Anyone trying to plan for a transition to ICD-10-CM (the United States' Clinical Modification of International Classification of Diseases, 10th Revision) must sometimes feel like Annie of the Broadway hit: "Tomorrow, tomorrow... You're always a day away!" Many standards-watchers expect the U.S. Department of Health and Human Services (HHS) to soon announce its intent to replace ICD-9, the present U.S. mortality and morbidity coding standard, with ICD-10, a far more precise system. But many also admit to having expected this notice before--even as early as 1996, when passage of the Health Insurance Portability and Accountability Act (HIPAA) pushed ICD-10 to a lower position on the HHS priority list.
Today, a spokesperson for HHS's Center for Medicare & Medicaid Services (CMS) says only that the agency continues to study the costs and benefits of a proposed rule. The expert consensus, however, is that healthcare IT leaders should be getting ready for ICD-10. "It's not a question of if," says John Quinn, Health & Life Sciences, Accenture, New York, who also chairs the American National Standards Institute's HL7 technical steering committee, "it's a question of when. This has been in development since the 1990s. It's not something that was developed last night on the back of a napkin."
"It's coming," agrees Washington, D.C.-based Richard Gundling, vice president of the Healthcare Financial Management Association (HFMA). "I don't have a good estimate of the date, but sooner rather than later."
What 10 has that 9 doesn't
Developed by the World Health Organization (WHO), the United Nation's public health arm, ICD-9 has been the U.S. standard for reporting on patient diagnosis and inpatient procedures since 1979. It's used both for studying clinical issues and for coding reimbursement claims to insurers. ICD-9 uses three to five alphanumeric characters to generate codes for classifying diagnoses and three or four numeric characters for medical procedures. That's not enough digits for the evolving understanding of disease and the multitude of new procedures developed during the past three decades.
"We've outgrown the coding system," Quinn says. "There isn't enough room to code all the detail that medical science has come up with. And the rest of the world has gone ahead of us."
ICD-10 is based on a classification logic that differs significantly from ICD-9. It utilizes up to seven alphanumeric characters with the mathematical potential to generate billions of codes. Currently, its diagnostic classification set (ICD-10-CM) includes about 120,000 codes, almost 10 times the number now used under ICD-9. And its procedure usage set (ICD-10-PCS) includes more than 200,000 codes, roughly 50 times as many as ICD-9.
The WHO first released ICD-10 in 1993. The National Committee on Vital and Health Statistics (NCVHS), the National Center for Health Statistics of the Centers for Disease Control and Prevention, and CMS (then HCFA) promptly contracted with 3M Health Information Systems, Salt Lake City, to develop detailed plans for migration to ICD-10. In 1999, the United States adopted ICD-10 for reporting deaths. However, the United States is still using ICD-9 for coding morbidity statistics, the only industrialized country to do so.
Support for the switch
The most aggressive advocate for a switch to ICD-10 is the American Health Information Management Association (AHIMA), Chicago. Linda Kloss, its executive director, says, "When you look across healthcare, it's remarkable how much we rely on coded data. It's ubiquitous. It frames our statistics. It drives reimbursements. It's a key component in research and drug trials. And the quality of the data simply isn't what it could be and should be."
Several professional organizations support a shift to ICD-10, including the Chicago-based HFMA, American Medical Association and American Hospital Association. In late July, testifying for AHIMA before a Congressional subcommittee, Kloss urged Congress to prod HHS to act on ICD-10 by early in 2006. She stated that the greater specificity of ICD-10 is needed to:
- Support initiatives to enhance healthcare quality and promote patient safety, including disease management and pay for performance
- Improve monitoring of public health issues, including bioterrorism threats
- Provide the basis for more accurate reimbursement rates
"I think one of the big benefits," says Nancy A. Larson, vice president of 3M Health Information, "is being able to drive fairer payments because you have more specificity in coding. The system can capture procedures involving medical device innovations that can't be accommodated in ICD-9, permitting more accurate reimbursement."
For one of the nation's largest healthcare organizations, the Veterans Health Administration (VHA), Washington, D.C., a shift to ICD-10 can't come soon enough. Director of Health Data and Informatics Gail Graham points out that the VHA is both a provider and a payer, one that pays out more than $2 billion per year and accepts outside claims electronically.
"We use these codes for a lot of other purposes than billing," Graham says. "We use them to evaluate clinical interventions. We have an electronic health record, and ICD-10 gives you the granularity to make clinical reminders a little more sophisticated than they are today. Some of the ICD-9 codes lump several different diagnoses in one bucket. We try to tailor reminders to the patients as much as we can, and we really hope that ICD-10 will help us refine that."
Enthusiasm well under control
One major stakeholder in coding issues, the Chicago-based Blue Cross Blue Shield Association (BCBSA), remains unconvinced. A study it commissioned concluded that the total cost of migrating from ICD-9 to ICD-10 could reach $14 billion over a three-year period. BCBSA also contends that everyone in healthcare is already coping with other pending or expected HIPAA regulations and doesn't need another technical change at present.
"Now is not a good time for HHS to mandate ICD-10 as a required code set under HIPAA," says Joel Slackman, managing director for BCBSA's Office of Policy and Representation. "All of the algorithms and computer programs to track fraud are based on current ICD-9 data. It would take a two- to three-year period at least to go back and update all the systems developed to track fraud and patterns of abuse. And all of those costs are going to divert resources, money and programming expertise away from the top priority, which is establishing a nationwide network of health information technologies. Investing in ICD-10 is going to pull money out of those activities."
BCBSA is not saying ICD-10 should never be implemented, Slackman clarifies. But at the moment, it doesn't believe that ICD-10 "rises to the priority that would justify spending several billion to $14 billion over the next few years."
Advocates for ICD-10 cite a study published in 2004 by the RAND Corp., Santa Monica, Calif., conducted at the request of NCVHS. RAND estimated total costs of migrating to ICD-10 at somewhere between $425 million and $1.2 billion. Short-term costs of implementing a new code set involve IT system changes, training (primarily of coders, but also of physicians), and loss of productivity during the transition, according to RAND. Its analysts concluded that the total expense of IT system changes will likely be somewhat higher for payers than for providers, and that costs of reduced productivity, although much smaller than system change or training costs, will hit physicians hardest.
RAND also estimated benefits of $700 million to $7.7 billion. Among other benefits, RAND expected the increased precision of ICD-10 codes to reduce miscoded, rejected and improper reimbursement claims and lead to a better understanding of the value of new procedures, to the ultimate benefit of both providers and payers.
BCBSA believes that RAND both underestimated costs and overestimated benefits. By way of historical comparison, Slackman notes that the average cost to small and medium-size health plans of adopting HIPAA transaction and code set standards was roughly 10 times initial HHS estimates. He adds that existing contracts based on code definitions and their associated reimbursement rates often will have to be renegotiated, "an expensive and time-consuming exercise for payers and providers alike."
Slackman concedes that "a real three years" between an initial notice of proposed rule-making and mandatory compliance should be adequate for most organizations. (A notice of proposed rule-making issued this fall could have a compliance deadline of October 2008.) But he cautions that providers may face more complex transition problems than payers because their IT systems must meet more kinds of needs.
AHIMA's Kloss takes strong issue with the BCBSA position. "It's not clear what we're waiting for," she says. "They [the payer industry] are really one of the beneficiaries of better data. When you consider that all claims adjudication is based on a system that was designed 35 years ago, there certainly is more knowledge to be gleaned from coded diagnostic procedure data than they can access using today's archaic data tools."
Detour or down-ramp to the EMR?
BCBSA's position is that ICD-10 is a diversion of resources from promoting wider use of electronic medical records (EMRs), whose internal data standard is the Systematized Nomenclature of Medicine (SNOMED). Those on the other side of the argument say that use of ICD-10 is an almost essential step toward wider use of EMRs.
ICD-9 and ICD-10 are output systems--ways of coding data for purposes like measuring quality of care, assessing demand for medical services and, of course, reimbursement. SNOMED, on the other hand, is an input system that is designed for electronic use within EMRs to capture highly granular clinical detail during the entire course of patient care. The National Library of Medicine entered into a five-year agreement in 2003 to license use of SNOMED Clinical Terms (SNOMED-CT) in the United States.
Although the systems are designed for different purposes, mapping SNOMED codes to ICD-10 data will be easier and far more useful than to ICD-9 data, experts say. "This is a substantial change in syntax," Quinn says. "The difficult part in writing specs for an electronic health record system is: What does the hospital's electronic health record system have to do in order to interact with the environment outside the hospital--the regional and national networks--to create what we're calling the portable electronic health record? Clearly ICD-10 is one of the pieces on the path. The worst thing we could do is implement electronic health records with ICD-9 and then find out that we need to be chipping around the foundation."
Technical issues
By the time this article appears, HHS may well have proposed rules on claims and referrals attachments, a technically demanding problem. Quinn says that healthcare organizations and vendors can, in the process of complying with that standard, complete many of the tasks needed to prepare for migrating to ICD-10. "Why," he asks rhetorically, "would you want to do it twice?"
3M Health's Larson and others advise CIOs to conduct gap analyses to evaluate all IT systems that hold or transmit data with ICD-9 codes. A key objective is to determine how much redundancy, if any, is needed to continue access to old data. Timetables for new system purchases and software upgrades can be planned accordingly.
From a technical perspective, Larson says, changing a four-digit numeric character to a seven-digit alphanumeric is a straightforward process "similar to the conversion required for Y2K." However, many (perhaps most) payers and medical researchers will have to maintain dual data storage systems to avoid loss of legacy data coded under ICD-9 standards.
The VHA's Graham expects the technical difficulties associated with changing codes to be fairly minor, except for crosswalking from more precise data coded under ICD-10 to older, less precise data coded under ICD-9. CMS has already published a program to crosswalk back (mostly a matter of re-aggregating ICD-10's more granular data). However, it is difficult, often impossible, to disaggregate ICD-9 data into ICD-10 codes. But Graham doesn't see these difficulties as an argument for continuing to collect imprecise data.
The same is true of training, says Mary Johnson, a VHA health information management specialist. "An ICD-9 code may be mapped to four ICD-10 codes," she says. "It'll be a challenge to coding staff everywhere, because there is such a difference in coding format. They're going to need to know more anatomy, physiology and medical terminology. Most of us have had those courses in the past, but you kind of forget those skills because you don't use them." Even so, Johnson says she was convinced of the overall superiority of ICD-10 when she first heard about it. "That was in 1982," she says. "As a professional, I've been waiting for this a long time."
No one seems to doubt that software vendors can reconfigure their products within a two-year time frame. Most of the larger firms that market products internationally already have experience with ICD-10. Nevertheless, products designed for countries with radically different ways of paying for healthcare will require redesign for use in the United States. Any organization considering new software for EMRs should insist that potential vendors explain how they plan to make the transition to ICD-10 once HHS sets a deadline.
Only as efficient as your tools
"You know how it is when you've got a really old computer system," says Kloss, summing up AHIMA's perspective on ICD-10. "Then upgrading is a big deal. That's what this is like. We've got to help people understand that this is really part of our transition to an electronic healthcare system. An [electronic health record] ultimately is as good as the data tools that are in it. The idea that we're going to invest tens of millions of dollars into an electronic health system and then dump our data out into ICD-9... It's, well, ludicrous."
Fred D. Baldwin is a freelance writer in Carlisle, Pa.
The ABCs of ICDs
As medicine and technology advance and diagnostic terminology evolves, the International Classification of Diseases has to undergo periodic revision to accommodate the changes. ICD-10 reflects the many advances in scientific and clinical knowledge and the modifications they entail since ICD-9 went into effect in 1979.
New and different
The changes encompassed in ICD-10 are too complex and numerous to be easily specified, but in general, they involve the following:
- ICD-10 contains three additional chapters, and other chapters have been rearranged. It is published in three volumes rather than the two volumes for ICD-9.
- Cause-of-death titles have been altered and conditions have been regrouped.
- Some coding rules have been changed.
- ICD-10 offers more granularity and finer distinctions. Some distinctions, now deemed irrelevant, have been dropped. It contains about 8,000 categories, compared with 4,000 categories in ICD-9, largely to provide more clinical detail for morbidity applications.
- Information relevant to ambulatory and managed care encounters has been added and injury codes have been expanded.
- The number of codes needed to fully describe a condition are reduced through the creation of combination diagnosis/symptom codes.
- ICD-10 uses alphanumeric codes rather than the completely numeric codes of ICD-9, and the field size is longer.
Tips from HHS
The change in field size alone may require adjustments in computer systems, to say nothing of the ramifications of new codes and altered rules. In 2004, Santa Monica, Calif.-based RAND Corp. prepared a technical report for the U.S. Department of Health and Human Services, which offers some advice to ease the transition to ICD-10 code sets:
- Select a date by which everyone in the organization must make the transition. Everything associated with discharges on or before the selected date will be submitted in ICD-9 and anything later in ICD-10. This will avoid keeping two sets of books any longer than needed to resolve old reimbursement claims.
- Allow adequate time for the transition, which will probably take two to three years. According to experts, just moving the decision to make the switch to ICD-10 through the approvals process can take a year.
- Prepare and circulate a reliable, understandable crosswalk between ICD-9 and ICD-10 codes to aid the changeover. Translating every ICD-10 code into ICD-9 code for processing will not be feasible over the long term, but it will postpone major systems changes until they can be scheduled to meet several requirements at once.
- Consider coding some diagnoses and procedures in both ICD-9 and ICD-10 to determine which codes are interpreted similarly. This will help in crosswalk development, both in practice and in theory, and will assist analysts in interpreting before-and-after changes in health statistics.
Adapted from "ICD-10 Comparability Ratio" by Tong Zheng at http://health.utah.gov/ibisq/comp_ratio.htm, "The Costs and Benefits of Moving to the ICD-10 Code Sets" by Martin Libicki and Irene Brahmakulam at www.rand.org/publications/TR/TR132/TR132.pdf, and the Centers for Disease Control and Prevention at http://www.cdc.gov/nchs/about/otheract/icd9/abticd10.htm.
Corrine Charais is senior manuscript editor at Healthcare Informatics.
