How We Calculate Senior Care Costs

1. Base data: the 2020 Genworth state table

Genworth (now CareScout, a Genworth company) has run an annual Cost of Care Survey since 2004. The survey covering 2020 (fieldwork June–August 2020, published December 2020) is the most recent year for which Genworth published a complete, freely available, state-by-state table: all 50 states plus DC, across six care categories. Starting with later survey years, state and metro-area results moved behind an interactive ZIP/city/state lookup tool with no downloadable full table. Only national topline figures remain freely published for those years.

Official announcement: Genworth 17th Annual Cost of Care Survey press release. The full state table itself is no longer hosted as a static file by Genworth/CareScout; we cite an archival mirror of the original published PDF, cross-checked line-for-line against that press release's national figures.

2. Bringing 2020 dollars forward

To avoid quietly showing you six-year-old prices, we escalate the 2020 state figures using Genworth/CareScout's own published national median costs for their 2024 survey (fieldwork July–December 2024). For each care category, we compute one multiplier (2024 national median ÷ 2020 national median) and apply it uniformly to every state's 2020 figure for that category.

This is our modeling step, not a Genworth-reported number, and it carries a real assumption: that each state's price growth roughly tracks the national average. It won't be exactly right everywhere. That's a limitation we're disclosing, not hiding. The rows we publish are labeled period 2024 to reflect the anchor year of this escalation, not today's date.

3. The 2025 survey: a real state table again

CareScout's survey covering 2025 (fieldwork July–November 2025, published March 2026) resumed publishing a complete, free, state-by-state table: all 50 states, across five care categories, with monthly, daily, and annual figures. Unlike the escalated period-2024 rows above, these are CareScout's own published per-state medians, not a modeled estimate. We publish them as-is, labeled period 2025, alongside the period-2024 rows rather than replacing them, so the site now shows a real two-year trend.

One documented wrinkle: CareScout's 2025 methodology states plainly, "due to price convergence, CareScout has streamlined its reporting; homemaker and home health aide services are now collectively classified as ‘non-medical caregiver.’" Since our in-home-care and home-health-care pages were themselves sourced from Genworth's separate homemaker / home health aide lines through 2024, we apply that single merged 2025 figure to both: the publisher's own disclosed category merge, not a fabricated duplicate. The 2025 survey also introduces a new "Private Duty Nurse" (skilled LPN/RN home visit) category; we haven't mapped it to a page yet because no existing care-type page fits it cleanly (see plan/NEEDS-HUMAN.md).

District of Columbia completed the 2025 survey (named in the methodology narrative, with its own Washington-Arlington-Alexandria region row) but has no row in the published state table. Rather than guess a DC figure, DC keeps its period-2024 estimate; every other jurisdiction has both years.

Nationally, cost growth slowed markedly in the 2025 survey: most categories rose 1–5% year over year (versus 7–10% in the 2024 release), and adult day health care was the lone decliner, falling about 5% to a $24,700 annual national median. Official announcement: CareScout Releases 2025 Cost of Care Survey Results.

4. Metro city estimates use BLS OEWS MSA wages

For Tier 1 and Tier 2 Census places that resolve to a metropolitan statistical area with published OEWS wages, we apply a wage factor from BLS OEWS May 2025 wages for Home Health and Personal Care Aides (SOC 31-1120), covering every MSA in the free BLS metro download that reports this occupation:

city median = state median × (MSA median wage ÷ state metro-weighted median wage)

Both the MSA wage and the state denominator come from the same free BLS metro download (public domain). We do not relabel a state figure as a city estimate. Place→CBSA matching uses committed census-place maps, Census list1 county→CBSA delineations, and OEWS principal-city titles. Places without a cited join are left without a city row rather than invented.

5. Ranges (p25 / p75) from real distributions

When BLS publishes a non-degenerate wage distribution (25th percentile strictly below the median, 75th strictly above), we scale that shape onto the care-cost median to publish a p25–p75 range. If a side of the wage distribution is missing or equal to the median (common where wage floors compress the lower half), that side stays null. We never invent a spread, and the site suppresses degenerate "$X–$X" copy.

Nursing homes additionally use Genworth's private-room median as an upper band when it is strictly above the semi-private median. Two genuine cited room-type rates, not a fabricated percentile split.

6. Memory care (modeled premium band)

Genworth/CareScout do not publish a free, complete state-by-state memory-care table. Rather than invent local medians, we model memory care from the same local assisted-living estimate plus a documented industry premium band of 20–30% (higher staffing ratios, secured units, dementia programming). For each geography with an assisted-living figure:

memory p25 = AL median × 1.20 · median = AL × 1.25 · p75 = AL × 1.30

That band is the method, not a surveyed percentile. Pages label these figures as modeled. The premium range is consistent with industry underwriting summaries of NIC-linked assisted living vs memory-care asking rates (about 20–30% higher for memory care). See MMCG Invest underwriting case study (NIC-linked spreads). If a paid or free complete state memory-care table becomes available, we will replace this model and bump the methodology version.