April 6, 2006
Chairman Johnson, Congressman Stark, members of the Health Subcommittee, ladies and gentlemen, good afternoon. I am Gloryanne Bryant, corporate director for coding and Health Information Management (HIM) compliance with Catholic Healthcare West (CHW). I speak to you today not only from my position at CHW, but also as one of 50,000 health information management professionals throughout the country and industry who are interested in quality information for quality healthcare.
I am here today to urge you to move forward with legislation introduced by Chairman Johnson, HR 4157, the “Health Information Technology Promotion Act.” Specifically, I am here to ask you all to support HR 4157 and ensure this bill is acted upon with deliberate speed so our healthcare industry can make use of the best possible disease and procedure classification systems as soon as possible.
I have been involved in the coding and management of coding of healthcare data for over 27 years. Since 1993, when the National Committee on Vital and Health Statistics (NCVHS) declared our US disease and procedure classification system the International Classification of Diseases, Version 9, Clinical Modification (ICD-9-CM) “broken,” I have been involved with several of the groups that provided input for the replacement of this classification system.
In the late 1990s, I was a tester of the ICD-10-PCS (procedure coding system), the classification system anointed to replace the existing inpatient procedure codes in ICD-9-CM. It was a successful test and the ICD-10-PCS system has been maintained and ready to go since that time.
In recent years, I have been involved with the testing of the ICD-10-CM (ICD-10-CM) classification system for diagnoses. This testing not only proved the accuracy of the classification system, it also showed how simple the training for ICD-10-CM and ICD-10-PCS could be. This testing was done with the American Hospital Association (AHA) and my professional association the American Health Information Management Association (AHIMA), and the test report is available on the AHIMA Web site. I have also participated in meetings of the ICD-9-CM Coordination and Maintenance Committee, which is charged with overseeing the existing ICD-9-CM. All this activity has given me significant insight on the needs, issues, and problems surrounding the upgrading of ICD-9-CM.
Working with ICD-9-CM on a daily basis reaffirms that it is outdated, broken, inefficient, and nothing but an albatross to our healthcare system. I strongly support upgrading ICD-9-CM to ICD-10-CM and ICD-10-PCS. Yes, upgrading our coding system will require change, but change that is not insurmountable.
Working closely with other professionals at Catholic Healthcare West, I have developed a three year transition plan for CHW to use for the planning, implementing, and training that will need to occur with any upgrade. I have also educated my own staff and others about ICD-10-CM and ICD-10-PCS, the actual use of ICD-10 classification, and the important issues of planning, implementation, and training. You have to understand that the training materials we use today will be considered quite crude once the go-ahead is indicated for ICD-10 classifications. Even so, the coders I have trained have found the ICD-10 systems easy to use, and it enables them to present a complete and accurate report.
The following are some key points I want to share:
- ICD-9-CM is obsolete and the new version, ICD-10, is ready for implementation.
- ICD-10-CM and ICD-10-PCS, the ICD-9 upgrades, are needed to improve the quality of health information. Specificity equates to precision, not complexity.
- ICD-10 is needed to support interoperable electronic health records (EHRs) and a nationwide health information network (NHIN).
- Action has to occur now so that we can receive the benefits of this improved data by the end of this decade.
It is my intention to provide you with the necessary information on the benefits of ICD-10-CM and ICD-10-PCS to move ahead with HR 4157 so we can achieve the benefits these classification systems provide and better the healthcare of all individuals.
Why is ICD-10 necessary for pay-for-performance, accountability, quality reporting, and more?
Quality and Pay-for-Performance
I was asked to address why the ICD-10 classifications are necessary for quality of care monitoring, pay-for-performance, and other areas of healthcare accountability.
Increased detail and better depiction of severity allows improved linkage between a provider's performance and the patient's condition and a better ability to measure quality. It is hard to measure quality of care, or a provider's performance in addressing risk factors and effectively treating a patient's condition, if the relevant diagnostic or procedural code includes multiple conditions. For example, if 2 conditions with different treatment protocols are assigned to the same code, how will we evaluate the provider's performance in treating one of these 2 conditions? Severity can also be an issue -if all we know is that a patient has a decubitus ulcer, and not whether it involves skin only or all the way down to bone, how will we be able to measure the effectiveness of a wound management program -or the cost of treating decubitus ulcers? It is obviously much more difficult, and expensive, to treat a deep ulcer than a superficial one.
Many quality measures, such as HealthGrades and AHRQ's quality indicators, rely on ICD-9-CM codes. If these codes are too general or ambiguous, or not reflective of modern medicine, it will be impossible to produce accurate quality reports or pay providers accurately for performance. Situations have already occurred whereby hospitals have complained about erroneous quality conclusions based on ambiguous or poor ICD-9-CM codes -for example, if a code includes conditions that have variable quality implications (i.e., some conditions don't indicate a quality problem and other conditions do, but both sets of conditions are classified to the same code, the conclusion, or assumption, often goes to the worst case scenario -i.e., if you can't distinguish which condition the patient had, assume he had the condition with the adverse quality implication and "ding" the provider).
ICD-10-CM greatly expands the codes for medical complications and medical safety issues -allowing for much better capture of this information for use in quality measurement, P4P, and to assess the effectiveness of medical error prevention programs.
Also, if there is a disconnect between the outdated classification of a condition in ICD-9-CM and the modern clinical classification of a disease process, it is difficult to relate modern treatment protocols and performance measures to the relevant code in ICD-9-CM (it is like comparing apples with oranges -current performance measures and evidence-based medicine protocols are linked to a diagnostic structure or diagnostic or procedural distinctions that do not exist in ICD-9-CM) .
It is difficult to evaluate the outcomes of new procedures and emerging healthcare conditions when precise codes are lacking. For example, in ICD-9-CM, many procedures are not differentiated by approach. For basic quality measures such as mortality, rates can vary widely depending on the approach used.
CMS has even acknowledged that it would be difficult to implement a severity-refined DRG system without ICD-10 because the inability to collect a finer level of detail would limit the usefulness of the DRG refinements.
ICD-10-CM will open new opportunities in injury research and trauma services evaluation. To further research in the area of prevention and treatment of injuries, we must be able to more accurately classify the nature of the injuries sustained and correlate the nature of injury with the mechanism of injury, treatment, and outcome. ICD-10-CM will provide a much-improved ability to accomplish this task. Major improvements relative to injury prevention and treatment include: addition of a 6th character that expands the flexibility of the system and allows for incorporation of more specific injury codes; separation of many previous “multiple injury” codes into separate codes and elimination of certain “illogical” injury codes, especially with regard to head injuries; incorporation of terminology commonly used by clinicians; inclusion of laterality code which affords the opportunity to identify bilateral injuries and provides an unprecedented ability to more accurately study patterns of injury and how they relate to the underlying mechanism of injury.
The improvements in ICD-10-CM have important implications for our ability to rate severity of injuries. Several different classification systems are being used by trauma clinicians to rate severity of injuries for the purpose of benchmarking, quality improvement activities, and research. These systems require an independent review of the medical record, which is becoming increasingly cost-prohibitive. The use of ICD-10-CM as a basis for rating the severity of injuries would obviate the need for these alternative scaling systems.
ICD-9-CM lacks specificity regarding the extent of injury and uses terminology and severity parameters that are either outdated or inconsistent with other widely used clinical classifications. ICD10-CM would address many, if not most, of these inadequacies and bring us closer to a universal classification of injuries.
While some criticize the size of the ICD-10 code sets or the number of codes available for use, these code sets have been expanded from ICD-9-CM to reflect 21st century medicine and diseases. They provide the detail needed if healthcare providers are to be paid fairly in the future. Currently, hospitals are supplying quality monitoring data. The only way to get knowledge from this data is to also look at the diagnoses codes that are included on the claim.
This quality data is looked at against diagnoses represented with up to nine codes (codes are limited by Medicare in order to accommodate paper billing), but these are diagnoses codes first renewed (from ICD-8-CM) 30 years ago. Since then, or should I say, since the 1970s, our medical knowledge represented by diagnosis, technology and procedure codes has expanded greatly. Interestingly enough, today’s ICD-9-CM classification has to ignore this expansion of healthcare knowledge and treatment because it is running out of the codes necessary to provide the necessary and accurate representation.
In ICD-9, our medical coders find that they are unable to code accurately because the codes available do not have the level of specificity that matches the information and clinical documentation in the health record. In these cases, the coder often has to make use of an ambiguous code, which is frustrating and potentially expensive. When the claim is submitted someone may call back asking for additional detail, which involves getting data from the medical record, copying it, and sending it out.
Accountability is the hallmark of a medical coder. If we do not find the information in the record, we do not code it. Therefore, it is very frustrating to find the information in the record and then either be unable to code it, or have to use either a vague code or choose between any numbers of vague codes. In behavioral health, we are forced to use a “cheat sheet” to change record information into acceptable codes. Why? The codes in ICD-9-CM are so far behind the advances in behavioral health that we have no correct codes. Technically, we are not supposed to change codes, but with the blessing of CMS, it is something that we do or we would not get paid.
We have so many requests for additional detail that we have had to hire photocopying companies and other outsourcing resources to handle the requests. This is common throughout the healthcare industry. In addition to the costs and time associated with handling these requests, the plans also incur costs and dedicate employee time. Why? Someone has to read and interpret this information. Then, and only then, is a payment processed. Meanwhile, besides the cost of processing the additional information, an organization also has a cost of carrying a receivable.
Many of these efforts and expenses are incurred because we do not have a classification system that can represent the information that is in the patient’s record. The more that we get into sending data for quality monitoring or other reasons--often demanded by Congress--the greater this problem of equity, cost, and accuracy will be.
The collection of quality monitoring data is not the end of the story as we are really just starting this process. Next in line are the pay-for-performance programs. In this environment, the provider is financially affected if there is an issue between the quality indicator and the actual state of the patient’s health and procedures, which to judge correctly will mean getting the additional clinical data mentioned previously. Again, without the ICD-10-CM and ICD-10-PCS upgrades, providers are going to expend additional resources to provide health plans with enough data so we get paid accurately under P-4-P. If we cannot improve the content of our codes, we will all lose. It is a vicious cycle and I do not see it changing soon. Each day, more and more information is sought relevant to the claim, and for decades we have neglected to upgrade the crucial clinical information on the claim itself. We need to upgrade to ICD-10.
Medical necessity and infections are two other areas of healthcare that I believe are of interest to this subcommittee. Under medical necessity, a provider has to demonstrate that there is a need for the patient to receive certain care. Again, we are forced to manually process claims to ensure that we send parts of the health records to prove the medical necessity. While moving to ICD-10 classifications will not eliminate this problem, there will be a substantial decrease in the need to review additional parts of the record if the claim can carry more detailed coding--both in the codes used and in the number of codes reported. This problem, which is another resource burner for many hospitals and physicians, could be greatly alleviated with the detail available in the upgrades from ICD-9-CM to ICD-10.
In 2007, hospitals are required to begin reporting “diagnoses present on admission.” This data is to provide CMS with information that eventually may lead to some care not being reimbursed – situations where it is deemed the institution’s fault that additional care was needed. For instance, was the infection, or other problem, present when the patient was admitted, or did it arise during the stay? While some believe that continuing to use ICD-9-CM is not a problem, it will be clear when conflicts arise due to coding that may or may not be reflective of the details involved with infections or similar issues that arise. If we cannot implement a more contemporary classification system until 2009, it might perhaps be better for Congress delay this requirement until ICD-9-CM is upgraded.
I have repeatedly mentioned using medical records to make up for missing or vague codes. This problem is also becoming more prevalent in the area of medical procedures and technology. A few years ago, Congress asked CMS to ensure it is keeping up with new technology. Again we are running out of codes in the area of inpatient technology and procedures. So, we are merging technologies into essentially group codes, and then, to get proper reimbursement, we are sending additional “attachment” detail to the claims adjudicator.
It has been suggested by some that there are plenty of codes left to describe new technology and procedures. When citing the number of codes, these same critics fail to mention that once we run out of sequenced codes, the ICD-9-CM Coordination and Maintenance Committee will have to assign codes in other chapters (associated with body systems). This will essentially eliminate the ability to monitor such data by computer.
I can mention other areas where the detail in ICD-10-CM and ICD-10-PCS could greatly improve healthcare and its administration by CMS and others. It is important to recognize that the depth of clinical data continues to drive major healthcare decisions, for payers, researchers, regulators and Congress.
I know there is an expectation by some that once we have a standard electronic health record (EHR), we will not need to have such a classification system. However, this suggestion ignores the fact that the detail in the EHR will be too granular to be used for all the secondary purposes that require providers to submit data to the government, health plans and others. Classifications, or the codes that make up the classification system, provide this data and make it usable for a variety of purposes that a copy of the record, paper or electronic, just cannot do. Until we have a good classification system implemented, the value of the EHR development will not be experienced by the patient or the population.
Fraud and Abuse
I mentioned ambiguous codes. Using ICD-9-CM, coders often have to make choices in codes they assign because there may not be an accurate code for the diagnosis or procedure that is reflected in the record. It is almost like the story of the lady and the tiger. If you make the wrong choice, the tiger – be it the health plan, the Medicare carrier, or the Department of Health and Human Services (HHS) Office of the Inspector General (OIG) – comes out and accuses you of fraud. Some believe more detailed coding system will increase fraud, but I believe exactly the opposite will happen. When we have codes that can actually match the 21st century data in the medical record then we will not have to choose the lady or the tiger. The code needed to reflect the data present will become much easier to ascertain. In fact, ICD-10-CM was designed to eventually allow for what we call “computer assisted coding,” where the computer itself does the coding from the detail in the electronic record and the coder becomes more of an editor, validator, or monitor of the system.
Last summer, the AHIMA Foundation of Research and Education (FORE) undertook a fraud study for the Office of the National Coordinator for Health Information Technology (ONC) and the HHS OIG. The oversight committee was made up of fraud experts from health plans, providers, healthcare associations, and government including the Department of Justice and the HHS Inspector General’s office. Among its “guiding principles” and recommendations the reports called for “standardized reference terminology and up to date classification systems that facilitate the automation of clinical coding are essential to the adoption of interoperable EHRs and the associated IT enabled healthcare fraud management programs.” The reports from this study (“Report on the Use of Health Information Technology to Enhance and Expand Health Care Anti-Fraud Activities” and “Automated Coding Software: Development and Use to Enhance Anti-Fraud Activities”) are posted on both the Office of the National Coordinator’s Web Site and the AHIMA/FORE Web Site.
PHRs and Claims Data
In recent months we have also seen a significant increase in the importance of consumers having a personal health record (PHR). Educating consumers on the value of using a PHR is a high priority in my profession, and we welcome providers, health plans, and others who are looking to improve healthcare through the use of PHRs. However, even as we move to provide this data to our patients/consumers, another emerging trend is to build or sponsor PHRs that take their clinical data from the claim forms – namely, the ICD-9-CM and other classifications (CPT® in ambulatory claims) data.
I and many others in health information management and medical informatics are deeply concerned about this trend for all the reasons I have cited – the vagueness and limitations of today’s claims data. If we begin to populate these PHRs with claims data based in ICD-9 and do not warn the owner and users of the limitations, we could have significant negative impact on the owner of the personal health records. In addition to the limitations of the ICD-9-CM coding system, the ICD-9-CM codes on claims are truncated, as noted, at 9 codes. This means the information is not only potentially vague, but also may not reflect clinical information that is important but has been truncated in the claims process. In addition, codes on claims often reflect coding that has been altered to meet the health plan or payers reimbursement instructions. Personally, I believe that it is unsafe to develop these PHRs without:
- Diagnostic and procedure detail that can only be provided in a contemporary classification;
- Industry agreement for consistency in coding so codes are not changed for reimbursement purposes; or
- A clear statement that the information in such records does not represent all the diagnoses or procedures that potentially were identified in any episode of care.
Example: How ICD-9-CM does not reflect modern medicine.
I wanted to share some examples of the coding problems we are experiencing with regard to its reflection of modern medicine. Two that recently were addressed at the March 2006 ICD-9-CM Coordination and Maintenance Meeting included non-Hodgkin’s lymphoma and secondary diabetes mellitus
Non-Hodgkin’s Lymphoma
There are more than 30 subtypes of non-Hodgkin’s lymphoma. The request was to update the non-Hodgkin’s lymphoma codes to allow for more current classification. This involves the creation of several new codes for specific types. Currently, “non-Hodgkin’s lymphoma” is indexed to 202.8x (with the “x” referring to a fifth character for the specific site). However, there is no space for expansion in category 202. So, it has been proposed that category 200, which is currently limited to lymphosarcomas and reticulosarcomas, be expanded to also include several new codes for subtypes of non-Hodgkin’s lymphoma. To maintain consistency with longitudinal data, non-Hodgkin’s lymphoma, not further specified as to subtype, would continue to be classified to code 202.8x. This code includes lymphomas other than just non-Hodgkin’s lymphoma. Due the limited codes left in the system, it is not being proposed that a code be created specifically for non-Hodgkin’s lymphoma not further specified as to subtype.
So now, until we upgrade ICD-9-CM to ICD-10-CM, codes for non-Hodgkin’s lymphoma subtypes will be distributed between two entirely different categories, with no unique code for non-Hodgkin’s lymphoma not specified as to subtype. This will have a serious impact on data retrieval and,as I noted before, coding accuracy. Non-Hodgkin’s lymphoma codes will not be grouped together, which means data analysts and coders could miss identifying all the related codes. The non-Hodgkin’s lymphoma cases that are still classified to code 202.8x will not be able to be specifically identified as non-Hodgkin’s lymphoma cases because this code includes other types of lymphomas as well. Data retrieval for lymphosarcoma and reticulosarcoma will also be affected because category 200 will no longer be limited to these conditions. This impacts not only physicians and hospitals, but also our cancer registries, researchers, and others. Meanwhile, ICD-10-CM has numerous specific codes for non-Hodgkin’s lymphoma that are organized into appropriate categories.
Secondary Diabetes Mellitus
The diabetes classification in ICD-9-CM is also outdated -in ICD-10-CM, the diabetes codes reflect the American Diabetes Association's current clinical classification.
In ICD-10-CM, there are distinct categories of codes for diabetes mellitus due to underlying conditions and drug or chemical induced diabetes mellitus. In ICD-9-CM, these conditions are all classified to code 251.8 – other specified disorders of pancreatic internal secretion. It has been proposed that a new category of codes be created in ICD-9-CM to capture secondary diabetes mellitus. However, in an effort to conserve codes, an attempt is being made to cover both diabetes mellitus due to an underlying condition and diabetes mellitus due to drugs or chemicals into a single category of codes. This will result in confusion and coding errors due to the differences in coding diseases that are due to an underlying condition and those caused by drugs. All the requested information is in the medical record(s), and it is much clearer from both a coding and data analysis perspective to distinguish drug-induced diseases from those caused by an underlying condition. However, but this cannot be done under ICD-9-CM.
There are many examples highlighting the differences between ICD-9-CM and the classifications originally designed in the 1990s (ICD-10-CM and ICD-10-PCS) and maintained for conversion that show the differences in detail and organization. I would invite you to contact the Centers for Disease Control and Prevention (CDC)-National Center for Health Statistics (NCHS – custodian of ICD-9-CM, volumes 1&2 and ICD-10-CM) and the Centers for Medicare and Medicaid Services (CMS -custodian of ICD-9-CM volume 3, and ICD-10-PCS) for a much more detailed look at the differences between the two generations of classification systems. You can also contact the American Hospital Association (AHA) and AHIMA who constitute the provider and professional organizations overseeing the coding guidelines for ICD-9-CM.
Other Examples
A few other examples where ICD-9-CM does not reflect modern medicine include:
- Myeloproliferative disorders and myelodysplastic syndrome – classified as neoplasms of uncertain behavior in ICD-9-CM, but now recognized as hematologic malignancies
- Many conditions are classified according to outdated thinking. Examples include neuromuscular disease, essential tremor, epilepsy, transverse myelitis, stroke.
- Alzheimer's disease codes in ICD-10-CM are more reflective of current medical knowledge.
Why is ICD-10 necessary to keep the US in concert with the rest of the world?
It should be no surprise to the subcommittee that we live in a small world. In recent months, the media has reported numerous stories related to the avian flu outbreak and the potential for pandemic outbreaks in the US as well. This has generated attention in Congress, the Administration, the states, and localities. Unfortunately, the US would have difficulty in tracking a pandemic outbreak and comparing our data internationally.
Does the US have a diagnosis code for avian flu? No.
Does the World Health Organization have such a code? Yes.
Does the US have codes for West Nile Virus, SARs, or potential bioterrorism? We do now, but we didn’t have them at the time of the US outbreaks or for our first anthrax incidents when they occurred.
Why is the US behind? Essentially because we are not on the same ICD-based system that most of the world is on, including all other industrial nations. They have converted to ICD-10 while we continue to linger and plod along with a system designed and implemented in the 1970’s.
The ICD-10 codes are different than ICD-9-CM. Instead of five characters in version 9, version 10 has seven, and instead of being only numeric in version 9, version 10-based codes are alphanumeric. So, when a new disease is recognized by the WHO, the US has to take the code and figure a way to renumber it and then put it in our coding system. As we discussed before, some of our chapters are out of codes. This is not an easy task and it is getting harder all the time.
Avian flu, SARs, and other diseases get a lot of press, but we must not forget that information must be transferred for research and public health purposes hundreds of times daily. When this exchange is between a US public health department, the Centers for Disease and Prevention (CDC), or a research team and another ICD-10 country it means that someone must translate the codes. Because we have not moved to ICD-10 based classifications there has not been a lot of development of electronic translators or maps. What does this mean in today’s environment? It means doing this by hand.
This problem grows because if it is ICD-10 information coming in to the US, the receiver has to map it to the vague ICD-9-CM codes. Obviously, some researchers choose to just work in ICD-10 and they can do so if they do not have to look at information contained in both classification systems. Yet doing business this way or using translators is not without a cost. I suspect the pharmaceutical companies, major researchers, health data organizations, and other healthcare companies doing business internationally can give you the economic impact that occurs because of these differences.
This is a good point to note that ICD-10-CM has been restructured to facilitate not only computer-assisted coding, but also to work hand-in-hand with electronic health records that will have a vocabulary base. The most common vocabulary is SNOMED-CT®, which is the designated vocabulary the federal government will use. This means that ICD-10 has now become the base system for future versions. Future versions may be expanded to reduce the differences between the US clinical modifications and the international code as well as to work to smooth the interrelationship with other classifications such a functional status codes. The longer we remain on ICD-9-CM the longer we can not get the benefits of this classification, which has been designed to work with the standard EHR that Congress, the Administration, and many in the industry want to see in place.
US Mortality Coding
The problem is not just between the US and other countries. In 1999, the US upgraded its mortality reporting system to use ICD-10. This was done in part because of our international agreements (through the CDC’s NCHS) and the need to look at mortality on an international basis. So today, in 2006, each of the US states, district, and territories report mortality data monthly ICD-10 on a monthly basis. This leads to the question of whether we can easily look at US mortality data versus our morbidity data. The obvious answer is no, not without the same mapping or conversion process that we must use for international data.
Crosswalks
Crosswalks or mapping are terms we use to describe the connections, or paths, between classifications and vocabularies. There are several needs for a mapping associated with ICD-9-CM and ICD-10-CM or ICD-10-PCS. We have already discussed the need to map between data from the US and other countries or between our current morbidity and mortality systems. Such a map is also needed for the purpose of maintaining a longitudinal patient record to ensure that data in ICD-9-CM and one of the ICD-10 classifications can be obtained for a variety of reasons including clinical care, research, fraud monitoring and so forth. A third reason for a map might be to permit a healthcare plan or payer to accept claims with ICD-10-CM or PCS data, but map it back to ICD-9-CM so that the claim can be adjudicated on an older system that was not converted.
This last use is a practice used in Canada to alleviate some of the implementation issues during that country’s conversion. It is not a recommended practice because when you map back from ICD-10 to ICD-9, you lose most of the detail I have spoken of and consequently the initial benefits and savings that might come from simpler claims administration. This is not beneficial for either the plan or the provider. A similar use of this process is the function of fraud monitoring. For fraud monitoring, a plan might carry a three-year rolling set of claims data in order to detect fraud. To do this would require a three year period where you would keep data in ICD-9-CM, but once the three years is completed, you would have built a 3-year history in ICD-10 classifications. Technically, you could also map the ICD-9-CM data to ICD-10. Again, these are not the optimal solutions, but they are feasible and they do allow for the need to work through an initial implementation period.
I must indicate that both of these ICD-10 classifications have been maintained since they were originated. The update for ICD-10-PCS was just announced in the Federal Register, and the update for ICD-10-CM is due in June. The mapping between ICD-9-CM and ICD-10-CM will also be forthcoming shortly from the CDC-NCHS.
I noted that mapping occurs between classifications and vocabularies and that vocabularies would serve to be the data base for standard electronic health records. Therefore, mapping must occur between ICD-10-CM and SNOMED-CT and ICD-10-PCS and SNOMED-CT. Mapping has been completed and verified under the National Institute for Health’s (NIH) National Library of Medicine (NLM). At a February 2006 meeting of the NCVHS, the NLM did announce that it is ready and waiting to map between the ICD-10 classifications and SNOMED-CT, but it cannot do so until the two classifications are officially adopted by the HHS.
Conversion Costs
The question of conversion costs often arises as we discuss the upgrading of ICD-9-CM. In 2003 the NCVHS asked this question of the Rand Corporation and through Rand’s report was told that costs for the US (in 2003) were in the range of $425 million to $1.1 billion. Rand also noted that the anticipated benefits would be between $700 million and $7.7 million because of all the advantages of detailed information – many that I have mentioned before.
Significant losses in coding productivity and accuracy are resulting from the use of ICD-9-CM terminology that is not consistent with current medical practice. For example, coding professionals must often consult physicians for clarification regarding the appropriate code. Time is wasted when coding professionals have to try to determine the “best” code when none of the options seems appropriate. Often, the wrong code is selected due to the inability to determine the best ambiguous code or conflicts between ICD-9-CM terminology and terminology used in the record. Many of the coding questions that arise stem from the ambiguity and inconsistency of ICD-9-CM and the outdated terminology that is not reflected in current medical record documentation. As the obsolescence of ICD-9-CM continues to increase, these problems will grow even more.
You should also be aware that recent graduates of medical schools and medical terminology courses find ICD-9-CM particularly difficult to use because they are not taught the outdated terminology used in ICD-9-CM.
Increased costs are also incurred due to the extended time required to code cases where more and more disparate conditions and procedures are classified to the same code.
As long as we continue to use ICD-9-CM, the coding process will be heavily labor-intensive -i.e., manual coding process. The use of computer-assisted coding tools is limited with ICD-9-CM due to the code ambiguity, lack of precision, and inconsistent terminology and definitions. No matter how sophisticated electronic applications become, their use in the coding process will be limited as long as ICD-9-CM is in use. Once ICD-10 is implemented, the use of electronic coding tools will grow dramatically and these applications will become increasingly sophisticated -greatly facilitating the coding process and reducing the manual labor involved.
As I have noted, essentially the systems’ change created by ICD-9-CM being upgraded to the ICD-10 classification is one of an expanded field. The other change is moving this larger field from numeric to alpha-numeric. So, the systems that have to have the field changed, technically, are any software that has the ICD-9-CM code, and any electronic data base that contains ICD-9-CM codes. Currently, most of the systems that have extensive coding in software or in the data base are in hospitals, health plans, and reporting and research organizations. This is not to say that they do not exist in clinic, physician practices, and ancillary services. But, as you know, there is a substantial healthcare provider population that is still essentially paper-oriented, and even many of the organizations I mentioned may have only one or two systems, not a full EHR or administrative system. While software vendors will make the change to the software (and many international firms have experience with ICD-10), database upgrades will involve both the owners and vendors.
The longer we wait to make the upgrade to ICD-9-CM, the more expensive such a conversion will be, if for no other reason than the government and healthcare industry are pushing various entities into purchasing electronic health programs, records, reporting mechanisms, and the like. We have set goals and projects, even in Congress, to get providers to send information electronically. Healthcare entities who electronically expand their systems, purchase replacement systems, and so forth will have to make the ICD-9-CM upgrade to ICD-10 as a retrofit. If you have purchased the software or are beginning to build the database you will have to retrofit your system. Anyone who has renovated a house or office knows the renovation is often more costly than building in the first place. The same occurs for electronic systems.
The need to “retrofit” ICD-10-CM and ICD-10-PCS into a greater number of system applications, declines in coding productivity and accuracy due to difficulties in trying to use a failing coding system, and the implementation of “band-aid” approaches to keep ICD-9-CM afloat and attempt to meet healthcare data demands as much as possible. For example, increasingly, CMS has been forced, due to space constraints in the ICD-9-CM procedural coding system, to disrupt the hierarchical structure by starting to use available codes in unrelated code series, with the result being an unstructured, haphazard coding system. A hierarchical structure assists in defining coding concepts by placing them into organized, distinct groupings. Disruption of this hierarchical structure causes the complexity of using the system and mining coded data to increase dramatically and leads to declines in coding accuracy. Additionally, the US will continue to incur the costs of maintaining ICD-10-CM/PCS as well as ICD-9-CM for all of the years between now and implementation.
I must point out the obvious: the standards for ICD-10-CM and ICD-10-PCS are available now. The standards for the transactions that must also occur, the conversion to ASC X12 Version 5010 and the NCPDP upgrades are also known. If products (systems and software) were built from this point forward that could handle both code sets, then buyers would not have to retrofit later. It is not unusual to buy a product that is ready for a future change such as the high definition televisions that are sold in anticipation. As a prudent buyer I would like to be able to purchase new software and other systems that have the capacity to handle ICD-10 classifications. I understand, however, that vendors are reluctant to include the ICD-10 classifications in existing and near future products. It costs money to make such conversions in systems, but they have heard rumors of upgrading ICD-9-CM since the mid1990s. So, why should they run the expense?
The vendors want a formalized notification so they can move forward. At this stage, the notification must come from the federal government. It is important to express that HR 4157 explicitly addresses the notification issue by calling for a “notice of intent” to be sent out by the Secretary within 30 days of the passage of this provision. This is absolutely necessary.
Delay
Last summer, my professional association, AHIMA, issued a statement calling for the implementation of the ICD-10 classifications by October 2008. After your hearing on this issue last July, it became apparent that such an implementation could not occur until October of 2009, and only if the bill is passed. Without the bill, under HIPAA, the earliest conversion could occur would be 2011 or 2012.
I have already pointed out a number of examples of what happens when we have no codes in our diagnoses area, or if we have no sequential codes in the procedures area. Our data becomes more vague and more suspect. There will be more and more calls for additional information from the medical record. Quality data, injury data, all of this will be suspect because procedure and diagnoses codes are taken off the claim, and we will not have the detail to provide an ample picture of the patient’s health without considerable manual efforts. Congress continues to call for better payment systems, report cards, and other measures of care and healthcare value, but the cost of providing such data with a classification system that does not represent 21st century medicine will increase rapidly or no data will be provided because providers will not have the money to provide data outside of the existing claims system.
The United States is the last hold-out in the industrial world to convert to an ICD-10 based system. What is the cost to research as we go through the manual efforts I have described year after year? Perhaps we could use maps, but we are mapping between much greater detail from other countries, and a rather vague and somewhat violated system of codes. How long does our public health system have to hope that we have codes to describe and track international outbreaks? How easily will we be able to share biosurveillance data?
We are currently in the midst of a massive effort to ensure standard electronic health records and to provide a nationwide health information network. Certainly our electronic records will improve clinical care and to rapidly transfer records across the country will be great. But, what about all the other data transmitted through the network – the secondary data? If we cannot clean up our secondary data by upgrading our ICD-9-CM system, then it will be like sending polluted water through a new pipe system.
I see a day, in 2010, when my coders can code accurately with ICD-10 and not have to guess at the code, or merge it into a catch-all code – a day when the ICD codes truly represent the information our clinicians have entered into the record.
I see a day when a standard electronic health record can provide the initial ICD-10 coding information because of a standard map between its SNOMED-CT base and our ICD system, and coders will become validators of the process and not the process itself.
I see a day when my organization can send a claim and have it processed because all the information needed is in the ICD-10 codes and very few claims will result in a request for more information. And, because this claim’s information is so valuable, it can be used in quality and injury monitoring programs and where healthcare can be reimbursed, on the basis of outcomes and quality.
I see a day when public health and researchers can trade and use data internationally, in the form of ICD-10 based codes for monitoring and for instructions when an outbreak occurs.
Few of us are using the electronic standards for personal computers and similar devices we bought 10 years ago, let alone 30 years ago. Why do some feel comfortable doing so with our data standards?
HR 4157 calls for a review of priorities. Should the provider and payer identifiers be put aside to get better clinical data? We are not running out of identifiers at present and current identification numbers work in our systems. Should the potential for a HIPAA claims attachment that will only affect 2 percent (according to the National Uniform Billing Committee) of claims be delayed until we implement a classification system that can eliminate the need for attachments for many more claims?
These are questions that must be answered, and I congratulate Chairman Johnson and the subcommittee for providing leadership on this issue. I would be pleased to answer any questions you might have and I also point you to my professional association, the American Health Information Management Association, for a response to any further questions. Thank you.
Gloryanne Bryant, RHIA,
CCS Corporate Director, Coding HIM Compliance
Catholic Healthcare West
185 Berry Street, Suite 300
San Francisco, CA 94107
Telephone: (415) 438-5721
E-Mail: gbryant@chw.ecu
American Health Information Management Association
1730 M Street, Northwest
Suite 409
Washington, DC 20036
Telephone: (202) 659-9440
E-Mail: dan.rode@ahima.org
Web site: www.ahima.org