In 1855, European statisticians gathered in the stately chambers of France's Corps Législatif. There on the banks of the Champs-Elysees, they made the world's first formal attempt to list every single way a person could die.
More than 150 years later, in a beige and windowless hotel ballroom thousands of miles away, hundreds of American medical coders are diligently chipping away at the exact same task. They've set out to master the nuances of the sprawling ICD-10, formally known as the Tenth Edition of the International Classification of Diseases.
“It gives me a heart attack just thinking about it,” Louisa Reolubin said with a sigh. The grandmother, at a three-day training for medical billers, uses green and pink highlighters to mark important new instructions on how to use the code set.
Reolubin is one of the country's 186,000 medical coders who work in the back offices of hospitals. It's her job to translate doctors' scribbles into diagnosis codes. Those are sent to insurance companies, which use them to determine how much to pay hospitals for care.
For decades now, Reolubin — and the rest of the American health-care system — has relied on an older version of this same medical compendium, ICD-9.
With 14,000 codes, ICD-9 seems puny by comparison. The new manual explodes that code set to 68,000 much more granular and detailed terms to define — very exactly and specifically — what ails us.
Americans spend $2.8 trillion each year on medical care. These codes determine how that fortune — nearly one-fifth of the nation's economy — gets divvied up among thousands of hospitals and the doctors who work there. It is how the federal government, and most private health insurers, assess a value for each patient visit.
The prospect of quadrupling the number of medical codes used in those calculations has touched off a heated debate over whether more specificity is an onerous layer of bureaucratic red tape or a valuable chance to better understand and treat complex medical conditions.
The codes in ICD-10 can seem absurd in their granularity, replete with designations for seemingly impossible situations.
There are different numbers for getting struck or bitten by a turkey (W61.42 or W61.43). There are codes for injuries caused by squirrels (W53.21) and getting hit by a motor vehicle while riding an animal (V80.919), spending too much time in a deep-freeze refrigerator (W93.2) and a large toe that has gone unexpectedly missing (Z89.419).
At a conference in San Francisco, the AAPC, an organization for billing and coding professionals, sold shot glasses inscribed with “F10.950”³ — the code for an unspecified alcohol-induced psychotic disorder. “Give ICD-10 a shot!” it says in blue script.
Hospitals and insurers have fought the new codes, calling them a massive regulatory burden that will cost them billions of dollars to implement without improving patient care. For years, their protests succeeded: The federal government has twice delayed implementing the new code set, which was initially set for 2008.
ICD-10 proponents contend that adding specificity to medical diagnoses will provide a huge boon to the country. It will be easier for public health researchers, for example, to see warning signs of a possible flu pandemic — and easier for insurers to root out fraudulent claims.
“How many times are people going to be bitten by an orca? Probably not very many,” said Lynne Thomas Gordon, chief executive of the American Health Information Management Association. “But what if you're a researcher trying to find that? You can just press a button and find that information.”
Gordon notes that the United States relies on the last edition, written in the 1970s. “We are so far behind we can't compare data with other countries,” she said.
Still that doesn't mean the transition will be easy. Hospitals and insurers say they have spent billions updating technology and training medical coders to make the change. And after watching the botched rollout of Healthcare.gov last year, industry officials have expressed concern that the government hasn't done enough testing.
“The worst job you can have in health care right now is being a hospital administrator,” Angela Boynton director of ICD-10 adoption and training at UnitedHealth Group, who spoke at the San Francisco AAPC training session. “You're shaking in your boots not just about training and testing but also that your entire in-patient payment system is being overhauled by Medicare.”
Gordon wears a lapel pin, sometimes, that says Z56.6. It's the code, she explains, for “other physical and mental strain related to work.”
The ICD-10 manual is thick, about the size of a phone book. Printed in minuscule type on newsprint-thin paper, it weighs five pounds and includes more than 1,100 pages of medical procedures and ailments. The index alone — the guide to figuring out where to find the right code — is 421 pages.
Two key factors help explain the explosion in medical codes. First, ICD-10 adds in the ability to differentiate between left and right sides of the body. This can help insurers, for example, to root out fraud. A hip replacement on both the left and right side might not raise any red flags, but two hip replacements on the left side probably would.
Second, the new codes categorize whether a trip to the hospital was the first round of treatment or a subsequent encounter. This is important for reimbursement purposes, as first visits to the doctor tend to require more resources.
Whether this specificity improves the medical system is a subject of fierce debate in health technology circles. Opponents argue that the new larger set will slow productivity, making it more difficult for veteran billers to find the right code in a sea of parrot injuries and turkey bites.
Most other industrialized nations transitioned to ICD-10, which the World Health Organization published in 1992, more than a decade ago. The switch can take years because most countries come up with a slightly modified version of the code set that best suits their needs.
When Canada adopted ICD-10 in 2001, one study of a Toronto hospital system showed that productivity fell by half. Before ICD-10, medical coders could get through 4.62 charts in an hour. Right after the transition, that fell to 2.15 charts per hour. One year later, productivity had partially rebounded to 3.75 charts per hour.
“If you look at Canada's transition, there were some longer term cost impacts that went well beyond the transition itself,” said Michael Nolte, chief operating officer of technology firm MedAssets. “There's some evidence that there will be a long-term effect.”
One study funded by the American Medical Association estimated that it could cost doctors' offices $56,000 to $8 million to transition to ICD-10, depending on the size of the practice. The AMA, one of the larger groups opposed the switch, is still petitioning the federal government to reverse course.
“Adopting ICD-10, while it may provide benefits to others in the health-care system, is unlikely to improve the care physicians provide their patients and takes valuable resources away from implementing delivery reforms and health information technology,” the trade group wrote in a Feb. 12 letter to Health and Human Services Secretary Kathleen Sebelius.
Others contend that the change in productivity won't be as dramatic — that opthalmology coders could just stick to the ophthalmology section, for example, and don't have any reason to get bogged down in codes about parrots. Health insurers don't care if a bite came from a parrot or a turkey — they just want to know what type of medicine they're paying for when the hospital treats it.
“No individual has to use the whole thing,” said Martin Libicki, a researcher at RAND Corporation. “If you're working with an eye doctor, God knows why you'd learn the codes for broken legs. But if someone showed up with a broken leg, you would just look it up.”
Libicki authored a major RAND Corporation study in 2004 — when the Bush administration was first studying the transition — that estimated the potential benefits of switching to ICD-10 outweighed the costs by as much as $4.5 billion.
Much of this comes from increased specificity in coding, which makes it easier to accurately pay hospitals for the care they provide and reduces opportunity for fraudulent billing.
“If you have ICD-10, you have an enormous increase in precision,” said Richard Averill, senior vice president of clinical and economic research at 3M Health Information Systems, recalled. He has worked in the medical coding world for decades, and his company has a key federal contract to help run the ICD-10 transition. “Yes, there's an adjustment, but two years later you've gotten rid of a lot of that paper chase.”
In a more precise coding system, researchers see the potential to better track the quality of medical care that patients receive. Billers can denote whether a visit to the hospital is a first, second or later trip — which could indicate the severity of the condition.
Nearly everyone agrees that there is at least one compelling reason to switch to ICD-10: As new medical technologies have come online and demanded new codes, ICD-9 has run out of space. The capacity for noting cardiology procedures (assigned, in ICD-9, by codes that begin with “37”³) was exhausted in the early 2000s. That created a patchwork scenario, where new cardiology codes show up elsewhere in the code set, with little rhyme or reason.
“The consequence is very disruptive,” said Christopher Chute, a professor at the Mayo Clinic and expert on medical classification. “It's like they're renovating a city, and assigning addresses at random. That makes it a lot more difficult to find the right house.”
In 2012, Chute wrote an article in the journal Health Affairs advocating for delaying the ICD-10 implementation. He has serious doubts about whether the new codes will improve the medical system. But he also doesn't see any better option right now: The code set the country currently uses has no space left to grow.
When European statisticians and doctors wrote the first International Statistical in 1855 — the first draft of the medical code system used today — they were seeking an international standard to use in public record keeping for death certificates.
Even in 1855, there were fights about how to categorize medical knowledge. Drafters quarrelled about the best way to group maladies.
“There was a question of, are you going to have the diseases of the heart and the lung, or are you going to have diseases like inflammation and cancer, and which mode would dominate,” Chute said. “They settled on anatomy, and that's still how we organize ICD today.”
That first draft included 139 causes of death, grouped into larger categories like “stillbirths” and “ill-defined diseases.”
Updates ensued, decade by decade, as the world's scope of medical knowledge grew. ICD-3, published in 1922, included the first mention of parasitic diseases. In 1938, ICD-5 made room for deaths caused in transportation accidents and a decade later ICD-6 included the first mention of personality disorders.
In 1979, the United States adopted the ninth edition of the International Classification of Diseases, which the World Health Organization had published two years earlier.
“That was a non-event,” Averill said. “It was not being used to the expansive purposes that the coding system is being used for today.”
That all changed when Medicare started to lean heavily on ICD codes to figure out how much hospitals should be paid. Picking the right code became crucial to how much a doctor is reimbursed. If medical billers don't submit the right code, they could end up in a back-and-forth with Medicare or a private health plan over whether the doctor's treatment was justified.
Private health insurers quickly followed Medicare's lead, and soon ICD was transformed from a public-health record-keeping system into a $2.8 trillion industry's billing system.
“It's now equally important for private payers, in terms of a backbone of how bills get paid,” Nolte said. “It's just as fundamental.”
When the Centers for Medicare and Medicaid Services first explored a move to ICD-10 more than a decade ago, health insurance plans began diligently preparing. Medical billing trainers started developing their ICD-10 curriculum. No one imagined they would still be getting ready a full decade later.
Nobody in the medical community is quite sure what will happen on Oct. 1, when the federal government flips the switch on this new system.
“I think it will be a non-event in the same way Y2K was,” Gordon, the woman with the workplace stress pin, said. “I have such confidence in our health-care providers. They're not going to enjoy it, but they'll be ready.”
Others aren't quite as sanguine.
“The difference is Y2K was only a technical issue,” Nolte, of MedAssets, said. “You didn't have to ask anybody to do anything different. But here you have a culture change, where you're teaching thousands of people to do something that's somewhat foreign to them.”
The federal government has undergone a massive data mapping project, figuring out which codes from ICD-10 will replace each and every code from ICD-9. Technology firm 3M, where Averill works, has one of the major contracts to complete that process.
This month the agency announced it would hold a testing week in March, where hospitals can check if their new ICD-10 claims make it to the federal government. While those are the only testing plans in place for the moment, the agency says it's confident that it will be able to handle the new codes come October. More of the concern tends to center on smaller, private health insurance plans, which don't have the resources of the federal government to prepare.
Regardless what happens this year, even more change is in the works: In 2007, Chute, at the Mayo Clinic, began leading the World Health Organization's efforts to develop the Eleventh Edition of the International Classification of Diseases, or ICD-11. He expects that to come into use in the United States sometime around 2022.
“I think we've had about tens of thousands of person hours put into this already,” Chute said. The effort relies on hundreds of committees with thousands of doctors around the world, each leading experts in their medical specialities.
Reolubin made it through the San Francisco training. It's nearly certain she won't be around for the next ICD upgrade — and she says, only partially joking, she thinks sometimes about skipping this one, too.
“I keep telling my boss, 'I'll just retire,' ” she says, as she highlights her new, ICD-10 code book. “I've done this for long enough.”