Commercial aviation made fatal accidents extraordinarily rare by investigating failures, protecting candid reporting, publishing findings, and redesigning systems. Medicine should pursue the same near-zero standard for preventable death and injury.
Why should Americans accept tens of thousands of deaths from medical error when nearly ten million scheduled passenger flights operate in the United States each year with remarkably few fatal accidents?
The comparison is not perfect. Patients are already sick, human biology is less predictable than an aircraft, and some complications cannot be prevented. But aviation did not become safe by eliminating human fallibility. It became safe by building systems in which an individual mistake is detected, contained, investigated, and prevented from killing people.
That should be the objective of American medicine: near-zero preventable death and near-zero serious preventable harm.
Combined with aggressive drug prevention, automatic automobile intervention, fall prevention, faster access to effective treatments, genetic engineering, malpractice reform, AI documentation, and appropriate end-of-life care, the United States could plausibly save between 93,000 and 169,000 lives annually.
A reasonable central planning estimate is approximately 146,000 lives per year.
The quantifiable opportunity
| Reform | Potential annual benefit |
|---|
| Near-zero preventable hospital deaths | 22,000–98,000 lives |
| Central medical-error planning estimate | Approximately 75,000 lives |
| Matching Singapore’s drug-use-disorder mortality rate | Approximately 70,421 lives |
| Matching Singapore’s drug-related disability rate | Approximately 6.17 million healthy life-years |
| Automatic emergency braking | At least 360 lives and 24,000 injuries |
| Preventable hospital morbidity among Medicare patients | Potentially more than 1 million preventable harm events |
| Bedside urinals and nighttime fall prevention | Share of 41,000 annual older-adult fall deaths not yet measured |
| Palliative and nonbeneficial-care reform | Less pain, delirium, restraint, intensive treatment and unwanted hospitalization |
| Faster drug approval, genetic engineering and AI-supported medicine | Potentially very large, but not yet quantifiable |
The lower combined mortality estimate is approximately:
Using 75,000 preventable medical deaths as a central planning estimate:
Using the upper Institute of Medicine estimate:
These are ambitious prevention scenarios, not guaranteed forecasts. But they establish the scale of preventable death that America continues to tolerate.
1. Make preventable medical death as rare as an airline crash
The Federal Aviation Administration handles more than 9.8 million scheduled passenger flights annually. FAA air-traffic statistics
Commercial aviation is not literally error-free. The January 2025 collision near Washington, D.C., killed 67 people, demonstrating that no complex system permanently reaches zero risk. National Transportation Safety Board
But fatal commercial-airline accidents have become extraordinarily rare relative to the number of flights. Aviation achieved that result through:
Mandatory reporting of accidents and serious incidents
Confidential, nonpunitive reporting of near misses
Preservation of data and physical evidence
Independent investigation
Public findings and probable-cause determinations
Safety recommendations directed to regulators, manufacturers and operators
Standardized procedures and checklists
Redundant systems
Simulation and recurrent training
Continuous collection of operational data
Redesign after failure
Medicine needs the same structure.
The medical-error death count
The exact number of Americans killed by preventable medical error remains disputed because death certificates rarely identify error, reviewers disagree about preventability, and many affected patients have serious underlying illnesses.
A 2020 systematic review estimated approximately 22,165 preventable inpatient deaths annually. It found that about 3.1% of hospital deaths were probably preventable. Systematic review
The Institute of Medicine’s landmark estimate was 44,000 to 98,000 deaths annually. A former director of the Agency for Healthcare Research and Quality later proposed approximately 75,000 preventable in-hospital deaths as a reasonable estimate. AHRQ Patient Safety Network
The much-publicized estimate of more than 250,000 deaths has been criticized for extrapolation and classification problems. It is therefore not used in this article’s headline total.
Using the evidence-supported range of 22,000 to 98,000, reducing preventable medical deaths to near zero would save somewhere within that range annually. The 75,000 figure is a reasonable central planning benchmark, not a settled death count.
The morbidity is even greater
Death represents only the visible peak of medical injury.
The HHS Office of Inspector General reviewed a national sample of hospitalized Medicare patients and found:
25% experienced some form of harm.
12% experienced an adverse event involving prolonged hospitalization, permanent disability, life-saving intervention, or death.
Another 13% experienced temporary harm requiring medical intervention.
Physician reviewers judged 43% of the harm events preventable.
The overall preventable-harm rate was approximately 13% of hospitalized Medicare patients.
The injuries included medication complications, delirium, pressure injuries, infections, surgical injuries, bleeding, kidney injury, strokes, respiratory failure and failures to recognize deterioration. HHS-OIG hospital-harm report
The study represented roughly one million hospitalized Medicare patients during one month. A simple annualization of the 13% preventable-harm rate suggests approximately 1.5 million preventable-harm admissions per year among Medicare patients alone.
That is a planning extrapolation rather than a count of unique patients. It may include repeat admissions and seasonal variation. Nevertheless, it reveals the scale of avoidable morbidity.
Create a National Medical Safety Board
Every fatal, permanently disabling, or potentially catastrophic medical event should be reported immediately to an independent National Medical Safety Board.
The board should investigate:
Diagnostic delay or failure
Medication errors and interactions
Failure to review laboratory or imaging results
Wrong-patient and wrong-procedure events
Inadequate monitoring
Failure to respond to deterioration
Infection-control failures
Surgical and procedural complications
Staffing and supervision failures
Defective equipment or software
Communication and handoff failures
Patient falls, elopements and suicides
Errors caused by fatigue, distraction or excessive documentation
Information known to the family but never obtained by the treatment team
The investigation should preserve:
The original medical record
Medication-administration records
Monitor and ventilator data
EHR audit logs
Staffing schedules
Laboratory and imaging timelines
Device and software logs
Relevant audio or video
Interviews conducted promptly after the event
Protect the investigation—and publish the lessons
Aviation obtains information that would otherwise remain concealed by providing confidential and nonpunitive reporting systems.
The FAA’s Aviation Safety Reporting System is confidential and nonpunitive. Its Aviation Safety Action Program encourages pilots, mechanics, dispatchers, flight attendants and other personnel to report safety problems so corrective action can occur before an accident. FAA voluntary-reporting programs
Medicine should do the same.
The additional safety investigation—its interviews, deliberations, causal analysis and draft recommendations—should be privileged against discovery and use in civil litigation. Federal law already provides limited privilege for qualifying patient-safety work product, but the protection is fragmented and incomplete. HHS patient-safety protections
The privilege should not conceal the original medical record, falsify what happened, or deprive an injured patient of independently existing evidence. It should protect the additional investigation conducted to discover why the event occurred.
After completion, the health department should publish a de-identified report containing:
What happened
The sequence of events
Direct and contributing causes
Similar prior events
Corrective actions
Responsible institutions for implementation
Deadlines
Subsequent compliance
Evidence that the correction actually worked
That is how aviation turns one disaster into prevention across an entire industry.
Investigate near misses, not only deaths
Waiting for a fatality is too late.
A nurse catches the tenfold medication error before administration. A pharmacist identifies a dangerous interaction. A laboratory result is discovered shortly before the patient deteriorates. A surgeon stops before operating on the wrong side.
These are free warnings from the system.
Near misses should be reported confidentially, aggregated nationally and analyzed by AI to identify recurring patterns. Reporters acting in good faith should receive protection from retaliation and discipline. Recklessness, intentional falsification, impairment and deliberate concealment would remain outside the protection.
The objective is not to prove that physicians and nurses never make mistakes. The objective is to ensure that one person’s mistake cannot pass through multiple layers and kill the patient.
2. Replace the malpractice lottery with patient compensation
The present malpractice system discourages candid investigation while consuming money that should compensate injured patients.
A major study found that for every dollar paid in compensation, another 54 cents was consumed by administrative expenses, including attorneys, experts and courts. New England Journal of Medicine study
America should replace conventional medical-injury litigation with an administrative compensation system:
Independent expert review
Published compensation schedules
Rapid payment for medical expenses, lost earnings and disability
Lifelong-care funding for catastrophic injuries
Legal fees paid separately and strictly limited
Compensation based on avoidable injury rather than success in a litigation lottery
Separate professional discipline for reckless, dishonest or repeatedly incompetent conduct
Sweden, Denmark and New Zealand demonstrate that patients can seek compensation through administrative systems without conventional malpractice litigation. Commonwealth Fund review
The purpose of medical-injury compensation should be to care for injured patients—not to provide a jackpot for a few successful plaintiffs while most injured patients receive nothing.
3. End defensive medicine and knowingly unnecessary billing
Care ordered primarily to create a litigation defense is legally motivated rather than medically motivated.
Evidence-based clinical pathways should create liability safe harbors. Physicians should remain free to depart from a pathway when the patient’s condition justifies it, but the reason should be documented.
Insurers and regulators should also stop requiring visits, tests and procedures that do not improve diagnosis, treatment or outcome.
Repeated, knowing billing for medically unnecessary treatment should face escalating enforcement:
First substantiated pattern: education, repayment and corrective plan.
Second substantially similar violation: formal warning, audit and monitored compliance.
Third knowing repetition: mandatory referral for fraud prosecution, civil penalties and program exclusion.
Good-faith medical disagreement is not fraud. A documentation defect is not automatically fraud. But federal law already prohibits claims a provider knows or should know are false. HHS-OIG fraud guidance
CMS estimated $28.83 billion in Medicare fee-for-service improper payments for fiscal 2025. CMS correctly warns that improper payments are not synonymous with fraud; many arise from insufficient documentation. CMS payment-integrity data
Enforcement should focus on deliberate extraction of payment for care known to be unnecessary—not on criminalizing clinical judgment.
4. Apply the complete Singapore drug strategy
Singapore combines prevention, strict enforcement, compulsory rehabilitation, testing, supervision, aftercare and severe punishment of commercial trafficking. Some people caught solely for drug use are routed into rehabilitation without a criminal conviction. Singapore Central Narcotics Bureau
Harmonized 2021 Global Burden of Disease data show an age-standardized drug-use-disorder death rate of 19.52 per 100,000 in the United States and 0.13 in Singapore.
If the United States matched Singapore’s relative rate:
Annual deaths would decline from approximately 70,893 to 472.
Approximately 70,421 lives would be saved annually.
Approximately 6.17 million disability-adjusted life years would be preserved.
A disability-adjusted life year represents one healthy year lost through death or disability. Global Burden of Disease study and country tables
An American version should combine severe, proportionate punishment for organized commercial trafficking with compulsory recovery pathways for addicted users. Treatment in correctional facilities is essential: medication treatment in jails has been associated with a 52% reduction in fatal opioid overdose after release. National Institutes of Health
This comparison does not prove that punishment alone caused Singapore’s lower death rate. The effective strategy is the entire system: prevention, detection, rapid intervention, rehabilitation, supervision and enforcement.
5. Prevent nighttime toileting falls
Older-adult falls cause approximately:
Falls are also a major cause of traumatic brain injury, hip fracture, disability, fear and institutional placement. CDC fall statistics
A study of hospitalized older adults found that 34% of falls were toileting-related and at least 44% of those occurred at night. In 80% of nighttime toileting falls, the patient was moving without the recommended assistance. Toileting-fall study
Bedside male and female urinals should be routinely offered to people with:
Nocturia
Previous falls
Weakness or impaired balance
Orthostatic hypotension
Brain injury or cognitive impairment
Sedating medications
Urgency or incontinence
Recent surgery
Need for assistance with transfers
The receptacle can be emptied into the toilet, avoiding an unnecessary nighttime journey. Female urinal devices and bedside commodes should be readily available rather than treated as unusual equipment.
The precise number of falls and deaths preventable through bedside urinals has not been measured nationally. Hospitals and residential facilities should therefore implement prospective trials measuring falls, fractures, hospital transfers and patient satisfaction.
The absence of an expensive commercial sponsor should not prevent testing an inexpensive preventive intervention.
6. Require vehicles to intervene when a crash becomes imminent
An automobile should not merely warn a driver when a collision is highly probable. It should brake automatically when the driver fails to respond.
NHTSA estimates that its automatic-emergency-braking standard will prevent at least:
NHTSA automatic-braking rule
Real-world research has found that forward-collision warning with automatic braking reduces rear-end crashes by approximately 50% and rear-end injury crashes by 56%. Insurance Institute for Highway Safety
The next generation should combine emergency braking with:
Pedestrian and cyclist detection
Blind-spot intervention
Lane-departure control
Driver-impairment and inattention detection
Intelligent speed assistance
Automatic crash notification
The driver would ordinarily remain in control. The system would intervene only when validated sensors determine that death or serious injury is becoming highly probable.
7. Stop paying clinicians to type
Medical personnel should spend encounters observing, listening, examining, deciding and explaining—not functioning as clerks.
With patient consent, encounters should be recorded and converted by AI into:
The clinician should review and approve the final note. The AI should draft; the licensed professional should remain responsible.
A 2025 multicenter study found that clinician burnout declined from 51.9% to 38.8% after 30 days of ambient AI-scribe use. Clinicians also reported less after-hours documentation, lower cognitive burden and better attention to patients. JAMA Network Open
The mortality effect has not yet been measured. But removing clerical distraction creates more time to detect symptoms, review risks, obtain information from families and identify deterioration.
8. Pay for skill and outcome—not elapsed time
A highly skilled clinician may identify the central problem in two minutes. A less skilled clinician may spend 20 minutes without solving it.
Payment should reflect:
Time may remain one factor, but it should not define value.
Existing value-based programs have produced mixed results, partly because many reward documentation and process measures rather than meaningful outcomes. Value-based-payment review
Outcome measures must be risk-adjusted, clinically meaningful and resistant to gaming. A physician should not be punished for accepting difficult patients, and a patient should not be denied care because poor outcomes would damage a provider’s score.
9. Create a new pharmaceutical bargain
The standard utility-patent term is generally 20 years from filing. U.S. Patent and Trademark Office
A replacement pharmaceutical bargain could provide:
Five years of effective post-approval market exclusivity.
Presumptive approval of a complete application within six months unless the FDA issues a formal, evidence-based objection.
Immediate withdrawal authority for serious safety problems.
Mandatory postmarketing surveillance.
Rapid generic and biosimilar competition after exclusivity.
Government prizes and purchase commitments for important discoveries.
Expanded public financing of trials and genetic research.
The FDA already uses a six-month review goal for priority applications, compared with ten months for standard review. FDA priority review
Faster approval can get effective treatment to dying patients sooner. Shorter exclusivity can make treatment affordable sooner. But shorter patent protection must be paired with public research funding, milestone prizes and purchase commitments so society does not reduce the incentive to develop future treatments.
The number of lives saved cannot be known in advance because it depends on therapies that have not yet been discovered.
10. Surge investment in genetic engineering of familial disease
Gene editing is no longer theoretical. The FDA has approved the first CRISPR-based treatment for sickle-cell disease. FDA gene-therapy announcement
National research should prioritize:
Familial hypercholesterolemia
Hypertrophic cardiomyopathy
Inherited arrhythmias
Hereditary cancer syndromes
Familial kidney and liver diseases
Monogenic diabetes and obesity
Neurodegenerative conditions
Hemoglobin disorders
Familial hypercholesterolemia may affect roughly one in 250 people. Without treatment, approximately half of affected men experience a heart attack by age 50 and 30% of affected women by age 60. Early diagnosis and treatment can reduce coronary-disease risk by about 80%. CDC familial-hypercholesterolemia guidance
Screening and conventional treatment can save lives immediately. Gene editing may eventually replace decades of treatment with a durable correction.
11. End painful treatment that cannot benefit the patient
A patient who is irreversibly dying should not be subjected to painful procedures that cannot restore consciousness, function, discharge, or meaningful survival.
The reform should not be based on the unsupported claim that every dying patient receives $300,000 of futile care. Nor should all last-year-of-life treatment be classified as waste. Many treatments provide real comfort or meaningful additional life.
The standard should be medical benefit:
Early prognosis discussions
Default palliative-care consultation for defined high-risk conditions
Clear identification of the treatment goal
Independent ethics and medical review when disagreement occurs
A reasonable opportunity for transfer
No obligation to provide treatment incapable of accomplishing its physiological purpose
Aggressive relief of pain, dyspnea, anxiety and delirium
A meta-analysis of randomized trials found that palliative care improves quality of life and symptom burden, although it did not demonstrate a significant overall survival benefit. JAMA meta-analysis
The morbidity prevented includes pain, delirium, restraints, repeated resuscitation, invasive lines, treatment complications, prolonged ventilation and dying in circumstances the patient would never have chosen.
The final prevention total
A near-zero medical-harm system, Singapore-level drug outcomes and automatic collision intervention together produce a potential annual mortality reduction of:
Low estimate: approximately 93,000 lives
Central planning estimate: approximately 146,000 lives
Upper established estimate: approximately 169,000 lives
The morbidity reduction would include:
6.17 million healthy life-years preserved from drug-use disorders
Potentially more than one million preventable Medicare hospital-harm events
At least 24,000 automobile injuries
Uncounted falls, fractures and traumatic brain injuries
Less pain and delirium at the end of life
Fewer adverse effects from unnecessary tests and procedures
Earlier access to effective treatment
Reduced disability from familial disease
More clinician attention directed toward patients
The numbers should not be treated as guaranteed savings or added without attention to overlap. They are a statement of achievable scale.
America should stop treating medical injury as an unavoidable cost of doing business.
Aviation’s lesson is not that pilots never make mistakes. It is that every accident and near miss must make the entire system safer.
That should become the governing rule of medicine:
Investigate every serious error. Protect the investigation. Publish every lesson. Correct every identified hazard. Measure whether the correction works. Continue until preventable medical death becomes as rare as a fatal commercial-airline accident.
