by Erica E. Remer, MD, FACEP, CCDS
Auditors target multiple conditions which for clinical validation denials (CVD). Personally, I found acute kidney injury (AKI) and malnutrition the most commonly defensible targets. On the other hand, I often agreed with auditors on their CVDs for pneumonia and urinary tract infections. Other frequent CVDs included sepsis, encephalopathy, and respiratory failure.
Certain diagnoses are susceptible to CVDs for the following reasons:
- DRG downgrade: Auditors target medical records with only one CC/MCC because doing so downgrades the DRG and results in less reimbursement.
- Empiric treatment: Providers start antibiotic therapy early for patients who present with signs of infection (e.g., fever, leukocytosis, altered mental status) on as they seek the infection’s source. Sometimes the physician cannot identify the source or the etiology of the disease turns out to be a different than originally expected. Clinicians also need to be careful to not propagate the original debunked diagnosis via copy and paste, so it gets wrongly coded.
- Documentation consistency: Physicians should document their medical decision making process throughout the patient’s stay. Best practice is to document when a diagnosis is initially considered (may be in uncertain format), when the diagnosis is definitively ruled in, and when the condition resolves. The physician should recap this information in the discharge summary. Only mentioning a diagnosis once in the medical record, while permissible, raises an auditor’s interest and begs the question of whether the condition really was present.
- Pursuant to a query: When providers need to be queried to make a diagnosis codable, and they agree without supplying any clinical support, a red flag goes up—and I mean like waving one in front of an auditor like a bull, and not just signaling peril up ahead.
Finally, auditors target diagnoses with uncertain or emerging clinical guidelines. Clinical guidelines change, but it takes time for medical practice to adjust (e.g., malnutrition, sepsis). If a guideline is not universally adopted (for example, discussions regarding the new Sepsis-3 definitions) some variability in medical practice is allowed. That doesn’t necessarily mean a provider is wrong if he or she does not follow the latest guideline. As long as the provider is within the acceptable range of practice, he or she just needs to demonstrate the clinical considerations of the case and the auditor should accept the diagnosis. If a provider deviates from clinical criteria or guidelines, he or she should document the mitigating circumstances (e.g., on beta-blockers, previous antibiotic therapy, contamination).
On the other hand, the provider needs to consider established guidelines. If the average, prudent similarly qualified practitioner wouldn’t call an asymptomatic deviation of sodium by 1 mEq/L, hyponatremia, neither should you.
Similarly, AKI has criteria of change within the previous 48 hours, or deviation from a baseline from seven days ago, but a provider could make a convincing argument that the patient’s serum creatinine (SCr) is always X and the acute derangement is likely to have occurred since the onset of symptoms 36 hours ago. It would be quite serendipitous to randomly have a baseline drawn within seven days of an index visit if AKI preceded hospitalization, wouldn’t it? However, if the SCr is only off by 0.1 mg/dL, you are hard-pressed to spin that as AKI.
While the coder may not really the arbiter of clinical validation, coders should be empowered to refer questionable records for a clinical review either by the CDI specialist, the attending, or a physician advisor/champion, and coders need to know which conditions are vulnerable and which clinical indicators to consider.
Absence of abnormal clinical indicators does not mean the condition is definitively not present; it means the encounter needs clinical review and the condition may require more documentation to support it.
Editor’s note: This article, written by Erica E. Remer, MD, FACEP, CCDS, founder and president of Erica Remer, MD, Inc., Consulting Services, first appeared in its entirety, in JustCoding. Advice given is general. Readers should consult professional counsel for specific legal, ethical, clinical, or coding questions. Contact her at email@example.com. For the third part of this article, return to the blog next week!
by Erica E. Remer, MD, FACEP, CCDS
Clinical validation denials (CVD) result from a review by a clinician, such as a registered nurse, contractor medical director, or therapist, who concludes, retrospectively, that a patient was not really afflicted by a condition documented in the medical record and coded by the coder.
If a coder assigns a code for a condition not really present, and removing that code assignment results in a lower-weighted DRG, then it is reasonable for a payer to expect the overpayment back.
However, if the condition was indeed present, medical personnel invested time, energy, supplies, and other resources, the hospital is entitled to reimbursement. Therefore, it is not appropriate to remove a diagnosis which was genuinely present. It is also unfair to remove legitimate diagnoses, because this results in the downgrade of severity of illness and complexity of management, and falsely deflates the quality measures assigned for that patient’s care.
It has never been reasonable or compliant for a coder to infer medical conditions from clinical indicators, and it is not reasonable to expect a coder to decide that a condition doesn’t exist if the provider documented it.
The assignment of a diagnosis code is based on the provider’s diagnostic statement that the condition exists. The provider’s statement that the patient has a particular condition is sufficient.
Although the provider’s statement may be sufficient to code a particular condition, it does not mean the condition indisputably exists.
The adage is, “if you didn’t document it, you didn’t do it.” But we all know that is not true. No physician documents every thought or action, and it would be absurd to expect them to do so. The dilemma arises when a physician documents a condition without giving adequate evidence as to what led him or her to make the diagnosis.
Concurrently, a documented diagnosis which is not really present could lead other healthcare providers down an erroneous path. Conversely, not providing clinical support for a valid diagnosis sets the stage for future denials.
The legal definition of the standard of care is managing a patient at the level at which the average, prudent, similarly qualified practitioner in a given geographic medical community, would be providing medical care under the same or similar circumstances. There are clinical guidelines and scores which may assist a clinician in making diagnoses, but prudent practitioners also bring their past experience, knowledge, and judgment into play.
Medicine is an art, not a science, and not every patient “reads the textbook.” Many patients’ lives have been saved by a clinician who followed his or her gut. Just because a patient doesn’t strictly meet clinical criteria, doesn’t preclude him or her from having the medical condition that an astute provider diagnosed.
Is there a standard of documentation similar to that legal definition of a standard of care? The purpose of patient record documentation is to foster quality and ensure continuity of care. It is clinical communication. The fallacy is that documentation needs to be expansive and long; it just needs to convey to the subsequent healthcare provider (as well as the coder, utilization/case manager, auditor, lawyer, etc.) what the provider was thinking and why.
What coders can do
Sensible, qualified, and experienced coders or CDI specialists may read documentation and have concerns that a diagnosis is not supported by the clinical indicators.
Do they just unfailingly code a documented condition because the Official Guidelines for Coding and Reporting say that the provider’s statement is sufficient, or do they query the physician? Which conditions are prone to this? What clinical indicators should they be considering? How does one broach this subject with the physician?
The ACDIS/AHIMA’s Guidelines for Achieving a Compliant Query Practice recommend generating a query when the health record documentation “provides a diagnosis without underlying clinical validation.” It notes that “the focus of external audits has expanded in recent years to include clinical validation review,” and instructs coders to follow CMS and Coding Clinic guidelines, and to “query the physician when clinical validation is required.”
The CMS Statement of Work for the Medicare Fee-For-Service Recovery Audit Program 2013 notes that “clinical validation is beyond the scope of DRG (coding) validation, and the skills of a certified coder.”
I concur that a certified coder who has identified a diagnosis which needs more clinical support does not have the ability to validate it, but has the skills to recognize the necessity for validation.
If your institution has CDI specialists, then you have someone with the appropriate credentials in place to generate a query. The physician is the one who performs the validation by responding to the query in the affirmative, and by providing their clinical evidence for the diagnosis in question.
Editor’s note: This article, written by Erica E. Remer, MD, FACEP, CCDS, founder and president of Erica Remer, MD, Inc., Consulting Services, first appeared in its entirety, in JustCoding. Advice given is general. Readers should consult professional counsel for specific legal, ethical, clinical, or coding questions. Contact her at firstname.lastname@example.org. For the second part of this article, return to the blog next week!
Q: In terms of coding blood transfusions, does the documentation of which intravenous (IV) site used has to come from the physician in the progress note, or can this particular information be extrapolated from nursing notes, orders, etc.? As far as I can tell, a blood transfusion is usually administered to whatever peripheral IV line/site is available, unless otherwise contraindicated, or a specific physician order directs not to.
A: There are a number of procedure codes within ICD-10-PCS that represent procedures performed by staff not considered to be an independent provider. Depending upon the organization, this could be excisional debridements, peripherally inserted central catheter (PICC) line placement, irrigations, dressing changes, physical therapy activities, swallow studies, various radiology procedures, etc.
We have direction from an AHA Coding Clinic® that tells us that we may assign procedure codes based on the documentation of a non-physician professional when the professional provides the specific service. This applies to procedure coding only where there is documentation to substantiate the code. (AHA Coding Clinic® for ICD-10-CM/PCS First Quarter 2014).
So, in your example of the blood transfusion, if there are appropriate orders to administer a blood transfusion and the nursing documentation supports the assignment of the procedure code, you can certainly assign the code based on the nurse’s documentation.
Editor’s Note: Laurie L. Prescott, RN, MSN, CCDS, CDIP, AHIMA Approved ICD-10-CM/PCS Trainer, and CDI Education Director at HCPro in Danvers, Massachusetts, answered this question. Contact her at email@example.com. For information regarding CDI Boot Camps visit www.hcprobootcamps.com/courses/10040/overview.
By Michelle A. Leppert, CPC
Ah, the Fourth of July, picnics, parades, and pryotechnics. What could be better? Well, not having your family and friends end up at Fix ‘Em Up Clinic the next day would be a good start. Alas, holidays here in Anytown never go off without a hitch, so let’s see who has wandered in with a holiday malady.
Doug was running around with a lit sparkler and one of the sparks flew into his eye. So what kind of injury does Doug have? If the little metal shaving from the sparkler is still in his eye, he may have a foreign body in the cornea (T15.0-) or a foreign body in the conjunctival sac (T15.1-). Alternately, he could have a corneal abrasion without a foreign body (S05.0-).
The code we choose will ultimately depend on the physician’s documentation [more]
“Encephalopathy is a great big monster,” says Timothy N. Brundage, MD, CCDS, medical director of Brundage Medical Group in Redington Beach, Florida. Coders and clinical documentation improvement (CDI) specialists want physicians to document encephalopathy, when appropriate, because it is an MCC.
By definition, encephalopathy is a global cerebral dysfunction in the absence of structural brain disease, Brundage says. “That definition is very nebulous.”
Unfortunately, providers often describe encephalopathy instead of diagnosing it, says Cheryl Ericson, MS, RN, CCDS, CDIP, [more]
Q: When I started CDI, I was told that when a complication code happens to be the reason of admission, along with another condition also contributing to the admission, the complication code takes precedence over the other condition code. Is this correct, and is there any written evidence, like a Coding Clinic that tells me to do so?
A: The code set offers us direction within the alphabetic index and tabular list related to sequencing with notes that instruct us to code first, or code also, for example. There is instruction within the Official Guidelines for Coding and Reporting as to how to interpret the directional notes, found in Section I. The coding conventions is often the first place I check.
And, of course, the AHA Coding Clinic also gives us guidance. There is a hierarchy of what piece of guidance supersedes the other. The instructions within the index and tabular list (coding conventions) is the highest, followed by the Official Guidelines of Coding and Reporting, and, last, the AHA Coding Clinic.
In the Official Guidelines of Coding and Reporting Section II: Selection of Principal Diagnosis, Section G states:
“Complications of surgery and other medical care when the admission is for treatment of a complication resulting from surgery or other medical care, the complication code is sequenced as the principal diagnosis. If the complication is classified to the T80-T88 series and the code lacks the necessary specificity in describing the complication, an additional code for the specific complication should be assigned.”
Thus, your understanding of how these should be sequenced is absolutely correct and now you are able to state where you accessed this instruction.
Editor’s Note: Laurie L. Prescott, RN, MSN, CCDS, CDIP, AHIMA Approved ICD-10-CM/PCS Trainer, and CDI Education Specialist at HCPro in Danvers, Massachusetts, answered this question. Contact her at firstname.lastname@example.org. For information regarding CDI Boot Camps visit www.hcprobootcamps.com/courses/10040/overview.
Q: Can CDI programs use the information on ambulance forms or trip tickets to abstract from if the information is pulled into or reiterated in the ED or H&P documentation? Our staff doesn’t want to miss criteria that would diminish our ability to substantiate the true severity of illness of some patients, but I have been informed that coders are not allowed to code from ambulance papers or information.
A: There are a few issues to consider here. First, can you code from EMT documentation, such as trip sheets? No. Although these documents are often included in the health record, these documents are not “owned” by the hospital. They are usually classified as external correspondence. If the claim is selected for complex review, the EMT trip sheet cannot be released. As such, it can’t be used to support code assignment.
There is one caveat to this statement. In ICD-10-CM, when implemented, the code for Glasgow coma requires a character that indicates when the assessment was made, which can include those made by an EMT. Coding Clinic from 1st Quarter 2014 states:
“. . . If the EMT documents the patient’s initial GCS core in the field, can the EMT’s documentation be used? Coders are concerned there is no official advice or guideline that allows the use of nonphysician documentation for Glasgow coma scores. . . “The response was, “It would be appropriate to use the pre-hospital report containing the EMT’s documentation and other nonphysician documentation to determine the Glasgow coma score.”
Second, there could be an issue with how the provider is reiterating the EMT findings in the health record. The provider is expected to provide a history of present illness as part of the history and physical. However, conditions not related to the current episode of care should not be reported. The documentation by the provider needs to clearly show the conditions that exist at the time of admission, rather than just listing an overall history.
Sometimes a coder’s perspective is different than a clinician’s regarding what they define as a history of a condition. Often, if a provider fails to carry a diagnosis throughout the health record, and doesn’t include it in the discharge summary, it may not be perceived as reportable by a coder. Many coders begin the coding process with the discharge summary, because it is the final word of the attending provider. However, it is important to note that Coding Clinic 1st Qtr. 2014 states “documentation is not limited to the face sheet, discharge summary, progress note, history and physical, or other report designed to capture diagnostic information. This advice only refers to inpatient coding.”
Just because the provider doesn’t mention a diagnosis more than once does not mean it isn’t reportable. Oftentimes, the provider’s focus changes daily, so they may not feel the need to summarize conditions that are no longer a focus of their efforts. If there is a disagreement between CDI and coding, it is best to clarify with the provider, assuming the totality of health record supports the condition as reportable.
If the provider only mentions the condition(s) in the history and physical, it might be helpful to query for the status of the condition to see if it should be reported. For example, if the provider, in their history and physical, documents “early clinical sepsis” and it is never documented again, be sure there are clinical indicators that support it as a reportable diagnosis. If there are clinical indicators to support it as a reportable condition than your query may be as follows:
Please clarify the status of the condition “early clinical sepsis” as documented in the H&P in this patient who presented with (give specific s/sx) and was treated with (give specifics) or had the following diagnostics (give specifics), etc. Was the “early clinical sepsis”
- Confirmed and ongoing
- Confirmed and resolved
- Ruled out
- Without clinical significance
- Unable to determine
Also note, the multiple choice format would only work well if your organization maintains the query as part of the health record so it would need to be validated by the provider. If the provider responds, by confirming the diagnosis (either ongoing or resolved), it would be reportable. If the provider responds with any other choice, it would not be reportable.
Keep in mind that you can use clinical indicators obtained from EMT documentation to query the provider if there appears to be an undocumented, reportable condition relevant to the current episode of care, if the current provider documentation doesn’t support code assignment.
Editor’s Note: Cheryl Ericson, MS, RN, CCDS, CDIP, AHIMA Approved ICD-10-CM/PCS Trainer, Associate Director for Education at ACDIS and CDI Education Director at HCPro in Danvers, Mass, contributed to this post.
First into the Fix ‘em Up Clinic today is Jeff. He took part in a s’more eating contest at camp last night. I’ve personally never understood the appeal of burned marshmallows, but Jeff, well he was so determined to claim the s’mores title that he ate a few marshmallows that were a little too hot. As in, they were on fire. And while fire eating is fine for professionals, for a kid at camp, it’s not such a great idea.
Dr. Sunni Daze examines Jeff and documents burns to the mouth, pharynx, tongue, and lips. The burns of the mouth, pharynx, and tongue are easy. One code covers all three and it does not specify degree of the burn. Since this is Jeff’s initial visit, we would report T28.5XXA.
The lip burns require a little more information. We need to know what degree of burns Jeff suffered on his lips. Fortunately for him, Dr. Daze notes the burns are first degree, so we would report T20.12XA (burn of first degree of lip[s]).
ICD-10-CM does not include separate codes for the upper and lower lip, so T20.12XA covers one lip or both.
We also find the following note under pretty much all of the burn codes:
- Use additional external cause code to identify the source, place and intent of the burn (X00-X19, X75-X77,X96-X98, Y92)
We definitely need an X00-X19 code, which in Jeff’s case is X10.1XXA (contact with hot food, initial encounter).
The X75-X77 codes are for intentional self-harm. Overeating burning marshmallows doesn’t quite qualify as planning to hurt yourself. Jeff just got caught up in the moment.
The X96-X98 are codes for assault. Again, not applicable in Jeff’s case.
For our place of occurrence, we’ll use Y92.833 (campsite as the place of occurrence of the external cause). Notice we do not need a seventh character for this code.
Why in the world is Mr. Grinch so mean? Maybe the problem is his health. Let’s see if we can diagnose the Grinch’s health woes.
First, he is as cuddly as a cactus. What does that mean? He’s covered in spines? A better explanation is he suffers from eczema, which causes redness, skin edema (swelling), itching and dryness, crusting, flaking, blistering, cracking, oozing, or bleeding. Great, we have a condition. Now we need a code. And ICD-10-CM has a lot of codes for eczema to specify the type, including:
- B00.0, eczema herpeticum
- H01.13, eczematous dermatitis of eyelid (subcategories identify which eye and which lid)
- H60.54, acute eczematoid otitis externa (again, we need to specify right, left, bilateral)
- L20, atopic dermatitis (with an associated list of subcategories)
What else is wrong with the Grinch? One friend calls him “a bad banana with a greasy black peel.” A black peel (skin) could be necrosis and we need to know the exact location in order to code for it. If Mr. Grinch suffers from skin necrosis only, we would report I96 (gangrene, not elsewhere classified).
Or maybe he has necrotizing fasciitis (M72.6), in which case we need to use an additional code (B95.-, B96.-) to identify causative organism.
Further reports claim Mr. Grinch’s head is full of spiders. Literally? Let’s hope not. I don’t think medical science could do much for him. Perhaps Mr. Grinch is suffering from a psychological condition that makes his behavior erratic.
He could be suffering from borderline personality disorder (F60.3, note this code excludes antisocial personality disorder [F60.2]).
Or maybe he’s bipolar. If that’s the case, we need a lot more specific information. For example, is he in a manic or depressive phase? Is he displaying psychotic symptoms? Is he in full remission or partial remission?
The Grinch also suffers from termites in his smile, better known as dental caries. Well, we still need to know what kind of dental caries:
- K02.3, arrested dental caries
- K02.5-, dental caries on pit and fissure surface
- K02.6-, dental caries on smooth surface
- K02.7, dental root caries
- K02.9, dental caries, unspecified
Although several of these categories includes more specific subcategories, sadly, termite-induced is not one of our choices.
The Grinch also suffers from seasickness, T75.3- (motion sickness). Two things to note here. One, we need a seventh character to identify the episode of care. To make sure it shows up in the seventh position, we also need two placeholder Xs. We also need to report an additional external cause code to identify vehicle or type of motion (Y92.81-, Y93.5-).
Mr. Grinch is further described as “a crooked jerky jockey.” A-ha! He suffers from scoliosis! Oh dear, we need some more specific information to choose between our scoliosis codes (isn’t that always the case?):
- M41.2-, other idiopathic scoliosis (with subcategories specifying the spinal region affected)
- M41.3-, thoracogenic scoliosis
- M41.4-, neuromuscular scoliosis (this is scoliosis secondary to cerebral palsy, Friedreich’s ataxia, poliomyelitis and other neuromuscular disorders and we need to code also the underlying condition)
- M41.5-, other secondary scoliosis
- M41.8-, other forms of scoliosis
- M41.9, scoliosis, unspecified
Mr. Grinch reported consuming a three-decker sauerkraut and toadstool sandwich with arsenic sauce (yuck!). Perhaps he was poisoned by the arsenic. In order to code for the poisoning, we need to know whether it was:
- Accidental, T57.0X1
- Intentional self-harm, T57.0X2 (not likely given the patient)
- Assault, T57.0X3 (much more likely)
- Undetermined, T57.0X4 (always a popular fallback)
Don’t forget to include the seventh character for the encounter. Hopefully, Dr. Seuss can successfully sift through the Grinch’s conditions and prescribe the correct treatment to get him back in the holiday spirit!
Q: Do you predict coder productivity will decline as a result of ICD-10? If so, what do you think the declines will be six months after implementation?
A: These are just my predictions, but I think that inpatient cases are going to drop to 2.5–3 records per hour. Currently we’re upwards of 3–3.5 per hour in non-teaching/tertiary environments.
On the ambulatory surgery side, I think those are going to drop to 5.5-6.5, and I really think it will be closer to the 5. HCPro’s 2011 Coder Productivity survey results show coders completing 6 -7 cases per hour at the time. So the reason I give these estimate is because we’re going to have more of a challenge with the surgeons being able to provide coders the information needed. So I really do think it will be the lower end of that range.
And if you’re one of those facilities that codes today in both ICD-9 and CPT® and if you can continue that practice in ICD-10 and CPT, then you’re going to have more of a reduction, closer to 4 cases per hour just because of the two different thinking patterns for the two coding classifications.
For non-interventional radiology outpatient testing cases, we’re averaging approximately 25–30 per hour right now. I think we’ll that also go down slightly to a range of 23–26.
Editor’s Note: Rose T. Dunn, MBA, RHIA, CPA, FACHE, FHFMA, chief operating officer of St. Louis–based First Class Solutions, Inc., answered this question during the February 29-March 2, 2012 “JustCoding Virtual Summit: ICD-10-CM and ICD-10-PCS, ” and was originally published on JustCoding.com.