The Great Cost Debate

The cost of the switch to ICD-10-CM (and its counterpart ICD-10 Procedure Coding System or ICD-10-PCS) is a subject of vigorous debate.

Prior to the proposed rule for mandating the switch, there were two studies that looked at the cost (RAND and Nolan) and a third analysis that examined the two studies (Hay). The proposed rule performs a detailed impact analysis of the costs and benefits of the switch (including drawing on both of the two prior studies), devoting 100 of 162 pages to it. This impact analysis examines all three of the prior analyses.

The first study (warning: pdf) was performed by RAND. The National Committee on Vital and Health Statistics commissioned RAND to perform this study.

Per the authors of the study, their charge was solely to evaluate the costs and benefits of switching to ICD-10-CM for diagnoses and to ICD-10-PCS for procedures: …we are asking not whether ICD-10 is the best coding system to switch to…

The RAND report states …our best guess is that the cost of conversion will run $425 million to $1,150 million in one-time costs plus somewhere between $5 million and $40 million a year in lost productivity.

The RAND report estimates the benefits of switching to be in the range of $700 million to $7.7 billion.

However, the estimate of benefits comes with a key assumption the authors admit does not hold today with ICD-9-CM: However, to realize those benefits, providers must use the full codes, use them correctly, and use them in a fashion that is neutral to the reimbursement system. ICD-9-CM is by no means always completely, correctly, or neutrally exploited.

Thus in the worst-case scenario, RAND estimates a net loss of $450 million, and in the best-case scenario, RAND estimates a net benefit of $7.275 billion.

The second study (warning: pdf) was performed by the Robert E. Nolan Company. The Nolan report, as it is known, only studied the costs of the switch and did not estimate benefits. The Blue Cross and Blue Shield Association commissioned the Nolan report.

The Nolan report states …Our estimate concludes that key segments of the health care industry would incur significant expenditure of between $6 to nearly $14 billion during a two- to three-year implementation period.

However, it did not consider costs that would be incurred by some significant segments of the health care industry, including nursing homes, clinical labs, durable medical equipment vendors, and several types of payer organizations such as third-party administrators, clearinghouses, and many small-to-medium-sized insurers.

Nolan also compared the switch to other recent, major upheavals in health care information technology: Y2K and compliance with the Health Insurance Portability and Accountability Act (HIPAA). The American Hospital Association estimated that Y2K cost hospitals alone $8 billion, as cited in Nolan. Although fewer information systems manage ICD codes than dates, and although systems that do manage ICD codes have fewer ICD fields in the database than date fields, ICD code sets are much more complex to convert and test. Furthermore, the $8 billion affected hospitals alone, not physicians, payers, and so on.

The Hay report summarized RAND and Nolan to derive yet a third estimate of the costs of the switch. America's Health Insurance Plans commissioned the Hay report.

The Hay report concludes that A reasonable preliminary estimate of the total cost to the healthcare system would be $3.2 to $8.3 billion.

Finally, we come to the proposed rule, which takes all three analyses and comes up with yet a fourth estimate, which we described in our last post: $849 million to $3 billion with a “primary estimate” of $1.64 billion.

The proposed rule also estimates benefits, and under its more optimistic estimate of costs (relative to Nolan and Hay), even it does not see any benefit accruing to the health care system until 2018 (that is, it estimates that the net cumulative benefit will not exceed the net cumulative cost until 2018, 7 years after the switch takes place).

This table summarizes the reports and their cost estimates (in millions of dollars):

	RAND	Nolan	Hay	NPRM
Low	425	5,700	3,200	849
High	1,150	13,900	8,300	3,000
Primary estimate				1,640

Regardless of the strengths and weaknesses of any of these given estimates, the fact is that the estimates vary wildly, and they are just that, estimates. No one has a reliable prediction that can be trusted to within plus or minus a few hundred million dollars. The cost of switching is high.

To put these estimates in perspective, the health care industry spent $1.1 billion on electronic health records (EHRs) in 2005, with a projected growth to $4.8 billion by 2015. Given limited resources, will the ICD-10 switch impede EHR adoption, just as Y2K and HIPAA did?

HHS proposes rule to require ICD-10-CM by 2011

On Friday, August 15, the Department of Health and Human Services issued a notice of proposed rule making that would require the adoption of ICD-10-CM for the purposes of diagnosis coding. They proposed a date of October 1, 2011 for the switch.

The proposed rule is available here (warning: pdf).

The proposal recommends an "all at once" change over. There would be no phased approach, or time period where both ICD-9-CM and ICD-10-CM are used concurrently.

The proposed rule would affect all "covered entities" under the Health Insurance Portability and Accountability Act. Determining whether a health care provider, physician, insurance plan, or claims "clearinghouse" is a covered entity is a bit complex (see the 10pp pdf file the Centers for Medicare and Medicaid Services provide for making the determination). However, the bottom line is that the vast majority of physicians, physician practices, hospitals, health plans, and claims clearinghouses are covered and thus affected.

A great deal of the proposal is made up of a detailed analysis of the costs and benefits of the switch. The proposal itself does not provide a total cost, but one can easily add up the costs in Table 10. The cost estimate ranges from $849 million to $3 billion, with a "primary estimate" of $1.64 billion.

We'll have a lot to say about various myths that people cite in favor of a switch to ICD-10-CM in coming posts. However, suffice it to say that many of these myths are reproduced in this proposal as arguments in favor of ICD-10-CM. We'll expose these myths.

All in all, the notice of proposed rule making is a colossal mistake. Again, if we will expend billions of dollars to change our diagnosis coding system, we ought to switch to a better system than ICD-10-CM.

What is this all about, anyway?

The coding of diagnoses by the health care system initially supported the goals of education and quality improvement (Dr. Slee, whom we mentioned in yesterday's post, was an early pioneer in health care quality improvement efforts).

However, in 1983 the government stepped in and changed that situation radically. It implemented a prospective payment system within the Medicare program. This system uses diagnosis-related groups (for hospital billing), which in turn are based on ICD-9-CM codes (at least for now).

Thus, all health care providers (physicians, hospitals, etc.) who accepted Medicare as payment for services (nearly all of them) were thus required to submit ICD-9-CM codes for the diagnoses that were relevant to a patient visit or hospital stay. If they wanted to get paid, that is.

Thus, beginning in 1983 ICD-9-CM took on the role of supporting billing, or "administrative" purposes (a euphemism for billing and payment) at the national level.

Because most health insurance companies mimic the programs and projects of Medicare, most of them followed suit. Thus, providers were required to submit ICD-9-CM codes to all third-party payers, not just Medicare.

The net result is that a change in the system for coding diagnoses has huge ramifications that impact nearly every physician, hospital, nursing home, and so on in the United States. And patient. How your diagnoses are encoded affect you, too.

In future posts, we will look at studies of the costs of changing the diagnosis coding system. We will also look at uses of "administrative" data that were never imagined prior to the implementation of DRGs in Medicare. We'll explain how diagnosis coding affects you. And we'll spell out the dysfunction current in the system, and why ICD-10-CM will not address most of it.

The Tyranny of the Diagnosis Code

Dr. Vergil Slee and associates nicely summarize the core problems with ICD-9-CM and ICD-10-CM, and why the upgrade to ICD-10-CM is only a marginal improvement:

Slee VN, Slee D, and Schmidt HJ. The Tyranny of the Diagnosis Code. North Carolina Medical Journal, 2005;66(5):331-7.

Lest you think that this paper is unimportant, no less an authority than Dr. David Kibbe gave his approval in a commentary published in the same issue as Slee et al.'s paper (available via the same link above).

Dr. Kibbe is Director, Center for Health Information Technology of the venerable American Academy of Family Physicians (AAFP). The AAFP is the premier specialty society for family physicians in the United States, and represents 95,000 physicians.

Dr. Kibbe also shudders at the thought of using ICD-9-CM and ICD-10-CM as the foundation of the information we use to improve the quality of health care in America.

However, he stops short of endorsing a course of action to adopt an alternative to diagnosis coding that is vastly improved over ICD-9-CM and ICD-10-CM, citing a lack of political and "economic" will in America to do so.

That's a shame. The opposition of the AAFP to ICD-10-CM might have helped reverse that situation.

And if you are wondering what authority Dr. Slee may have to comment on these matters, the following is a brief biography, taken from additional commentary he made with a colleague to the House Ways and Means Committee on the issue of the switch to ICD-10-CM:

Vergil Slee, MD, MPH, FACP, FACHE (Hon) was responsible for the first deployment of ICD in hospitals as a tool for diagnosis indexing, a task for which ICD was admirably suited at that time (1955). In 1975 he represented the U. S. at the WHO conference which designed ICD-9. In 1976 he became President of the Council on Clinical Classifications which, in collaboration with the U. S. National Center for Health Statistics, developed ICD-9-CM (1978). He has analyzed ICD-10-CM (Reference 6) and has written extensively on the expanding demands on medical record information (Reference 7), demands which have destroyed ICD’s suitability for diagnosis input.

A fundamental requirement

If we're going to have a database of health care data, then the codes we use to refer to diseases ought to be in a form that we can import into a database.

In other words, we require a file that associates each code with a disease in a way that the computer can reliably extract and work with the codes.

For example, a common, machine-readable file is a comma- or tab-delimited file:

12345,Coronary artery disease
23456,Diabetes mellitus
34567,Asthma

However, neither ICD-9-CM nor ICD-10-CM are available in such a format. In fact, both disease classifications are released in a file format that makes it impossible to import codes directly into database tables.

The National Center for Health Statistics—which maintains the disease classifications in ICD-9-CM (it also includes a classification of procedures, which is a long story for another post) and ICD-10-CM—releases ICD-9-CM in a rich text file format. It releases ICD-10-CM in portable document format (pdf).

Here is a portion of ICD-10-CM from its 2,392 page, 23MB pdf file:



The net implication is that you cannot get ICD-9-CM or ICD-10-CM, from its source, in a format that you can import into a database table.

Thus, neither ICD-9-CM nor ICD-10-CM meet the most basic requirement for moving health care information into the modern era: we need disease codes in a format for use in computers, not for printing!

The purpose of this blog

The sole purpose of this blog is to advocate for a better alternative to diagnosis coding than ICD-10-CM (the International Classification of Diseases, Tenth Revision, Clinical Modification). The United States government, in particular the Centers for Medicare and Medicaid Services (CMS), is considering whether to 'switch' from coding with ICD-9-CM to ICD-10-CM. At present, CMS requires that all health care providers (hospitals, physicians, etc) who bill Medicare or Medicaid for services rendered, encode the diagnosis or diagnoses necessitating those services with ICD-9-CM.

The essence of our argument against the change is that ICD-10-CM, despite superficial improvements over ICD-9-CM, is not a sufficient advance to warrant the cost. The basic structure of ICD-10-CM is the same as that of ICD-9-CM, which was created in the era of punch cards and batch jobs. The science of medical terminology and ontology has advanced so far beyond the structure and content of ICD-9-CM and ICD-10-CM, that 'upgrading' to ICD-10-CM will preserve the Dark Ages of healthcare information.

Over the coming months, we will present our arguments against the upgrade to ICD-10-CM and in favor of a better alternative. Stay tuned.