By Dr. Kate Scannell, Syndicated columnist
First Published in Print: 09/18/2011
Sherlock Holmes famously partnered with his Doctor Watson to help solve the most baffling mysteries of his day. Soon, doctors in clinics will be partnering with another famous Watson, ostensibly to help solve diagnostic conundrums for patients.
Last week, IBM and health insurer WellPoint Inc. jointly announced that supercomputer Watson -- best known for trouncing human contestants on the TV game show "Jeopardy!" -- was being put to work as a doctor's aide. Or as an insurance agent. Maybe both. It's hard to compute just yet.
According to the announcement, Watson's new role is to provide nearly instant medical diagnoses for patients and to recommend treatment options. Reportedly, he (or she?) will be able to sort through patients' records and vast amounts of medical literature -- and insurance claims data -- to provide these solutions within seconds, thusly improving patient care and saving money.
Still, WellPoint stresses, Watson will not be used to make decisions about reimbursing patients and doctors for the cost of care, nor will it be used to deny patients expensive treatments -- all a bit hard to compute when Watson's recommendations are so intimately informed by the insurance company's claims data. Presumably, there will exist some kind of hard-drive partitioning in Watson's soul that will completely separate its medical recommendations from the monetary or profit concerns of its health-insurer parents.
So what happens if a human doctor disagrees with Watson's diagnosis or therapeutic recommendation? Some frustrated doctors might be tempted to pull the plug on him. But a corporate spokesperson for Watson cautions us not to worry -- that if a doctor veers from Watson's guidance, the insurance company will not automatically base its coverage decision on Watson's lone opinion. Rather, it would ask one of the clinicians working for Watson's boss to review the case.
In fairness, it is possible -- but as yet unproven -- that Watson could improve clinical efficiency and reduce diagnostic error. He could reliably select the most effective treatment options and provide authoritative guidance about rare diseases. That's all good for patients, doctors and society at large.
Still, as we know, Watson is only as good -- or bad -- as the data he's fed. And even he consults Wikipedia. The decisions he makes are rooted in his programming, and his programming is controlled by human choices. Time will tell whether Watson puts us in jeopardy or becomes the saving new face -- or screen -- of medicine.
Intriguingly, Watson's official clinical debut in early 2012 will occur at several cancer centers across the country. That's an interesting and tricky place to begin his new job. "Interesting" because of the high personal and societal costs of cancer care. To that point, the National Institutes of Health projects that by 2020 medical expenditures for cancer will reach at least $158 billion. And "tricky" because great controversy exists about the "value" of many cancer treatments. Is a $100,000 cancer drug "worth" its cost to insurers when it might prolong a patient's life by a week, or a month, or ... what?
In essence, "WWWD?" -- What would Watson do? -- when faced with a terminally ill patient's plea for a costly last chance at a long-shot hope?
Personally, I think that Watson's greatest opportunity for clinical celebrity will become manifest through his skill in locating billing codes for insurance claims.
Currently, hospitals and doctors use a billing system -- the "ICD-9" -- that includes about 18,000 codes that can be used to document and bill insurers for provided care. For example, after seeing a patient with sinus problems, a doctor might search the ICD-9 and choose Code 461.0 for "Acute maxillary sinusitis," which she sends to the insurer as a billable claim for her services.
As a mere human doctor, I must admit that keeping "up to code" with this existing system is already daunting. Still, a modernized new system -- the "ICD-10" -- is scheduled to take effect in 2013, and it greatly expands the number of possible codes to 68,000.Each code is three-to-seven characters in length. As the American Medical Association cautions: "Due to the increased number of codes, the change in the number of characters per code, and increased code specificity, this transition will require significant planning, training, software/system upgrades/replacements, as well as other necessary investments."
I, for one, will be embracing Watson's authority for coding. Even if I ever actually encountered a patient who had been "struck" by a parrot or a chicken (and would admit to it), I doubt I would have imagined the existence of relevant new codes with which to bill (W6102XA -- Struck by parrot; and W6132XA -- Struck by chicken).
Reading the new coding system is not for the hypochondriac or faint-of-heart. It suggests too many possible human harms and infirmities, the stuff of nightmares.
You could, for example, be crushed by reptiles in an "initial encounter" (W5983XA) -- as well as in a "subsequent episode" (W5983XD) -- and, perhaps not surprisingly, suffer some sort of bad sequela to boot (W5983XS).
Separate codes exist for initial -- and subsequent!!! -- encounters with avalanches, mudslides, landslides, tornadoes, volcanic eruptions, hurricanes, dam collapses, blizzards, dust storms, and even "prolonged stay in a weightless environment."
Almost 100 codes exist for "terrorism." Fully 70 codes delineate possible occasions for "spontaneous" rupture of tendons. How, you might nervously wonder, does anyone obtain an open bite of the trachea (S11025A) or vocal cord (S11035A)? Numerous codes delineate various poisoning deaths or injuries, including harms from laxatives.
Obviously, the massive new coding system could prove very inspirational for mystery and crime writers seeking unique story lines involving esoteric human wrongdoing. Watson -- unplugged from his day job -- could become quite the scribe if allowed to work creatively with the codes. Try it yourself at home -- see how rewarding it can be to randomly pick a few ICD-10 codes from the bunch and string them together in a story:
The R452 (Unhappy) and R4581 (Low self-esteemed) medical examiner had yet to arrive at the crime scene. But her usual R464 (Slowness and poor responsiveness) could be explained -- this time -- by the V309XXA (Unspecified occupant of three-wheeled motor vehicle injured in collision with pedestrian or animal in traffic accident) she'd encountered en route.
Still, hardscrabble Detective Buzz Blick was not full of R4582 (Worries). While K131'ing (Biting his lip and cheek) and staring at the dead body covered with bird feathers and orthodontically marked by unique M2632 (Excessive spacing of fully erupted teeth) -- he realized what had happened.
The answer arrived as forcefully as a hoof strike to his head (W5532XA) -- again (W5532XD) and yet again (W5532XS) -- and it presented itself as incisively as a bite from a pig (W5541XA) or a dolphin (W5589XS). When he put it all together, it made perfect sense -- no R462 (Strange and inexplicable behavior) here! The oddly nibbled toast in the dead man's hand suggested a recent X151XXS (Contact with hot toaster) or, perhaps, an X101XXS (Contact with hot food) -- it all depended on your personal code. And the spots of green paint across the body spelled out an obvious X9502XA -- assault by paintball gun discharge.
Clearly, the vic' had been fatally Z563'd -- pecked to death by turkeys -- turkeys R4587 and R463 (impulsive and overactive, respectively) with a craving for wheat bread and a mean need for artistic self-expression.
=====================================================
Kate Scannell is a Bay Area physician and the author of the novel "Flood Stage."
Copyright 2011, Kate Scannell