P = 0.017) and 90-day readmissions from 19.6% to 4.7% (P < 0.001). 14
What This Approach Is
This framework is pragmatic population health: starting with specific populations, defining realistic success, and organizing delivery efficiently through structured referrals, targeted subsidies, and selective ownership. It recognizes that health systems overreach when they attempt to reverse decades of poverty, food engineering, and structural social determinants through clinical programming—and that this overreach diverts resources from the core clinical services where these organizations can actually improve outcomes. It rejects the persistent limitations of global capitation across incompatible populations. It succeeds because accountability is focused, outcomes are realistic, and delivery is organized around what actually works.
This approach explicitly succeeds where PACE works, where disease-specific programs work, and where integrated multispecialty groups with focused risk contracts work. It is not theoretical. It has evidence from organizations executing all 3 elements consistently.
What This Approach Is Not
This framework explicitly rejects full capitation across entire patient populations or health systems. Global capitation across diverse populations creates financial incentives for cost-shifting and undertreatment and is not aligned with the existing care delivery model. Focused population accountability is fundamentally different from system-wide capitation.
It rejects upstream social determinant interventions as the primary operational strategy for health systems, assuming clinical risk. Health systems should not take financial accountability for outcomes driven by social factors outside their delivery competency. These interventions have legitimate and demonstrated value at the policy and community level, but they represent overreach when health systems treat them as core operational strategy within risk-bearing arrangements.
It rejects weak value-based contracts that penalize process failures while preserving primary cost control elsewhere. Real accountability means that dollars and doctors share consequences.
It rejects defaulting to Tier 3 (direct service ownership) without rigorous Tiers 1 and 2 analysis. Most gaps can be addressed through referral optimization or capacity building.
It rejects unmeasurable outcomes and process proxies. Real outcomes track whether patients improved. Not whether services were accessed.
Conclusion
Population health works when it combines 3 executable elements: defining the specific population, establishing what success realistically looks like, and organizing delivery efficiently to achieve it. This is not new. PACE succeeds this way. Disease-specific programs in integrated systems succeed this way. Focused risk arrangements in multispecialty groups succeed this way.
What has failed is attempting universal improvement across incompatible populations, measuring compliance instead of outcomes, and building services without analyzing existing capacity.
The technical competency exists. The financial models exist. The clinical knowledge exists. What remains is committing to specific populations, defining achievable outcomes, and organizing delivery efficiently around those 3 elements—staying within reach rather than overreaching into services, risk, and interventions that health systems cannot control. That is how population health can actually work.
Footnotes
Author Disclosure Statement
No competing financial interests exist.
Funding Information
No funding was received for this article.