Author: Doug Zweibel

ICD-10 Readiness Spotlight: Assessing Medical Necessity Risk

Doug Zweibel

INTRODUCTION

Medical Necessity is a term that refers to the Medicare policy whereby Medicare coverage is limited to items and services that are reasonable and necessary for the diagnosis or treatment of an illness or injury.

There are two main categories which define “what is allowed and under what circumstances.”

National coverage determinations (NCDs) are made through an evidence-based process, with opportunities for public participation. These policies affect all providers. In the absence of a national coverage policy, an item or service may be covered at the discretion of the Medicare contractors (MAC) based on a local coverage determination (LCD). These policies only affect providers in the geographic region covered by that MAC.

WHAT DOES THIS MEAN IN PLAIN LANGUAGE?

Simply put, there must be medical justification for everything we do to, with, or for a patient. For example, if a patient presented with a headache, dizziness, nausea and vomiting, one would have to question why we did an X-ray of their foot.

But everything is not always so clear. For example, if a patient had a colonoscopy, isn’t it normal to also do a biopsy of the colon? It depends on the provider documentation as recorded in the patient’s medical record. When it comes to medical necessity, provider documentation is everything.

If, during the colonoscopy, the MD documented that she discovered a polyp or cancer or other lesion of the colon, then yes, doing a biopsy would be justified.   However, if the doctor documented the findings of the colonoscopy as “normal,” then there would be zero justification to have performed a biopsy. What happens when there is no documented justification to provide a service? The answer is that we fail medical necessity and we do not get paid.

The next obvious question is “how can we tell” if procedure or treatment X will be allowed (pass medical necessity) according to the Medical Necessity rules? That’s pretty easy too. We can look up the publicly available NCDs and LCDs previously cited for that service or procedure.

Each NCDs and LCDs includes the CPT© code(s) for a service and the corresponding diagnosis code(s) that support coverage of that service. CPT© codes are a numeric value for every possible procedure. For any given procedure, there might a dozen or even scores of qualifying diagnosis codes. Sometimes, a combination of two or more diagnosis codes is required.

WHAT DOES THIS HAVE TO DO WITH ICD-10?

You may already know, or have read, or have heard that CPT© codes will not be impacted at all by the transition from ICD9 to ICD10.  That is correct.  However, the “justification” part of the NCDs and LCDs is based on diagnosis codes, which are only represented by using ICD9 (or soon, under ICD10) codes. So this is a big concern.

HOW TO PROCEED

In preparation for ICD10, NYP wants to know ahead of time if any of the procedures we do, which pass medical necessity edits today, might not do so in the future, under ICD10.

Testing for this starts with a rather “mechanical” process.

First, we rank all of the procedures we do (outpatient) by (a) volume by CPT© code and (b) total annual reimbursement per CPT©. This gives us an idea of where to start by prioritizing our work.

Next, we look at the current NCD/LCD for that particular CPT© code and identify all of the ICD9 diagnosis codes that are listed as “acceptable diagnosis codes” for that procedure.

The third step is convert each ICD9 code listed under that CPT© code’s NCD/LCD into ICD10. Since there are so many more ICD10 diagnosis codes as compared to ICD9, it is fully expected that the number of resulting “converted to” ICD10 diagnosis codes can (and probably will) be much larger than what is listed under ICD9.

Next, we look at the published NCD/LCD policies which have already been “converted” to ICD10 diagnosis code by Medicare or the MACs and compare “their” list to “our” list. We look for discrepancies: what does “our” list include that “their list” does not.

The fifth step involves “back mapping” those ICD10 codes that have been “lost” (will no longer count under the projected NCD/LCS using ICD10) to ICD9 codes.

The next step is create a report of all of the NYP cases which have the “lost” ICD9 code and the CPT© code under scrutiny, and to have them actually coded in ICD10 and see if the ICD10 code produced actually matches the “lost” code or not. If the answer is no, then we can stop because there won’t be a negative impact. But if the answer is yes, then we go to step seven.

Step seven means first adding up all of the cases which will “no longer pass medical necessity” to determine the potential reimbursement impact. If there is a large volume of such cases, we proceed with the “non-mechanical” portion of analysis. We want to further analyze each scenario and see what can be done, if anything, to mitigate the risk of not getting paid in the future for some we have been paid for in the past. This can include feedback with the providers or feedback directly to Medicare or possibly to GNYHA – or maybe to all three.

A LOT OF WORK

Medical necessity is an existing area of risk for NYP and all providers under ICD-9. ICD-10 certainly has the potential to exacerbate that risk as we move towards the implementation date of October 1, 2015. Assessing the impact of that additional risk is a key readiness activity. Mitigating that risk will continue to be of great importance to NYP’s revenue cycle and financial performance.

 

References:

http://www.cms.gov/Medicare/Coverage/DeterminationProcess/

http://www.cms.gov/Medicare/Medicare-Contracting/Medicare-Administrative-Contractors/MedicareAdministrativeContractors.html

http://www.cms.gov/Medicare/Medicare-Contracting/Medicare-Administrative-Contractors/MACJurisdictions.html

The Benefits of 10: Coding Preventable Medical Error

Doug Zweibel

by Rhonda Butler, senior clinical research analyst with 3M Health Information Systems.

See full article here.

Preventable errors in hospital care are the third leading cause of death, after cancer and heart disease. Updated estimates in a 2013 study in the Journal of Patient Safety say that between 210,000 and 440,000 people die in US hospitals every year because some preventable harm was done to them. The Office of the Inspector General published its own 2010 estimate only for Medicare patients, determining that poor hospital care was a contributing factor in the deaths of 180,000 of them. Even the universally accepted and much older number (published in 1999) of 98,000 is widely considered a “crisis” and a “national priority” in health care. And remember, these numbers don’t even attempt to quantify the financial and human toll of poor quality care that doesn’t kill people, but just costs us money and time and causes needless suffering.

Why can’t we accurately track something this important? Because ICD-9 lacks the detail to allow complications and errors in medical care to be easily entered in the permanent medical record so these events can be accurately measured. And as the saying goes, if you can’t measure something, you can’t understand it, and if you can’t understand it you can’t control it.

ICD-9 codes are pitifully inadequate for measuring the occurrence of preventable medical error. Instead of talking about the dumbest ICD-10 codes, we should be demanding the ICD-10 codes we need to start measuring and understanding and controlling these totally unnecessary healthcare costs.

Here are just a few examples that show the difference between what you can say about complications, adverse effects, and errors using ICD-9 and what you can say using ICD-10.

ICD-10 says T36.0X5A Adverse effect of penicillins
ICD-9 says 995.29 Unspecified adverse effect of other drug, medicinal and biological substance

ICD-10 says T88.4 Failed or difficult intubation
ICD-9 says 999.9 Other and unspecified complications of medical care, not elsewhere classified

ICD-10 says T88.51XA Hypothermia following anesthesia, initial encounter
ICD-9 says 995.89 Other specified adverse effects, not elsewhere classified

ICD-10 says T86.822 Skin graft (allograft) (autograft) infection
ICD-9 says 996.79 Other complications due to other internal prosthetic device, implant, and graft

ICD-10 says R57.1 Hypovolemic shock [shock from severe dehydration]
ICD-9 says 785.59 Other shock without mention of trauma

ICD-10 says Z20.6 Contact with and (suspected) exposure to human immunodeficiency virus [HIV]
ICD-9 says V01.79 Contact with or exposure to other viral diseases

ICD-10 says I97.711 Intraoperative cardiac arrest during other [non-cardiac] surgery
ICD-9 says 997.1 Cardiac complications, not elsewhere classified

ICD-10 says J95.71 Accidental puncture and laceration of a respiratory system organ or structure during a respiratory system procedure
ICD-9 says 998.2 Accidental puncture or laceration during a procedure, not elsewhere classified

What matters about codes is not how many there are, but how precise they are. If heaven forbid you or someone you care about suffers the consequences of a preventable medical mistake, look at the way it will be entered on the permanent medical record in ICD-10, and then look at how that exact same condition will be recorded in ICD-9. Lobbyists are fighting tooth and nail to keep using the ICD-9 system. Letting a few animal bite codes stop us from using the power of ICD-10 to measure and understand and begin to control preventable medical errors—really, how dumb is that?

ICD-10: Defining Clearer Boundaries

Doug Zweibel

by Rhonda Butler, senior clinical research analyst with 3M Health Information Systems.

See full article here

What makes a species distinct enough that it gets its own unique name?  Like any classification system, Linnaeus’ conceptual framework for organizing and naming living things is an exercise in drawing boundaries.  Similar things are grouped together, initially by laying out general boundaries—is it animal, vegetable, or mineral?—and making progressively finer distinctions.

All classification systems work in basically the same way, because all classification systems are products of the human mind. Classification is a profoundly human endeavor. We invent systems that allow us to organize and codify our understanding of the world and ourselves.

ICD-10 is a classification system for understanding the ways our own bodies can break or be broken, and what we can do to try and fix them. ICD-10 has more codes than ICD-9 simply because it makes finer distinctions.

To the extent possible, classification systems try to ask for the same amount of detail in drawing a next level boundary. Often this is depicted in graphic form like a tree structure—the kind of thing we see in a company’s org charts all the time. As it evolved with each annual update, ICD-9 did not make a serious attempt to be systematic in the detail it added, or consistent in the types of information it classified to the same level boundary.

ICD-10 is both systematic in its application of detail and consistent in the type of information it classifies to the same level boundary. For example, the American Congress of Obstetricians and Gynecologists (ACOG) asked the CDC’s National Center for Health Statistics (NCHS) to add detail to ICD-10 that tells them whether a pregnant woman is in the first, second or third trimester of her pregnancy. So that information about trimester of pregnancy was added systematically to the obstetrics branch of the ICD-10 tree.

Systematic and complete application of detail to an entire branch of a classification system obviously increases the number of codes by a factor of the amount of meaningful detail—three trimesters = three times as many pregnancy diagnosis codes.

And that is of course the point of any good classification, to draw boundaries that are meaningful to the people in that field. Imagine telling ACOG that all trimesters are the same, that the information they asked for in ICD-10 is unnecessary and a burden to physicians, and that they can track quality, outcomes, and do sophisticated clinical research without knowing the trimester of the patient. Maybe the opponents of ICD-10 can tell them.

ICD-10 Readiness Spotlight: Dual Coding

Doug Zweibel

In the context of healthcare, “coding” refers to the process of converting words or phrases from the medical record into standardized codes that capture the conditions of a patient (diagnosis codes) and what we did to treat those conditions (procedure codes). In the inpatient world, these codes are analyzed to yield a single Diagnosis Related Group (DRG), which is the basis of reimbursement for Medicare, Medicaid, and many other payers. The codes are also used for various types of research, e.g. on outcomes and other measures of quality. The current code set is ICD-9 (ICD stands for International Classification of Diseases), and the next version will be ICD-10.

A tool called an encoder helps the coder to arrive at each diagnosis or procedure code by asking them a series of questions about the case. Next, another tool called a grouper analyzes the codes to yield a DRG for that inpatient discharge. Because the DRG is the basis for how we get paid, getting the ICD codes correct is vital to ensure appropriate reimbursement.

Dual coding refers to the assignment of both ICD-9 and ICD-10 codes to a given patient visit during a single coding session. Because ICD-10 is almost always more complex than ICD-9, the encoder provides additional prompts to the coder in order to gain the additional information that is required to produce ICD-10 codes.

Dual coding is a transition-period function that will help us prepare for ICD-10. Some motivating factors for dual coding are listed below. As part of the ICD-10 project, NYP has coded more than 5,000 inpatient cases and hundreds more outpatient cases in ICD-10 to satisfy these goals.

  1. Coder practice and proficiency. NYP employs dozens of coding professionals to translate the care documented in patients’ medical records into billable and reimbursable diagnosis and procedure codes. This group is as affected by the transition to ICD-10 as any in the organization. The more time they are provided to practice and master coding actual claims in ICD-10, the greater chance for NYP success.
  2. System and interface testing. Dual coding allows us to determine whether our systems and interfaces are capable of handling ICD-10 codes, and to take corrective action if there are issues. This applies to both our internal systems and to external systems used by payers, clearinghouses, registries, etc.
  3. Financial impact analysis. Dual coding a case generates two DRGs – one for ICD-9 and one for ICD-10.   If the DRGs (or other metrics such as Severity of Illness, or SOI) don’t match, further investigation is warranted. Sometimes a DRG shift is due to human error, and so highlights an educational opportunity. Other times, correctly coding the chart in ICD-9 and ICD-10 results in two different DRGs, which means our reimbursement could change. In these cases, we may comment to Medicare to request that they alter their methodology.
  4. Claims testing. We want to be sure that our payers are ready to process ICD-10 coded claims, and that they get the same DRGs that we do. If our DRGs don’t match, we need to explore why.
  5. Quality metrics drift. Dual coding cases that fall within the metric population for various quality metrics, e.g. core measures and patient safety indicators, allows us to see whether the metric population may shift under ICD-10, which could result in changes to our performance on those metrics.
  6. Medical Necessity testing. Dual coding can help to identify cases that met medical necessity in ICD-9 but do not in ICD-10, or vice versa.
  7. Clinical documentation improvement. The best way to prepare providers for the additional document requirements of ICD-10 is with actual examples. When the coder is responding to the encoder’s prompts during a dual coding session, if they are not able to respond to a question because of inadequate documentation, this indicates an opportunity to provide education to providers.