Methodology
How the score is calculated. Four inputs, publicly sourced, updated quarterly. Every number on this page is verifiable by anyone with a browser.
What the score measures
The US Health Clock produces a single number representing the net health output of the United States healthcare system. Not spending. Not access. Not life expectancy. The aggregate of what the system produces relative to what it costs and what it hides.
A system that spent average money and produced average outcomes would score 0. Zero is not good. Zero is parity: you get what you pay for. The US scores below zero because it spends far more than average and produces far less, and because a documented share of the harm it causes is not counted in official statistics.
What the score does not do: it does not indict individual doctors, nurses, dentists, or patients. It measures the aggregate design. Individual actors can be excellent inside a system whose aggregate score is negative. The score indicts the design, not the people inside it.
Input 1 Parity baseline
A healthcare system that spends the OECD average per capita and produces median-ranked outcomes scores 0. This is the honest neutral. It is not "fine." It is the baseline from which adjustments are made.
The OECD average per-capita healthcare expenditure is approximately $5,500 USD (2023 data, PPP-adjusted). The median outcome rank across OECD nations, measured by life expectancy, infant mortality, and preventable death rate, is approximately 15th.
| Data point | Value | Source |
|---|---|---|
| OECD avg per-capita spend | ~$5,500 (2023, PPP) | OECD Health Statistics 2025 |
| OECD median outcome rank | ~15th | OECD Health at a Glance 2025 |
Input 2 Spend-vs-outcome adjustment
The United States spends approximately $13,500 per capita on healthcare, roughly 2.5 times the OECD average. It ranks approximately 39th in health outcomes among comparable nations.
This is the core disparity. A system spending 2.5 times more should, at minimum, produce above-average outcomes. Instead it produces outcomes ranked in the bottom third. The adjustment quantifies the gap between what was paid for and what was delivered.
spend_ratio = US_per_capita / OECD_avg_per_capita
rank_ratio = US_outcome_rank / OECD_median_rank
adjustment = -(spend_ratio * rank_ratio) * weight
spend_ratio = 13,500 / 5,500 = 2.45. rank_ratio = 39 / 15 = 2.60. Combined disparity = 6.38. Weighted contribution to final score: approximately -18 to -22 points.
| Data point | Value | Source |
|---|---|---|
| US per-capita spend | ~$13,500 (2023, PPP) | CMS National Health Expenditure Data |
| US outcome rank | ~39th | Commonwealth Fund Mirror, Mirror 2024 |
Input 3 Uncounted-harm correction
Medical error is the third leading cause of death in the United States, responsible for an estimated 251,000 deaths per year. It has no ICD code. It does not appear in CDC mortality tables. It is not counted in the statistics that determine the system's reported performance.
This is not a disputed estimate. It was published in the BMJ in 2016 by Makary and Daniel at Johns Hopkins, using extrapolation from four large studies of US hospital admission data. The CDC acknowledged the methodological gap but has not added an ICD code for medical error.
When the third leading cause of death is invisible in the data, every outcome metric built on that data is understating harm. The uncounted-harm correction adds these deaths back into the calculation.
Dental-deletion adjustment
A documented share of the top-1 cause of death (cardiovascular disease) and top-2 cause of death (cancer) has upstream oral-infection contribution. The system's billing-silo separation between medicine and dentistry means this contribution is structurally uncounted. Peer-reviewed evidence for oral-systemic pathways includes:
- Periodontal disease and cardiovascular risk: AHA scientific statement 2012, updated review 2024. Association established; Mendelian randomization studies in progress.
- Oral bacteria in atherosclerotic plaque: Multiple studies identifying P. gingivalis and F. nucleatum in cardiac tissue.
- Oral infection and systemic inflammation: Chronic periodontal infection elevates CRP, IL-6, and TNF-alpha, all independent cardiovascular risk markers.
The dental-deletion adjustment applies a conservative multiplier reflecting the share of cardiovascular and cancer mortality with documented upstream oral contribution. The exact weight is disclosed in the data appendix.
| Data point | Value | Source |
|---|---|---|
| Medical error deaths/yr | ~251,000 | Makary & Daniel, BMJ 2016 |
| ICD code for medical error | None | WHO ICD-10/ICD-11 code sets |
| Oral-systemic association | Established | AHA Scientific Statement 2012; Lockhart et al. |
Input 4 Denial-and-access correction
Insurance payers collect premiums in exchange for coverage. When claims are denied at rates that exceed actuarial norms, the payer is collecting for non-delivery. This is a measurable subtraction from the system's output.
Marketplace denial rates range from 2% to over 80% depending on the plan and state. Plans at the high end of this range are statistically indistinguishable from fraud: collecting premiums while denying the majority of claims. These rates are not estimates. They are published by KFF from CMS transparency data, and confirmed by state attorney general settlements.
| Data point | Value | Source |
|---|---|---|
| Marketplace denial rates | 2% to 80%+ | KFF Marketplace Denial Rate Data, 2023 |
| AG settlement totals | Varies by state | State attorney general settlement records |
Final calculation
The four inputs are combined with transparent weights:
score = baseline(0)
- spend_vs_outcome_adjustment
- uncounted_harm_correction
- denial_access_correction
current score = -43
Range under reasonable weight variation: -30 to -60. The exact digit depends on how the inputs are weighted. Under no weighting does the score reach zero.
Under no weighting does the score reach zero. Even removing the dental-deletion adjustment and using the most conservative medical error estimate (98,000/yr from the 1999 IOM report instead of 251,000/yr from the 2016 BMJ study), the spend-vs-outcome disparity alone pushes the score negative. The system costs 2.5 times more and delivers less. The other corrections make it worse, not negative in the first place.
The axis explained
The homepage displays a vertical axis from -99 to 300. These are not arbitrary bounds.
- -99 is the theoretical floor: a system producing maximum harm with zero benefit. No real system reaches this.
- 0 is parity: spend average, get average. The system's own definition of success, "absence of sickness," lives here.
- 100 is the system ceiling: the maximum the paid, synthetic, maintained healthcare system can produce. The best outcomes money can buy. It is still a ceiling.
- 300 is full human capacity. Physical, immune, intellectual, emotional, and perceptual capacity operating at the level observable in documented cases. Not lifespan. Capacity. This is the space the system cannot take you to, but it is measurable, and the interventions that move you there are free.
About Number Needed to Treat (NNT)
NNT is a standard clinical metric. It answers: how many people must receive this intervention for one person to benefit?
NNT = 1 means every person treated benefits. NNT = 100 means 99 out of 100 people receive the intervention with no benefit, while being exposed to its risks and costs. NNT = ∞ means the trial showed no benefit on the measured endpoint.
The homepage compares NNT values across interventions the system sells (high NNT, high revenue) and interventions the system does not sell (low NNT, low or no revenue). Every NNT on the homepage is sourced to a Cochrane review, a peer-reviewed trial, or a primary government document. The sources are listed on each row.
View the NNT comparison on the homepage
Update schedule
Inputs are reviewed quarterly. The OECD data, CMS expenditure data, KFF denial rates, and mortality estimates are checked against their most recent public releases. If any input changes by more than 5%, the score is recalculated and the homepage is updated.
Last reviewed: April 2026.