Coders are encouraged to query if not sure.
The ratio of observed-to-expected deaths is considered a measure of hospital quality. Recording accurate metrics depends on capturing all relevant diagnoses to optimize the expected mortality denominator. The explanatory variables groupings of mortality modeling for institutions include both definitive diagnoses and signs/symptoms.
A colleague is having an issue with some relatively new significant variables in the groupings of a specific healthcare quality improvement company. The question was posed to me, “are weakness or tachycardia really independent secondary diagnoses, or should they be considered ‘inherent to’ the definitive diagnosis?” There is a difference of opinion between the coders and the clinical documentation improvement (CDI) staff in her organization.
This quandary is very thought-provoking. The questions to ponder are:
- Is a symptom or sign an independently codable secondary diagnosis if it could be caused by the definitive diagnosis? Remember, a symptom is subjective (e.g., shortness of breath); a sign is observable or measurable by the clinician (e.g., tachypnea).
- Must the healthcare provider (HCP) make an associative linkage? Do they have to explicitly document that the symptom and the definitive diagnosis are not linked?
- May the coder discount the sign or symptom without querying the HCP?
- What constitutes “integral to” or “inherent to?”
The ICD-10-CM Official Guidelines for Coding and Reporting (FY 2019) instruct us in I.B.5. that signs and symptoms that are associated routinely with a disease process should not be assigned as additional codes, unless otherwise instructed by the classification.
- N40.1, Benign prostatic hyperplasia with lower urinary tract symptoms. We are directed to use an additional code for associated symptoms, such as R39.12, Weak urinary stream, or R39.11, Urinary hesitancy. These symptoms are routinely associated with BPH, but it is useful clinical information to specify which the patient is manifesting.
- Sequelae of cerebral infarctions (I69.-) can be coded with acute cerebrovascular accident as per a Type 2 Excludes instruction. These codes give an added dimension of information; the patient is having a stroke, and here are the manifestations of that stroke.
- The codes for migraines have a Type 1 Excludes for R51, Headache. Most migraines involve headache, so there is no extra information gained from coding the symptom. There are some migraine types that do not routinely include headache, such as cyclical vomiting or abdominal migraine, and in fact, those latter types would most likely meet the Type 1 Excludes exception, permitting one to use the symptom code if the conditions were unrelated and concomitant.
- An example of signs and symptoms routinely associated and not coded at the same time because they are superfluous is cough/fever with J18.9, Pneumonia, unspecified organism. These symptoms are universally accepted as being typically associated. The fact that the patient is experiencing a cough or fever with pneumonia adds nothing to the story.
Guideline I.B.6. indicates that additional signs and symptoms that may not be associated routinely with a disease process should be coded when present. For example:
- If a patient with chest pain and palpitations was evaluated and determined to have gastroesophageal reflux disease (GERD) causing the chest pain, the palpitations would be an additional R code because that symptom is atypical for GERD. In contradistinction, R07.9, Chest pain, unspecified, would not be coded.
The ICD-10-CM Coding Handbook by Nelly Leon-Chisen, RHIA, of the American Hospital Association, notes in Chapter 13 that:
“Unless otherwise instructed by the classification, codes from chapter 18 are assigned as secondary codes only when the symptom or sign is not integral to the underlying condition, and when the presence of the sign or symptoms makes a difference in the severity of the patient’s condition and/or the care given.
Codes from chapter 18 are not assigned when they are implicit in the diagnosis or when the symptom is included in the condition code.”
It then refers to combination codes, and how an additional sign/symptom code would be redundant and inappropriate. For instance:
- R40.2, Coma would not be coded with K70.41, Alcoholic hepatic failure with coma, because the coma is explicitly included in the combination code.
- Pleural effusion is not usually coded in heart failure. That is, unless it requires specific treatment, like a thoracentesis (i.e., when the presence of the sign or symptoms makes a difference in the severity of the patient’s condition and/or the care given). Then you utilize J91.8, Pleural effusion, in other conditions classified elsewhere, along with the I50.- code.
Let’s consider general signs and symptoms, which may be associated with many different underlying conditions. Weakness, fatigue, headache, dizziness, nausea, and vomiting are some examples. If you Google just about any condition, you will find one of these listed under possible associated signs and symptoms. Do you code them with all other conditions? Do you not code them with any definitive diagnosis? Do you query the HCP every time to make the determination?
Merriam-Webster Unabridged defines “inherent” as “structural or involved in the essential character of something.” “Integral” means “of, relating to, or serving to form a whole; essential to completeness.” Renal failure may present with weakness as a symptom, but is it considered integral or inherent to renal failure?
We should ask ourselves: “Does this condition (e.g., renal failure) routinely present with weakness?” and “Do patients with weakness usually have this as an underlying diagnosis?” If the answer to one of these is no, we need to delve a little deeper.
Does weakness meet the definition of a secondary diagnosis? Is it being clinically evaluated, are diagnostic studies being performed, is there therapeutic treatment, is it increasing the need for nursing care and/or monitoring, or extending length of stay? If the weakness, independent of the underlying diagnosis, is not warranting any attention by the clinician, it is not a valid secondary diagnosis.
How can we tell? There are clues. If the HCP is documenting it as “weakness due to renal failure,” and then it drops off their list, but they continue to copy and paste the renal failure going forward, they do not think it is a separate, codable entity. If the HCP documents weakness as a separate line item on their assessment and plan list without linkage, they may think it is relevant. If they evaluate it with electromyography (EMG) or order therapy specifically for the weakness, e.g., physical therapy, then it is my opinion that it is permissible to make it a distinct diagnosis.
Maybe we can ask ourselves these questions:
- Is this not usually/typically/routinely related to the underlying condition?
- Are there multiple diagnoses that could be causing the same symptom, such that it cannot be proximately related to any single one (i.e., multifactorial)?
- Is the symptom out of proportion to the usual presentation of the potentially underlying condition?
- Does the symptom persist despite successful treatment of the suspected underlying condition?
- Is the symptom requiring its own evaluation or treatment, or fulfilling any other secondary diagnosis criterion?
Any affirmative response should suggest that the symptom may not be inherent or integral to the accompanying condition.
The coding abstraction should tell the story of the patient encounter. Every sign and symptom that the patient presents with is not necessary to tell the story if a given sign or symptom doesn’t contribute anything above and beyond the definitive diagnosis.
Let’s return to the reason for this discussion. Conditions that risk-adjust for mortality do so because it has been observed that those specific conditions, present on admission, render the patient more likely to die during the hospitalization. I posit that there is a difference between subjectively feeling weak (i.e., a symptom) and objectively having muscle weakness requiring assistance in activities of daily living (i.e., a sign).
By the way, coders reading this article, the code for the symptom of feeling weak should be R53.1, but the code used for the sign of demonstrable physical weakness should be M62.81, Muscular weakness (generalized).
The grouping titled “debility” includes the code, R53.1, Weakness. The debility set also includes R54, Age-related physical debility (also known as frailty), R53.0. Neoplastic-related fatigue, R53.2, Functional quadriplegia, Z73.6, Limitation of activities due to disability, and Z74.01, Bed confinement status. This latter group of conditions indicate a more debilitated patient who has diminished physiologic reserve and is known to be at a higher risk for falls, a prolonged stay in the hospital, need for home health or skilled nursing care, and death. I think the risk-adjustment group is supposed to be selecting for those.
As I write this article, I am afflicted with a really nasty cold, courtesy of my husband, for Hanukkah. I felt weak this morning when I got out of bed. I am confident that if I went to the hospital, my chief complaint of “weakness” would not be clinically significant, nor would it rate a code on my final abstract.
What if the coder can’t tell?
My recommendation is that if the HCP documented a symptom, sign, or condition throughout the entire stay and there was compelling evidence that it independently met the criteria for a secondary diagnosis, code it. It would not be unreasonable to query a diagnosis of “weakness” for a more definitive diagnosis, like age-related frailty, however.
Focus on the admitting physician’s H&P’s list of diagnoses. The ED often uses signs or symptoms initially, and then the definitive diagnoses are determined over the course of the hospitalization. Tachycardia or leukocytosis in the ED may transition to volume-depletion dehydration or an infection on the floor by the attending. Those preliminary signs would not be fodder for final diagnosis status on the abstraction.
If the patient expired, and a mortality review conference determines that a POA-Y sign or symptom was documented early and repeatedly, and it is found on the mortality model, code it if your CDI professional believes it is legitimate. If you are not sure, query. Don’t discount it independently. Give the clinician the opportunity to decide if it is valid.
Do not go fishing for risk-adjusting conditions for the sake of gaming the system. Public health statistics are only as good as the data input. The point of this activity isn’t to make your institution look better; it is to identify patients who are at higher risk of dying so we can avoid preventable deaths.
Listen to Dr. Remer every Tuesday on Talk Ten Tuesday, 10 a.m. ET.