Overtreatment
Does all that extra care actually make people healthier?
About a quarter of US health spending is estimated waste, roughly 760 billion to 935 billion dollars a year across six categories. Guideline-discordant services stay common: more than half of low-risk cataract patients still get needless pre-op testing.
The problem
Overtreatment is a value and safety problem across the U.S. care landscape: hospitals and clinicians can deliver more visits, procedures, tests, or spending without improving outcomes. The strategic risk is that low-value utilization consumes capacity and exposes patients to harm while crowding out services that would improve access or health.
The recommendation
Treat low-value care reduction as a clinical-quality and capacity-release program. The recommended approach is to identify high-intensity practice patterns, align payment with demonstrated benefit, and reinvest avoided waste into services that improve access, prevention, and outcomes.
How big, how common
The dollar scale of waste, how much is truly clinical, how routinely low-value services still get delivered, and how much is recoverable.
Where the waste sits: overtreatment is only the fourth-largest bucket
Bars show the high estimate for each category, with the low-to-high range in the label gutter. Clinical overtreatment, highlighted, is dwarfed by administrative complexity and pricing failure.
Read it this way Overtreatment or low-value care, highlighted, is only the fourth-largest of six categories at up to 101.2 billion dollars, well behind administrative complexity at 265.6 billion and pricing failure at up to 240.5 billion. This argues against treating clinical overtreatment as the main driver of the 760 billion to 935 billion dollar total. The larger dollars sit in administrative and pricing failure. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
⊞ data table⬇ CSV
| Category | Low estimate (USD billions) | High estimate (USD billions) |
|---|---|---|
| Administrative complexity | 265.6 | 265.6 |
| Pricing failure | 230.7 | 240.5 |
| Failure of care delivery | 102.4 | 165.7 |
| Overtreatment or low-value care | 75.7 | 101.2 |
| Fraud and abuse | 58.5 | 83.9 |
| Failure of care coordination | 27.2 | 78.2 |
Shrank, Rogstad, and Parekh, Waste in the US Health Care System, JAMA · 2019 · source
How often low-value services get delivered anyway
Share of eligible patients who received a service that guidelines advise against.
Read it this way More than half of Medicare cataract patients, 53.0 percent, received pre-op testing the guidelines advise against, and even the lowest rate shown, imaging for low-back pain at 27.2 percent, means more than one in four eligible patients got a discouraged service. These four bars come from separate studies with different populations and years, not one representative sample of all low-value care. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
⊞ data table⬇ CSV
| Low-value service | Delivery rate (percent) | Population and period |
|---|---|---|
| Pre-op testing before cataract surgery | 53 | Medicare, 2010 to 2011 |
| Antibiotics for viral infection | 43.8 | Adults 20 to 64, 2010 to 2011 |
| PSA screening in men 70 and older | 38.6 | Medicare, 2016 to 2018 |
| Imaging for low-back pain | 27.2 | Medicare 66+, 2007 to 2011 |
Peer-reviewed low-value-care studies (Tan 2015, Fleming-Dutra 2016, Chen 2015, Kim 2022) · 2011 to 2018 · source
Three independent waste estimates over time
Each point is a separate published study, not a continuously tracked series. The dashed lines show each study's low and high bounds.
Read it this way The midpoint traces 750 billion dollars in 2009, a dip to 734 billion in 2011, then a rise to 847.5 billion in 2019, but because each point comes from a separate study with its own definitions and methods, read this as three independent snapshots agreeing waste is large and persistent, not as a smooth continuous trend. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
Caveat Each study used different data years, category definitions, and methods, so read these as three snapshots of a persistently large estimate, not a smooth trend.
⊞ data table⬇ CSV
| Study year | Low (USD billions) | Midpoint (USD billions) | High (USD billions) | Study |
|---|---|---|---|---|
| 2009 | 750 | 750 | 750 | Institute of Medicine, Best Care at Lower Cost |
| 2011 | 476 | 734 | 992 | Berwick and Hackbarth, JAMA |
| 2019 | 760 | 847.5 | 935 | Shrank, Rogstad, and Parekh, JAMA |
IOM 2012, Berwick and Hackbarth 2012, Shrank et al 2019 · 2009 to 2019 · source
The geography of overtreatment
For the same price- and demographic-adjusted Medicare population, care intensity varies widely by region, yet higher spending buys no survival benefit.
Adjusted Medicare spending per enrollee, by region
Each dot is one Hospital Referral Region. Spending is fully adjusted for price, age, sex, and race, so the 1.72-fold spread is care intensity, not prices or sicker patients. The line marks the national average.
Read it this way The spread from Miami at 13,678 dollars down to Santa Cruz at 7,967, a 1.72-fold difference, is fully adjusted for price and demographics, so it reflects care-intensity choices rather than sicker patients or higher local prices. Only the file's named extreme regions are plotted, so this shows the range of variation, not where a typical middle-of-the-pack region falls. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
Caveat Only the file's named extreme regions carry values. A full map of all 306 regions is not in this dataset.
⊞ data table⬇ CSV
| Hospital Referral Region | Adjusted Medicare spending per enrollee (USD) |
|---|---|
| Miami, FL | 13678 |
| Munster, IN | 13622 |
| Monroe, LA | 13619 |
| Los Angeles, CA | 13514 |
| Wichita Falls, TX | 13402 |
| McAllen, TX | 13372 |
| National average | 10786 |
| Anchorage, AK | 8251 |
| Burlington, VT | 8202 |
| Grand Junction, CO | 8101 |
| Honolulu, HI | 8090 |
| Santa Cruz, CA | 7967 |
Dartmouth Atlas of Health Care, Geographic Variation · 2018 · source
Physician visits in the last six months of life, by region
Physician visits per decedent in the last six months of life. A patient in the highest-intensity region saw physicians about five times as often as one in the lowest, with no survival benefit.
Read it this way McAllen's 61.0 visits per decedent versus Ogden's 11.2, about a five-fold difference, echoes the same pattern the spending chart shows: more intensity with no documented survival gain. These are 2011 figures for the file's listed extreme regions only, and the national average had already fallen to 26.3 by 2018, so the true gap today may be narrower than this snapshot suggests. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
Caveat These are the file's 2011 endpoint regions. Per-region 2018 values are not in the dataset. The national average fell from 28.8 in 2011 to 26.3 in 2018.
⊞ data table⬇ CSV
| Hospital Referral Region | Physician visits per decedent, last 6 months of life (2011) |
|---|---|
| McAllen, TX | 61 |
| Los Angeles, CA | 60.4 |
| Newark, NJ | 57.5 |
| National average | 28.8 |
| Appleton, WI | 12.1 |
| Idaho Falls, ID | 11.9 |
| Ogden, UT | 11.2 |
Dartmouth Atlas of Health Care, Geographic Variation · 2011 · source
The highest-spending regions delivered far more care
Care delivered in the highest-spending regions versus the lowest, as a rate ratio. The reference line marks equal intensity. Individual utilization measures range from 1.52 to 2.36 times the low-spending regions.
Read it this way New inpatient consults and inpatient visits show the widest gaps, at 2.36 and 2.13 times the low-spending regions, while ICU days and inpatient days sit lower but still clear the equal-intensity line at 1.0. This chart shows utilization rate ratios only. Whether that extra care improved outcomes is what the next chart answers, not this one. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
⊞ data table⬇ CSV
| Utilization measure | Highest vs lowest quintile rate ratio |
|---|---|
| New inpatient consults | 2.36 |
| Inpatient visits | 2.13 |
| ICU days | 1.55 |
| Inpatient days | 1.52 |
Fisher et al., Regional Variations in Medicare Spending, Annals of Internal Medicine · 2003 · source
But no better survival
Mortality relative risk for each 10 percent increase in regional spending. Every cohort sits at or just above 1.0, meaning more spending bought no survival benefit. Exact values and confidence intervals are in the table.
Read it this way Every cohort's relative risk sits at or barely above 1.0, from 1.003 for hip fracture to 1.012 for colorectal cancer, meaning a 10 percent spending increase bought no measurable survival benefit in any group studied. The general-population confidence interval of 0.99 to 1.03 crosses 1.0, so that estimate alone cannot rule out a true effect of zero. Use this chart to distinguish more care from better care, and to see why the recommendation targets high-intensity, low-benefit utilization rather than across-the-board cuts.
Caveat Values cluster at 1.0 by the design of the finding. The exact relative risks and 95 percent confidence intervals are in the table below.
⊞ data table⬇ CSV
| Patient cohort | Mortality relative risk per 10% more spending | 95% CI |
|---|---|---|
| Colorectal cancer | 1.012 | 1.004 to 1.019 |
| General population (MCBS) | 1.01 | 0.99 to 1.03 |
| Acute myocardial infarction | 1.007 | 1.001 to 1.014 |
| Hip fracture | 1.003 | 0.999 to 1.006 |
Fisher et al., Regional Variations in Medicare Spending, Annals of Internal Medicine · 2003 · source
Why this matters
The two largest waste categories, administrative complexity at 265.6 billion dollars and pricing failure at up to 240.5 billion, sit outside clinical decision-making entirely, in billing, contracting, and price-setting, so efforts aimed only at reducing unnecessary tests and procedures target a bucket less than half the combined size of those two. At the same time, guideline-discordant services remain routinely delivered, more than half of Medicare cataract patients still receive pre-op testing the guidelines advise against, showing the narrower clinical-overtreatment bucket still has real room to shrink through guideline adherence.
Recommended actions
- Direct primary reform attention at administrative complexity and pricing failure, the two largest waste categories, rather than treating clinical overtreatment as the main lever.
- Pilot Choosing Wisely style guideline-adherence programs targeting the four documented low-value services, starting with cataract pre-op testing at 53.0 percent delivery, the highest of the four.
- Monitor regional care-intensity ratios from the Dartmouth Atlas as a proxy for recoverable overtreatment, since high-intensity regions show no survival benefit over low-intensity ones.
- Treat the 760 billion to 935 billion dollar total as a directional order-of-magnitude estimate, not a precise figure, since three independent studies over a decade used different methods and produced a wide range.
- Measure against the 191 billion to 286 billion dollar recoverable-savings estimate rather than the full waste total when setting a policy target, since not all identified waste is recoverable in practice.
The recommendation
Therefore, treat low-value care reduction as a clinical-quality and capacity-release program. The recommended approach is to identify high-intensity practice patterns, align payment with demonstrated benefit, and reinvest avoided waste into services that improve access, prevention, and outcomes.
Demographic slice none. Choosing Wisely / waste studies report national or setting-level rates.
Sources
- Shrank, Rogstad, and Parekh, Waste in the US Health Care System, JAMA · 2019
- Tan et al, Imaging for Older Patients with Acute Low Back Pain, JGIM · 2015
- Fleming-Dutra et al, Inappropriate Antibiotic Prescriptions, JAMA · 2016
- Chen et al, Preoperative Testing in Cataract Surgery, NEJM · 2015
- Kim et al, Low-Value PSA Testing, JAMA Network Open · 2022
- Institute of Medicine, Best Care at Lower Cost · 2012
- Berwick and Hackbarth, Eliminating Waste in US Health Care, JAMA · 2012
- Dartmouth Atlas of Health Care, Geographic Variation · 2018
- Fisher et al., Regional Variations in Medicare Spending, Annals of Internal Medicine · 2003