Is Obesity Really the No. 1 Driver?

Verdict: no published source supports 70% specifically. The figure appears to be a softened version of the widely repeated 90%.

Total U.S. national health expenditure is currently over $5.3 trillion per year (CMS National Health Expenditure data). 70% of that would be roughly $3.7 trillion.

No peer-reviewed analysis has produced that number for chronic disease treatment. The closest modeled estimate, a projection averaging about $2.2 trillion a year in medical costs from 2024 to 2039, works out to roughly 42% of today's spending and nearer 25 to 30% of projected spending over that window.

Verdict: this is the real error, and it is not a small one. "Caused by" and "spent on people who have" are different claims.

The RAND finding is that 60% of Americans have at least one chronic condition, and that group accounts for 90% of spending, including their broken bones, their eyeglasses, their childbirths, and their antibiotics.

PolitiFact rated the "caused by / goes to treating" phrasing False in February 2026 when a different federal official used it. RAND's lead author says it misrepresents the report.

Why it matters: the inflated version implies that treating chronic disease would free up 70–90% of health spending. The honest version implies something more modest and more interesting: most health care dollars flow through people who are chronically ill, which is an argument about who to care for, not what to cut.

Verdict: true on one yardstick, false on another, and unprovable in principle on a third. This is the strongest part of the statement, and it is still overstated.

Where it holds. Bolnick and colleagues (Lancet Public Health, 2020) applied a single consistent attribution method to 84 modifiable risk factors across $2.7 trillion of 2016 personal health spending. High BMI ranked first at $238.5 billion, ahead of high blood pressure, high blood sugar, dietary risks, and tobacco. Overall, 27% of that spending, over $730 billion, traced to modifiable risk factors collectively. Among Americans 65 and older the attributable fraction was higher still at 45.5%, and 86.7% of all attributable spending occurred in people 45 and older. That is exactly why Medicare is the payer making this argument.

Where it fails. Switch from dollars to deaths and obesity is not first. The Global Burden of Disease analysis of the United States ranks high BMI fifth among attributable deaths, behind high systolic blood pressure, high fasting plasma glucose, tobacco, and dietary risks. Switch again to total years of healthy life lost and high BMI moves to first; age-adjust that same measure and it drops to second, behind drug use. One dataset, one year, one country, four different answers depending on which column you read.

Why no yardstick can settle it. Risk factors are not independent of one another. Obesity drives disease partly by raising blood pressure and blood sugar, so the burden attributed to obesity and the burden attributed to those two overlap on purpose. Bolnick's team states it plainly: attributable spending estimates for individual risk factors "cannot be aggregated." Their own numbers demonstrate it. High BMI ($238.5B) plus high blood pressure ($179.9B) plus high blood sugar ($171.9B) sums to $590.3B, yet the study's joint estimate for those three plus three additional metabolic risks is only $508.0 billion. Naive addition overshoots by more than $80 billion across a smaller set of risks. A ranking built on overlapping quantities cannot support a definitive "number one."

The caveat that gives a hostile reader ammunition: if instead you line up CDC's individually-sourced fact-sheet figures, obesity (~$173B) sits behind physical inactivity ($192B) and smoking (>$240B). Those are different methods and different dollar-years, and CDC never presents them as a ranking, but they are published on a federal website. (CDC also lists alcohol at $249B; that one isn't health care spending at all, it's total economic cost in 2010 dollars.)

One more thing worth saying out loud: the Bolnick estimates come from 2016 spending data, and nobody has redone the attribution analysis on newer figures. The strongest single piece of evidence for the CMS Administrator's claim is now a decade old. That does not make it wrong. It does mean it should be cited with its age attached.

Verdict: unproven, and the demonstration isn't designed to prove it.

The Medicare GLP-1 Bridge runs 18 months, through the end of 2027. Health policy analysts at KFF have publicly doubted that any total-cost-of-care savings can be demonstrated in that window, and CMS's own policy director conceded 18 months is short for spend analysis.

A Milliman actuary's reply to a colleague modeling return on investment for semaglutide, quoted at the policy forum: "In health care there is no ROI. The best you can do is try to work through this cost-shifting scenario."

Savings also depend on people staying on treatment. One surgeon at the forum warned from his own institution's experience: measure the six months after the medication stops, not just the year on it.

Ask: "Can we track how I'm actually doing day to day, things like pain, energy, and mood, not just my weight?"

The standard uses a brief validated questionnaire (EQ-5D-5L) covering mobility, self-care, usual activities, pain, and anxiety or depression. It takes about a minute to complete.

Ask: "Has what I can physically do changed? What about my mood, and my relationship with food?"

The standard uses the BODY-Q modules, covering physical function, psychological function, social function, eating behavior, and sexual function.

These are the domains patients consistently rank highest, and the ones least likely to appear in a chart.

Ask: "Should I be screened for sleep apnea, and if I've been treated, has it improved?"

The standard uses STOP-BANG, an eight-question screener. In the SURMOUNT-OSA trials, tirzepatide reduced breathing interruptions by roughly 25 to 29 events per hour versus about 5 on placebo.

This is one of the clearest examples of an outcome that matters enormously to how you feel and is almost never tracked as a measure of obesity treatment success.

Ask: "How have my blood pressure, A1c, cholesterol, liver numbers, and kidney function changed since we started?"

These are already in your chart. The gap is that nobody routinely reads them as a report card on obesity treatment.

This is also the most persuasive documentation you have if a treatment gets denied or discontinued.

Ask: "Should we be checking vitamin D, B12, ferritin, and folate while I'm losing weight?"

The standard includes all four. Deficiencies rise during active weight loss. Unpublished data presented at the 2026 Obesity Medicine Association meeting showed rates climbing from 13% to 22% over twelve months on GLP-1 therapy.

Ask: "Are we tracking whether I'm keeping my strength, not just losing weight?"

The standard uses grip strength measured with a hand dynamometer: cheap, fast, and reproducible.

Losing weight while losing function is not success. Grip strength is the simplest guardrail against that.

Ask: "Can we track waist circumference alongside weight?"

Height, weight, and waist circumference. This is the category most people think is obesity care. It is one of nine.

Waist circumference often improves meaningfully even when the scale stalls, which makes it the better companion measure during a plateau.

Ask (if you've had bariatric surgery): "How are my results and any complications being tracked over time?"

The standard applies the Clavien-Dindo classification. This category applies only to people who've had surgery.

Ask (where relevant): "How does this affect my cycles, fertility, or a future pregnancy?"

The standard tracks fertility, menstrual irregularity, and pregnancy outcomes as a population-specific category.

Worth raising early, before starting treatment rather than after.