Wednesday, September 21, 2011

The Challenges of ICD10 Implementation

On October 1, 2013, the entire US healthcare system will shift from ICD9 to ICD10.   It will be one of the largest, most expensive and riskiest transitions that healthcare CIOs will experience in their careers, affecting every clinical and financial system.   It's a kind of Y2k for healthcare.

Most large provider and payer organizations, have a ICD10 project budget of $50-100 million, which is interesting because the ICD10 final rule estimated the cost as .03% of revenue.  For BIDMC, that would be about $450,000.   Our project budget estimates are about ten times that.

CMS and HHS have significant reasons for wanting to move forward with ICD10 including
1) easier detection of fraud and abuse given the granularity of ICD10 i.e. having 3 comminuted distal radius fractures of your right arm within 3 weeks would be unlikely
2) more detailed quality reporting
3) administrative data will contain more clinical detail enabling more refined reimbursement

Large healthcare organizations have already been working hard on ICD10, so they have sunk costs and a fixed run rate for their project management office.   At this point, any extension of the deadline would cost them more.

Most small to medium healthcare organizations are desperate. They are consumed with meaningful use, 5010, e-prescribing, healthcare reform, and compliance.   They have no bandwidth or resources to execute a massive ICD10 project over the next 2 years.

Vendors have told me such things as "I have been amazed at how much we (and our third-party partners) are spending on getting prepared for ICD10 – and it's not what you would expect (extending data tables, new code lookup tools, etc.)  It's a whole host of clinician assistance tools, new documentation workflows, new kinds of provider-facing decision support to maximize coding revenue while guarding against RAC audits - all simply for billing!"

In my CIO role, not any state or federal role, I will continue to listen to concerns about ICD10, sharing them broadly on my blog and with government leaders who will listen.

The Wall Street Journal recently published an article about the granularity of ICD10.

One of my staff posted this response, which is very thoughtful:

While nice-to-have, ICD-10 comes at a time when substantial cuts await providers. The "super committee" is deliberating on these now for Medicare and Medicaid. Adding more administrative overhead to the U.S. healthcare system is untimely and will ultimately impact clinical care. Our health care system already has twice the administrative overhead of other advanced nations. We arguably have the most complex medical reimbursement system in the world. ICD-10 makes it worse.

When HHS published the requirement for ICD-10 in the January 16, 2009 Federal Register, they estimated transition costs for health care providers to be 0.03% of patient revenues. For a $1B medical center, this would be $300,000. Based on experience at our hospital and that of my colleagues at other hospitals, they missed it by a factor of 10 or more.

When a regulation of this magnitude is published, various laws and executive orders require a Regulatory Impact Analysis. Some requirements are intended to protect small businesses and non-profits from burdensome, unfunded federal mandates. The marginal cost estimate published in the Federal Register for ICD-10 was $2.966 billion over the period 2011 to 2025. Two-thirds of this was transitional cost. The benefits were estimated at $4.540 billion.

HHS has a tradition of low-balling cost estimates. Further evidence can be seen with recent estimates of HITECH privacy regulations.

If Congress was doing its job of regulatory oversight, they would sponsor hearings to learn what payers and providers are actually spending on ICD-10 conversion. Costs for consulting services alone run into the millions. This does not count the application software conversion, training and education, and other "in-house" costs. At our medical center, we would be paying $380,000 according to HHS estimates. Instead, the marginal cost of ICD-10 will be in excess of $5m. For multi-hospital systems, the costs may exceed $100m.

A Congressional review of transition costs would turn the regulatory impact assessment on its head. Costs could easily become double the estimated benefit savings.

With ICD-10, the government is perpetuating a reimbursement system that is far too complex. We spend more than any other country on healthcare administrative overhead. The Medicare Claims Processing Manual, for example, is over 4,000 pages in length. The reimbursement system needs simplification to bring the cost of this function in line with other industries.

Recently, HHS began promoting a "global payment" initiative. This had the potential for simplifying reimbursement, but they over-laid it on top of the existing system. Instead of substitution, it was additive. You bill as usual and then have a settlement process that adds one more layer of administrative overhead.

Unfortunately, there are too many activities within and outside the government whose livelihood depends on perpetuating this complex system. It is akin to the Internal Revenue Code. There are also groups who promote ICD-10 for its more granular health care research potential. This is laudable, but not affordable. There is no "free lunch". Every dollar spent on administrative overhead is one less dollar spent on clinical care.

What's needed is a fresh look at the reimbursement system. While ICD is used in other countries, it is not used for reimbursement purposes. Rather, it is used for health statistics and reporting. Using it for reimbursement adds an entirely different dimension. Because it is used for reimbursement, the U.S. version requires numerous extensions. Read this as more codes and more complexity.

Our health care system needs to change. If we are going to cut cost, let it be overhead, not clinical care.


Dave said...

Terrific piece, John. The key statement of your post is "What's needed is a fresh look at the reimbursement system." You couldn't be more correct. I'd encourage your readers to Google "Do it yourself health reform" and/or read the piece I wrote about "The Most Important Organization In Silicon Valley That No One Has Heard About" (

Healthcare providers/innovators such as Samir Qamar (MedLion), Garrison Bliss (Qliance), Brian Koniver (Organic Medicine), Brian Forrest (Physician Care Direct) and others have done exactly what you suggested - i.e., come up with a better care & payment model that has shown to dramatically lower costs AND improve outcomes by completing avoiding what I call the present reimbursement model - a Gordian Knot designed by Rube Goldberg.

ICD10 will have the effect of accelerating an existing trend of insurance-free practices, particularly for day-to-day medicine. We don't use insurance for our car tune-ups. Why would we incur the insurance bureaucracy for the health equivalent of a tune-up?

If anyone wanted the detail on the positive outcomes ref'ed above, I have a PPT that goes into detail I'm happy to share. Email me at dave [at] avado {dot} com.

Chris Tonozzi, MD said...

Wow, excellent analysis. This brings it home. The momentum builds for a single payer . . .

JGF said...

I wondered when you'd put this out. It's been brewing for some time.

My preference about six years ago, when the NLM paid Rand to evaluate ICD-10, was to keep hacking with ICD-9 (could be done) and accelerate a SNOMED-backed version of ICD-11.

As challenging as ICD-10-CM is, I can think of a number of ways to manage it.

ICD-10-PCS however -- that's a whole new ballgame. Nothing in common with ICD-10-CM or ICD-9 procedures save the name.

It's very neat if you understand how 3M worked within the constraints of simple data stores, but it's hard to imagine HCIT making use of it. I also wonder who owns the IP for ICD-10-PCS.

Your bottom line on the consequences of piling hacks atop our current schemes for healthcare accounting cost is absolutely right.

What a mess. Glad you got it off your chest.

Anonymous said...

I couldn't agree more. Glad you wrote it. Now we should all spread it around and make sure everyone else hears about it.

conor dowling said...

you hit the nail on the head with "The reimbursement system needs simplification ... While ICD is used in other countries, it is not used for reimbursement purposes".

The line is usually "have expansive terminology for clinical care and succinct classification for billing". There's lip service for SNOMED for the first, but ICD 10 CM is what exactly? (I wrote a comparison of their granularities in fishy about ICDs).

Problem is, has the horse bolted? Can the classification-to-rule-them-all be stopped?

Anonymous said...

We have taken action and the result was the AMA resolution, which importantly contains a clause that asks the AMA to work with informatics organziations to find a suitable replacement for ICD-9.

ICD-9 is dying, but going to ICD-10 will put us in the same position at the end of this decade as we are now - behind the rest of the world, which will be on ICD-11 or its replacement.

Thanks for the post - I hadn't seen it until today.

Anonymous said...

Yes to a single payer system. Have to stop the gluttony of insurance companies. The fact we are still using dual coding systems (CPT and ICD) at all is a testimony to the inefficiency in the system. Coding in healthcare should have been about epidemiology and statistics as originally designed, and not bastardized into a reimbursement systems that is as screwed up as the implementation date. Why else would be 'still' be trying to figure out the DRG system today!? As an experienced HIM professional and consultant, unfortunately I see few winners in this debacle, unless some true and significant change occurs in the reimbursement model for sure. Good points John and Dave