Americans consume so much healthcare that they don't need.
As it turns out, this is true in a lot of places, and we have excellent evidence that's the case.
Thread.
Britain has universal healthcare via their National Health Service.
In this system, the doctors are paid very poorly. Junior doctors—known as "resident doctors" since September of this year—have gone on strike about this several times in recent years.
In 2016, a dispute between the government and medical unions about new junior doctor contracts came to a head and the junior doctors didn't like the terms they were offered.
So, five strikes took place across all English public hospitals between January and April of that year.
During the protests, the vast majority of junior doctors did not report for duty.
As a result, more than 100,000 outpatient appointments and more than 25,000 fewer planned admissions had to be canceled.
Senior doctors and nurses also had to be redeployed to emergency services.
Hospitals acted to mitigate the impacts of the strikes, requiring some of the junior doctors on roster for emergency services to stay out, calling freelance locum doctors into the NHS, canceling holidays and study leave for staff groups, and asking private doctors for help.
But this wasn't enough to save the NHS' volume, and the backlog of appointments the strike generated was vast.
The strike reduced emergency arrivals and admissions, as well as elective admissions considerably:
When the strikes took place, patients who still came in had different characteristics, meaning still going to the hospital on a day where there were fewer doctors available was selective.
Go ahead and read these to see how. For reference, Charlson Score is a comorbidity index:
Given elective patients were older, emergency ones were younger, etc., a strategy to identify the impacts the strike had on patients given the reduced volume of care caused isn't obvious
So, these authors leveraged the proportion of junior docs at a hospital as an exposure index
Taking this interaction out, the impact of strikes on patient characteristics is no longer significant, and that result is precise enough with small enough coefficients that we're probably fine to go ahead with using this instrument.
So let's check: What happens to patients when they're heavily exposed to a strike?
In terms of readmissions within 30 days and mortality... nothing, not even when you stratify by exposure level or control for the severity of patient condition!
The volume of care provided by the NHS is reduced by strikes, but not so much that patients are harmed.
That means there's unnecessary care happening.
This study's conclusions, by the way, are not unique. There's actually a large literature on the effects of doctor strikes.
In 2008, Cunningham et al. provided a review, in which they noted that doctor strikes with variables lengths, participation, and so on, from Jerusalem to Los Angeles had similar non-effects, or even potentially positive effects on patient mortality!
The amount of care people consume might not just be so high it's wasteful, but so high it's harmful.
Meta-analytically, the impact of doctor strikes all the way through 2021 seems to be... bupkes. It just doesn't matter when doctors go on strike.
This finding holds up in low-middle income countries, for strikes that happen for nurses and other staff too, across many sites, and even up to 250 days of striking in one study.
Care volumes are definitely affected, appointments are missed, prescription numbers decline, etc.
And yet, people carry on, and maybe even get a little better off.
Now there's obviously important care doctors need to be there to provide, but most of the time people are visiting the doc, it's just not providing them or the healthcare system any value: It's payment for nothing
There are a lot of other ways we can see that people consume too much care, aided by plenty of different designs, like RCTs comparing more and less extensive screening protocols.
But to some extent, it should be obvious that people consume too much healthcare that's way too low-value.
Consider @robinhanson's explanation for a variety of stylized facts about overprovisioning of care, to explain why it's a superior good:
You can also look at simpler data to see this, like the data showing that the health share of consumption does rise very rapidly with income, and thus the reason the U.S. spends so much on healthcare is primarily because it's very rich.
Relatedly, if you take a look at health expenditures per capita versus life expectancies, you actually see evidence of nonlinearities, such that past some level of spending, the superior good status of healthcare gets ugly because it stops generating returns.
We can go on, talking about ineffective but common treatments, overprovided medicines and overly long therapies and surgeries, and more, but I think my point is clear:
People consume too much healthcare, and it doesn't benefit them to do so.
To cut costs, they could spend less.
If you want to see a country like America cut its costs, you can eliminate all the inefficiencies, and then you'll still have to deal with the fact that Americans consume too much healthcare.
How much? I think Hanson and Cutler are right, at about 30-50% and increasingly more.
Oh and this is the vital role of insurers in the healthcare system: they are the hidden rationers that keep it running.
Healthcare in the U.S. is more expensive than in peer countries primarily because Americans consume so much more care, not because the number of administrators is out of wack
One way we know this is through private equity
When PE firms buy hospitals, they drastically cut admin:
PE acquisition leads to a momentary decline in the numbers of core workers (nurses, physicians, pharmacists, etc.), but this fizzles out in the long run.
By cutting a massive part of the administrative bloat (they also cut admin wages!), PE firms are able to turn hospitals around without sacrificing anything for patients or providers.
Anesthesiology was one of the medical specialties that was the most likely to cause a surprise bill, because patients usually don't select their anesthesiologist.
This meant lots of patients got saddled with out-of-network care, even at an in-network facility.
This has, to some extent, been reduced recently thanks to state-level protective legislation, and also to the federal No Surprises Act that went into effect in January of 2022.
Unfortunately, there are still exceptions aplenty, so surprise bills still exist.
When people report an excess of '5s' and '0s' for quantities like 'IQ', 'height', 'age', etc., it indicates error and lying
Anesthesiologists report an excess of anesthesia times ending in increments of 5. As it just so happens, the ones who do this the most profit the most too
A given anesthesia case generates compensation units based on self-reported time providing patient care.
One unit is 15 minutes of care, and insurers like to pay to the exact minute, so they usually require unrounded time reporting.
And yet, some anesthesiologists round a lot.
The anesthesiologists who report the most rounding to times ending in increments of 5 (95th percentile '5-reporters'+) end up costing the healthcare system much more than expected.
They bill for significantly longer anesthesia times than they ought to without extreme rounding:
The vast majority of crime is done by just a few individuals.
The vast majority of crime is also done in just a few locations.
For example, in New York City, 75% of the violent crime happens on less than 10% of the streets.
When it comes to property crime, there's less concentration than there is for violent crimes.
This makes sense, because property to harm should be more spread out more than violent people are.
If you throw different crimes together to look at the geographic concentration of 'all crimes', the concentration is reduced even further, because different types of crime occur in imperfectly overlapping places.
The results of a new large-scale international educational assessment just came out.
This one focuses on student mathematics and science achievement and, once again, America came in near the top of the chart!🧵
First up, here are mean scores for fourth-graders:
Next, here are fourth-graders' mathematics scores.
This is, theoretically, the most trainable test, and we see that Americans do well, but Asian countries take the cake.
What about the science side of things?
This is less trainable than mathematics, and Americans end up doing quite a bit better, with even American Whites managing to outperform some Asian countries.