Reference
Caregiving by the numbers
Caregiving statistics get tossed around without sources, mixed across years, or quietly borrowed from a vendor's marketing. Here they are the honest way: each figure with its source, its year, and a confidence level — and anything vendor-commissioned is labeled, so you know what to trust. It’s the same data behind our Price Index and Tech Plan.
The scale: who needs care
The aging population driving all of it.
61+ million (~18% of US population)
Americans age 65+
Up from 57.8M (17.3%) in 2022.
78+ million (~22% of US population)
Projected Americans 65+ by 2040
The demographic driver behind rising caregiving demand.
~28% (community-dwelling)
Older adults 65+ living alone
59% live with a spouse/partner.
23%
Sandwich-generation adults
Have a parent 65+ AND are raising a child <18 or financially supporting an adult child (Oct 2021 survey).
54%
Sandwich generation — adults in their 40s
36% of those in their 50s, 27% of 30s are also sandwiched.
1 in 4 (~14 million); ~3 million ER visits/yr
Older adults who fall each year
Falls are the leading cause of injury death in 65+ (~43,000 preventable-fall deaths in 2024, National Safety Council). The fall → ER → rehab cascade is the demo's core journey and the case for PERS / fall-detection.
75% (stay in current home); 73% (stay in community)
Adults 50+ who want to age in place
n=3,090, summer 2024; preference is even higher among 65+. The demand-side anchor for Tugboat's 'help them stay home' thesis and the Open Mic 'choose how they live as they age' prompt.
~15% (range 11–23% by condition)
30-day hospital readmission rate, older adults
National all-cause average ~14.7%; elderly-Medicare reviews span 11–23% depending on condition. Many readmissions are preventable — the 'next emergency' the Care Graph is built to anticipate. Rate varies by source → cite as a range, medium confidence.
78%
Adults 65+ who own a smartphone
vs 90% of adults 50–64 and 97% of under-50. Same Pew data: 67% of 65+ now get news on a mobile device; ~44% own a tablet (2021). The strongest counter to 'older adults can't use the software' — supports Tugboat's affordable-help-is-software thesis over the robot narrative.
30% (2025), up from 9% in 2023
Adults 50+ who have used AI
FULL PROGRESSION: 9% (2023) → 18% (2024) → 30% (2025) — more than tripled in two years, so cite the trajectory, not just the endpoint. Employment splits it sharply: 42% of EMPLOYED adults 50+ have used generative AI vs 21% of those not employed (relevant to the working-sandwich-caregiver personas). 58% of AI users have used a branded platform (ChatGPT, Copilot). ⚠️ THE CAVEAT THAT MATTERS MOST TO US: trust, privacy and data security are the top reasons for hesitating, cited by about half — which is direct evidence for Tugboat's nothing-you-type-is-saved promise on the free tools, not a reason to hide the AI. Surfaced via Laurie Orlov's 'AI and Older Adults — What's Now and Next in 2026' (Jan 2026), the same analyst behind the shipped Take `orlov-ai-older-adults-now-next` (whose angle is hesitation-is-unfamiliarity, NOT these numbers — the two are complementary, not duplicative). Pairs with demo-65plus-smartphone as the second half of the 'older adults won't use it' rebuttal.
The caregivers — and the value of their work
The invisible workforce holding the system up.
63 million
US family caregivers
About 1 in 4 US adults. A 45% rise — nearly 20 million more caregivers — since 2015. Released 2025-07-24. Counts ALL family caregivers, including the ~4M caring for children. This is Tugboat's headline caregiver figure. The narrower 59M (pop-caregivers) is the nested subset caring for an adult 18+, and is the figure the >$1T valuation attaches to — the two are nested, not contradictory.
59 million
US family caregivers of adults 18+
Up from ~53M (2020). The NESTED SUBSET of the 63M headline (pop-caregivers-all): 63M counts all family caregivers including ~4M caring for children; 59M counts those caring for an adult 18+. Not contradictory — nested. Use 59M only where the >$1T valuation or adult-care specificity is the point; otherwise lead with 63M.
~20.1% (about 1 in 5)
US adults who are current caregivers
Government public-health surveillance (Caregiver Module, 2021–2022, 47 states + Puerto Rico). Independent, non-AARP corroboration of the '~1 in 5 adults' prevalence. NOT in conflict with the '1 in 4 adults' in the 63M headline (pop-caregivers-all): CDC measures prevalence of caring for an ADULT, while AARP/NAC counts ALL family caregiving including care for children. Different instruments, different denominators, same nesting.
Over $1 trillion/yr
Value of unpaid family caregiving
Surpasses 2024 private health spending ($967B) and Medicaid ($932B). Up from ~$600B in prior editions.
49.5 billion hours/yr
Total unpaid family-care hours
Equals work of 23.8M full-time workers.
23.8 million full-time workers (~17% of US full-time workforce)
Caregiving workforce equivalent
~3.3 million by 2040 (~2.9M even under optimistic assumptions)
Projected US long-term care worker shortfall
DERIVATION: ~3.4M ADDITIONAL long-term care workers needed by 2040, minus fewer than 100,000 expected to enter the field = ~3.3M shortfall; ~2.9M even assuming employment returns to pre-pandemic levels AND immigration exceeds current projections. ⭐ THE FRAMING THAT MATTERS: today's entire paid LTC workforce is only ~3 million, so the shortfall EXCEEDS the existing workforce — closing it means building a second, larger LTC workforce from scratch in 14 years. Primary source = Andrew R. Olenski, Ph.D. & David C. Grabowski, Ph.D., NEJM Perspective, published 2026-07-04, N Engl J Med 2026;395:107-110, DOI 10.1056/NEJMp2604122; the authors also note that in every area where policy could help, federal policy is moving the wrong way. ⚠️ PROVENANCE: NEJM is PAYWALLED ($55) and its public abstract says only 'millions' — the specific figures come from LTC News's read of the paper (James Kelly, updated 2026-07-05), which names both authors and links the DOI. Cite LTC News as the reader-accessible source per the paywall rule. One unresolved conflict: a July 2026 web summary attributed '2.5 million under optimistic scenarios' rather than 2.9M; LTC News is the more trustworthy reading (internally consistent arithmetic + direct DOI link), but confirm against the paper itself if full text is ever obtained. Pairs with the PHI projection (short ~1M home-care workers by the early 2030s) in content/atlas/united-states.mdx — two independent sources, two horizons, same direction.
$7,242/yr on average (~26% of income)
Family caregiver out-of-pocket spending
78% of caregivers face regular out-of-pocket costs; ~half had ≥1 major financial setback (took on debt, stopped saving, couldn't afford food). By condition: dementia caregivers ~$8,978/yr, mental-health ~$8,384/yr. Housing (rent/mortgage/assisted living/home mods) = >50% of caregiver OOP spend. The $7,242 / ~26%-of-income figure is restated by AARP/NAC 'Caregiving in the U.S. 2025'. Feeds the finance pillar.
What care costs
The price of hired human care — the benchmark everything else is measured against.
$35/hr (~$80,080/yr at 44 hrs/wk)
Non-medical home aide
Part-time ~20 hrs/wk ≈ $36,400/yr.
$95/day (~$24,700/yr)
Adult day health care
$6,200/mo (~$74,400/yr)
Assisted living
~$7,250/mo (~$87,000/yr)
Memory care
Derived, not a direct survey line — verify before public use.
$114,975/yr
Nursing home (semi-private room)
$129,575/yr
Nursing home (private room)
~$216,000–$324,000/yr
24/7 in-home care
Composite behind the 'families can't self-fund' contrast.
$257.0 billion (2023)
Medicaid long-term services & supports spending
Home- & community-based services ≈ 45% of LTSS (~$116B, FY2020), with the mix steadily shifting toward home care. The public-payer backdrop behind the Appeals tool + finance pillar.
The affordability gap
Why most families can't simply buy their way out.
~$56,680/yr
Median 65+ household income
~$200,000
Median 65+ retirement savings
Anchor: the median family cannot self-fund the human-care alternative.
What the affordable tech actually costs
The here-now stack — a fraction of the cost of human care.
~$28–$65/mo
Medical alert / PERS
Fall-detection add-on +$8–15/mo.
~$30–$60/mo + ~$100 upfront
Automatic pill dispenser (Hero)
NOT suitable beyond early-stage dementia (cannot verify ingestion).
~$45–$65/mo all-in
GPS locator (dementia/wandering)
$249 + ~$30–$40/mo
Companion device (ElliQ)
WA Medicaid began reimbursing it Mar 2026 — first US state.
$20,000 or $499/mo
Home humanoid (1X NEO)
Price is verified; CAPABILITY is marketing — relies on remote human teleoperators; not a validated eldercare product.
$3.9B (2026), projected ~$9.8B by 2033
Elder-care assistive robots market size
Syndicated 3rd-party research (NOT a vendor-commissioned promo survey) — treat as an Estimate. Corroborated within ~5% by independent Persistence Market Research ($9.4B/2033, same 14.2% CAGR); full methodology paywalled. For Tugboat's POV: reinforces the affordable-robot-isn't-a-robot thesis — even the robot market is small + facility-skewed.
What the tech can (and can't) do
The evidence, not the marketing.
~70–80% (controlled ~90–98%)
Fall-detection real-world accuracy
The 24/7 human monitoring relay is the actual safety mechanism, not the sensor.
Agitation ↓ (SMD −0.36), anxiety ↓; no effect on cognition/depression/QoL
Companion-robot clinical effect
Peer-reviewed; effects real but modest. Vendor '95% loneliness' claims are uncontrolled.
80.7% overturned — but only 11.5% of denials are appealed
Medicare Advantage denials overturned on appeal
MA insurers fully/partially denied 4.1M (7.7%) of ~53M prior-auth requests in 2024; of the small share appealed, 80.7% were overturned. The evidence base for the Appeal Drafter: most people never appeal — and when they do, they usually win.
Dementia, specifically
The condition that defines so much of the caregiving journey.
7.4 million
Americans 65+ living with Alzheimer's
$818 billion (2026)
Total US cost of dementia (USC cost model)
Peer-reviewed (US Cost of Dementia Project, Julie Zissimopoulos PI). USC's $818B = $222B medical/LTC + $237B unpaid family care + $23B forgone earnings + $320B lost quality of life. SUPERSEDES the older Alzheimer's Association Facts & Figures total of ~$409B (health + LTC only; alz.org/alzheimers-dementia/facts-figures) — the AA figure isn't wrong, just narrower scope. USC is canonical here; cite the AA ~$409B only when you specifically mean health-and-LTC-only.
$237 billion/yr
Value of unpaid dementia family care (USC model)
The 'invisible ledger' line — 5.2M caregivers, 6.8B hours, mostly in prime earning years. NOTE — the Alzheimer's Association Facts & Figures reports a higher unpaid-care figure (19+ billion hours, ~$446.3B, 2025; alz.org/alzheimers-dementia/facts-figures); the gap is methodology + scope (USC uses a narrower caregiver definition + its own wage valuation). USC is canonical here; both are authoritative — don't mix, cite each with its own source.
6.8 billion hours/yr
Unpaid dementia care hours (USC model)
From 5.2M unpaid caregivers; valued at $237B. The Alzheimer's Association Facts & Figures reports a higher 19+ billion-hour figure (2025; alz.org/alzheimers-dementia/facts-figures) under a broader scope/method. USC is canonical here — pair each hours figure with its own source.
5.2 million
Unpaid dementia caregivers (USC model)
USC cost-model count. The Alzheimer's Association Facts & Figures cites ~13 million dementia caregivers (family/friends providing any unpaid dementia care; alz.org/alzheimers-dementia/facts-figures) — a broader definition than USC's primary-caregiver cost model. USC is canonical here; cite the source that matches the claim.
Survey signals — read with caution
Useful directionally, but vendor-commissioned. Not gospel.
90% (20% severe)
Caregivers reporting burnout symptoms
VENDOR-COMMISSIONED (LogicMark sells safety devices) — directional only, never cite as gospel.
77%
Caregivers open to AI health monitoring
VENDOR-COMMISSIONED — same caveat. Same survey: 73% report financial strain.
How to read these
- Sources first. Government and peer-reviewed figures (Census/ACL, AARP, the Alzheimer’s Association, CareScout/Genworth, published RCTs) carry the most weight.
- “Vendor-commissioned” means the number comes from a company that sells a related product. Treat it as a directional signal, not a fact.
- “Estimate” means derived or composite — useful for scale, worth verifying before high-stakes use.
Figures are point-in-time and drift; confirm against the original source before relying on one. This page is information, not medical or financial advice. Spot an error or have a better source? Tell us — accuracy is the point.