Risk Stratification in Healthcare: Models, Levels & Examples

Risk stratification is the process of categorizing patients by their likelihood of experiencing adverse health events — hospitalization, emergency department visits, disease progression, or high-cost care — so that clinical teams can match care intensity to patient need. 

The American Academy of Family Physicians describes it as a two-step process that combines objective claims or EHR data with subjective clinical assessment to identify patients who need more focused care management.

In value-based care models, risk stratification determines who receives preventive outreach, who gets assigned a care manager, and who needs intensive interdisciplinary support.

Let’s look more into risk stratification, and explore:

• Common risk tiers and how they translate to care actions
• Common mistakes that weaken stratification programs
• How risk stratification differs from risk adjustment
• The two-step process for stratifying patients
• What data feeds a risk stratification model
• Real-world examples by patient type

What data goes into risk stratification?

Risk models pull from multiple sources — and the accuracy of the output depends entirely on the completeness and recency of the input.

Data type	Examples	What it adds
Demographics	Age, sex, geography	Baseline population-level risk context
Clinical history	Chronic conditions, comorbidities, ICD-10 diagnoses	Complexity and disease burden
Utilization	ED visits, admissions, readmissions	Signals instability or care gaps
Claims data	Diagnoses, procedures, pharmacy fills	Population-level patterns across payers
EHR data	Vitals, labs, medications, problem lists	Real-time clinical detail
Medication data	Polypharmacy, adherence gaps	Safety and adherence risk
Social drivers	Transportation, housing, food insecurity, health literacy	Barriers that clinical data alone misses
Care-team input	Functional status, family support, recent deterioration	Context that no algorithm captures

Johns Hopkins ACG describes risk grouping based on predictive cost, clinical, social, and behavioral factors.

The strongest stratification models combine objective data (claims, labs, utilization) with subjective context (clinician observations, social drivers, patient goals) because neither source is complete on its own.

In practice, the most common failure is relying exclusively on claims data. 

Claims lag behind clinical reality by weeks or months. A patient whose diabetes is newly uncontrolled shows up in EHR data immediately but may not appear in claims until the next billing cycle — by which point the window for early intervention has narrowed.

What are the common risk stratification levels?

Risk categories vary across organizations, but most healthcare stratification programs use some version of a tiered model — typically three to six levels.

Tier	Meaning	Typical care approach
Low risk	Generally healthy or stable chronic conditions	Preventive care, routine screenings, annual wellness visits
Rising/moderate risk	Early warning signs — uncontrolled condition, missed visits, increasing utilization	Chronic disease coaching, care-gap closure, medication adherence support
High risk	Multiple comorbidities, frequent utilization, significant care needs	Assigned care manager, longer visits, medication reconciliation, post-discharge follow-up
Complex/very high risk	Serious instability, repeated admissions, end-of-life needs, severe social barriers	Intensive interdisciplinary support, home-based services, advance care planning

AAFP uses a more granular six-level model. Aledade segments patients into low, rising, high, and catastrophic risk groups. NACHC describes grouping patients into high, medium or rising, and low-risk groups as a common starting framework.

The exact number of tiers is less important than whether each tier triggers a specific, documented care action. A risk score without a matching intervention is just a dashboard metric.

How does the two-step stratification process work?

AAFP describes step one as sorting patients by objective data from claims or the EHR, then adjusting based on subjective considerations from the care team.

Step 1 — objective scoring

Pull historical data to assign preliminary risk tiers. Inputs include:

• HCC risk scores
• Prior authorization history
• Lab values outside normal range
• Medication burden and adherence gaps
• ICD-10 diagnoses and comorbidity counts
• Claims-based utilization (ED visits, hospitalizations, readmissions)

Step 2 — clinical and social adjustment

The care team reviews the objective tier and adjusts based on information algorithms cannot capture:

• Did a recent life event change the patient’s care trajectory?
• Is the patient engaged and improving, or disengaged and deteriorating?
• Are social barriers (transportation, food insecurity, housing) increasing risk?
• Is the patient functionally declining despite stable claims data?

Step 3 — assignment and reassessment

The adjusted risk level gets documented in the EHR or care registry and linked to a care model. Risk levels should update after hospitalizations, new diagnoses, ED visits, social changes, or significant improvement.

AHRQ emphasizes that practices should identify which patients are likely to benefit from more intensive care management rather than applying the same coordination level to everyone. 

The purpose of stratification is not labeling — it is matching resources to need.

What does risk stratification look like in practice?

Concrete examples show how stratification translates from data to care decisions.

Patient profile	Objective data	Contextual data	Likely tier	Care response
Healthy adult, overdue screenings	No chronic disease, no recent utilization	Good access, stable support	Low risk	Preventive reminders, annual wellness
Type 2 diabetes, rising A1C	A1C above goal, 2 missed visits in 6 months	Transportation barriers	Rising risk	Outreach call, diabetes education, care-gap closure
CHF with recent ED visit	2 admissions in 90 days, polypharmacy	Lives alone, limited food access	High risk	Assigned care manager, pharmacist review, weekly follow-up
Frail older adult, multiple conditions	Recent fall, 3 specialists, declining function	Limited family support, cognitive changes	Complex	Interdisciplinary care plan, home support, advance care planning

The rising-risk tier is where many organizations have the largest return on investment — because early intervention can prevent the costly hospitalizations and ED visits that high-risk patients already experience. 

Many programs over-focus on high-risk patients and under-invest in the rising-risk group, where the care trajectory is still modifiable.

How does risk stratification support value-based care?

In fee-for-service, every visit generates revenue regardless of outcomes. In value-based care (VBC), organizations share financial accountability for the health of a population — which makes identifying who needs what level of care a financial imperative, not just a clinical one.

Risk stratification supports VBC by:

• Matching outreach intensity to patient complexity
• Supporting efficient use of care management staff
• Prioritizing care-gap closure for quality measure performance
• Providing data for shared savings calculations and performance reporting
• Identifying patients who would benefit from care management before avoidable hospitalization

Aledade frames risk stratification as a way to segment patients by medical complexity and allocate resources to reduce avoidable costs while improving outcomes. 

ACOs, Medicare Shared Savings Program participants, and value-based primary care organizations use stratification as the operational backbone of population health management.

How is risk stratification different from risk adjustment?

The two terms get conflated constantly in value-based care conversations, but they serve different purposes.

Dimension	Risk stratification	Risk adjustment
Primary purpose	Guide care, outreach, and resource allocation	Adjust payment or performance comparisons for patient complexity
Primary users	Care teams, population health managers, practice leaders	Payers, CMS, ACOs, health plans
Main inputs	EHR, claims, labs, utilization, SDOH, clinical judgment	Diagnoses, demographics, model-specific factors
Main output	Care tier or risk category	Payment or performance risk score (RAF)
Main risk	Under-identifying patients who need help	Overcoding or unsupported diagnosis documentation

CMS-HCC risk adjustment estimates expected costs for Medicare Advantage enrollees and assigns risk scores that affect plan payment. 

Clinical risk stratification categorizes patients by care needs to guide operational decisions. A patient can have a high HCC risk score (expensive diagnosis mix) but low clinical risk (stable, well-managed conditions) or the reverse.

The danger of confusing the two — treating HCC coding as a clinical risk exercise can lead to documentation patterns that inflate payment risk scores without improving care delivery.

What mistakes weaken risk stratification programs?

Here are a list of mistakes to weaken risk stratification programs:

Mistake	Why it matters	Better approach
Relying only on claims data	Claims lag behind clinical changes by weeks or months	Combine claims, EHR, and care-team input
Ignoring social drivers	Clinical risk may underestimate total risk	Screen for transportation, housing, food, support, literacy
Treating risk as static	Patient conditions change constantly	Reassess after hospitalizations, new diagnoses, and social changes
Focusing only on high-risk patients	Rising-risk patients may be more impactable	Build interventions for every tier, not just the top
Scoring without workflow	Risk scores do not improve outcomes by themselves	Link each tier to documented care actions and team roles
Overcoding or unsupported diagnoses	Creates compliance and payment risk	Document accurately based on clinical evidence
Not measuring outcomes	Cannot tell if the program works	Track ED visits, hospitalizations, care gaps, patient experience, and cost

A 2025 systematic review notes that population risk stratification tools are used to tailor interventions, prioritize resources, and proactively manage high-risk people with chronic disease in primary care — but the review also found wide variation in how well those tools actually predict outcomes across populations.

The strongest programs treat stratification as an ongoing operational cycle — score, adjust, act, measure, and re-score — not a one-time data exercise.

How do you implement risk stratification in a practice?

NACHC summarizes an approach that starts with compiling a patient list, sorting by condition, stratifying into target groups, and designing care models for each risk group.

A practical implementation checklist:

For most primary care organizations, the first implementation cycle reveals as much about data quality and workflow gaps as it does about patient risk. 

Missing problem lists, outdated medication records, and incomplete social screening all surface during stratification — and fixing those issues improves the accuracy of future risk scoring.

You stratify patients by clinical risk. MedHeave stratifies claims by revenue risk.

Identifying high-risk patients is a clinical priority. Identifying high-risk claims — the ones most likely to be denied, downcoded, or stalled in AR — is a revenue priority. Most practices do one well but not both.

• MedHeave monitors every claim from submission through final payment
• Denials are caught through ERA monitoring and reworked the same day
• Dedicated account managers report weekly on AR aging, denial trends, and collection rates
• High-risk claim patterns (prior auth failures, modifier issues, payer-specific quirks) get flagged before submission

Talk to MedHeave about building revenue cycle operations that catch payment risk before it becomes lost revenue.

Frequently asked questions

Here are some frequently asked questions on this topic: