ICD-10: Medical Voyeurism or Just A Lust for Complexity?

We were going to write about ICD-10 anyway, but a recent article in The Wall Street Journal (subscription required) beat us to it.  But first, what is ICD-10?  Those of you who may not be fluent in Medispeak deserve a 60 second explanation.

“ICD” is the abbreviation for the International Classification of Diseases.  It’s a tabulated and coded compendium of the various things that can go wrong with people and cause morbidity and mortality.  The idea of categorizing and classifying illness is an old one, and the first iteration of the ICD was published as the International List of Causes of Death in 1893.  Just as we have universal classifications of plants, animals, fungi, bacteria and everything else, the purpose of the ICD is supposed to be to help people organize information so that we can better understand the world and improve our lives.  The ICD is, for example, used to compile information on the causes of death and disability.  It’s used to find and track epidemics of diseases, compare the distribution of illnesses to geographic locations and match resource use to specific diseases.  The invention of computers marked an enormous change in the use of the ICD; their ability to process and tabulate enormous numbers of codes simultaneously made it easy and painless ask increasingly detailed – and even esoteric – questions, as long as enough data was available in enough detail.  Want to know if there’s a relationship between smoking and cancer?  An analysis of ICD codes could tell us that.  And there is one other important use for ICD coding; the federal government and private health insurers require that the appropriate ICD codes be included as a component of each and every medical claim submitted for payment.

The World Health Organization (WHO) took over responsibility for maintaining and updating the ICD in 1948, and it’s been through many versions since then.  ICD-10 is, as the name implies, the tenth version to be published, and the WHO released it in 1994.  One of the big changes between ICD-10 and ICD-9 is the ability to add lots of new diagnoses.  ICD-9 codes consisted of up to five numbers.  Theoretically this allows it to represent up to 100,000 different diseases and conditions.  In contrast, ICD-10 is represented by a combination of up to seven letters or numbers, giving it the ability to code for over 78 billion different conditions.  Despite this, the WHO version of ICD-10 is relatively manageable and consists of only 16,000 different codes.

Here in the United States, the medical community is generally still using ICD-9, but because of its legal jurisdiction over all aspects of medical information under the Health Insurance Portability and Accountability Act of 1996 (HIPAA), the federal government has the power to require that every physician, medical facility and provider in the country use whatever version of the ICD it mandates.  Eager to keep America’s healthcare system in line with the rest of the world, the federal government decreed that all U.S. healthcare providers must switch over to ICD-10 by October 1, 2013.

Hey, what could be wrong with that?  Lots, if you leave the implementation of the code set to be used up to a bunch of people with a penchant toward the academic and the bureaucratic.

You see, every country is free to modify WHO’s version of the ICD to meet its own particular needs.  Canada has its own version called ICD-10-CA.  Australia uses ICD-10-AM, while Germany uses ICD-10-GM.  So the committee that the Department of Health and Human Services (HHS) set up to adapt ICD-10 for domestic use felt free to add lots of new codes to the U.S. version.  Some 85,000 of these were created for to represent procedures done in the hospital, a subset called ICD-10-PCS.  But while U.S. clinicians only had to deal with 16,000 diagnostic codes in ICD-9, HHS packed the ICD-10 diagnostic set with some 70,000 codes and called it ICD-10-CM.  In total, American healthcare providers are now required to accommodate up to 155,000 new codes for billing and documentation purposes into their businesses.  The recent article in the Wall Street Journal tells the story:

“Indeed, health plans may never again wonder where a patient got hurt. There are codes for injuries in opera houses, art galleries, squash courts and nine locations in and around a mobile home, from the bathroom to the bedroom.

Some doctors aren’t sure they need quite that much detail. ‘Really? Bathroom versus bedroom?’ says Brian Bachelder, a family physician in Akron, Ohio. ‘What difference does it make?’…

Some codes could seem downright insulting: R46.1 is ‘bizarre personal appearance,’ while R46.0 is ‘very low level of personal hygiene.’

It’s not clear how many klutzes want to notify their insurers that a doctor visit was a W22.02XA, ‘walked into lamppost, initial encounter’ (or, for that matter, a W22.02XD, ‘walked into lamppost, subsequent encounter’).

Why are there codes for injuries received while sewing, ironing, playing a brass instrument, crocheting, doing handcrafts, or knitting—but not while shopping, wonders Rhonda Buckholtz, who does ICD-10 training for the American Academy of Professional Coders, a credentialing organization.

Code V91.07XA, which involves a ‘burn due to water-skis on fire,’ is another mystery she ponders: ‘Is it work-related?’ she asks. ‘Is it a trick skier jumping through hoops of fire? How does it happen?’”

Much of the new system is based on a World Health Organization code set in use in many countries for more than a decade. Still, the American version, developed by the Centers for Disease Control and Prevention and the Centers for Medicare and Medicaid Services, is considerably more fine-grained.

Y93.J4: A code for injuries received while playing brass instruments.

The WHO, for instance, didn’t see the need for 72 codes about injuries tied to birds. But American doctors whose patients run afoul of a duck, macaw, parrot, goose, turkey or chicken will be able to select from nine codes for each animal, notes George Alex, an official at the Advisory Board Co., a health-care research firm.

There are 312 animal codes in all, he says, compared to nine in the international version. There are separate codes for “bitten by turtle” and “struck by turtle.”

Of course for the really curious even these codes still don’t tell the full story for those struck by turtles.  Were they struck vertically as the turtle fell from the sky, horizontally as a turtle was hurled at them, or bowled over by a turtle sprinting at top speed?  And while the Wall Street Journal stops there, let us assure you that it’s not for lack of content.  Take a look at this:

V9540XA  Unspecified spacecraft accident injuring occupant, initial encounter

V9541XA  Spacecraft crash injuring occupant, initial encounter

V9542XA  Forced landing of spacecraft injuring occupant, initial encounter

V9543XA  Spacecraft collision injuring occupant, initial encounter

V9544XA  Spacecraft fire injuring occupant, initial encounter

V9545XA  Spacecraft explosion injuring occupant, initial encounter

V9549XA  Other spacecraft accident injuring occupant, initial encounter

Someone, somewhere in the HHS ICD committee is far more interested in space travel than the lowly problems of the average Medicare recipient.

And lest anyone be concerned that the ICD-10-PCS system is any less rigorous in its specificity, never fear.

“’You have millions of transactions flowing in the health-care system and this is an opportunity to mess them all up,’ says Jeremy Delinsky, chief technology officer for athenahealth Inc., which provides billing services to doctors.

Medicare officials say they believe many big insurers and hospital systems are making preparations, but there may be some issues with smaller ones that won’t be ready.

With the move to ICD-10, the one code for suturing an artery will become 195 codes, designating every single artery, among other variables, according to OptumInsight, a unit of UnitedHealth Group Inc. A single code for a badly healed fracture could now translate to 2,595 different codes, the firm calculates. Each signals information including what bone was broken, as well as which side of the body it was on.”

Predictably, the people that created the U.S. version of ICD-10 claim that it’s all needed, and that they got lots of input from real, live doctors.  Well, sort of.  Here’s the HHS response to the “myth” that it was developed “without clinical input”:

“The development of ICD-10-CM/PCS involved significant clinical input. A number of medical specialty societies contributed to the development of the coding systems.”

Any doctor these days can tell you that few “medical societies” have anything in common with practicing physicians these days, except that they stick up clinicians for membership dues and fees for mandated certification and continuing medical education.  But really, why did someone go to all of this trouble to find out if patients were “struck by orcas” multiple times and frankly, why should we care?

Let’s return to planet Earth for a minute and think about these codes and how they’re used to deliver or improve healthcare services in real life.  Let’s say that we have a patient in front of us in the clinic or the emergency room.  Do we need to know the patient’s ICD-10 codes?

Well, no.  Seriously, we don’t.  What we’re really like to know are the patient’s past and current diagnoses and treatments.  We don’t need the ICD code for that unless that’s the specific method used to store the information.  We could really manage just as well with a simple list of these things spelled out in plain English or medical terms.  In fact, that method has worked successfully for hundreds of years.  If someone wishes to represent those pieces of information by one code or another it’s all well and good, but no doctor on Earth really thinks in terms of their patient manifesting a “W5801XA”.  Instead, he just wants to know that they were bitten by damned alligator.  But if doctors don’t care about this, then who does?  How about epidemiologists and medical researchers?

There is no question that having large amounts of accurate coded data readily and rapidly available for computer processing would come in handy for medical epidemiologists and researchers who are trying to track down epidemics and public health hazards.  But the value of this information is limited by two competing factors: detail and reliability.  In a situation in which one is relying on others to collect information as a byproduct of their normal activities, the speed and accuracy of collection will be inversely related to the amount of detail we are asking them to describe.  “You were struck by a bird?  What kind of bird?  A brown bird.  How big was it?  What other markings did it have?  What color was its beak?”  It’s easy to understand that the more detail we request, the less abundant and accurate the information will become as a result of two factors.  The first is the ability of the patient to recall the details.  The second is the limited time, patience and energy of the clinician asked to compile and code the information.  Under these circumstances, asking for more and more data rapidly becomes a double-edged sword.  If we demand high levels of detail, there is good reason to question its accuracy.  If we’re not holding a gun to each doctor’s head, more and more of the data will be coded as such-and-such an injury, “unspecified”.

Even so, it’s hard to imagine that the needs of a relatively few epidemiologists and researchers is sufficient to justify the unbelievably expensive and complex disruption that switching to HHS’ version of ICD-10 is going to entail.  When all is said and done, we’re talking about an investment of literally hundreds of billions of dollars.  Every single piece of medical record, billing and accounting software in the industry has to be re-programmed, and none of the programming is the same as that already being done for the versions of ICD-10 used by WHO or any other country.  Every medical encounter form in every medical clinic has to be changed.  Every clinician and every biller needs to learn the new codes.  Sure, no one is likely to need all of them, but the sheer number and complexity of them is daunting.  So who the heck would want to create such a system, and why?

There are really only two possible suspects: medical voyeurs and those who stand to benefit from increasing the complexity (and ultimate cost) of the U.S. healthcare system.

Medical voyeurism is self-explanatory.  Somewhere in America, there is always going to a substantial group of people – be they academics, bureaucrats or tabloid journalists – who get a thrill out of being able to peek into the statistics and write articles about the number of people mauled by Macaws or put into the hospital when their water skis somehow burst into flame.  It is a big mistake to underestimate the power and influence of people who wish to collect and analyze data, simply on the off chance that there might be something interesting in it.  But who could possibly benefit from making coding more complex?  The answer is simple: health insurers, especially the federal government itself.

As we’ve extensively documented many times before, the RBRVS-based medical payment system invented at the behest of Congress and used by virtually all public and private U.S. insurers is a travesty of the first order.  In its current form it is hopelessly complex, prone to abuse by criminals and payers alike, inefficient and expense.  Both the ICD and CPT system represent major components of this system.  But while proponents of ICD-10 claim that the increasing levels increasing level of detail provided will make it easier to ensure “quality and affordability”, the reality is that increasing levels of complexity always increase rather than decrease the potential for abuse by payers and criminals alike.  There is no scam that cannot be perpetrated and hidden more easily in a complex system than in simpler one, especially by insurers intent on requiring that every “i” be dotted and “t” crossed before paying providers for their services.

As it is, public and private insurers both play games with reporting requirements and paperwork to avoid meeting their financial obligations.  In a recent post, Dr. Rich over at The Covert Rationing Blog accurately describes how Medicare uses the process of physician certification and re-certification as an excuse for denying payments owed for medical services legitimately provided in good faith.  Medical billing is already an arms race between the billing software used to submit claims based upon elaborate coding algorithms designed to maximize payments, and equally sophisticated software designed to deny payment.  All ICD-10 is going to do is take this to the next level, further impoverishing real healthcare in the process.  Insurers will ask for more detailed codes in order to justify payment, and considerable expense will be generated in providing them.  For patients, providers, businesses and taxpayers, it’s a no-win situation.

In January of 2011, President Obama pledged to reduce the amount of regulation sandbagging the U.S. economy, and “ordered a government-wide review of regulations, both old and new, in a broad push to curtail rules that retard job creation and economic growth.”  Frankly, the wholesale simplification of ICD-10 would be a good place to start.