Nearly 12 million Americans provide unpaid care for someone living with Alzheimer's disease or another dementia, contributing an estimated 19 billion hours of care in 2024 alone, valued at $413.4 billion. Family caregivers bear an estimated 70% of the total lifetime cost of dementia care, which runs to roughly $405,262 per person in 2024 dollars. Yet almost none of this activity is captured in the structured clinical or claims data that health systems use to plan care, allocate resources, or measure outcomes. Caregiving is, in effect, the largest single component of the dementia care system that most health data infrastructure cannot see.

A Workforce Operating Largely Outside the Record

The scale of this invisible workforce is difficult to overstate. Caregivers report significant personal cost: 59% rate the emotional stress of caregiving as high or very high, and 38% report high physical stress; among employed caregivers, 57% have had to arrive late, leave early, or take time off, and 16% have taken a formal leave of absence. Peer-reviewed economic modeling projects that the replacement cost of unpaid family caregiving, currently $96 billion to $182 billion annually, of which 44% is attributable to dementia care, will grow to $277 billion to $571 billion by 2060, with dementia's share rising to 53%.

That burden is not distributed evenly. Black dementia caregivers were found to be 69% less likely than White caregivers to use respite care in 2015 data, though the gap has since narrowed as overall respite utilization remains low across both groups. A 2025 analysis published in JAMA Network Open projects that the economic burden of Alzheimer's and related dementias borne by Black and Latino families, driven substantially by unpaid caregiving, which accounted for about 55% of their 2020 burden, will grow from $113 billion in 2020 to $1.7 trillion by 2060, eventually surpassing the burden projected for White families. An AARP analysis further found that out-of-pocket caregiving costs consume up to 44% of annual income for Hispanic and Latino caregivers, compared with 34% for Black caregivers and 14% for White caregivers, disparities that reflect structural differences in access to paid support rather than differences in caregiving need.

What Structured Investment Looks Like

The largest single federal investment aimed at closing this gap illustrates both the scale of need and the limits of any one program. In 2024, the Administration for Community Living awarded the Alzheimer's Association a five-year, $25 million grant, the largest in the Association's history, to establish the Center for Dementia Respite Innovation (CDRI), administered jointly with the University of Minnesota School of Public Health and USAging. In its second year, CDRI awarded more than $4 million to 41 community-based organizations across 26 states, selected from more than 200 applicants, to fund new and pilot dementia-specific respite models. Notably, CDRI is explicitly designed to generate data that informs public policy, not simply to fund direct services, the program's own materials describe a mandate to "collect data and study the outcomes of these innovative respite projects". Even so, $4 million spread across 41 organizations represents a fraction of the national respite-care need implied by 12 million unpaid caregivers.

The professional care workforce faces a parallel, compounding shortage. PHI, the paraprofessional healthcare research organization, projects 8.9 million total direct-care job openings nationally between 2022 and 2032, driven by both new job growth and high turnover in a physically and emotionally demanding, historically low-wage sector. Specialist capacity is thinner still: federal workforce projections point to a shortage of roughly 27,000 geriatricians by 2025, and roughly 30% of geriatric medicine fellowship positions went unfilled in the 2022-23 cycle even as the American Geriatrics Society estimates 30,000 geriatricians will be needed by 2030.

Building the Data Standards to Close the Gap

The technical building blocks for integrating caregiver and social-care data into clinical infrastructure are beginning to take shape, even if adoption remains early-stage. The clearest current example is the CMS Innovation Center's GUIDE Model, which requires participating health systems to formally assess and report a patient's dementia stage, caregiver status, and level of caregiver burden as part of a national payment model, with caregiver burden tracked as one of three non-claims-based quality metrics reported annually, and a dedicated payment for respite services built into the model's core structure. This represents a concrete instance of caregiver data entering claims and clinical infrastructure through a federal payment mechanism rather than remaining a purely qualitative or anecdotal concern.

At the standards layer, the Gravity Project, an HL7 FHIR Accelerator that began in 2019 with Robert Wood Johnson Foundation funding, has built consensus terminology for social-determinants-of-health data across 17 domains, published as the HL7 SDOH Clinical Care Implementation Guide and now referenced in CMS Medicaid policy, the ONC's HTI-1 Final Rule, and the US Core Data for Interoperability. Complementing this, the PACIO Project's FHIR implementation guide for functional status explicitly includes data fields for "supporting caregivers" alongside patient functional assessments, an early but concrete technical path for representing caregiver information in interoperable records. The Zarit Burden Interview, a validated instrument for measuring caregiver strain with short-form versions suitable for routine screening, offers a ready-made clinical tool that could be standardized within these emerging data structures. It is worth noting plainly that no major commercial EHR vendor has yet announced native, standardized caregiver-burden screening in broad production use — this remains an emerging area of infrastructure, built on the GUIDE model and Gravity Project foundations, rather than a mature, widely deployed practice.

A persistent structural gap sits underneath all of this: Medicaid home- and community-based services providers have historically been excluded as "eligible professionals" under federal EHR incentive programs, meaning many HCBS organizations never received incentive payments to acquire health IT systems in the first place. Data quality analyses of Medicaid's Transformed Medicaid Statistical Information System have found considerable missingness and cross-state inconsistency in HCBS-related data elements, though CMS's Ensuring Access to Medicaid Services final rule will require states to report against a standardized HCBS Quality Measure Set beginning in 2028, a multi-year runway toward the kind of standardized, cross-state HCBS data that could eventually be linked to clinical records.

Structural Incentives Worth Tracking

Two federal tax mechanisms illustrate how policy design intersects with caregiver data and workforce participation, independent of any partisan framing. The Section 45S federal tax credit, available to employers who voluntarily provide paid family and medical leave, was made permanent in 2025 and now allows eligibility after six months of employment rather than one year. Separately, the Credit for Caring Act, reintroduced in the current Congress, proposes a tax credit of up to $5,000 annually per family caregiver for qualified care-related expenses. Both represent structural levers that, if paired with better caregiver identification in clinical and claims data, could be more precisely targeted toward the caregivers who need them most.

Treating Caregiving as Infrastructure, Not Anecdote

The throughline across this evidence is that caregiving is currently treated as a social fact that health systems acknowledge but rarely measure systematically. The GUIDE Model, the Gravity Project's SDOH standards, and the PACIO Project's caregiver data fields each represent a piece of the technical foundation needed to change that — but none, individually, constitutes a complete solution, and none has yet achieved broad adoption. For health data leaders, the opportunity is to treat caregiver status, caregiver burden, and respite utilization as structured data elements worth capturing with the same rigor applied to any other clinical risk factor — not because caregiving is a clinical diagnosis, but because it is one of the largest determinants of whether a care plan for a person with dementia actually succeeds.