Diagnosis Pointers on the CMS 1500: How Box 24E Works

Diagnosis pointers on the CMS 1500 are letters in box 24E that tie each service line to a diagnosis code listed in box 21. Box 21 holds up to 12 ICD-10-CM codes on lines A to L. Each service line points to one or more of those letters, with the primary diagnosis for that service first.

How CMS 1500 diagnosis pointer letters map to box 21

A diagnosis pointer on the CMS 1500 is a reference letter, not a code. Box 21 lists the claim's diagnosis codes on lettered lines A to L. Box 24E on each service line repeats the letter of the diagnosis that explains why that service was done, so the code itself is written once and referenced many times.

The NUCC 1500 Claim Form Reference Instruction Manual sets the rules for both boxes on the 02/12 version of the form:

• Box 21 takes an ICD indicator, 0 for ICD-10-CM or 9 for ICD-9-CM, between the dotted lines at the top right.
• Box 21 holds no more than 12 diagnosis codes, entered left justified, with no narrative description.
• The decimal point in each code is left out "because it is implied". A code printed as E11.9 in a code book is entered as E119.
• Box 24E takes the letter or letters from box 21, never the diagnosis codes themselves.

The letter is only meaningful next to the box 21 list it came from. "B" on a service line means whatever code sits on line B of that claim's box 21. On a different claim, B points somewhere else.

Worked example: four codes, three service lines

Everything in this example is fictional, including the patient, Dana Example, and claim number CLM-0000-1234. The diagnosis codes are real ICD-10-CM codes, with descriptions as listed in the NLM Clinical Tables ICD-10-CM search. They show how pointers work, not how to code a visit.

Box 21, with the ICD indicator set to 0:

Box 24, with the procedure codes left out and each service described in words:

Line 1 lists A first because diabetes was the main reason for the visit in this example, and the other letters follow. Line 2 points only to the two conditions the test relates to. Line 3 needs one pointer. The Medicare column differs because Medicare asks for a single letter per line.

How many diagnosis pointers can one CMS 1500 service line carry?

One CMS 1500 service line can carry up to four diagnosis pointers. The NUCC manual gives box 24E room for "4 characters in the unshaded area", and each pointer is one letter. Medicare is stricter and asks for only one letter per line.

The NUCC instruction says that when several diagnoses apply, "the primary reference letter for each service should be listed first, other applicable services should follow." The letters go left justified with nothing between them. A line pointing to diagnoses A, C and D is entered as ACD.

The Medicare Claims Processing Manual, chapter 26 gives a different instruction for Medicare claims. It says to "enter only one reference number/letter per line item". It adds that if two or more diagnoses are required for a procedure, the provider "shall reference only one of the diagnoses in item 21."

So the limit depends on the payer:

• Under NUCC's general instructions, up to four letters, from A to L, primary first.
• Under Medicare's instructions, one letter, the primary one.
• Under another payer's rules, check that payer's own billing guide. NUCC says its manual is a general guide and that users should follow payer instructions.

A claim with more than 12 diagnoses cannot point to all of them. NUCC says the claim must be split, and the services tied to the extra diagnoses are billed on a separate claim.

Diagnosis pointer order and the primary diagnosis on a service line

The first diagnosis pointer on a service line names the primary reason for that service. NUCC asks for the primary letter first, and Medicare asks for only the primary letter. Either way, the first position is the one both sets of instructions use to relate the service to its primary diagnosis.

Order matters at two levels, and they are easy to confuse:

• The order of codes in box 21. Chapter 26 tells Medicare billers to "enter the diagnoses in priority order" in box 21. Line A is the claim's top diagnosis.
• The order of letters in box 24E. This is per service line. The primary diagnosis for one service may not be line A. In the worked example, line 3 points to D because the back exam relates to the back pain code, even though diabetes is line A for the claim.

One slip is to copy "A" into every line because A is the claim's main diagnosis. That tells the payer every service was done for the first code, which may not be true for a test or procedure aimed at a different condition.

Diagnosis pointer errors that get a CMS 1500 rejected

Diagnosis pointer errors on the CMS 1500 usually break the link between box 24E and box 21. Chapter 26 marks box 24E as a required field for Medicare, and claims with incomplete or invalid information "may be returned to the submitter as unprocessable". The mistakes below each break a published instruction.

• Numbers instead of letters. The older 08/05 form used numbers. Chapter 26 says that on version 08/05 the reference "will be either a 1, or a 2, or a 3, or a 4", while on version 02/12 it "will be a letter from A-L." A 1 or 2 in 24E on the current form is a leftover habit.
• Codes instead of letters. NUCC says ICD codes "must be entered in Item Number 21. Do not enter them in 24E." An E119 typed into 24E fills all four characters and points to no box 21 line.
• Commas, spaces or dashes. NUCC says "Do not use commas between the letters." A range written as A-C is not a list of letters, and it takes three of the four characters.
• Pointers to empty lines. A letter must match a line in box 21 that holds a code. If box 21 stops at C, a pointer to D refers to a blank line.
• Too many letters for the payer. Four letters may be fine under NUCC's general rules but break Medicare's one-letter instruction.
• Wrong page on a multi-page claim. NUCC says only the diagnosis codes on the first page may be used, and they must be repeated on later pages. Letters on page two still refer to the page one list.

The rest of the service line, including dates in 24A, place of service in 24B and the rendering provider in 24J, is covered column by column in reading one CMS 1500 service line across columns 24A to 24J. The rendering NPI in 24J and the billing NPI in 33a have their own guide to NPI and tax ID fields on claim forms.

What does a diagnosis pointer look like after extraction?

After extraction, a diagnosis pointer should stay a letter that still resolves to a code. Store the box 21 list as lettered slots, and store each service line's 24E entry as an ordered list of letters. Keeping both lets anyone rebuild which code each service pointed to, in the original order.

A flat record that keeps only "ABCD" as a string loses meaning when it leaves the claim. A record that swaps the letters for codes too early loses the order and the link to the page. A layout that keeps both looks like this, using the worked example's line 2:

Four checks keep extracted pointers honest:

1. Every pointer letter must match a filled box 21 slot.
2. The number of letters must fit the target payer's rule, four for NUCC and one for Medicare.
3. On a multi-page claim, letters must resolve against the first page's box 21.
4. The ICD indicator must match the code set in box 21, since a 0 with ICD-9 codes or a 9 with ICD-10 codes makes every pointer suspect.

Scanned claims make these checks harder, because a faint letter in 24E or a code that drifts across lines in box 21 reads badly. The guide to CMS 1500 OCR covers how paper forms are read. Assigning the diagnosis codes in the first place is a separate job, covered in the post on medical coding automation.

Schema Care reads diagnosis and procedure codes from CMS-1500 claims and scores each field it extracts. A pointer letter or a box 21 code below your team's threshold goes to a person, who sees it highlighted on the scanned page. The document types Schema Care reads are listed on the homepage.