Preventive Care in the Modern Era

Every physician I have worked with can name a patient they lost between visits. Not lost to death. Lost to the gap. The medication that was adjusted but never confirmed. The lab that was ordered but never drawn. The blood pressure that drifted for three months because no one called.

The physician knew what that patient needed. The care plan said it plainly. But between the visit that created the plan and the visit that would have caught the drift, no one owned the space in between.

I have watched this pattern repeat across practices of every size. A nurse logs into the EHR at 7:45 AM with 340 attributed patients who have two or more chronic conditions. She knows at least 40 of them had something change last month. She also has prior authorizations, inbox messages, referrals, and a schedule full of same-day tasks the physicians need handled before lunch. By 10 AM, the chronic care work has not started. By Friday, it will not have started.

The patients who needed a call this week will not get one. The 20 minutes of documented non-face-to-face time required to bill 99490 will not happen, because no one had 20 minutes.

In Chapter 1, I described how the Annual Wellness Visit produces a prevention plan that sits in the chart producing nothing. The same structural failure applies here, only the stakes compound monthly. The AWV is one visit per year. Chronic Care Management is a monthly obligation to every enrolled patient. When no one owns the follow-through, the failure is not occasional. It is systematic.

Chronic Care Management is a billing code attached to an operations challenge. Most practices treat it as a clinical program. That framing is why adoption remains low and revenue stays uncaptured.

Where This Code Came From

CMS introduced separate payment for Chronic Care Management in the CY 2015 Medicare Physician Fee Schedule Final Rule, effective January 1, 2015. The foundational code, CPT 99490, created a distinct billable service for non-face-to-face care coordination provided to patients with multiple chronic conditions. Before 2015, this work happened in most primary care practices without reimbursement. Physicians and their staff coordinated care between visits, reconciled medications by phone, followed up on lab results, and managed transitions. None of it generated revenue.

CCM services fall under the Medicare Physician Fee Schedule, governed by 42 CFR Part 414. CMS has expanded the code family multiple times since launch. In 2017, complex CCM codes 99487 and 99489 were added for patients requiring at least 60 minutes of clinical staff time and higher-complexity medical decision-making. In 2019, CPT 99491 was introduced for 30 minutes of CCM provided directly by a physician or other qualified health professional. In 2020, CMS added the add-on code G2058, later replaced by CPT 99439 in 2021. In 2022, CMS revalued the entire CCM code family and added CPT 99437 as a physician-time add-on, acknowledging what it called a "longstanding concern about undervaluation of care management under the physician fee schedule."

CMS rarely admits undervaluation. The 2022 revaluation was a signal: the agency views chronic care management as infrastructure, not a bonus.

In 2025, CMS introduced Advanced Primary Care Management (APCM) under three new G-codes (G0556, G0557, G0558), bundling elements of CCM and other care management services into a tiered model without time-based thresholds. APCM cannot be billed in the same month as CCM. Chapter 2 of this series covers APCM in depth. The choice between CCM and APCM is operational, not clinical. The patient's needs are the same. The billing model differs.

Also in 2025, CMS eliminated the consolidated G0511 code used by Rural Health Clinics and Federally Qualified Health Centers, requiring these providers to bill individual CCM CPT codes like all other fee-for-service practices. After a transition period ending July 1, 2025, G0511 is no longer payable.

The CY 2026 PFS Final Rule, published November 2025, finalized an approximately 10% increase in CCM reimbursement. The base code 99490 rises from approximately $60.49 to $66.30 nationally. CMS continues investing in chronic care infrastructure.

What CCM Requires

The eligibility criteria have not changed since 2015. A patient must have two or more chronic conditions expected to last at least 12 months or until the patient's death. Those conditions must place the patient at significant risk of death, acute exacerbation or decompensation, or functional decline. A comprehensive care plan must be established, implemented, revised, or monitored.

CMS does not provide an exhaustive list of qualifying conditions. The Chronic Condition Warehouse includes 27 categories. Common qualifying conditions include diabetes, hypertension, heart failure, COPD, chronic kidney disease, depression, hyperlipidemia, and arthritis. Any condition meeting the duration and risk criteria can qualify.

The billing requirements vary by code.

Non-complex CCM (99490): At least 20 minutes of clinical staff time per calendar month, directed by a physician or qualified health professional. Services are non-face-to-face. Clinical staff can perform the work under general supervision. One add-on code (99439) covers each additional 20 minutes in the same month.

Physician-directed CCM (99491): At least 30 minutes provided personally by the physician, NP, or PA per calendar month. One add-on code (99437) covers each additional 30 minutes. Cannot be billed in the same month as 99490.

Complex CCM (99487): At least 60 minutes of clinical staff time per calendar month for patients whose conditions are at acute risk of exacerbation and require substantially more complex medical decision-making. One add-on code (99489) covers each additional 30 minutes. Cannot be billed in the same month as 99490 or 99491.

All CCM codes share the same operational requirements.

Patient consent, written or verbal, documented in the chart. The patient must be informed that only one practitioner can bill CCM per month and that cost-sharing applies.

A comprehensive care plan, established, monitored, and updated. The plan must address all health issues and be available to all providers in the care team.

A certified electronic health record with structured recording of demographics, problem lists, medications, and allergies.

24/7 access for patients to reach a health care professional.

Care management activities including transition management, specialist coordination, medication reconciliation, and follow-up on referrals and tests.

The 20-minute threshold is where most practices fail. Not because the work takes less than 20 minutes. Because no one tracks the time, documents the activities, or completes the work in a structured way that survives an audit.

The Revenue Model

Current 2025 national average reimbursement rates:

• 99490 (non-complex, 20 min clinical staff): approximately $60.49 per patient per month
• 99439 (each additional 20 min, add-on to 99490): approximately $45.93
• 99491 (30 min physician/NP/PA time): approximately $82.16
• 99437 (each additional 30 min, add-on to 99491): approximately $57.58
• 99487 (complex, 60 min clinical staff): approximately $131.65
• 99489 (each additional 30 min, add-on to 99487): approximately $70.52

These are national averages. Geographic Practice Cost Indices adjust rates by location.

Here is what the math looks like for a mid-size practice. A panel of 2,000 Medicare patients. Approximately two-thirds have two or more chronic conditions. That is roughly 1,300 potentially eligible patients. At a 20% enrollment and billing rate, 260 patients receive monthly CCM. At $60.49 per patient per month, that is approximately $15,727 per month or $188,729 per year in base CCM revenue alone.

If 30% of those patients qualify for an additional 20-minute increment (99439), add approximately $3,577 per month. If the practice stacks CCM with other care management services, such as RPM or BHI, downstream revenue compounds. CCM and RPM can be billed concurrently. CCM and BHI can be billed concurrently. The services must be distinct and separately documented.

These are revenue projections, not profit figures. The primary cost inputs are staff time, technology infrastructure, documentation overhead, and supervision. A 2017 CMS-commissioned study by Mathematica estimated that a practice would need approximately 131 CCM-enrolled patients to recoup the cost of a full-time RN, or 76 patients for an LPN, at 2015 reimbursement rates. Rates have increased since then. The staffing math has improved, but the operational challenge remains: someone has to do the work consistently, every month, for every enrolled patient.

Why Adoption Remains Low

CCM has existed for a decade. The revenue is real. The clinical logic is sound. Most eligible patients do not receive it.

A 2024 study in the Journal of the American Geriatrics Society (Jang et al.) examined Medicare claims from 2015 to 2019 using a 5% random sample of fee-for-service beneficiaries aged 65 and older. Among approximately 1.1 million beneficiaries with two or more chronic conditions, CCM utilization increased from 1.1% in 2015 to 3.4% in 2019. Nearly 75% of fee-for-service beneficiaries were potentially eligible. Fewer than 4% received CCM services.

More recent data shows acceleration. CMS claims data analyzed by Prevounce showed approximately 1.3 million patients received CCM services in 2023, with a 25% year-over-year increase in patients and 20% increase in providers. An Avalere Health analysis found an average annual growth rate of 7.4% from 2019 to 2023, with the largest single-year increase (23.4%) between 2022 and 2023, following CMS's revaluation of the code family.

Growth is real. The gap between eligible and enrolled remains enormous.

A 2017 qualitative study in the Journal of General Internal Medicine (O'Malley et al.) interviewed 71 respondents across billing and non-billing practices about their CCM experience. The barriers they identified were not clinical. Enrolling patients and obtaining consent was time-consuming. Time tracking was burdensome. Documentation requirements added friction. Practices with fewer resources struggled most. Many enrolled far fewer patients than their eligible panels could support and billed for only a few months per patient.

The Avalere analysis also found documentation gaps: not all CCM claims included the minimum two chronic condition diagnoses required for eligibility, suggesting that some practices lack the operational infrastructure for proper coding.

CCM adoption is heavily concentrated by geography. The Jang et al. study found the Southern region had the highest utilization growth, with Georgia at 6.1% and Texas, Florida, and Mississippi each at 5.8% in 2019. New Hampshire, Montana, North Dakota, Alaska, and Vermont had rates below 0.5%.

This variation likely reflects the concentration of outsourced CCM vendors rather than patient need. Rural practices face compounded challenges: smaller panels, fewer staff, less technology infrastructure, and higher patient complexity. The elimination of G0511 adds billing burden for RHCs and FQHCs that previously used a single code for multiple care management services.

These are workflow failures. The clinician knows who qualifies. The system does not reliably convert that knowledge into monthly, documented, billable care management.

The Evidence

The most comprehensive evaluation of CCM's impact is the CMS-commissioned Mathematica study examining the program's first 18 months (2015–2016). Compared to a matched control group, CCM recipients had significantly higher rates of advance care planning, more visits with their primary care physician, and greater use of home health care. The CCM group was hospitalized at lower rates, used emergency departments less frequently, and cost Medicare approximately $74 less per beneficiary per month. Excluding patients who received only one month of CCM, the savings rose to approximately $95 per month. The authors estimated net savings to CMS exceeding $38 million after accounting for CCM payment costs.

These findings carry important caveats. Practices that implemented CCM in 2015 were disproportionately motivated and well-resourced. Whether the same savings hold in practices without that level of infrastructure investment remains uncertain. CCM recipients tend to be more engaged than the general eligible population. Some portion of the observed cost reduction likely reflects selection effects rather than the intervention itself. Association, not causation.

A 2020 study in Preventing Chronic Disease (Crespo and Shrader) examined a community health worker-based CCM model in rural Appalachia among 137 patients with uncontrolled diabetes. Mean A1C decreased from 10.2% to 8.5% after 12 months. ED visits decreased by 22% and hospitalizations by 30% from year one to year two. Limited by small sample size, single-disease focus, and a rural context that may not generalize.

A 2020 study in Population Health Management (Kvedar et al.) reported outcomes across 26,500 patients using a standardized CCM platform. Two-thirds had six or more chronic conditions. The study demonstrated that a technology-enabled CCM model could operate at scale and identified predictors of hospitalization. Limited by the absence of a concurrent control group.

A 2025 study in the American Journal of Public Health evaluated an expanded CCM team model (adding a pharmacist and community health worker to the standard provider-nurse team) among 134 Medicare patients with uncontrolled type 2 diabetes or hypertension. The CCM cohort showed significant improvements: 41% improved A1C versus 12% of controls, and average systolic blood pressure decreased 17 points. Medication review identified at least one potential adverse drug event in 80% of patients. Limited by convenience sampling, small sample size, and four-month duration.

The aggregate evidence supports an association between CCM services and improved clinical process measures, increased primary care engagement, and reduced utilization in some settings. The evidence on cost reduction is concentrated in well-resourced, early-adopter practices. The strongest evidence supports CCM's role in closing care gaps, increasing advance care planning, and improving chronic disease control metrics. Practices should build their financial case on the revenue and quality improvement side, where the data is most robust.

What Ownership Looks Like When Technology Supports the Workflow

The operational failure in CCM is predictable. The eligible patient exists in the EHR. The care plan exists. No one initiates the monthly touchpoint. No one tracks the time. No one documents the activities. No one bills the code.

Here is what changes when the workflow has an owner.

Step 1: Eligibility Identification. The system ingests EMR data, problem lists, and claims history. It identifies every patient with two or more qualifying chronic conditions and generates a ranked eligible panel, updated continuously. Owner: automated system. Output: eligible patient list with current care plan status.

Step 2: Consent and Enrollment. An AI agent contacts eligible patients by phone to explain CCM services, obtain verbal consent, and document the consent in the EHR. The call follows a scripted protocol meeting CMS requirements, including disclosure of cost-sharing and single-provider billing. Owner: AI agent with clinical staff oversight. Output: documented consent in the chart.

Step 3: Monthly Care Management Contact. The AI agent conducts the monthly non-face-to-face touchpoint. It reviews medication adherence, screens for symptom changes, confirms upcoming appointments, and identifies care gaps. It documents the interaction in the EHR with time stamps. Owner: AI agent. Output: structured care management note with documented time meeting the 20-minute threshold.

Step 4: Care Plan Review and Escalation. The system flags changes requiring clinical review: medication discrepancies, new symptoms, missed labs, overdue screenings. The care plan is updated and queued for physician review. Owner: system generates; physician reviews and approves. Output: updated care plan with physician sign-off.

Step 5: Closed-Loop Follow-Through. Any issue requiring clinical intervention becomes a tracked task with a deadline and an owner. Referrals are sent. Labs are ordered. Follow-up calls are scheduled. Nothing sits as an intention. Every action item has a closed loop. Owner: system tracks; clinical staff and physician execute. Output: resolved action items with documentation.

Step 6: Billing and Documentation Verification. At month's end, the system verifies that time thresholds are met, documentation is complete, and consent is current. It generates the claim. Owner: automated system with billing staff review. Output: clean claim with correct CPT code and supporting documentation.

The physician stays in control of every clinical decision. The physician reviews the care plan. The physician approves escalations. The physician decides what interventions are appropriate. The system owns the execution that makes those decisions happen reliably, every month, for every enrolled patient.

Who This Model Misses

Every model has boundaries. Naming them protects the people who trust it.

CCM requires patient consent, and consent creates friction. Patients who are skeptical of phone calls from unfamiliar numbers do not consent. Patients who have been contacted by scammers posing as Medicare representatives do not consent. Patients who do not understand cost-sharing implications decline when they hear about a copay. The consent barrier disproportionately affects patients who are isolated, have low health literacy, or have reason to distrust the healthcare system.

The Jang et al. study found an unexpected racial pattern: CCM utilization was higher among Hispanic and Black beneficiaries than among White beneficiaries. This may reflect the concentration of CCM billing in Southern states with larger minority populations, or the disproportionate burden of chronic disease among Black and Hispanic Medicare beneficiaries. The finding does not mean access is equitable. Overall utilization remains below 4% of eligible beneficiaries across all groups. The structural barriers documented in Chapter 1 apply: transportation, language, trust deficits from historical discrimination, and the compounding effect of social determinants on chronic disease management.

CCM's primary delivery mechanism is the telephone. If the patient does not speak English or the language available from the care management team, the monthly touchpoint either does not happen or requires an interpreter, adding time and cost. AI phone agents must have validated multilingual capabilities or the model excludes non-English-speaking patients by design. Health literacy compounds the language barrier. Patients with cognitive impairment require caregiver involvement in every interaction.

The monthly care management contact depends on patient-reported information. Patients may underreport symptoms, overreport adherence, or decline to answer questions about depression, substance use, or social determinants. The care plan is only as complete as the data that feeds it. A patient who reports taking all medications as prescribed when she is not generates a care management note that looks clean but misses the problem. The system documents what the patient says. It does not independently verify what the patient does.

AI does not build trust. It does not resolve the patient's suspicion that this call is a scam. It cannot fix a medication access problem when the patient cannot afford the copay. It cannot make a specialist accept a referral when the referral network is broken. It cannot make a patient answer the phone. The model assumes a functioning care delivery system underneath. When the system itself is broken, adding operational efficiency to the care management layer does not solve the upstream problem.

Even with technology support, a physician must review care plans and approve clinical decisions. If the practice enrolls 500 patients in CCM, someone must review 500 care plan updates per month. The system reduces the time per review. It does not eliminate the review. The physician remains the bottleneck. Practices considering CCM must model staffing capacity before projecting revenue. The revenue exists. The question is whether the staff capacity exists to earn it.

The published literature on CCM outcomes is concentrated in early-adopter practices, large vendor platforms, and motivated provider networks. Whether CCM produces similar outcomes in practices without dedicated infrastructure is not established. The evidence on cost reduction is associational, not causal, and is shaped by selection effects. Practices should build their financial case on the revenue and care quality side, where the data is strongest.

What Changes This Week

First, pull your eligible panel. Run a report in your EHR for Medicare patients with two or more chronic conditions on their active problem list. Count them. That number is the ceiling for your CCM program.

Second, count your current CCM billing. Pull the number of 99490 claims your practice submitted in the last 12 months. Divide by your eligible panel. That percentage is your capture rate. If it is below 10%, the gap is operational, not clinical.

Third, audit 10 charts. Pull the last 10 patients for whom you billed CCM. Check whether consent is documented, the care plan is current, activities and time are documented, and the claim is supported. Count how many of the 10 would survive an audit.

Fourth, assign ownership. Name one person responsible for CCM operations this month. Not the physician. An RN, an MA, a care coordinator. One person whose job includes making the monthly calls, documenting the time, and flagging issues for physician review.

Fifth, measure monthly. Track three numbers: patients enrolled, patients contacted this month, and claims submitted. If the gap between enrolled and contacted is wide, the operational infrastructure is failing. The clinical knowledge is not the problem.

The Execution Layer Pear Health Builds

The model described in this chapter is what Pear Health operates.

Pear identifies every CCM-eligible patient on the practice's panel by ingesting EMR data, problem lists, and claims history. The AI phone agent calls patients to obtain consent, conducts monthly care management touchpoints, reconciles medications, screens for care gaps, and documents every interaction directly into the EHR with time stamps. When an issue requires clinical review, the system escalates with a structured alert. The physician reviews. The physician decides. The system closes the loop.

Every enrolled patient receives a monthly contact. Every care plan update has a documented owner. Every claim is generated from verified documentation. The 20-minute threshold is met not because someone remembered to track time on a spreadsheet, but because the system was built to track it from the first second of the call.

Pear does not replace the nurse at 7:45 AM. Pear takes the chronic care work off her list so the patients who need a call this month actually get one. The clinical judgment stays with the physician. The monthly execution becomes infrastructure instead of aspiration.

If your practice has hundreds of CCM-eligible patients and bills for fewer than 50, the gap is not knowledge. The gap is workflow. We should talk. Not about what CCM could look like. About what changes in your operations next week.