MDM Data Column: How to Count It

The data column is the only part of the MDM table with arithmetic in it, and arithmetic is auditable in a way that clinical judgement is not. A reviewer cannot easily second-guess whether you thought a patient’s chest pain was concerning. A reviewer can count, precisely and without argument, whether the four “data points” you claimed were actually four.

That is why this column produces more downgrades than the other two combined. Almost none of them come from clinicians inventing data. They come from counting rules that are narrower than they look: tests that are not unique, sources that are not external, notes that are not a discussion, and interpretations that were already paid for somewhere else on the claim.

Verified against the American Medical Association’s CPT Evaluation and Management (E/M) Code and Guideline Changes, effective 1 January 2023 — the E/M Services Guidelines section “Amount and/or Complexity of Data to Be Reviewed and Analyzed,” Table 1, and the definitions of terms — together with the AMA’s published E/M revisions FAQ guidance, and CMS Medicare Learning Network booklet MLN006764, Evaluation and Management Services (May 2026 revision). CPT codes, descriptions and other data are copyright the American Medical Association.

The Structure, and Its One Genuine Trap

Data is assessed in three categories:

Tests, documents, orders, or independent historian(s) — where each unique test, order or document is counted to meet a threshold number.
Independent interpretation of tests not separately reported.
Discussion of management or test interpretation with an external physician, other qualified health care professional, or appropriate source, not separately reported.

The thresholds sit in Table 1 as follows.

Data level	Requirement	Category 1 count needed
Minimal or none	—	—
Limited	Must meet the requirements of at least 1 out of 2 categories	Any combination of 2
Moderate	Must meet the requirements of at least 1 out of 3 categories	Any combination of 3
Extensive	Must meet the requirements of at least 2 out of 3 categories	Any combination of 3

Now the trap. The category numbers do not mean the same thing at every level.

At the limited (low) level there are only two categories: Category 1 is “Tests and documents,” and Category 2 is “Assessment requiring an independent historian(s).” The table adds a pointer at that row: “For the categories of independent interpretation of tests and discussion of management or test interpretation, see moderate or high.”

At the moderate and extensive levels there are three: Category 1 becomes “Tests, documents, or independent historian(s)” — the historian has folded into Category 1 — Category 2 becomes independent interpretation, and Category 3 is discussion.

So “Category 2” means the independent historian at low and independent interpretation at moderate and high. Auditors see notes claiming a “Category 2” at the low level on the strength of an independent interpretation, which is not an available route at that level, and notes claiming an independent historian as a separate category at moderate, where it is simply one more Category 1 element. Read the row you are actually in.

Elements are summed across types

Within Category 1 the elements combine freely:

“A combination of different data elements, for example, a combination of notes reviewed, tests ordered, tests reviewed, or independent historian, allows these elements to be summed. It does not require each item type or category to be represented. A unique test ordered, plus a note reviewed and an independent historian would be a combination of three elements.”

You do not need one of each. Three tests ordered is three elements, as is one test, one note and one historian.

Counting Category 1: Where the Errors Live

The three Category 1 element types are: review of prior external note(s) from each unique source; review of the result(s) of each unique test; and ordering of each unique test. Each is fenced by a definition.

“Test”

“Tests are imaging, laboratory, psychometric, or physiologic data. A clinical laboratory panel (eg, basic metabolic panel [80047]) is a single test. The differentiation between single or multiple tests is defined in accordance with the CPT code set. For the purpose of data reviewed and analyzed, pulse oximetry is not a test.”

Two hard limits in three sentences. A panel is one test, no matter how many analytes print on the report — a comprehensive metabolic panel is one element, not fourteen. And pulse oximetry is carved out by name.

“Unique”

“A unique test is defined by the CPT code set. When multiple results of the same unique test (eg, serial blood glucose values) are compared during an E/M service, count it as one unique test. Tests that have overlapping elements are not unique, even if they are identified with distinct CPT codes. For example, a CBC with differential would incorporate the set of hemoglobin, CBC without differential, and platelet count.”

This is the rule that most often turns a claimed four into an actual two. Serial troponins are one test. Serial glucoses are one test. A CBC with differential plus a separately ordered haemoglobin plus a platelet count is one element, not three, because the elements overlap even though the codes differ. The relationship between repeat testing and separate reporting is a different question, handled by the modifier 91 repeat-lab rules — but a repeat that is separately reportable on the claim is still one unique test for MDM.

“Unique source”

“A unique source is defined as a physician or other qualified health care professional in a distinct group or different specialty or subspecialty, or a unique entity. Review of all materials from any unique source counts as one element toward MDM.”

This is the second great deflator. A 200-page hospital discharge packet — notes, labs, imaging, operative report — is one element, because it all came from one unique source. The AMA’s own FAQ guidance makes the same point with a cardiology example: reviewing a cardiologist’s records including an EKG and a chest x-ray counts as one unique source element. Practices that count each document in a records bundle are the ones who lose the whole column on review.

“External”

“External records, communications and/or test results are from an external physician, other qualified health care professional, facility, or health care organization.”

And the person:

“An external physician or other qualified health care professional… is not in the same group practice or is of a different specialty or subspecialty. This includes licensed professionals who are practicing independently. The individual may also be a facility or organizational provider such as from a hospital, nursing facility, or home health care agency.”

Your own partner’s note from last month is not external. Your own note from last month is certainly not. Reviewing internal records is good practice and is not a Category 1 data element. Note the parallel with how the same phrase operates in the new-versus-established patient rule, where “same specialty and subspecialty… same group practice” does similar work in the opposite direction.

“Analyzed”, and when an order counts

“Tests ordered are presumed to be analyzed when the results are reported. Therefore, when they are ordered during an encounter, they are counted in that encounter. Tests that are ordered outside of an encounter may be counted in the encounter in which they are analyzed. In the case of a recurring order, each new result may be counted in the encounter in which it is analyzed. For example, an encounter that includes an order for monthly prothrombin times would count for one prothrombin time ordered and reviewed.”

Two consequences. You count the order at the visit where you place it — you do not have to wait for the result, and you do not get to count the same test again when the result comes back at the next visit. And a standing order for twelve monthly tests is one element, not twelve; each future result may be counted once, in the encounter where it is actually analysed.

The rule that overrides everything else

“Any service for which the professional component is separately reported by the physician or other qualified health care professional reporting the E/M services is not counted as a data element ordered, reviewed, analyzed, or independently interpreted for the purposes of determining the level of MDM.”

If you bill for reading it, you cannot also count reading it. This is the direct link between the data column and the professional/technical component split: appending modifier 26 to an ECG interpretation removes that ECG from your data count. The guidelines repeat the principle at the top of the MDM section for both interpretation and discussion, so it is not an obscure footnote.

Category 2: Independent Interpretation

“The interpretation of a test for which there is a CPT code, and an interpretation or report is customary. This does not apply when the physician or other qualified health care professional who reports the E/M service is reporting or has previously reported the test. A form of interpretation should be documented but need not conform to the usual standards of a complete report for the test.”

Three requirements and one concession. The test must have a CPT code; an interpretation must be customary for it; and you must not be billing (or have billed) for that test yourself. The concession is real and under-used: your documented interpretation does not have to look like a formal radiology report. “Chest film reviewed by me: no consolidation, no effusion, cardiac silhouette normal” is an interpretation. “CXR normal” is not — it records a conclusion without showing any interpretive act.

Category 3: Discussion of Management or Test Interpretation

This category has the most specific definition in the section, and nearly every clause of it excludes something practices try to count.

“Discussion requires an interactive exchange. The exchange must be direct and not through intermediaries (eg, clinical staff or trainees). Sending chart notes or written exchanges that are within progress notes does not qualify as an interactive exchange. The discussion does not need to be on the date of the encounter, but it is counted only once and only when it is used in the decision making of the encounter. It may be asynchronous (ie, does not need to be in person), but it must be initiated and completed within a short time period (eg, within a day or two).”

Interactive and direct. A message relayed by your nurse to their nurse is not a discussion.
Not chart notes. Documenting your plan in a progress note the consultant will later read is not an exchange.
Asynchronous is fine. Secure messaging back and forth counts, provided it completes within a day or two.
Once only. One discussion cannot be counted at two encounters.

The counterparty must be external, or an “appropriate source”:

“An appropriate source includes professionals who are not health care professionals but may be involved in the management of the patient (eg, lawyer, parole officer, case manager, teacher). It does not include discussion with family or informal caregivers.”

A call to a patient’s teacher about school-based management counts. A call to the patient’s daughter does not — that conversation may well count toward total time, but it is not a Category 3 data element.

The Independent Historian

“An individual (eg, parent, guardian, surrogate, spouse, witness) who provides a history in addition to a history provided by the patient who is unable to provide a complete or reliable history (eg, due to developmental stage, dementia, or psychosis) or because a confirmatory history is judged to be necessary. In the case where there may be conflict or poor communication between multiple historians and more than one historian is needed, the independent historian requirement is met. It does not include translation services. The independent history does not need to be obtained in person but does need to be obtained directly from the historian providing the independent information.”

The gate is the patient’s inability to give a complete or reliable history, or a judged need for confirmation. A spouse who simply adds detail to a fully competent patient’s account is not an independent historian. An interpreter never is. And the history must come to you directly from the historian, not relayed.

Worked Counting Examples

The claimed six that is actually three

An established patient with worsening fatigue. The clinician reviews a hospital discharge summary from an admission two months ago, plus the CBC, comprehensive metabolic panel and chest radiograph contained in that same discharge packet; orders a TSH and a ferritin; and reviews serial home blood-pressure readings.

The tempting count is six. The correct count:

Discharge summary, CBC, CMP and chest film all came from one hospital — one unique source, one element.
TSH ordered — one element.
Ferritin ordered — one element.
Home blood-pressure readings — not a test in the CPT sense and not from an external source; zero elements.

Three Category 1 elements. That reaches moderate data, not extensive. Extensive would need a second category as well — an independent interpretation or a qualifying discussion.

The two elements that reach extensive

A patient with a new pulmonary nodule found incidentally. The clinician orders a CT chest (one Category 1 element), reviews the outside radiology report from the imaging centre (a second Category 1 element from a unique external source), personally reviews the images and documents her own interpretation without billing for it (Category 2), and telephones a thoracic surgeon to agree a surveillance interval (Category 3).

Category 1 has only two elements, so it fails its own threshold of three. But extensive requires two of three categories, and Categories 2 and 3 are both independently satisfied. Extensive data. The lesson is that Category 1 is not the main route to the top of this column — for many clinicians it is the hardest one.

Frequently Asked Questions

Do labs ordered at a preventive visit count?

Not toward the preventive service. The AMA’s E/M guidance is that laboratory tests ordered under the preventive medicine codes cannot be counted toward MDM, because those codes were not affected by the office-visit revisions. If a separate problem-oriented service is also performed and billed, the data belonging to that service is counted for it. See the preventive-versus-problem visit guide.

Does reviewing my own previous note count?

No. Prior notes count as Category 1 elements only when they are from a unique external source, meaning a different group practice or a different specialty or subspecialty.

Is ordering a test worth the same as reviewing one?

Yes — both are single Category 1 elements. But you cannot count both for the same test across the ordering and reviewing encounters; the order is counted at the encounter where it was placed.

Does the data column alone decide the level?

Never. Two of the three MDM elements must be met or exceeded, so a high data column with low problems and low risk yields a low visit. The overview guide covers how the columns combine.

What if I discussed the case with my own partner?

It is not a Category 3 discussion, because a partner in the same group practice and same specialty is not external. It may be excellent care and it may count toward total time; it is not a data element.

How should the note evidence a discussion?

Name the person and their practice or specialty, record that the exchange was direct and interactive, give the date, and state the substance and what it changed. Four short clauses convert an uncountable line into a defensible element.

Do the same rules apply in the inpatient setting?

The definitions are common to the E/M guidelines and apply across the families that use MDM. The examples in the risk column shift by setting, and each code family has its own introductory guidelines. For inpatient status questions specifically, the two-midnight rule and condition code 44 address a separate decision from level selection.

A Counting Discipline That Survives Review

The practices that do well on this column tend to do the same small things. They write the source next to every reviewed item, so uniqueness is visible on the page. They avoid the phrase “labs reviewed” in favour of naming what was reviewed and from where. They record their own interpretations as interpretations rather than as conclusions. And before claiming an element they ask one question: is this already being paid for elsewhere on the claim? If the answer is yes, it leaves the data column — and if two other columns already carry the level, that costs nothing at all.