@robgraymdi
iAccount based inUnited States
About this account
- Account based in
- United States
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
â & Microsurgeon. Mayo Clinic, Rush & Brown. âI treat it like itâs my own handâ Opinions my own & this isnât medical advice
Chicago's North Shore
Joined April 2015
- Tweets6.9K
- Following989
- Followers2.3K
- Likes34.2K
Rob Gray𧢠retweeted
You could be an excellent surgeon ten minutes from someoneâs house. They may have absolutely no idea you exist.
Go live.
âI replace hips and knees here in Dallas. Today, Iâll explain what you should ask before agreeing to surgery.â
Explain your approach. Discuss recovery. Explain when surgery makes sense and when you recommend another option.
Let people hear how you think.
Then tell them where to find you.
Medicine. But with computers!
What. Could. Go. Wrong?
Rob Gray𧢠retweeted
The number one reason healthcare is so expensive is not tax-exempt hospitals.
It is not insurance companies.
It is the incentives created by government.
At an independent physicianâs office, Medicare makes one payment.
For services delivered through a hospital outpatient department (HOPD), Medicare makes two payments.
Here is an example:
Lumbar epidural in a physicianâs office: $256.
The same procedure through a hospital outpatient department: $741.
Nearly three times the payment! Itâs utterly ridiculous to be $40 trillion in the hole and then pay 3x as much for the same thing.
Same physician.
Same patient.
Same procedure.
Different owner.
Higher bill.
Changing ownership triples the payment.
That payment gap gives hospitals a government-funded reason to acquire physician practices.
The more practices a hospital acquires, the more care it moves through the higher-paying system.
This arithmetic is not difficult. Congress simply finds it inconvenient. Government rewards hospitals for acquiring physician practices.
Congress banned physician-owned hospitals and froze most existing ones at their 2010 capacity.
Apparently, physicians are trustworthy enough to perform the surgery. Owning the hospital is where Congress draws the line. Congress approves lawyers, MBAs, and Felons. Not physicians.
Tax-exempt hospital systems pay no federal income tax, no state income tax and no property tax.
Your grandmother pays property tax.
She made the rookie mistake of not calling her house a hospital campus.
Government also created 340B.
In 2025, 340B covered entities purchased $100 billion in discounted outpatient drugs.
That is what they purchased.
It does not tell us what they billed.
It does not tell us how much they made.
The hospital buys the drug at the 340B price.
The hospital bills the employer the full contracted allowable amount.
The patient pays a deductible or coinsurance calculated from that amount.
The patientâs name unlocks the discount.
The hospital keeps it.
Acquire an oncology practice.
Extend the discount.
Keep the spread.
In 34 states and Washington, D.C., hospitals, ambulatory surgery centers and radiology centers must ask government for permission to open or expand.
The incumbent hospital gets to object.
Naturally.
Who should decide whether a community needs competition if not the company facing it?
Then politicians pretend to be surprised when independent physicians disappear, hospital markets consolidate and employer premiums rise.
Hospitals and insurance companies are not innocent.
Hospitals lobby to preserve these advantages. Insurers pass the resulting expense into next yearâs premiums.
Everyone follows incentives.
Government chose these incentives.
The people you elected wrote them into law, protected them for years and now hold hearings to discover who could possibly have done this.
Want lower healthcare costs?
Pay the same amount for the same service.
Let physicians own and expand hospitals.
End Certificate of Need laws.
Make 340B discounts follow the patient.
Americans must demand that government stop subsidizing consolidation and picking winners and losers.
Washington pays hospitals more to consolidate, blocks physicians from competing, exempts the winners from taxes and then lectures America about greed.
Rob Gray𧢠retweeted
EVERY LIE HIRES A BUREAUCRAT.
Every scam writes a rule.
Every theft installs a camera.
Every fraudulent claim creates another audit.
We complain when bureaucracy grows. We rarely ask what fed it. Someone lied, so now everyone needs documentation.
Someone stole, so now everyone gets searched. Someone abused the freedom they received, so now nobody receives it.
The bureaucrat did not always cause the problem. Sometimes the bureaucrat is the receipt.
Dishonesty does more than harm the person who believed the lie. It raises the cost of freedom for every honest person who comes afterward.
If we want fewer rules, we need more people who keep their word. If we want less supervision, we need more people who behave properly when nobody watches.
People who govern themselves require less government from everyone else. Every act of dishonesty makes freedom more expensive.
Exactly how cheaply do we value life?
Lack of âskin in the gameâ explains a lot of our health outcomes.
It is already egregious that doctors accept $1M downside risk for a payment of $200 or even $2000.
Everyone who thinks they can scale away doctors (who are only 6-8 cents of the HC dollar, btw) has no answer for how to cover the scaled risk.
The volume required to make $20 visits profitable, even if only AI, creates $ billions of liability.
It would be an existential threat to even the largest companies. Theyâd be uninsurable.
Rob Gray𧢠retweeted
Underrated life advice: Wake up early and work out. It creates evidence that you have the power to take an action to achieve a desired outcome. That has ripple effects into every area. It rewires your brain. A lot of problems in life are solved by waking up early and working out.
Love this.
But donât worry residents, you jobs are safe as attendings too.
Youâre much easier to sue than a machine.
Soon, radiology residents will replace radiology AI models.
Pedrini et al (2026) looked at 3 months' worth of CT scans (2,153) interpreted by an on-call resident. 15.4% of them had intracranial hemorrhage (ICH).
The residents had a sensitivity of 96.4% and specificity of 99.6%. The same studies were given to directly to a commercial AI software, which generated heatmaps. The results and heatmaps were separately evaluated as part of the research protocol (aka not available at interpretation time) with sensitivity of 84% and specificity of 94.4%. Performance by AI improved with multiple hemorrhagic types or sites, but did not outperform the resident.
Of 12 FN reports by residents, 6 would have been caught by the model; the attending that overread the preliminary report obviously found all 12.
AI mislabeled 101 cases as FP for ICH. If these were autonomously read, this would have likely led to increased length of stay, follow-up imaging, or inappropriate treatment changes, including discontinuation or non-administration of thrombolytic therapy in patients with ischemia.
The study authors note that the AI FP "typically would not cause confusion with ICH for radiologists interpreting CT scans, as they are easily recognized as various hyperdense intracranial abnormalities not related to ICH."
Take-home points: 1) Clinical deployment studies are increasingly important for medical AI. The commercial model used here was good and well-validated. 2) Perhaps you should hire a Swiss radiology resident to read all your ICH cases, they seem pretty good.
link.springer.com/article/10âŚ
Rob Gray𧢠retweeted
Soon, radiology residents will replace radiology AI models.
Pedrini et al (2026) looked at 3 months' worth of CT scans (2,153) interpreted by an on-call resident. 15.4% of them had intracranial hemorrhage (ICH).
The residents had a sensitivity of 96.4% and specificity of 99.6%. The same studies were given to directly to a commercial AI software, which generated heatmaps. The results and heatmaps were separately evaluated as part of the research protocol (aka not available at interpretation time) with sensitivity of 84% and specificity of 94.4%. Performance by AI improved with multiple hemorrhagic types or sites, but did not outperform the resident.
Of 12 FN reports by residents, 6 would have been caught by the model; the attending that overread the preliminary report obviously found all 12.
AI mislabeled 101 cases as FP for ICH. If these were autonomously read, this would have likely led to increased length of stay, follow-up imaging, or inappropriate treatment changes, including discontinuation or non-administration of thrombolytic therapy in patients with ischemia.
The study authors note that the AI FP "typically would not cause confusion with ICH for radiologists interpreting CT scans, as they are easily recognized as various hyperdense intracranial abnormalities not related to ICH."
Take-home points: 1) Clinical deployment studies are increasingly important for medical AI. The commercial model used here was good and well-validated. 2) Perhaps you should hire a Swiss radiology resident to read all your ICH cases, they seem pretty good.
link.springer.com/article/10âŚ
Rob Gray𧢠retweeted
Some people have asked me why Iâm making fun of that JAMA viewpoint piece so much. Itâs because corporate propaganda is not benign. It influences both public sentiment and government policy.
To issue a press release on your companyâs website touting biased comparisons as evidence of your productâs efficacy is one thing. To publish what amounts to a corporate advertisement with the imprimatur of a prestigious peer-reviewed journal (albeit under the guise of a âviewpointâ piece) is entirely another â co-opting the legitimizing effects of journal publication to imply that somehow the scientific community has blessed your companyâs product.
The article in question will no doubt be used by its heavily conflicted authors to garner further investment for their companies and to lobby the government to shape CMS policy in their favor â giving further institutional credence, and potentially public money, to their proprietary technology. The AI hype cycle will continue with its air of inevitability, despite not having objectively demonstrated benefit to anyone apart from corporate shareholders.
This is a woefully under followed account
This claim. One surgery, one night in the hospital. Charges $166,166.13.
There are two pages to this claim but this is the main one. A lot of BS charges are the first page. The actual charge amount didn't bother me because I had an SCA (Single Case Agreement) in hand before the patient entered the hospital.
What bothers me is:
1. Left column - see the Revenue code 710? That's for recovery room. There are two listed. One for 8 "units" and one for 14 "units. Two different dates of service but only one surgery on one day.
2. Third line down. Over $101,000 for a procedure that Medicare reimburses at $10,000. Billed charges are 10x Medicare.
3. Fourth line down. $24,000+ for anesthesia.
Again, this is just charges and charges don't matter to me because we don't play the PPO or "percent of billed charges" games. We stayed one step ahead and knew the approximate payment for this claim beforehand.
What pisses me off is when a hospital thinks I'm stupid. Two lines of recovery room charges. Two different dates. Billed on a bill type 131 - Outpatient.
Pulled the medical records. The patient was admitted overnight and, of course, didn't stay in recovery for all that time.
Why does that matter? The Medicare reimbursement (which is my benchmark) is higher for this procedure on an outpatient basis than the applicable inpatient DRG (diagnosis related group). With the hospital admission, this becomes an inpatient stay.
The good news? I caught it. This claim has been denied in full and a corrected claim has been requested reflecting inpatient status. Anticipated payment will be less than $30,000.
The bad news? I pulled this hospital's MRF. The largest network discount I found was 45% off of billed charges.
What's $166K minus a 45% discount? $91,391 and THAT is the amount an employer with the largest PPO "discount" would pay on this claim.
How many times does this happen over the course of a year? A plan pays a minimum of $61,000 MORE than it should - again, IF they have the best "discount".
Other network "discounts" were 15% - that would make this PPO claim payable at $141,000.
AI wouldn't have caught this (I ran it through several LLMs and they all failed miserably).
An adjudication engine wouldn't have caught this. I've helped build adjudication engines and it's nearly impossible to code for some of the creativity that is going on these days.
I wouldn't have caught this without pulling the medical records.
I had the patient send me the medical records from the portal. The hospital has no idea I have them. Now I wait for the claim revision.
The moral of this story:
PLEASE
Stop being gamed by the system.
Stop letting brokers sell you a 45% discount on a charge that is inflated by 1000%.
Stop being passive.
OPEN YOUR EYES!
CEOs â if you're responsible for your company's health plan, fiduciary responsibility comes with it.
This is what can happen when nobody actually investigates what you're paying.
Rob Gray𧢠retweeted
This claim. One surgery, one night in the hospital. Charges $166,166.13.
There are two pages to this claim but this is the main one. A lot of BS charges are the first page. The actual charge amount didn't bother me because I had an SCA (Single Case Agreement) in hand before the patient entered the hospital.
What bothers me is:
1. Left column - see the Revenue code 710? That's for recovery room. There are two listed. One for 8 "units" and one for 14 "units. Two different dates of service but only one surgery on one day.
2. Third line down. Over $101,000 for a procedure that Medicare reimburses at $10,000. Billed charges are 10x Medicare.
3. Fourth line down. $24,000+ for anesthesia.
Again, this is just charges and charges don't matter to me because we don't play the PPO or "percent of billed charges" games. We stayed one step ahead and knew the approximate payment for this claim beforehand.
What pisses me off is when a hospital thinks I'm stupid. Two lines of recovery room charges. Two different dates. Billed on a bill type 131 - Outpatient.
Pulled the medical records. The patient was admitted overnight and, of course, didn't stay in recovery for all that time.
Why does that matter? The Medicare reimbursement (which is my benchmark) is higher for this procedure on an outpatient basis than the applicable inpatient DRG (diagnosis related group). With the hospital admission, this becomes an inpatient stay.
The good news? I caught it. This claim has been denied in full and a corrected claim has been requested reflecting inpatient status. Anticipated payment will be less than $30,000.
The bad news? I pulled this hospital's MRF. The largest network discount I found was 45% off of billed charges.
What's $166K minus a 45% discount? $91,391 and THAT is the amount an employer with the largest PPO "discount" would pay on this claim.
How many times does this happen over the course of a year? A plan pays a minimum of $61,000 MORE than it should - again, IF they have the best "discount".
Other network "discounts" were 15% - that would make this PPO claim payable at $141,000.
AI wouldn't have caught this (I ran it through several LLMs and they all failed miserably).
An adjudication engine wouldn't have caught this. I've helped build adjudication engines and it's nearly impossible to code for some of the creativity that is going on these days.
I wouldn't have caught this without pulling the medical records.
I had the patient send me the medical records from the portal. The hospital has no idea I have them. Now I wait for the claim revision.
The moral of this story:
PLEASE
Stop being gamed by the system.
Stop letting brokers sell you a 45% discount on a charge that is inflated by 1000%.
Stop being passive.
OPEN YOUR EYES!
CEOs â if you're responsible for your company's health plan, fiduciary responsibility comes with it.
This is what can happen when nobody actually investigates what you're paying.
Rob Gray𧢠retweeted
1977 Guidelines: Cut the fat. All of it. Fat is the problem.
Man: Right you are. What do I eat instead?
1977 Guidelines: Bread. Pasta. Cereal.
Man: Marvellous. How much?
1977 Guidelines: Six to eleven servings.
Man: Eleven. Lovely. I'll get a bigger bowl.
1980 Guidelines: Avoid too much fat, saturated fat and cholesterol.
Man: Already on it. Half a stone up, mind, but I'm sure that's the water.
1980 Guidelines: Do it again.
Man: Righto.
1992 Guidelines: Here's a pyramid. Bread at the bottom, fat at the tip.
Man: Handsome. Why's my food at the pointy end?
1992 Guidelines: Because there's less of it.
Man: Fair enough.
1994 Guidelines: Lose the yolks.
Man: The yolks. The good bit.
1994 Guidelines: The yolks.
Man: I'm hungry by ten every morning as it is.
1994 Guidelines: That's normal.
Man: Is it.
2003 Guidelines: Five a day.
Man: Five what?
2003 Guidelines: Just five.
Man: Of what.
2003 Guidelines: Five.
2005 Guidelines: The margarine has to go. It's the trans fat.
Man: You put me on the margarine.
2005 Guidelines: Try the spread.
Man: What's in the spread?
2005 Guidelines: Next question.
2015 Guidelines: Eggs are fine. Cholesterol is no longer a nutrient of concern.
Man: I have not eaten a yolk since Thatcher.
2015 Guidelines: Then you're in excellent shape.
Man: I am nineteen stone.
2019 Guidelines: Fasting glucose 103. That's prediabetes.
Man: I did every single thing you told me.
2019 Guidelines: Nobody is suggesting otherwise.
2020 Guidelines: It was the sugar.
Man: THE WHOLE TIME?
2020 Guidelines: We have never been clearer about anything.
Man: You have said that six times. I'm three stone up and on two tablets.
2026 Guidelines: Good news. Butter's back. Steak's fine. Whole milk, marvellous.
Man: [quietly] Forty-nine years.
2026 Guidelines: Keep the saturated fat down, obviously.
Man: [long pause]
Man: What do you think butter is made of?
2026 Guidelines: We'll come back to you.
Add to it, Rob Gray, 2016. Told you all that scaphoid nonunion was a stability problem, not a vascularity problem. Told you that most radial arm neuropathic pain was wartenbergâs syndrome and was cured by a brachioradialis tenotomy rather than a radial tunnel release or ACDF.
Told you to stop transposing ulnar nerves.
When you all realize Iâm right I want this gif in every chapter citing me.
Every name on this list was told the science was settled. Every one of them was right. Most were punished first.
James Lind, 1747: scurvy came from bad air and idleness, said the Navy. He put twelve sailors on six treatments and proved it was citrus. The Admiralty acted forty-eight years later, one year after he died.
Ignaz Semmelweis, 1847: doctors walked from the autopsy table to the delivery room without washing. He made them use chlorinated lime and deaths in his ward fell from eighteen per cent to two. His profession took it as an accusation, drove him out, and he died in an asylum of sepsis.
Alfred Wegener, 1912: continents do not move. He said they had. Geology laughed for fifty years and he died on the Greenland ice long before anybody stopped.
Joseph Goldberger, 1914: pellagra was infectious and every physician in America knew it. He proved it was diet, then swallowed material from infected patients in front of them to show it was not catching. He never fell ill. Nor did his wife, who volunteered.
Barbara McClintock, 1951: genes sit still on the chromosome. She showed they move, to a silent room, then stopped publishing on it for decades. She took the Nobel alone at eighty-one.
Frances Kelsey, 1960: a morning sickness drug landed on her desk at the FDA with thin safety data. The company pressed her six times. She refused six times, and thalidomide never reached America.
John Yudkin, 1972: fat was the killer, said the entire field. He named sugar in print. His career was dismantled by the men with the rival theory, and his book returned from the dead forty years later.
Bernard Fisher, 1970s: the radical mastectomy had been standard for eighty years. He proved the lesser operation matched it, and spared a generation of women a mutilation nobody needed.
Barry Marshall, 1984: ulcers are caused by stress. He drank a flask of the bacteria, gave himself the disease, cured it with antibiotics, and collected the Nobel twenty-one years later.
Christopher Ramsden, 2016: the largest diet trial ever run had gone unpublished since the seventies. He found the boxes in a dead researcher's basement. The men who swapped saturated fat for corn oil had lower cholesterol and more funerals.
None of them had better equipment than the people they were arguing with.
They had the same data as everybody else and the stomach to be hated for reading it properly.
Consensus is a headcount. Every name above lost the vote.
Rob Gray𧢠retweeted
Looking at our 68 five star reviews after 3.5 months, I honestly feel all the reward I need. The next step will be scaling this nationally and actually making an impact on the American healthcare system. Really appreciate the support everyone and if you love what weâre doing, tell the world! search.google.com/local/writâŚ
Rob Gray𧢠retweeted
all four of our boys âtwo about to go back to college.
what if we just all stayed here and didnât leave
Rob Gray𧢠retweeted
When all else fails, examine the patientâŚâŚ.;)
Can MRI accurately diagnose carpal tunnel syndrome, or does clinical evaluation still reign supreme?
In this commentary on Liu et al., Robert R. Slater Jr., MD, FACS examines research evaluating whether MRI measurements of median nerve size correlate with electrodiagnostic findings in patients with carpal tunnel syndrome.
The study found that an increased median nerve cross-sectional area at the carpal tunnel inlet was associated with abnormal electrodiagnostic results, but MRI demonstrated only limited diagnostic accuracy overall. Slater emphasizes that carpal tunnel syndrome remains a clinical diagnosis and suggests MRI may be most valuable when clinicians need to rule out other underlying conditions or lesions contributing to patients' symptoms.
Read the full commentary đ bit.ly/45GvEK2
And read the original study đbit.ly/45J7uP8
#JBJS
Not all investments are good investments. Not all good investments are good investments FOR YOU.
It is fine to gamble with 5% or so of your money. Itâs a lotto ticketâmassive upside, meaningless if lost.
More than that and youâre taking on levels of risk that are unnecessary
I often find HNW individuals get shiny object syndrome
They often invest in almost every:
- property
- syndicate
- private investment
- friends company
That comes there way
Then all the sudden they have no liquidity & they can't keep track of anything
And are waiting till the last second each year on small K1s
You do not need to invest in everything
It will just make your life more complex