ICD-10: What’ll it be, lumping or splitting? Part 2

In my last blog, I talked about the trade-offs between lumping and splitting as they apply to the annual update of the ICD-10-CM/PCS classification systems. This time I am going to talk about how lumping and splitting apply to Coding Clinic, the official source of coding advice published by the American Hospital Association.

Lumping and splitting are both terms that refer to an orientation toward the data. Basically, lumping means making fewer distinctions in the coded data and splitting means making more distinctions in the coded data. With Coding Clinic advice, the tricky bit is deciding whether, within the constraints of the existing coding rules, lumping or splitting produces better coded data in specific cases.

Lumping and splitting are often at the heart of the questions submitted to Coding Clinic. For instance, a question that asks whether a procedure should be coded separately or not is another way of asking, should we lump in this case or should we split?

The following are four examples, two each of lumping and splitting, and all are taken from the latest issue of Coding Clinic, 3^rd Quarter 2017.

Lumping Example #1, pp. 8-9

A newborn with a condition called gastroschisis (where the abdominal wall fails to close completely, and so some of the gastrointestinal contents are outside the abdomen) has been previously treated by placement of a “silo,” a device that allows for the gradual return of the stomach/intestines to their proper place. The treatment has been successful, and the infant is taken to the operating room, where the silo is removed and the anterior abdominal wall defect is closed with sutures.

The questioner asks, “How should the removal of silo and final closure of gastroschisis be reported? Should we report both the removal and closure, or is removal of the silo inherent to the surgical closure of the abdominal wall? What are the appropriate codes for this procedure?”

“Answer: Assign the following ICD-10-PCS code: 0WQF0ZZ Repair abdominal wall, open approach, for closure of the abdominal wall. The removal of the silo device is not coded separately, as it is inherent to the definitive procedure, which is the surgical closure of the defect in the abdominal wall. The silo is simply lifted out after having performed its function. The definitive procedure is to close the defect in the abdominal wall.”

In this case, the advice to the coder is: lump. The coded data would not benefit from an extra code here. Coding a necessary and preliminary step—lifting out the silo—before the definitive repair—closing the abdominal wall—would have given the false impression that something significant and distinct from the suture repair of the abdominal wall had occurred.

Lumping Example #2, p. 13

Another common example of lumping on page 13 of the same Coding Clinic issue describes a patient with ascites (excess fluid in the abdominal cavity) treated with “diagnostic and therapeutic paracentesis… both performed via a catheter.” (Abdominal paracentesis is just a fancy term for taking fluid out of the abdominal cavity using a hollow needle.) The questioner asks, "Is it appropriate to report two procedure codes for the diagnostic and therapeutic paracentesis?”

Coding Clinic’s answer is consistent with longstanding coding conventions, that the procedure is coded as therapeutic in cases such as this, and the diagnostic portion is not coded separately. if a single procedure has been performed in which a portion of the tissue or fluid removed is sent to a lab or pathologist for evaluation, that is not coded as a separate diagnostic procedure. Here’s why: sending fluid or tissue off for evaluation (testing for cancer or infection, for example) is something is not separate procedure performed on the patient, but simply a divvying up of tissue or fluid that was taken as a single procedure. For that reason, the “diagnostic paracentesis” is not assigned its own procedure code.

As Coding Clinic goes on to explain in the answer, there are situations where a diagnostic procedure would be coded separately, such as “a diagnostic drainage procedure that uses a different approach or samples a different site from the therapeutic drainage procedure.” In such cases, a separate procedure is performed on the patient and so it is assigned its own procedure code.

Splitting Example #1, pp. 20-21

When a key component of an operative episode is important to capture separately in the coded data, Coding Clinic will advise coders to split. Often these are more complex cases, where one code would give an incomplete picture of the what was done during surgery.

In this scenario, the patient’s bladder is removed (bladder cancer) and a type of urinary diversion procedure called an Indiana pouch is created. During the procedure, a portion of the ileum is harvested from the patient, cut and fashioned into a reservoir, and then reattached to the patient’s ureters to create a “neo-bladder.” An opening called a stoma is created on the abdominal wall so the ileocecal valve can be used as a natural, one-way valve to drain urine from the neo-bladder.

In this case, Coding Clinic advice says that two codes are necessary to fully capture the essentials of the case. 0TRB07Z Replacement of bladder with autologous tissue substitute, open approach is used to specify the creation of the Indiana pouch. 0T180ZC Bypass bilateral ureters to ileocutaneous, open approach, captures the fact that the urinary diversion procedure includes the bypass from both ureters to the ileal pouch, and also the bypass from the ileal pouch to the skin of the abdominal wall, forming the stoma. It would be impossible to capture the essentials of what was done in this very complex operation without assigning both PCS codes.

Splitting Example #2, p. 22

In this example, a bone flap was previously removed from the patient’s skull to relieve pressure from a traumatic head injury. The bone flap was stored in the patient’s abdominal wall (I know, this sounds weird, like they are afraid they’re going to lose track of it in some refrigerator otherwise, but it actually preserves the bone flap better that way.) while the intracranial pressure subsided, and now it is time take the bone flap out of the abdominal wall and reconstruct the skull, using titanium plates and screws to reattach it.

In this case, clearly two separate codes are required to describe this operative scenario accurately as coded data. One code specifies the surgical removal of the bone flap from the abdominal wall, and another code specifies repositioning the bone flap on the skull using the metal plates and screws.

So, when is lumping advisable, and when is splitting the better way to go? Well, naturally, that depends, as you can see from these examples. Since the whole point of coding is to produce useful coded data—we wouldn’t be doing it otherwise—Coding Clinic’s job is to consider the needs of the coded data within the constraints of existing coding guidelines and definitions.

Rhonda Butler is a clinical research manager with 3M Health Information Systems.