Changes to Diabetes Coding Rules Create Challenges

In Q1 2016, Coding Clinic published a question and answer about "Diabetes with Associated Conditions".

On the surface, this new rule seems great--here's an excerpt from AHA's Coding Clinic:


Question:

"The ICD-10-CM Alphabetic Index entry for 'Diabetes with' includes listings for conditions associated with diabetes...Does the provider need to document a relationship between the two conditions or should the coder assume a causal relationship?"

Answer:

"... the term "with" means "associated with" or "due to," when it appears in a code title, the Alphabetic Index, or an instructional note in the Tabular List,... The classification assumes a cause-and-effect relationship between diabetes and certain diseases of the kidneys, nerves, and circulatory system....

It is not required that two conditions be listed together in the health record. However, the provider needs to document the linkage, except for situations where the classification assumes an association ...When the provider establishes a linkage or relationship between the two conditions, they should be coded as such. .... The fact that a patient has two conditions that commonly occur together does not necessarily mean they are related...."

All of this sounds like it should make things easier for both physicians and coders.  However, the opposite can be true.

As coders, it's our job to educate physicians on coding rules.  But I think this Coding Clinic ruling, and other similar ones, as potential minefields.  Physicians, by the very nature of their jobs, have to keep current with a large amount of very complex information.  Educating physicians on a rule like this has the potential to backfire.

The physician is unlikely to go to the ICD-10 and check the code title, then to the Alphabetic Index, and  continue on to search for instructional notes in the Tabular list to determine if the term "with" is present.  That may lead them to assume that the rule applies in cases where it doesn't.

There is no harm in documenting the association between two diseases, but there can be tremendous harm (in an audit situation) in failing to document that association if the rule does not apply.

So, how can we best educate our physicians?  I think that explaining that there are some limited cases where a causal relationship is assumed is a good approach.  But with that explanation, physicians must also be told that the majority of the time, a  causal relationship is NOT assumed, and that trying to remember which is which is difficult for coders who do this all day long.  The safest approach is to state the causal relationship, which ensures that coders can code what the physician actually means.  Give them examples--and make sure those examples demonstrate how this can be done without significantly increasing their workload.  By simply adding a few words, the physician can convey their clinical judgement that one disease cause another--for example "CKD 4 2^o DM2" clearly shows the physician's diagnosis that type 2 diabetes mellitus caused the patient's stage 4 kidney disease.  And it means that doctors don't have one more thing to remember.

While telling our physicians about these rules in detail may seem like a good idea, it can increase the likelihood of errors.


What coding and educational challenges do you face?  Do you have ideas for helping providers navigate documentation issues?  Be sure to leave us a comment, or send ideas you would like to share to coding@scanhealthplan.com.

Encounter Data Filtering Logic Updates

As you probably know, CMS has provided guidance regarding the filtering of encounter data for risk adjustment purposes.  This doesn't mean that either Medical Groups or Health Plans should not submit certain encounters- CMS requires that ALL encounter data must be submitted to them.

Currently, encounter data represents 25% of the calculation of risk scores.  CMS plans to increase the weighting of encounter data-based risk scores over the next couple of years by moving to a risk score incorporating 50% of the encounter data/FFS-based risk score in 2018, a risk score incorporating 75% of the encounter data/FFS-based risk score for 2019, and a risk score of 100% encounter data/FFS-based risk score in 2020,

The logic that they use to determine what encounters are included is laid out in this 2015 memo from CMS to health plans.  At the same time, CMS released a list of procedure codes that would be used for filtering professional and outpatient hospital encounters.  

Recently, CMS released the list of 2015 codes and preliminary 2016 codes to be used for encounter data filtering.

When determining which encounters may be used in calculating risk scores, health plans and provider groups should refer to these documents, as well as reports received from CMS and health plans.  As CMS moves forward with using only encounter data for risk score calculation it will become more important than ever that your encounter data be accepted by the health plan and ultimately by CMS.  We will continue to provide you as much information as possible on our Encounter Data/ICD-10 Page.  CMS is currently hosting teleconferences with health plans, to keep them informed about changes to encounter data processing.  We'll post webinar slides like these, on duplicate record and demographic data fields processing, so you can stay up to date as well.

What other tools would be helpful to you?  Remember, if you have suggestions for HCC University, or the blog, you can contact us at coding@scanhealthplan.com.

Coding Clinic Clarifies Diabetes and Complications, Ketoacidosis, and Anti-MAG Polyneuropathy

Apparently the Q1 2016 "clarification" of diabetes with associated conditions confused many people.  In Q2, 2016, Coding Clinic furthers their clarification by stating:

"The subterm "with" in the Index should be interpreted as a link between diabetes and any of those conditions indented under the word "with." The physician documentation does not need to provide a link ..... These conditions should be coded as related even in the absence of provider documentation explicitly linking them, unless the documentation clearly states the conditions are unrelated.... For conditions not specifically linked by these relational terms in the classification, provider documentation must link the conditions in order to code them as related."

Coders will have to look to the Index and/or the code description for the term "with" in order to make the determination whether or not the physician must specifically link the diabetes to the complication.


Coding Clinic also addressed Ketoacidosis in Diabetes.  They noted that physicians should be queried if they do not specify the type (i.e. Type 1 or Type 2) of diabetes.  In most cases, when a physician does not state the type of diabetes, the default is Type 2, due to coding rules.  However, ketoacidosis occurs most frequently in Type 1 diabetes. Therefore, when the physician fails to state what type of diabetes the patient has, the physician is to be queried.   This presents problems when coders are reviewing a chart note that is months old, since addenda or late entries should generally be made within a 'reasonable' time frame.  It's especially important to inform physicians that they must state the type of diabetes when documenting ketoacidosis, to avoid coding problems later.

Finally, Coding Clinic was asked what the correct code assignment for Anti-MAG (anti-myelin-associated glycoprotein) polyneuropathy is.   Coding Clinic instructed that code G62.89, Other specified polyneuropathies, should be used.

Most of the other Coding Clinic entries for Q2 were related to procedural coding.

Given the ongoing confusion about diabetic complications, I think that we can expect more Coding Clinic comments on diabetes in the future.

Although Coding Clinic has provided a lot of instruction about the documentation requirements for diabetic complications, physicians don't usually access the Index when documenting in a medical record.  Short descriptions in EMRs don't always provide enough information for physicians to know whether or not linkage exists in the description or index.  Given these limitations, it cannot hurt for physicians to include the causal relationship in their documentation when present.

Really Common Coding Errors--and how to fix them!

Sometimes, it seems like we see the same list of "common coding errors" over and over.  Mine probably won't be that different, except that I thought something was missing from those lists--the SOLUTION!  So here's my list of common coding and documentation errors--and how to fix them.

Problem: The record doesn't contain a legible signature and/or credential
Solution: If you hand write your charts, make sure your name and credential are on your progress note letterhead.  If there's more than one person in the practice, have a check box (be sure to check it!) next to each name/credential.

Problem:  The status of a disease is unclear, due to use of "history of"
Solution:   Be sure to re-state any disease being assessed/treated on this date of service in the present tense, in your assessment.  Be sure to include the patient's current status.  For example, you might note "diabetes well controlled on diet, ck HbA1C 1 week prior to next visit.  f/u 3 mos.

Problem: The documentation doesn't match the ICD-10-CM diagnosis code chosen.
Solution:  This one comes in many flavors.  Some of the most common are things like "h/o CVA 2-20-13, no neurologic deficits" and the diagnosis code is for an acute CVA.  Another common issue is with amputations--the patient lost an arm in an accident 10 years ago, and the ICD10 code chosen is for a traumatic amputation.   The last section of the ICD-10 is Factors influencing health status and contact with health services (Z00-Z99), and this section contains codes representing statuses like amputations and history of many illnesses.

Problem:  The status of a patient's cancer isn't clear.
Solution:    If the patient is still undergoing treatment, then it is considered an active malignancy.  The ICD-10-CM, Chapter 2 m. "When a primary malignancy has been excised but further treatment,
such as an additional surgery for the malignancy, radiation therapy or chemotherapy is directed to
that site, the primary malignancy code should be used until treatment is completed.

When a primary malignancy has been previously excised or eradicated from its site, there is no
further treatment (of the malignancy) directed to that site, and there is no evidence of any
existing primary malignancy, a code from category Z85, Personal history of malignant
neoplasm, should be used to indicate the former site of the malignancy."

So, if  the patient is undergoing routine surveillance only, it should be documented and coded as a "history of" the cancer.

Problem:  The documentation is not sufficient to support a diabetic complication.
Solution:  Be sure to document the history, physical exam and treatment directed at the complication.
Problem: Chronic conditions are not documented in the medical record at least annually.
Solution:  This one happens a lot--and we often see signs (for example, medications ordered) that the disease exists.  But unless you're documenting the disease, it's current status, and anything you've advised the patient, it can't be coded.

Problem: Electronic medical record is not signed.
Solution:  Assign someone the task of following up on all records that have not been properly closed by authentication every week. Be sure that all staff are aware that they must close records by authentication immediately following the visit.

Problem: Diseases are coded without proper supporting documentation.
Solution:  According to the Official Guidelines for Coding and Reporting, Section IV. Diagnostic Coding and Reporting Guidelines for Outpatient Services, J.: "Code all documented conditions that coexist at the time of the encounter/visit, and require or affect patient care treatment or management."   If there's no documentation that they required (and received) treatment or management, or had an impact on your treatment of another condition, they should not be reported.

Problem:  The diagnosis code chosen was more specific than the documentation.
Solution:   Be sure you thoroughly document all illnesses, including their complications.  Always document the patient's  response (or lack of response) to treatment.

Problem:  The diagnosis code chosen was not at the highest level of specificity documented.
Solution:  This is the opposite of the problem above--you've documented the patient's condition and complications well--but chosen a code that doesn't reflect those complications.  If the patient has multiple complications of a disease, each one should be coded separately.

Coding Clinic Q1 2016 Addresses Diabetic Complications

In the Q1 2016,  AHA Coding Clinic addresses a number of issues--two of which are hot topics for coders in risk adjustment.

Most of the questions we get in the Coding Inbox (coding@scanhealthplan.com) deal with "Diabetes and ______________", and whether or not the ICD-10CM assumes a causal relationship.  Usually, this is because the physician has not made a link in the medical record, and the coder is unsure whether or not they can code the diabetic complication.

In ICD-9, the question also came up a lot--and most often, the answer was NO.

Coding Clinic took this subject on once again, for ICD-10CM..  In Section I. Conventions, general coding guidelines and chapter specific guidelines of the Official Guidelines for Coding and Reporting, the ICD-10 describes how the word "with" is to be interpreted:



15.     “With”

The word “with” should be interpreted to mean “associated with” or “due to”
when it appears in a code title, the Alphabetic Index, or an instructional note in the Tabular 
List.

In a code title, the association might look like this:

E11.31  Type 2 diabetes mellitus with unspecified diabetic retinopathy

This tells you that there is an association between the two diseases.
_____________________________________________________________

The word “with” in the Alphabetic Index is sequenced immediately following the main term, not in
alphabetical order.

In part, the Alphabetic Index under "Diabetes" looks like this:

Diabetes, diabetic (mellitus) (sugar) E11.9
- with
- - amyotrophy E11.44,- - arthropathy NEC E11.618
- - autonomic (poly)neuropathy E11.43
- - cataract E11.36
- - Charcot's joints E11.610
- - chronic kidney disease E11.22
- - circulatory complication NEC E11.59
- - complication E11.8
- - - specified NEC E11.69
- - dermatitis E11.620
- - foot ulcer E11.621
- - gangrene E11.52
- - gastroparesis E11.43
- - glomerulonephrosis, intracapillary E11.21
- - glomerulosclerosis, intercapillary E11.21
- - hyperglycemia E11.65
- - hyperosmolarity E11.00
- - - with coma E11.01
- - hypoglycemia E11.649
- - - with coma E11.641
- - kidney complications NEC E11.29
- - Kimmelsteil-Wilson disease E11.21
- - loss of protective sensation (LOPS) —see Diabetes, by type, with neuropathy
- - mononeuropathy E11.41
- - myasthenia E11.44
- - necrobiosis lipoidica E11.620
- - nephropathy E11.21
- - neuralgia E11.42
- - neurologic complication NEC E11.49
- - neuropathic arthropathy E11.610
- - neuropathy E11.40
- - ophthalmic complication NEC E11.39
- - oral complication NEC E11.638
- - periodontal disease E11.630
- - peripheral angiopathy E11.51

With this new Coding Clinic clarification, when the physician documents and another disease, you need to check the Alphabetic Index first, to see if the problem/complication is listed there, and then as always, go to the code itself.  If the condition is qualified by the term diabetes with ____, the you can code the complication.  The doctor does not have to state the complication--it is assumed.   If you don't find that causal relationship is supported, then the two diseases are coded as unrelated.

This clarification should help make coding more accurately reflect what the physician is trying to convey to the coder.  As always, we as coders have to be careful to ensure that we double check the ICD-10CM to be sure we're coding it right.

Coding Clinic Q1 2016 Addresses Heart Failure Coding

Hi everyone!  It's been a while since I've posted anything--I had an unexpected "vacation" from work.

While I was away, the AHA's Coding Clinic was busy clarifying ICD-10 coding for us.  There were many situations clarified in Q1 of 2016, but a couple of them warrant blog posts.

Heart failure coding has been a challenge for quite some time--with coders and physicians going bck and forth.  The physicians telling us that the patient has heart failure, the coders telling the physcians that it doesn't say that in the chart!

Earlier Coding Clinic guidance (Q1 2014) indicated that the coder could not assume diastolic heart failure, systolic heart failure or combined diastolic/systolic heart failure when the physician documented heart failure with reduced ejection fraction (HFrEF) or heart failure with preserved ejection fraction (HFpEF).  Thanks to input from the American College of Cardiology,  these contemporary terms for heart failure have been now been accepted by Coding Clinic.

Going forward, when the provider has documented HFpEF, HFrEF, or other similar terms noted above, the coder may interpret these as "diastolic heart failure" or "systolic heart failure," respectively, or a combination of both if the provider has documented it,  and assign the appropriate ICD-10-CM codes based on these definitions.

Just as an FYI--we've posted new information to HCC University.  The 2016 Risk Score Calculator is online, there's an updated document with all of the ICD-10 codes by HCC and the updated Companion Guides for encounter data submission.  Stay tuned for exciting HCC U news in the very near future!

Finally, remember, when you have coding questions, you can contact us at Coding@scanhealthplan.com.

Coding "Cardiac Dysfunction" in ICD-10

I was recently asked how to code "diastolic dysfunction" in ICD-9.  The inquirer asked if it should be coded as "heart failure, NOS based on something she had read.

In 2009, Q1, Coding Clinic noted that you could not assume heart failure in a patient with diastolic dysfunction, and that it should be coded as 429.9, heart disease, unspecified.  

The term dysfunction is very non-specific, and runs the gamut from very mild, to extremely severe.  There may be minimal dysfunction present, or life threatening dysfunction.  As with a lot of documentation, there's simply not enough information there to select a specific diagnosis code.  Therefore, the best you can do is choose a non-specific code, that is not more severe than the documentation.

In ICD-10, "heart dysfunction" is indexed to I51.89--Other ill-defined heart diseases.  This seems like it follows similar logic, and it makes sense--ill defined documentation corresponding to an ill defined heart disease.