E/M Spine Coding: Best Practices for Accuracy

When I audit a spine chart, I am answering one question. Does the documentation in front of me support the level that was reported? I am not deciding what level the visit should have been, and I am not filling in what the physician probably meant.

Spine makes that harder than most specialties. The clinical picture is usually complex, the physician almost always looked at imaging, and it is easy to talk yourself into credit the note does not actually give you. A coder who sees severe stenosis on the problem list and scores high complexity has made a clinical judgment. A coder who sees an MRI referenced and credits independent interpretation has written a sentence the physician did not write.

This works in both directions. If the record supports the level, I leave it alone, and I have defended plenty of level fours and fives that a payer wanted to reduce. If it does not, the answer is a query or a lower level. It is never a reconstruction of what the physician was probably thinking.

Coder scope and clinical judgment

Level selection falls right on the line between clinical judgment and coding validation, and it is easy to step over that line without realizing it. The guidelines assign clinical significance, severity, and the assessment of what the encounter required to the physician, and define the elements by reference to what the record documents. That division suggests a working boundary, though where exactly an organization draws it is a policy decision worth making deliberately.

Table scrolls sideways on mobile.

The two of three rule

Office and outpatient services 99202 through 99215 are selected on medical decision making or total time. MDM has three elements: problems addressed, data reviewed and analyzed, and risk of complications. The level reported is the one met or exceeded by two of the three.

That rule protects the physician and the practice equally. A low risk element does not disqualify a level when the other two are documented at that level. A strong risk element does not rescue a level when the problem and data elements are not in writing.

Source. The element definitions, the two of three rule, and the terms used throughout are set out in the Evaluation and Management Services Guidelines in the CPT Professional Edition. CMS aligned its own policy with those guidelines effective January 1, 2021 and restates them in the MLN booklet Evaluation and Management Services, MLN006764.

Element requirements by level

The data categories at each level

The category labels do not carry the same meaning from one level to the next, so a rule learned at one level will misfire at another. Each category's contents and the requirement at each level are set out here.

How the counting works inside each category

• Category 1 is a threshold reached by summing elements. Two elements at limited, three at moderate and extensive. The elements may be any mix of the choices listed, and no choice has to be represented. A test ordered plus an external note reviewed plus an independent historian is a combination of three. So is three tests ordered.
• Category 2 is satisfied by one qualifying item. A single documented independent interpretation meets it. At limited, where Category 2 is the independent historian instead, a single qualifying historian meets it.
• Category 3 is satisfied by one qualifying item. A single documented interactive discussion meets it.

So the categories do not carry the same weight. A documented independent interpretation meets Category 2 by itself. An order does not meet Category 1 by itself, since Category 1 needs three elements at moderate and two at limited.

Data element availability by level

Not every element is available at every level. Ordering a test, reviewing a test result, and reviewing an external note can be counted at limited, moderate, and extensive, in any combination that reaches the Category 1 threshold. Independent interpretation and discussion cannot. Both first appear at moderate, which is level four, and continue at extensive, level five.

In a spine encounter this comes up when the only data documented is the surgeon's reading of an outside MRI. At limited there is nowhere for that to be counted, so the data element rests on whatever ordering or reviewing the note also shows. At moderate and extensive the same interpretation meets Category 2 by itself.

The independent historian

The independent historian is the one item that changes category as the level changes. At limited it stands alone as Category 2. At moderate and extensive it is folded into Category 1 as a fourth choice in the combination. Nothing else relocates. Orders, test results reviewed, and external notes reviewed are Category 1 choices at every level where data is scored.

Because the label moves, two people can each describe Category 2 accurately and mean different things. One learned it at limited, where Category 2 is the independent historian. The other learned it at moderate, where Category 2 is independent interpretation. Each is correct about the level they learned. The guidelines define the categories per level, so the reliable approach is to establish the data level first and then apply the categories belonging to it.

Element one: problems addressed

A problem is addressed or managed when it is evaluated or treated at the encounter by the reporting physician, and the guidelines add that this includes consideration of further testing or treatment not elected because of a risk and benefit analysis or patient choice. They also state that notation that another professional is managing the problem, without additional assessment or care coordination documented, does not qualify, and that referral without evaluation or consideration of treatment does not qualify either.

Source. CMS states in MLN006764 that the record must support the level reported, and that the volume of documentation is not what determines the level. It also directs that the specific sign, symptom, or complaint establishing medical necessity be identified for every service billed.

The category comes from the status the physician documented, not from the diagnosis. Chronic lumbar spondylosis can be a stable chronic illness or a chronic illness with severe exacerbation. Same diagnosis code either way. The only thing separating them is the language in the note.

Severe exacerbation draws the most disagreement in the audits I run, and the disagreement is usually reasonable on both sides. One reviewer reads a note describing marked functional decline and objective neurologic change and concludes the record establishes severity. Another reads the same note, finds no severity language from the physician, and scores it as exacerbation without severity. The guideline aligns these by treating severity as a clinical characterization, which places it with the physician rather than the reviewer. That makes the coder's task confirming the characterization appears and that supporting detail accompanies it, and it makes the unresolved note a query rather than a judgment call.

The undiagnosed new problem

An undiagnosed new problem with uncertain prognosis supports moderate complexity. CPT defines it as a problem in the differential representing a condition likely to result in high risk of morbidity without treatment. Two pieces get dropped constantly.

A differential has to appear in the record. The category applies to a problem under active diagnostic consideration, which means the physician stated the possibilities. The morbidity of the untreated condition is the qualifying feature, so a new complaint with a benign documented course is an acute uncomplicated injury and scores low.

Source. The definition is in the CPT E/M guidelines. Earlier CMS guidance made the same point about documentation form, instructing that where no diagnosis is established, the assessment may be stated as a differential or in probable or possible terms. For outpatient reporting, the ICD-10-CM Official Guidelines still prohibit coding uncertain diagnoses, so the differential supports the MDM level without being reported as a diagnosis.

A new patient with low back pain is not automatically an undiagnosed new problem. I have watched coders build the differential in their heads, reasoning that a spine surgeon looking at new back pain has to be considering something serious. That is clinical territory. If the workup ordered points to a differential the note never states, query for it.

Element two: data reviewed and analyzed

Data is scored in three categories. Moderate needs one satisfied, extensive needs two. This element carries real weight in spine, and it is where I see the most credit taken for work that probably happened but was never documented.

Counting rules and the guideline language

These are the provisions that decide most of the counting, quoted or paraphrased close enough to the source that you can weigh them yourself. Some answer the question cleanly. Others leave room, and I have noted where.

Ordering and later reviewing the same test

CPT Assistant from November 2021 addresses this directly, stating that the physician is assumed to review the results of a test ordered, so no dual credit is available for both ordering and reviewing the same test. That points toward counting the test once, at the encounter where the order was placed, with no further element when the result comes back and is reviewed at the next visit.

The guidelines set out the timing directly. Tests ordered are presumed to be analyzed when the results are reported, so when they are ordered during an encounter they are counted in that encounter. Tests ordered outside an encounter may be counted in the encounter where they are analyzed. For a recurring order, each new result may be counted in the encounter where it is analyzed, and the guidelines give the example of an order for monthly prothrombin times counting as one test ordered and reviewed, with future results counting as a single test in a later encounter if analyzed there.

For a spine practice the practical read is that a study ordered at a visit is credited at that visit, and a study the patient arrives with, ordered elsewhere, is credited at the visit where it is analyzed.

Unique tests

The guidelines define a unique test by the CPT code set, state that multiple results of the same unique test compared during the service count as one, and state that tests with overlapping elements are not unique even where identified by distinct codes. The example given is a CBC with differential, which incorporates hemoglobin, CBC without differential, and platelet count. Tests are defined as imaging, laboratory, psychometric, or physiologic data, a clinical laboratory panel counts as a single test, and pulse oximetry is expressly excluded from being a test for this purpose.

Applied to spine imaging, four-view lumbar films reported under a single radiographic code read as one test rather than four. Cervical and lumbar MRI, distinct codes without overlapping elements, read as two. Lumbar MRI without contrast alongside lumbar MRI with and without is where the overlapping elements language does the work, and the more defensible reading treats them as one. Reviewers do differ on how far the overlap concept extends across imaging, which is worth settling in policy rather than case by case.

Tests considered but not ordered

The guidelines state that ordering a test may include those considered but not selected after shared decision making, and that these considerations must be documented. Two examples are given: a patient requesting imaging that is not necessary for the condition, where a discussion of the lack of benefit is required, and a test that would normally be performed but is not ordered because of the risk to that specific patient.

Both come up regularly in spine. A patient asking for an MRI on a two week old strain, and a study deferred because of an implanted device or renal function, each qualify where the reasoning is written down. The requirement is the documentation of the consideration, not the order itself.

Tests with no professional component

Laboratory studies and other results-only tests have no separate interpretation to render, so Category 2 does not arise for them at all. The guidelines address this directly, stating that tests which do not require separate interpretation and are analyzed as part of MDM do not count as an independent interpretation but may be counted as ordered or reviewed. In a spine practice this covers the preoperative laboratory panels, inflammatory markers, and hemoglobin A1c results that show up alongside imaging. They are Category 1 elements, and the count follows the unique test rules rather than the interpretation rules.

Ordering and independently interpreting the same test

The March 2023 errata added to the definition of independent interpretation that a test ordered and independently interpreted may count as both a test ordered and interpreted. On that language, the order is one of the elements making up the Category 1 combination and the interpretation meets Category 2. Both are cognitive work the practice is not being paid for through a procedure code, which is why both land in the data element.

Worth knowing that guidance written before the erratum often instructs coders to credit such a study once, so an internal tool or a training deck predating 2023 may still reflect the earlier approach. None of this reaches a study the billing exclusion has removed to begin with, and none of it applies at the limited level, where independent interpretation is not among the available items.

Ordering without an interpretation

Ordering of each unique test appears as a Category 1 choice at every data level where data is scored. The guidelines attach no further condition to it. Nothing in the language requires that the study also be interpreted, or reviewed, or that anything else happen to it at all. Where the billing exclusion does not apply, the order is a Category 1 element on the strength of the order itself.

Unique sources

Two reasonable approaches circulate here. One counts by the organization the records came from, which is intuitive because that is usually how they are identified when they come in. The other counts by practitioner. The guideline settles it with a definition that includes both concepts but subordinates location to relationship. A unique source is a physician or qualified health care professional in a distinct group or a different specialty or subspecialty, or a unique entity, and review of all materials from any one unique source counts as one element toward MDM. The alignment step is to identify who authored the records and their group or specialty, then treat the organization as a source only where no distinct practitioner is identifiable.

Source. The definitions of unique source and external are in the CPT E/M guidelines and are restated in the AMA CPT E/M revisions FAQ. Note that external is a separate defined term. External records are those from a physician or qualified health care professional not in the same group practice, or of a different specialty or subspecialty, and the guideline notes that the source may also be a facility or organizational provider such as a hospital, nursing facility, or home health agency. A facility can therefore serve as an external source, though the count is driven by the practitioner and the entity.

Records from prior encounters within your own practice, by physicians of the same specialty in the same group, are not external at all and generate no Category 1 credit as external notes. In a spine practice that pattern shows up constantly, because the prior visits a physician reviews are frequently their own.

Discussion

The definition states that discussion requires an interactive exchange, that the exchange must be direct and not through intermediaries such as clinical staff or trainees, and that sending chart notes or written exchanges within progress notes does not qualify. It may be asynchronous rather than in person, though it must be initiated and completed within a short time period, with a day or two given as the example. The discussion does not have to occur on the date of the encounter, but it is counted once and only when it is used in the decision making of that encounter.

Records forwarded to another practitioner and referrals placed therefore do not qualify. A secure message exchange can qualify, provided it is direct, initiated and completed promptly, and used in the decision making of the encounter. Given how much clinical communication now runs through messaging, this is worth addressing specifically in an organization's policy.

Identifying which data element the note documents

Ordering, reviewing a result, and independent interpretation are three separate data elements under the guidelines, each with its own definition and its own placement in the categories. Clinically they run together. A physician orders the MRI, the report comes back, they look at it, and they use it, and in conversation any of that gets called reviewing the imaging. Scoring the element means identifying which of the three the documentation supports, and the definitions are the reference for that determination.

Distinguishing review of a result from independent interpretation

These two get treated as interchangeable more often than any other pair in the data element, and in spine charts the difference between them comes up constantly. The definitions separate them, and applying the definitions comes down to identifying where the finding in the note originated.

Review of a result is documented when the note shows the surgeon working from findings the radiologist already determined. The radiologist read the study and wrote the impression, and the surgeon read that impression and used it in the plan. The guidelines credit this under Category 1 as review of the result of a unique test.

Independent interpretation is documented when the note shows the surgeon going to the study directly and arriving at a finding from it. The definition describes an interpretation of a test for which there is a CPT code and for which a report is customary, and calls for a form of interpretation to be documented, though not to the standard of a complete report. The two are separated by the origin of the finding rather than by the amount of detail in the note. A note can restate the radiologist's impression in great detail and still document a review, because the radiologist determined all of it.

An indication that the images were viewed is therefore the feature of the documentation that separates the two. Without it, a reviewer has no way to determine whether the surgeon arrived at the finding or read it in the report. Where the note does not settle the question, the conservative reading is review of a result, and a query is the route to the other determination.

When a single test can be counted twice, and when it cannot

The pairing rules run differently depending on which two are combined, including for the two that appear similar.

Two of these are worth committing to memory. Ordering and then reviewing is the same work described at two points in time. Ordering and then interpreting is two different kinds of work, one deciding the test was needed and one analyzing the study, and the guidelines count them separately.

Source. On ordering and reviewing, CPT Assistant from November 2021 states that the physician is assumed to review the results of a test ordered, so no dual credit is available for both ordering and reviewing the same test. On ordering and interpreting, the March 2023 AMA errata added to the definition of independent interpretation that a test ordered and independently interpreted may count as both a test ordered and interpreted.

When the double credit actually comes up

The both-count rule sounds broader than it plays out, and it helps to think through the situations where a surgeon would realistically order a study and independently interpret it at the same encounter.

The order and the interpretation fall away under different conditions, and both conditions come back to whether a procedure code has already paid for the work. Taking them one at a time.

The order is lost only when the surgeon, or a partner of the same specialty, reports the professional component. That is the narrow case the exclusion reaches. Every other arrangement leaves the order intact. If the practice reports only the technical component and an employed radiologist reports the professional component, the order still counts, because radiology is not the surgeon's specialty. If an outside facility handles the study entirely, the order still counts.

The interpretation is lost when the reporting physician, or a same specialty partner, reports the professional component. Whether it is also lost when a different specialty within the group reports it is not answered in the guideline text and is taken up below.

Three arrangements follow from those two conditions. Where the surgeon or a same specialty partner reports the professional component, nothing counts. Where an unaffiliated radiologist or facility reports it, both the order and the interpretation count, provided the surgeon documented a reading of their own. Where a radiologist employed by the group reports it, the order counts, and the interpretation depends on how the organization reads a provision the guidelines leave open.

So the double credit requires a study the surgeon ordered, billed by someone unaffiliated with the practice, and personally reviewed by the surgeon with their own findings documented. In a spine practice that usually looks like a study ordered and obtained the same day at an outside facility, or read by an unaffiliated teleradiology group, with the surgeon pulling up the images before the patient leaves.

Timing matters as much as billing here. Both credits attach to one encounter only if the order and the interpretation both occur at that encounter. The more common spine pattern runs across two visits: the MRI is ordered at one encounter and the images are reviewed and interpreted at the next. The order was counted at the first visit, so the second visit carries the interpretation alone. That is one Category 2 credit at the later encounter, not two elements, and no order credit carries forward.

Worth knowing that guidance written before the erratum often instructs coders to credit such a study once, so an internal tool or a training deck predating 2023 may still reflect the earlier approach. None of this reaches a study the billing exclusion has removed to begin with, and none of it applies at the limited level, where independent interpretation is not among the available items.

Independent interpretation of imaging

Spine physicians typically do look at the actual images rather than relying on the narrative. The surgical decision turns on findings a general report may not emphasize, so the images get opened. That work satisfies Category 2 when it is documented as an interpretation.

Being typical is precisely why this element is hard to audit. Everyone in the practice knows the surgeon reviewed the study, so writing it down feels unnecessary. Two conditions have to be satisfied before the documentation is even the question:

• The study was performed by another physician or qualified health professional
• The interpreting physician does not separately report the professional component

Source. Category 2 is defined in the CPT E/M guidelines, and the definition of independent interpretation was revised by the March 2023 errata. As revised, it does not apply where the physician reporting the E/M service is reporting, or has previously reported, the test. Note the previously reported clause: a surgeon who read and billed a study at an earlier encounter cannot claim independent interpretation of that same study later. The AMA has also published clarifications on MDM, time, and documentation in its CPT E/M revisions FAQ and in CPT Assistant, and the reference to consult before setting internal policy on a contested point.

The two-part test in the documentation

Once those are satisfied, the note itself has to show two separate things, and this is where most Category 2 credit falls apart. The physician has to indicate that the images were viewed, and the note has to contain an interpretation the physician arrived at from that viewing.

Each has to appear on its own. A note can say the images were reviewed and then report nothing but the radiologist's impression. A note can carry detailed findings lifted straight out of the report without the physician ever opening the study. Category 2 rests on both being present, and I see each version regularly.

Language that indicates viewing

These are the phrases I look for as evidence the physician was in the images rather than in the report:

• Naming the medium or platform, such as images reviewed personally on PACS or films reviewed in the office
• Naming specific sequences, planes, or views, such as sagittal and axial T2 sequences or standing lateral view
• Measurements the physician took, such as a canal diameter or a curve angle stated in the note
• A finding the report does not contain, or a characterization that differs from the report
• Comparison to a prior study performed at a different facility
• An explicit statement of disagreement with, or expansion on, the radiologist's impression

The guidelines mandate none of these and require no particular phrase. They state only that a form of interpretation should be documented, and that it need not meet the usual standards of a complete report. Which indicators an organization accepts is therefore left to policy. An attestation such as images independently reviewed and interpreted by me addresses the viewing component, though findings underneath it would still be needed for the second.

Language that indicates interpretation

The second component calls for a finding the physician arrived at. Restating the impression, however much detail is restated, and agreeing with radiology both leave the finding with the radiologist. An interpretation states the conclusion the physician drew from the study and, in the best notes, connects it to the clinical picture.

Of course the surgeon probably pulled up the images before the visit. Probably is not documentation. If the note says the study was reviewed and stops there, that is Category 1 and nothing more.

Studies the practice bills

Where the payment for the work already sits

The reasoning behind these provisions is worth understanding before the provisions themselves, because it makes them predictable rather than something to memorize.

A test follows a normal sequence. Someone decides it is needed and orders it. The test is performed. Someone interprets it. That is real work at every step, and the guidelines are concerned with one question about it: has the practice already been paid for that work somewhere else.

When the ordering practice reports the CPT code for the test, the answer is yes. Payment for deciding on the study, performing it, and interpreting it comes through that code. A radiograph reported as 73610 with modifier 26 is payment for the interpretation of that radiograph. Counting the same work again as an E/M data element would be asking to be paid twice for one thing, and the guidelines close that by excluding the study from the data element entirely, the order along with the interpretation.

When the practice does not report a code for the test, the answer is no. The physician still decided the study was needed, and if they went to the images and reached their own finding, they did interpretive work as well. No CPT code carries payment for either one. The E/M data element is where that work is recognized, and so it counts.

The 2023 errata follows from the same reasoning. Where the practice orders a study, an outside party performs and bills it, and the physician then independently interprets it, the order and the interpretation are two separate pieces of cognitive work and neither is being paid for through a procedure code. So both count. The errata was not creating an exception. It was confirming that two uncompensated pieces of work are two data elements rather than one.

The provisions that follow all rest on that principle. If the payment is in the CPT code, the work is not in the E/M. If there is no CPT code payment to the practice, the work is recognized in the E/M data.

What the guidelines say

This rule is commonly read as a restriction on Category 2 alone, which is a natural reading given that independent interpretation is the credit most obviously at stake. The guideline reaches further. It excludes the ordering and the performance or interpretation of diagnostic tests from level determination when the professional interpretation is separately reported by the physician reporting the E/M service.

Source. Three provisions bear on this and two of them name ordering. Under Services Reported Separately, the ordering and actual performance or interpretation of diagnostic tests are not included in determining the level of E/M services when the professional interpretation is reported separately by the physician reporting the E/M service. The definition of analyzed is more explicit still: any service for which the professional component is separately reported by the physician reporting the E/M services is not counted as a data element ordered, reviewed, analyzed, or independently interpreted for purposes of determining the level of MDM. A third provision, under the note that MDM may be impacted by role and management responsibility, states that where the physician is reporting a separate CPT code that includes interpretation and report, the interpretation and report are not counted. That one omits ordering, though the first two control.

Note on scope. All three condition the exclusion on the physician or other qualified health care professional reporting the E/M service. The guidelines carry no group practice or same specialty language in these provisions. Those phrases appear elsewhere in the E/M guidelines, in new versus established patient, initial versus subsequent services, and the definitions of external and unique source, but not here. Both provisions also name the professional component specifically rather than the test generally.

There is a reasonable argument on the other side, and it comes up every time I teach this. Only the interpretation is bundled into the professional component, the thinking goes, so the decision to order the test is separate work that should still count. The argument has a passage behind it. Under the heading addressing how MDM may be affected by role and management responsibility, the guidelines state that when the physician is reporting a separate CPT code that includes interpretation or report, the interpretation or report is not counted toward MDM. Ordering is not mentioned there.

Read against the rest of the guidelines, that passage does not create an exception. It answers a different question, which is what becomes of interpretive work when the physician reports a code that already contains it. The two provisions that speak to ordering both exclude it, and the definition of analyzed does so by name. Reading the whole set together, the work is already reflected in the valuation of the component being billed, and counting it again under data would pay for it twice.

Worth noticing that the drafters preserved ordering credit where they intended to. The same Services Reported Separately paragraph states that tests which do not require separate interpretation, such as results-only tests, do not count as an independent interpretation but may be counted as ordered or reviewed. The carve-out exists, and it was written for tests with no professional component rather than for tests the physician bills.

So a study whose professional interpretation the reporting physician or a same specialty partner bills is out of the data element completely. No Category 1 credit for the order, no Category 2 credit for the interpretation. The physician can still reference the study and its findings, and that reference does real work for the problem and risk elements, but it produces no data credit. The same reasoning extends to time, since time spent performing a separately reported service is excluded as well. Where the professional component is reported by a physician of a different specialty, the analysis changes, and that is taken up below.

For a surgeon reading and billing their own in-office films globally or with modifier 26, that removes both credits at every encounter.

When the technical and professional components are split

The exclusion is often applied more broadly than the guideline supports, on the understanding that a study the practice bills in any form is off limits. The condition is narrower than that. It attaches to who reports the professional interpretation, and specifically to whether that is the physician reporting the E/M service. If the practice bills only the technical component and a radiologist reports the professional component, the surgeon has not separately reported that interpretation and nothing is excluded.

The order counts under Category 1 in that arrangement. And Category 2 is not blocked either. Category 2 exists precisely for independent interpretation of a test performed by another physician. A radiologist interpreting the study has never disqualified the treating physician, since a radiologist interprets nearly every study. If the surgeon documents their own reading of the images, Category 2 is available on the terms described above.

The determining question is which entity is being paid to interpret the study. Where that is an unaffiliated radiologist or facility, the interpretation is available and the order counts, and the two are scored independently.

Two of these arrangements appear similar and are treated differently. Where the surgeon or a same specialty partner reports the professional component, both the order and the interpretation fall away, and the guideline text says so plainly. Where an employed radiologist reports it, the order clearly survives, and the interpretation is the contested piece, discussed further under the employed radiologist arrangement. The guideline conditions attach to the physician reporting the E/M service rather than to the group, so a literal reading leaves the interpretation available.

The employed radiologist arrangement turns on specialty rather than on employment. Same specialty plus same group is treated as the same physician, and radiology is a different specialty from orthopaedic surgery or neurosurgery, so an employed radiologist does not collapse into the treating physician. Some payers apply an entity-level reading instead and treat anything billed under the group as reported by the group. The guideline does not support that position, though it is worth confirming in writing with your major payers before the arrangement is built into a charge capture template.

Where both are available, the same test can be used in both places. That is worth stating carefully, because the two categories do not work the same way and it is easy to read them as equivalent.

So a single outside MRI that the surgeon ordered and then independently interpreted contributes one element toward the Category 1 combination and satisfies Category 2 on its own. Neither use depends on the other. The order counts whether or not anyone interprets the study, and a documented interpretation satisfies Category 2 whether or not the practice ordered it.

Source. The CPT E/M guidelines define the combination of data elements and state that a combination of different elements, such as notes reviewed, tests ordered, tests reviewed, or an independent historian, may be summed, that each item type or category does not have to be represented, and that a unique test ordered plus a note reviewed plus an independent historian is a combination of three elements.

When the group holds both components

An arrangement common in orthopaedic practice deserves separate treatment. The practice owns the equipment and reports the technical component. An employed radiologist reports the professional component. The surgeon documents their own reading of the study, and the question is whether Category 2 is available.

Start with the guideline text, because it is narrower than most guidance built on top of it. Every provision excluding a study from the data element conditions that exclusion on the physician or other qualified health care professional reporting the E/M service. The 2023 definition of independent interpretation reads the same way, stating that it does not apply where the physician who reports the E/M service is reporting or has previously reported the test. Nothing in these provisions extends to the group practice, and the guidelines use group practice language elsewhere where they intend it.

Applied literally, the surgeon retains both. The radiologist performed and reports the professional component, the surgeon reports neither, and the guideline conditions are met.

The counterargument is not in the guideline text but is worth weighing anyway. The group has already been paid to interpret the study, and a second physician within it then taking data credit for interpreting it again reaches the result the exclusion appears designed to prevent. Published guidance divides on this. Some states the rule at the practice level, phrasing it as you or your group practice billing the professional component, which closes the credit. Others states it by specialty, which leaves it open since radiology is a different specialty from spine surgery.

Two facts are worth verifying rather than assuming before any of this is decided. Whether the employed radiologist's professional component is reported under the group identification number or reassigned elsewhere. And whether the practice reports only the technical component, since the guideline provisions name the professional component specifically and technical component billing alone does not trigger them.

In-office radiograph, technical component only

This is the arrangement I get asked about most. The surgeon orders an ankle radiograph at today's visit. The practice bills 73610 with modifier TC. The employed radiology department reports the professional component under modifier 26. The surgeon documents no interpretation of their own today.

• The order counts. One element toward the Category 1 combination. The exclusion applies only where the physician reporting the E/M service is the one reporting the professional interpretation, which is not the case here. The radiologist is, and radiology is a different specialty, so the same specialty and same group rule does not treat the radiologist as the surgeon.
• Category 2 is empty. No interpretation was performed or documented at this encounter, so there is nothing to credit. That absence takes nothing away from the order.
• Net result: one Category 1 element for the order, and nothing under Category 2. Whether Category 1 is satisfied depends on the rest of the documentation at that encounter, since the combination needs three elements at moderate and extensive.

Change one fact and the answer changes with it. If the practice bills 73610 globally, or the surgeon reports 73610 with modifier 26 directly, or a same-specialty partner reports it, then the study is out of the data element altogether and the order goes with it. The variable is who reports the professional component, and nothing else in this analysis.

One timing note. Credit the order at the encounter where it was placed. When the surgeon reviews that same result at the follow-up visit, there is no second item, because the test has already been counted.

The 2023 errata

The March 2023 AMA errata and technical corrections, retroactive to January 1, 2023 and printed in the guidelines from the 2024 edition forward, made two separate additions to the definition of independent interpretation. Because they arrived in the same revision, they get read as if they compete with each other. They do not. They answer different questions.

• The first addition tightened the billing exclusion. Independent interpretation does not apply where the physician reporting the E/M service is reporting, or has previously reported, the test. That moved the exclusion inside the definition itself and extended it to studies the physician read and billed at an earlier encounter.
• The second addition resolved a counting question. A test that is ordered and independently interpreted may count as both a test ordered and interpreted. That addressed double counting on a study nobody at the practice is billing.

The second does not create an exception to the first. Billing the professional component still eliminates the interpretation, exactly as it did before 2023, and the errata arguably made that harder rather than easier by adding the previously reported clause. The two provisions apply in sequence.

This is a change worth checking your internal policy against. Audit tools written before 2023, and any reference material that predates the errata, still instruct coders to credit such a study once. That approach undercounts the data element on exactly the encounters where spine physicians do the most analytic work.

Element three: risk of complications

CPT states that risk includes decision making related to the need to initiate or to forego further testing, treatment, or hospitalization. The table reads decision regarding elective major surgery. It does not read decision to perform.

Source. The risk definitions, including the treatment of options considered and not selected and the distinction between minor and major procedures, are in the CPT E/M guidelines. The guidelines also tie the assessment of risk to the ordinary standards of a physician in the same specialty.

A documented analysis that concludes against surgery is still a decision regarding surgery. The coding value is real and so is the documentation requirement behind it. A note saying the patient is not a surgical candidate has recorded a conclusion, not an analysis. A coder cannot supply that analysis by reasoning that the surgeon obviously weighed the comorbidities.

Disease morbidity versus management risk

The guidelines address this directly, stating that the term risk as used in the problem definitions relates to risk from the condition, and that while condition risk and management risk often correlate, the risk from the condition is distinct from the risk of the management. So the morbidity of the untreated condition scores under problems. The risk of the management options scores under risk. Mixing them causes both undercoding and upcoding that will not survive review.

A patient with severe stenosis and progressive weakness managed conservatively may have high problem complexity, if the physician documented it that way. That does not make risk high on its own. Risk goes high when the record shows surgery was analyzed, risk factors were weighed, and a decision was reached. Two elements, two separate pieces of documentation.

Four encounters, scored

Four abbreviated notes, cut down to the language that drives selection. In each one, the level follows from what is on the page.

99212, straightforward

Nothing here needs interpretation. The physician characterized the course and I validated it.

99213, low

Notice the radiograph is credited as review of a test result, not as interpretation. The note repeats what the report said.

99214, moderate

99215, high, without a surgical decision

The word "severe" is in the physician's assessment and the supporting detail follows it. If the note carried the same clinical facts with no characterization of severity, I would query. I would not decide the facts speak for themselves.

Comparative documentation examples

Each pair is the same encounter written twice. The clinical facts are identical in both versions. The only thing that changes is whether the record communicates the physician's determination, and that is the entire question a coder is answering.

1. Status of a chronic condition

2. Exacerbation versus severe exacerbation

3. New complaint versus undiagnosed new problem

4. Prior treatment history

5. External records: one source or three

6. Referral versus discussion

7. Medication list versus prescription drug management

8. Surgical decision with comorbidities present

9. Declining to operate

10. Time-based selection

11. Longitudinal relationship for the add-on code

12. Viewing versus interpreting the images

When to query

Coders rarely go wrong on the notes that clearly support a level or clearly do not. It is the ones in between. Query when the record suggests work happened that the documentation does not capture.

• Status is missing. A chronic condition is addressed and managed but never characterized as stable, worsening, or at goal.
• Imaging may have been interpreted. The note carries detailed findings, but nothing indicates the physician viewed the study rather than reading the report.
• A discussion may have occurred. The plan reflects input that appears to come from another physician, with no documented exchange.
• Risk factors are listed but unconnected. Comorbidities appear in the history alongside a surgical decision, with no statement that they bore on it.
• Prior treatment is referenced generically. "Failed conservative care" without identification of what was tried and how the patient responded.
• The differential is implied by the workup. Studies ordered suggest a diagnostic question the assessment does not state.

A compliant query is non-leading. The physician is asked for the characterization rather than offered one to confirm. Outcome does not enter into it either. A query is equally appropriate whether the answer would raise the level, lower it, or leave it where it is.

Source. Query construction is governed by your organization's query policy, which should reflect current industry query standards and address outpatient practice specifically. Confirm which version of that guidance your compliance program has adopted, since these standards are periodically revised.

G2211 and longitudinal spine care

The E/M level covers the work of the encounter. It does not cover the work of carrying a treatment relationship across encounters. HCPCS G2211 is there for that, and it is one of the most commonly missed charges in specialty practices that qualify for it.

First-year national data showed the code reported on roughly 10.5% of eligible visits through the first three quarters, reaching about 19% by December, against agency projections of 40% to 50%.

Source. CMS implementation guidance for G2211 is in MM13473 and in MLN006764. CMS states that the complexity the code captures is not in the clinical condition but in the cognitive load of continued responsibility for the patient, and that the visit itself must be reasonable and necessary.

Not a fourth MDM element

The visit complexity G2211 describes does not enter the problem, data, or risk analysis and does not move the level. It is relational rather than clinical: the burden of retaining responsibility for a patient's care over time. Two consequences. A 99213 plus G2211 is an ordinary, appropriate combination. A 99215 for a one-time surgical consultation supports no add-on at all, however complex the visit was.

The relationship in the record

The descriptor covers two relationships: serving as the continuing focal point for all needed care, or providing ongoing care related to a single serious or complex condition. Spine practices generally qualify under the second. Eligibility is not restricted by specialty and no specific diagnosis is required. The coder validates whether the record identifies the condition under ongoing management and reflects continuity, not whether the physician feels responsible for the patient.

The modifier 25 restriction

Modifier 25. G2211 is not payable when the base E/M carries modifier 25, except where the same practitioner furnishes an allowed annual wellness visit, vaccine administration, or other Part B preventive service that day. That exception, effective for services on or after January 1, 2025, does not extend to procedures. An office visit with modifier 25 alongside an ESI, medial branch block, or trigger point injection does not support the add-on. In interventional spine practices that removes a large share of otherwise qualifying encounters.

The allowed services for that exception are listed in Attachment 1 to CMS Transmittal 13015. Also worth knowing: postoperative visits in the global period have no separately reported base code to attach to; effective January 1, 2026 the descriptor was revised and reporting expanded to home and residence codes 99341 to 99345 and 99347 to 99350; Medicare Advantage recognition varies and commercial and Medicaid payers are not required to recognize it; and the add-on is subject to coinsurance and deductible.

Audit this one both ways. Practices that never report it are missing payment for work already done. Practices that append it to every encounter, including one-time consults, are creating exposure. Neither pattern matches the relationship the code was written to recognize.

Common audit findings

Validation checklist

These are the questions I work through against a spine note. They are questions about the record rather than about the patient, and they are offered as a starting point for an organization building its own review tool.

• Does the note state a status for each problem credited?
• Is a management decision documented for each problem credited?
• Where severity or threat to function drives the level, did the physician characterize it?
• Is functional change described in the note rather than inferred from the diagnosis?
• Are prior treatments identified by type and response, or referenced generically?
• For external notes, is each source identified well enough to determine whether it is a distinct group, specialty, subspecialty, or entity?
• Is the correct category structure being applied for the level being scored, given that the independent historian falls under Category 2 at limited and Category 1 at moderate and extensive?
• For each study credited, is the source and date identified?
• Where Category 2 is credited, does the note indicate the images were viewed, and does it contain a finding the physician drew from that viewing rather than received from the report?
• For each test credited, is it clear what is being counted: the order, the review of a result, or an independent interpretation?
• For every study credited, is it known who reports the professional component, whether that physician shares a specialty with the treating physician, and whether the component is reported under the same group identification number?
• Where a study was both ordered and independently interpreted, has credit been taken for both rather than one?
• Where Category 3 is credited, is there a documented interactive exchange?
• Where a surgical decision drives risk, is the analysis documented rather than only the conclusion?
• Where the level rests on an element that is close, has a query been considered?

Sources of authority, and how much weight each carries

Guidance in this area arrives from several places, and the sources do not carry equal weight. Sorting them out is the step that turns an opinion into a position an organization can defend.

Where a determination is answered plainly in the guidelines or an erratum, there is little reason for an organization to adopt a different reading. Where the guidelines are silent or in tension, payer policy is usually the most useful thing to go after, since it turns an interpretive question into a documented understanding with the entity that will actually pay the claim.

For the arrangements discussed in this article that turn on billing structure, particularly the technical and professional split and the treatment of ordering, a written payer response carries more protection than any interpretation offered here or anywhere else.

Where your organization has to take a position

Several of the determinations in this article remain open under the guidelines. Reasonable, well credentialed people read them differently, and payers apply them unevenly. Protection on these comes from having a written position, the reasoning that supports it, the date it was adopted, and consistent application across every coder in the organization.

My position on each is set out with the reasoning behind it. Your organization may reach a different conclusion. The one approach to avoid is leaving the question unresolved for individual coders to settle chart by chart, since inconsistency across a panel turns a defensible interpretation into a pattern finding.

A written policy does something a correct answer alone cannot. It shows that the organization considered the question, chose a reading, and applied it deliberately. That is a materially different posture in an audit than a coder explaining their longstanding habit, and it holds even where a reviewer disagrees with the position taken.

Revisit these annually against the current CPT publication and any new errata. Two of the six changed by technical correction, and errata do not arrive with the visibility of a code set update.

Primary sources and further reading

Nothing in this article should be applied from the article itself. Each point traces back to a published source, and the source is the citation for an appeal, an internal policy, or a provider education session. These are the references I keep open when I audit spine charts.

Code set and coding guidance

• CPT Professional Edition, Evaluation and Management Services Guidelines. The controlling source. It defines problems addressed, unique test, independent interpretation, appropriate source, the data categories, the risk table, and total time. Any internal policy on MDM should quote this and not a summary of it. Verify against the current code year.
• AMA, CPT Evaluation and Management Code and Guideline Changes. The AMA publishes the E/M guidelines and definitions as a standalone document alongside each year's changes. It carries the full definitions of analyzed, test, unique, combination of data elements, external, discussion, independent historian, independent interpretation, appropriate source, and risk, along with Table 1. Reading the definitions in full is faster than working from any summary of them. 2023 edition
• CPT Assistant (AMA). The AMA's official coding guidance publication. When a determination is contested internally or by a payer, this is where to look before writing policy. Payers frequently cite it back.
• AMA, CPT Errata and Technical Corrections. Two are relevant to this discussion. The March 2021 correction added ordering to the list of activities excluded when the professional interpretation is separately reported. The March 2023 correction, retroactive to January 1, 2023, made two separate additions to the definition of independent interpretation. It specified that independent interpretation does not apply where the E/M reporting physician is reporting or has previously reported the test, which tightened the billing exclusion. And it specified that a test ordered and independently interpreted may count as both, which resolved a counting question on studies the practice is not billing. Neither addition creates an exception to the other. Both are carried in the printed guidelines now, though reference material written earlier still reflects the pre-correction wording. 2023 errata
• AMA, CPT Evaluation and Management revisions FAQs. Addresses MDM, time, and documentation questions not covered elsewhere, including chronic illness severity examples. ama-assn.org
• ICD-10-CM Official Guidelines for Coding and Reporting. Governs diagnosis reporting, including the outpatient prohibition on reporting uncertain diagnoses. Relevant to the undiagnosed new problem discussion, where the differential supports MDM without being coded.

Medicare policy

• CMS MLN Booklet, Evaluation and Management Services, MLN006764. Current version dated May 2026. States that documentation must support the level reported and that volume of documentation does not determine the level. Also carries the G2211 instructions. Note that the May 2026 revision removed the prior discussion of modifier 25 with minor procedures, so if your internal policy cites that passage it needs updating. cms.gov
• CMS MM13473. The implementation article for G2211, including the description of visit complexity as cognitive load rather than clinical condition.
• CMS Transmittal 13015, Attachment 1. Lists the Part B preventive services that permit G2211 alongside a base E/M carrying modifier 25, effective for services on or after January 1, 2025.
• CY2026 Medicare Physician Fee Schedule Final Rule, CMS-1832-F. Source for the revised G2211 descriptor and its extension to home and residence visits, the efficiency adjustment to work RVUs, and the Ambulatory Specialty Model. CMS fact sheet
• Medicare Claims Processing Manual, Pub. 100-04, Chapter 12. Global surgery package and the modifier instructions for 24, 25, and 57. Relevant to any spine practice reporting E/M during a global period or on the same date as an injection.
• National Correct Coding Initiative Policy Manual, Chapter 1. General correct coding policies, including the standards applied to modifier 25 and the separately identifiable E/M service.
• Local Coverage Determinations from your MAC. Spine practices should track LCDs on epidural injections, facet interventions, and imaging, since coverage criteria shape both the procedure and the documentation supporting the visit.

Documentation integrity and audit

• Current industry query practice standards. Provider query construction is governed by published industry guidance that is periodically revised. Confirm the version your compliance program has adopted, and that it addresses outpatient query practice rather than inpatient alone.
• Coding education and reference material. Practical treatment of independent interpretation and the data element, useful for seeing how a provision is being applied in the field. Readings divide on whether the Category 2 exclusion is stated at the practice level or by specialty, which is the distinction that decides the employed radiologist arrangement. Check any of it against the current guideline text, since older material may predate an erratum.
• OIG Work Plan and OIG audit reports on evaluation and management services. E/M coding accuracy has appeared repeatedly. Worth reviewing annually for whether specialty E/M or add-on codes have been added.
• CERT program improper payment data. Insufficient documentation is consistently a leading cause of improper payment, which is the same finding this article describes at the chart level.
• Comparative Billing Reports and your own MAC's provider education. Useful for benchmarking level distribution before an outside reviewer does it for you.

Closing thoughts

Spine E/M gets missed in both directions, and the cause is the same either way. The level is being decided by something other than the documentation. Undercoding happens when risk is treated as the controlling element and the documented problem complexity and analytic work get ignored. Overcoding happens when clinical facts are treated as a substitute for the physician's characterization of them.

The discipline is the same either way. Score each element against what the note says. Apply two of three without adjusting for how the visit felt. Credit what is written, decline what is not, and query the gap in between. The physician decides what the encounter required. My job is to confirm the record says so.

Where the guidance itself is unsettled, that discipline needs something behind it. The organization has to decide how it reads the guidance, write the decision down with the reasoning and the date, and hold every coder to the same reading. An interpretation a reviewer disagrees with can still be defended. Six coders interpreting the same provision six ways leaves nothing to defend.