False positives aren't inherently a problem. If you have accurate understanding of the sensitivity and specificity of a test, then you can take action to maximise the utility of the expected outcome.
This means, for example, not necessarily performing a biopsy just because something is there (but it's probably benign). The cost of invasive procedures such as biopsies, and risk-increasing tests such as CT scans has to be balanced against their benefits.
Of course, there is no risk with a full-body MRI scan. The machine itself is expensive, but otherwise it nominally only costs electricity, maintenance and operation. If full-body scans were, as you say, routine, the cost of the machine would be amortised and an economy of scale should prevail.
The problem is legal; if a doctor sees something on a scan and doesn't take action, they can be sued. This is the case even if there is negative expected utility from performing the test due to the low probability of an actual malignant condition. Due to this, it is not uncommon to hear doctors say that full-body scans do more harm than good.
Until the legal system starts to account for rational choices based on statistical information, this problem is going to persist.
> Of course, there is no risk with a full-body MRI scan.
Not quite true. While the scan itself isn't harmful, the strong magnetic fields, RF fields and noise have caused several accidents, some fatal. Compare for example http://www.auntminnie.com/index.aspx?sec=sup_n&sub=mri&pag=d... "The FDA's Manufacturer and User Facility Device Experience (MAUDE) database documented a total of 40 incidents for MRI in 2004. In 2009, the total more than quadrupled, to 193 reported adverse events."
> False positives aren't inherently a problem. If you have accurate understanding of the sensitivity and specificity of a test, then you can take action to maximise the utility of the expected outcome.
False positives can still cause huge psychological strain on the involved person(s).
> Until the legal system starts to account for rational choices based on statistical information, this problem is going to persist.
The legal system does account for this. To succeed in a malpractice action, a plaintiff must prove a defendant failed to meet the required standard of care, which means the defendant failed to exercise that degree of skill and learning ordinarily exercised under the same or similar circumstances by members of the defendant's profession.
If these "rational choices based on statistical information" you refer to are within the accepted standard of care, a doctor is not going to be successfully sued for doing so.
> False positives aren't inherently a problem. If you have accurate understanding of the sensitivity and specificity of a test, then you can take action to maximise the utility of the expected outcome.
In many cases, the optimal action is not to test in the first place. "Overdiagnosed" is an interesting book on the subject by Dr. H. Gilbert Welch. A recent piece by him in the NYT gives an idea of his thesis:
http://www.nytimes.com/2014/11/06/opinion/an-epidemic-of-thy...
If full-body scans were, as you say, routine, the cost of the machine would be amortised and an economy of scale should prevail.
That's not really true. The cost of the machine is amortised by having it be in use at all times. In my experience MRI's are already fully utilized. I've had 3 in the past year. Even during Diwali there was always someone immediately before and immediately after me, and I never have much choice about appointment times.
Google indicates you can pay up to $3M for a MRI machine and $100K/yr energy and required maint costs. Google indicates a median radiologist gets $300K/yr and to make the figures ridiculous I'll dedicate one radiologist full time to the MRI. Google indicates mean lifetime of a MRI machine is a decade. Google indicates a hospital remodel to make a MRI safe room and electrical and cooling and ventilation might exceed the cost of the machine maybe $4M. Google indicates budgeting 3 hours per patient is not totally unreasonable. There is maint downtime and testing and the radiologist probably wants to sleep occasionally, even if there are actual specific scans that only take 15 minutes.
So the total cost over the decade lifetime of a MRI machine
3e6 + 4e6 + 300e3 * 10 + 100e3 * 10
You don't need a calculator to add that up to a cool eleven million bucks for ten years of MRIs. However you divide that out and thats only three grand per day over that decade. Or if you assume a long term average of 3 hours per scan (a bit on the high side) thats $375 total cost per scan.
Google reports the average billing is somewhat under two grand, but my math shows about four hundred bucks. Even assuming only 50% utilization they can still only account for $800 per scan.
I'm thinking that despite the mantra / astroturf that MRI is really expensive to provide, its actually a huge profit center for a hospital. Not as bad as the proverbial $35 aspirin pill, but pretty profitable none the less.
Another way to run the numbers is if the fixed cost of having the machine and radiologist available is about $3K/day, and they get about $2K/scan in revenue, you may have been part of a long MRI train that day, but the average machine actually only scans 3 people every 2 days on long term average.
Sounds like a market that's either ready for disruption, or so well protected by regulation that it can't be disrupted.
A single machine is going to have multiple technicians and service more than 1 doctor.
A little bit of trying to find numbers suggests that 1 MRI machine might be feeding 20 doctors (if you've arranged for 56 scans a week to be the low end number and take into account specialization, this starts to make sense). I guess they are each spending a couple hours during many of their weeks.
You have labor at ~30% of cost, it looks like it should be more than 50%, and could easily be even more.
"You have labor at ~30% of cost, it looks like it should be more than 50%"
That's because I chose the highest possible machine price and highest possible room construction price I could find on google. Its quite possible an individual installation could get away with a merely average $2M machine and basically nil building remodeling costs, in which case the labor cost percentage would naturally dramatically increase.
I don't trust the google data I found of $100K yearly for energy and service contract. That sounds low. 10% of purchase price sounds believable for high tech devices.
I googled some MRI tech salary data and one tech for a decade is going to be $1M. So merely not installing marble and solid gold fixtures in the room would easily pay for a team of three full time techs and one radiologist doing absolutely nothing but supporting one machine. Which is pretty ridiculous, but I want to estimate high to make my overall conclusion strong.
None the less my overall conclusion stands that it would be a significant challenge to spend, both labor and capex, more than about a third of MRI sales price to provide the service. Looks like quite a nice profit center.
The price can certainly come down. It cost me 7500rs for an MRI, 10,000 with contrast (about $150). I'm only disputing that increasing demand will reduce prices due to amortization.
This means, for example, not necessarily performing a biopsy just because something is there (but it's probably benign). The cost of invasive procedures such as biopsies, and risk-increasing tests such as CT scans has to be balanced against their benefits.
Of course, there is no risk with a full-body MRI scan. The machine itself is expensive, but otherwise it nominally only costs electricity, maintenance and operation. If full-body scans were, as you say, routine, the cost of the machine would be amortised and an economy of scale should prevail.
The problem is legal; if a doctor sees something on a scan and doesn't take action, they can be sued. This is the case even if there is negative expected utility from performing the test due to the low probability of an actual malignant condition. Due to this, it is not uncommon to hear doctors say that full-body scans do more harm than good.
Until the legal system starts to account for rational choices based on statistical information, this problem is going to persist.