Why ICD-10?

Posted on March 24, 2014 I Written By

Kyle is Founder and CEO of Pristine, a company in Austin, TX that develops telehealth communication tools optimized for Google Glass in healthcare environments. Prior to founding Pristine, Kyle spent years developing, selling, and implementing electronic medical records (EMRs) into hospitals. He also writes for EMR and HIPAA, TechZulu, and Svbtle about the intersections of healthcare, technology, and business. All of his writing is reproduced at kylesamani.com

At least half a dozen folks have asked me to explain why HHS is mandating the transition to ICD-10. So I thought I’d write a blog post about the subject.

First, I’ll examine some of the benefits that proponents of ICD-10 site. Then, I’ll examine the cost of transition from ICD-9 to ICD-10.

There are about a dozen frequently cited reasons to switch from ICD-9 to ICD-10. But they can be summarized into three major categories:

1) The US needs to catch up to the rest of the world.

2) The more granular nature of ICD-10 will lend itself to data analysis of all forms – claims processing, population health, improved interoperability, clinical trials, research, etc.

3) ICD-9 doesn’t support the latest diagnoses and procedures, and ICD-10 does.

Regarding #1, who cares? Coding standards are intrinsically arbitrary. Sequels are not necessarily better than their predecessors.

Although #2 sounds nice, there are a lot of problems with the supposed “value” of more granular data in practice. Following the classic 80-20 rule of life (80% of value comes from 20% of activity), the majority of codes are rarely used. By increasing the number of codes six-fold, the system is creating 6x the opportunities to inaccurately code. There is no reason to believe that providers will more accurately code, but the chances of incorrect diagnosis are now significantly higher than they were before. Garbage in, garbage out.

Below are some specific examples of how increasing the number of codes will affect processes in the healthcare system:

Payers – payers argue that making codes more granular will improve efficiency in the reimbursement process by removing ambiguity. There is nothing further from the truth. Payers will use the new granularity to further discriminate against providers and reject claims for what will appear to be no reason. With 6x the number of codes, there are at least 6x as many opportunities for payers to reject claims.

Clinical trials – ICD-10 proponents like to argue that with more granular diagnosis codes, companies like ePatientFinder can more effectively find patients and match them to clinical trials. This notion is predicated on the ability of providers to enter the correct diagnosis codes into EMRs, which is a poor assumption. Further, it doesn’t actually address the fundamental challenges of clinical trials recruitment, namely provider education, patient education, and the fact that most patients aren’t limited to trials by diagnosis codes, but rather by other data points (such as number of years with a given disease and comorbidities).

Public health – ICD-10 proponents also claim that the new coding system will help public health officials make better decisions. Again, this is predicated on accuracy of data, which is a poor assumption. But the greater challenge is that the most pressing public health issues of our time simply don’t need any more granularity in diagnosis codes. Public health officials already know what the top 20 public health problems are. Adding 6x the number of codes will not help address public health issues.

Regarding #3, why do we need to reinvent the entire coding system and make the entire system more granular to accommodate new diagnoses and procedures? Why can’t we continue to use the existing structure and simply create new branches of the ICD tree using alphanumeric characters? Why do we need to complicate every existing diagnosis and procedure to support new diagnoses and treatments? We don’t. There are plenty of letters left to be utilized in ICD-9 to accommodate new discoveries in medicine.

Next, I’ll provide a very brief summary of the enormity of the cost associated with transitioning from ICD-9 to ICD-10. The root of the challenge is that a string of interconnected entities, none of whom want to work with one another or even see one another, must execute in sync for the months and years leading up to the transition. Below is a synopsis of how the stars must align:

EMR vendors – EMR vendors must upgrade their entire client base to ICD-10 compliant versions of their systems in the next couple of months to begin testing ICD-10 based claims. Given the timescales at which providers move, the burden of MU2 on vendors, and the upgrade cycles for EMR vendors, this is a daunting challenge.

Providers – providers don’t want to learn a new coding system, and don’t want to see 6 times the number of codes when they search for basic clinical terms. Companies such as IMO can mitigate a lot of this, but only a small percentage of providers use EMRs that have integrated with IMO.

Coding vendors – like EMR vendors, auto-coding vendors must upgrade their clients systems now to one that supports dual coding for ICD-9 and ICD-10. They must also incur significant costs to add in a host of new ICD-10 based rules and mappings.

Coders – coders must achieve dual certification in ICD-9 and ICD-10, and must double-code all claims during the transition period to ensure no hiccups when the final cut over takes place.

Clearinghouses – clearinghouses must upgrade their systems to support both ICD-9 and ICD-10 and all of the new rules behind ICD-10, and must process an artificially inflated number of claims because of the volume of double-coded claims coming from providers.

Payers – payers must upgrade their systems to receive both ICD-9 and ICD-10 claims, process both, and provide results to clearinghouses and providers about accuracy to help providers ensure that everyone will be ready for the cut over to ICD-10.

The paragraphs above do not describe even 10% of the complexity involved in the transition. Reality is far more nuanced and complicated. It’s clear from the above that the likelihood that all of the parties can upgrade their systems, train their staff, and double code claims is dubious. The system is simply too convoluted with too many intertwined but unaligned puzzle pieces to make such a dramatic transition by a fixed drop-dead date.

Lastly, switching to ICD-10 now seems a bit shortsighted in light of the changes going on in the US healthcare system today. ICD-10 is already a decade old, and in no way reflects what we’re learning as we transition from volume to value models of care. It will make sense to change coding schemes at some point, but only when it’s widely understood what the future of healthcare delivery in the US will look like. As of today, no one knows what healthcare delivery will look like in 10 years, let alone 20. Why should we incur the enormous costs of the ICD-10 transition when we know what we’re transitioning to was never designed to accommodate a future we’re heading towards?

At the end of the day, the biggest winners as a result of this transition are the consultants and vendors who’re supporting providers in making the transition. And the payers who can come up with more reasons not to pay claims. Some have claimed that HHS is doing this to reduce Medicare reimbursements to artificially lower costs. Although the incentives are aligned to encourage malicious behavior, I think it’s unlikely the feds are being malicious. There are far easier ways to save money than this painful transition.

The ICD-10 transition may be one of the largest and most complex IT coordination projects in the history of mankind. And it creates almost no value. If you can think of a larger transition in technology history that has destroyed more value than the ICD-9 to ICD-10 transition in the US, please leave a comment. I’m always curious to learn more.