Thinking on Paper: Guidelines for Documenting Medical Decision Making

The published guidelines don't tell the whole story. Here are the tools the guidelines should have included.

Robert L. Edsall and Kent J. Moore

HCFA's new documentation guidelines for medical decision making are more striking for what they leave unsaid than for what they say. They do tell you what a reviewer would look for in determining the level of your decision making. Unlike the guidelines for documenting history and exam, though, they don't tell you enough to figure out the level yourself. That means, ultimately, that they leave you unable to tell whether your documentation supports the E/M claim you submit. True, that's nothing new, but why do the other guidelines help where these don't?

Is the problem that thinking is mysterious and hard to quantify? Is it that only you can know what you were thinking? Could be. Pat Price, MD, medical director for Medicare Part B in Kansas, Nebraska and western Missouri, wrote in a newsletter for participating physicians, "While it may be possible for an astute assistant to determine the level of history and the level of physical examination performed, only the physician can determine the level of medical decision making." Hmm.

Obviously, that doesn't cut it. Somebody other than the physician has to be able to determine the level of decision making -- and if you're audited, somebody will. A reviewer, probably a nonphysician reviewer, will decide what level of medical decision making you've documented. In fact, as you'll see from the guidelines, the reviewer will be quantifying what you were thinking about -- not how or what you thought. And if a reviewer can do it based on what you've written down, then why don't the guidelines give you enough information to do it yourself?

We believe that the standards for determining the level of medical decision making are articulated in documents that HCFA is currently unwilling to share with physicians. FPM and the Academy are working to obtain the pertinent information. In the meantime, we do have a draft version of one document that casts light on the situation. In this article, we'll share what we've learned from that document in addition to squeezing what help we can out of the guidelines themselves.

As you know, the CPT manual considers the level of complexity of medical decision making to be a function of three variables:

This tripartite division is reflected in the guidelines. For short, we can call these three elements uncertainty, data and risk.

The guidelines follow CPT in recognizing four levels of each of these elements, and four corresponding levels of medical decision making overall. The level of medical decision making for a given visit actually depends on the highest two out of these three elements. If you find that confusing, remember that the level of medical decision making is always the same as the second-highest of the three elements if they're all different or the same as the tied elements if they're not (see "The elements of medical decision making"). That's the easy part, though. The challenge lies in properly documenting uncertainty, data and risk and in figuring out the level of each that a given note represents.

The number of potential diagnoses and management options you have to deal with in an encounter is apparently not as simple as it sounds. Counting the separate problems, potential diagnoses or treatment measures you mention to arrive at a number is part of it, to be sure, but the guidelines suggest that this factor depends on each of the following:

You can document the number of problems you deal with simply by listing them, as you probably do anyway. The guidelines do not require that you actually put down a number. The documentation of diagnostic uncertainty and management options is a little more involved, but it's also something you are probably doing already. If you understand what the guidelines are looking for in these areas, however, you may be more likely to cover all the bases.

How good a handle you have on the diagnosis of each problem is the crux of the issue here; according to the guidelines, "decision making with respect to a diagnosed problem is easier than for an identified but undiagnosed problem." Actually, the guidelines suggest the following spectrum of diagnostic uncertainty, from least to most uncertain:

While almost anything you say about a problem will suggest where it falls on this spectrum, you may save yourself trouble in an audit if you keep these categories in mind and always make it clear which category a given problem falls into. As with most of their other recommendations, the guidelines do not insist that you identify the category explicitly, only that it be clearly implied. For instance, the fact that you document ordering lab tests or referring a patient for a biopsy should make it clear enough to any reader that you're dealing with an undiagnosed problem for which the history and physical don't provide enough information. Similarly, if your assessment says, "IDDM -- controlled," it should be clear to any reviewer that you're dealing with a problem at the low end of the uncertainty spectrum (see "Just how clear was your implication?" below).

When the history and physical aren't enough to pin down a probable diagnosis, it's important to document what you did to get more information, in part because what you did can help indicate the degree of uncertainty you're dealing with. As the guidelines say, "The number and type of diagnostic tests employed may be an indicator of the number of possible diagnoses," and "The need to seek advice from others is another indicator of complexity of diagnostic or management problems." The guidelines ask, reasonably enough, that when you refer a patient, request a consultation or ask for advice, you specify whom you referred the patient to, requested a consultation of or asked for advice.

A caution: While noting that you've ordered a lab test can imply something about the diagnosis or diagnoses you are considering, it's probably worth taking the time to add an explicit comment to the assessment. Something as brief as "hypothyroidism vs. anemia" can be enough.

"You can't assume that we'll infer everything from a lab test," says Dr. Price. "You can't write 'EKG and a Chem-21 profile' and expect a reviewer to know what you were thinking about." Dr. Price recognizes that a test you order may indeed help a reviewer determine the level of medical decision making if it really does indicate your line of thinking, but to do so, it has to be fairly specific. "No credit will be given for ordering a large array of tests," he says.

The point, according to Dr. Price, is to convey that you have been thinking. "I see so many records that just have an impression -- a single word of diagnosis -- and a single line or single word that says what they've chosen as their treatment plan. What that doesn't indicate is that the physician has thought of a range of diagnoses and selected one based on the information above. This is what we all do in our heads, but now it has to be indicated on a piece of paper."

In addition to implying that the need to seek advice about treatment suggests a higher level of medical decision making, this section of the guidelines includes one point about treatment: "The initiation of, or changes in, treatment should be documented. Treatment includes a wide range of management options including patient instructions, nursing instructions, therapies and medications." The guidelines don't really get directly at the issue of how many management options are open to you in a given situation -- something you would expect to be important given that one element of medical decision making is the number of diagnoses and management options. Still, saying what option you picked will presumably imply something about the options that were open to you. In any case, if it's clear from your note that you're dealing with a self-limited problem that doesn't require any treatment, the fact that you had no management options to worry about will have obvious bearing on the level of decision making.

The guidelines give no help in defining what minimal, limited, multiple and extensive mean when applied to the number of diagnoses and management options. We provide the table shown in "Quantifying diagnoses and management options" as an effort to rectify the omission. The table is based on a draft of a documentation score sheet proposed for use by carriers. You may find it useful in determining how to code a given visit as well as in evaluating the adequacy of your documentation. Remember, though, that this is based on the draft of something that may change before it is fully implemented and may vary from carrier to carrier.

To get a feel for how the scoring system works, let's see how it applies to a sample note:

The patient's hypertension constitutes an established, previously diagnosed problem that is responding to management, so it rates one point. As a previously undiagnosed problem, the fatigue would rate three points, except that lab work is needed to rule out electrolyte abnormalities; that raises its score to four points, giving the encounter a total of five in this category -- more than enough to consider the number of diagnoses and management options extensive. Note, by the way, that the assessment section gives enough diagnostic information to establish the purpose of the lab work mentioned in the assessment; that's useful in helping the reviewer understand the thought processes involved.

The scoring system includes no provision for quantifying the uncertainty implied by the treatment options (is discontinuing the hydrochlorothiazide enough, or will the Corgard have to be discontinued in favor of another agent?), but since the maximum possible score of four points has already been reached, that doesn't matter in this example.

Several characteristics of this scoring system are important to keep in mind (you may want to refer to the table to see why each of the following is true):

When the guidelines talk about "the amount and complexity of data to be reviewed," they refer to information gathered from sources other than the history and physical -- lab tests, imaging, other diagnostic services, old records and history from sources other than the patient. Generally speaking, the guidelines ask that you record the decision to seek additional information and, if you have obtained the information, the results of your review of it. The rationale is that any such step implies an increase in the complexity and volume of the data you need to winnow. Specifically, they ask that you make sure the documentation reflects the following:

Again, the guidelines give no hint as to what constitutes minimal, limited, moderate or extensive review of data. Well, one hint: The bottom of the range is pinned down by the name of the lowest category, which is actually "minimal or none." Just keep in mind that none means "nothing beyond what you collected from the history and exam." So even if you document a comprehensive history and physical, you still have no "data to be reviewed" unless you've taken one of the steps outlined above.

The draft documentation score sheet mentioned above provides a point system that may be helpful in this respect (see "Quantifying the amount and complexity of data to be reviewed"). The same caveat applies: Since this information comes from a draft version, it may not be quite what your carrier uses. We believe that it is accurate enough to be useful, however, and FPM will provide updated information if it becomes available.

As you can see, the scoring system for this area is relatively straightforward. Note, though, that you can give yourself no more than one point for each of four types of data (lab, radiology, medicine and old records or additional history), no matter how much data there is of that type. One lab test counts as much as 20. The other opportunities for points involve documenting your digging into the data -- talking with the physician who performed a test or double-checking that physician's interpretation of the raw results, analyzing old records or taking history from someone who knows the patient. Where "direct visualization and independent interpretation of a specimen, image or tracing" is concerned, remember that this counts only in cases where you are checking another physician's interpretation. If you get a chest film in your office and read it yourself, you've earned yourself one point, not two, for review of data.

How does the data scoring system apply to the sample note we discussed previously? The only relevant action documented was the decision to order a CBC and SMA 7. Since both fall into the lab test category, the total score is only 1: minimal or none.

The guidelines consider risk to the patient in determining the level of medical decision making, not risk to you or your staff (risk of injury or infection, for instance) or malpractice liability. They're talking specifically about risk of significant complications, morbidity and mortality, and they recognize three sources of this risk: the presenting problems, any diagnostic procedures you choose and any management options you choose.

The guidelines include only a few points under documenting risk, but much of the information needed for assessment of the level of risk is already called for in other parts of the guidelines. These are the additional points made:

Clearly, the quantification of risk is not an exact science, any more than the quantification of uncertainty. The guidelines do attempt to contain the problem by specifying limited periods over which to estimate risk. Where risk is related to presenting problems, you are to assess the amount of risk the patient will be subjected to "between the present encounter and the next one." For instance, a patient who has recently tested HIV-positive faces high risk over the long term, but for the purposes of the documentation guidelines, you would assess the probably relatively low risk to the patient between now and your next scheduled visit. When risk is related to your decisions and actions -- that is, when it is related to diagnosis and management options -- the guidelines ask you to assess the risk "during and immediately following any procedures or treatment."

As with the other elements of medical decision making, the guidelines recognize four levels of risk: minimal, low, moderate and high. But while the guidelines provide no tools for quantifying uncertainty or data, they do provide a useful table for quantifying level of risk (see "Quantifying the risk of complications, morbidity and mortality" below). The table lists common examples of problems, diagnostic procedures and management options classified by level of risk. It is intended to serve loosely as a ruler against which to measure the risk inherent in problems and procedures not listed. It will probably not resolve every dispute you might have with a carrier about risk level, but it should help considerably.

To use the table in establishing a level of risk, find the categories in which the patient's presenting problems and your diagnostic and therapeutic actions seem to fit best. The highest individual level of risk determines the overall level. To see how it works, let's use the table in determining the level of risk represented by the encounter with the hypertensive patient documented above:

The presenting problem, hypertension with fatigue probably resulting from antihypertensive therapy, seems to fit in the moderate risk level as "one or more chronic illnesses with mild exacerbation, progression or side effects of treatment."
The diagnostic procedures ordered, CBC and SMA 7, seem to fall into the minimal risk level, as "laboratory tests requiring venipuncture."
The management option selected, modification of prescription drug therapy, seems to fall into the moderate risk category, as "prescription drug management."

Since the highest risk identified is moderate, that is the risk associated with the visit as a whole.

You may find it useful to examine the table for patterns and characteristics that will help impress the categories in your mind. For instance, under diagnostic procedures, note the four-step progression of imaging studies from chest X-rays and ultrasound to noncardiovascular studies involving contrast, to cardiovascular studies involving contrast and no identified risk factors, to cardiovascular studies involving contrast with identified risk factors. Similar patterns of increasing risk are evident in the columns for presenting problems and management options as well.

One remarkable characteristic of the table is that it puts prescription drug management on a par with elective major surgery as a moderate-risk intervention. Remembering that any visit that involves a prescription is at least a moderate-risk visit can simplify the process of establishing a risk level. You need only consider the entries in the high-risk row to see whether any characteristic of the visit justifies increasing the risk level from moderate to high.

Now that we've examined each of the three dimensions the guidelines use to measure medical decision making, let's try measuring another encounter. Consider the following note, from a visit with an established patient:

Nurse's notes: Age 50, T 98.6, Wt 156, BP 120/80

How would you evaluate this visit in terms of the level of medical decision making? If you wish, see what you come up with by referring to the tables concerning uncertainty, data and risk before reading the following analysis.

The documentation mentions two established, previously diagnosed problems (diabetes and coronary artery disease), which represent a score of two points. The mild obesity adds a third point, whether you count it as a minor problem or a third established problem. That the physician encouraged the patient to see an ophthalmologist does not affect this score, since there's no indication that the ophthalmologist was brought in to help diagnose a new problem; rather, it seems to be a routine follow-up visit. The score for this element of decision making is 3, then, representing multiple diagnoses or management options.

The note does not reflect an intention to go outside the history and physical for data. The score for this element is 0, representing minimal or no data to be reviewed.

With at least two stable chronic illnesses, the presenting problems constitute a moderate level of risk. No invasive diagnostic procedures are planned, so the risk in that category is minimal. Since insulin therapy figures in the management described, the level of risk involved in management is moderate. Since the highest level is moderate, the encounter represents moderate risk overall. (While two of the three categories came up moderate, one would have been enough.)

Using the standard table (see "The elements of medical decision making"), we can see that this visit rates at level three for both uncertainty and risk. The overall level of medical decision making for the visit is then level 3, moderate complexity.

Medical decision making seems to have a special role in determining the level of a patient encounter, even though it's supposed to be weighted evenly with the history and exam. Charles Colodny, MD, a family physician practicing in Libertyville, Ill., who represents the Academy on the AMA CPT Advisory Committee puts it succinctly: "The carriers are well aware that a physician intent on upcoding can increase the level of the history and physical very easily. Medical decision making is something else entirely. This is where they're going to be looking."

Dr. Price agrees wholeheartedly. In a newsletter for participating physicians in his region, he wrote, "It should be the complexity of the medical decision making process and the medical problem which is the most heavily weighted factor determining the E/M service level." Dr. Price made it clear in a subsequent conversation that he views the medical decision making component as a reality check on the other two key elements. While he recognizes that any two of the three can determine the overall level for an established patient visit, he says the physician who consistently "does a great history and a great physical on someone who has a cold" is asking for trouble.

The guidelines represent a system for relating documentation to level of service. And, like any system, they practically constitute an instruction manual for anyone intent on gaming the system. This uncomfortable realization may be part of the reason that important holes were left in the guidelines for decision making. Leaving part of the system unexplained certainly makes it harder to subvert the system. And when you think about it, the reason medical decision making is the element most likely to prevent upcoding is that it takes into account the nature of the presenting problem -- something outside the physician's control. The other aspects of decision making (the number of diagnoses and treatment options and the amount of data to be reviewed) are more within the physician's control and more susceptible to misrepresentation -- and they're the aspects for which the guidelines themselves don't provide "scoring" information.

But even if the obvious ethical considerations did not militate against subversion of the guidelines, good sense argues against it. A claim supported by documentation that obeys the letter of the guidelines while violating their spirit can still be challenged by the carrier on the grounds of medical necessity, as Dr. Price suggests. The trick here, and one that may be difficult to pull off, given the emotionally charged atmosphere of physician-carrier relations, is to push the guidelines exactly as far as accuracy requires -- no more, no less.

Rather than an invitation to game the system, these guidelines are an encouragement to stop short-changing yourself. You are careful to provide the services the patient needs. If you follow the guidelines in documenting just as carefully the services you have provided, you can go on with assurance to claim precisely what you are owed. Surely that's fair enough.

Robert Edsall is editor-in-chief of Family Practice Management. Kent Moore is the AAFP's manager for reimbursement issues and a contributing editor to Family Practice Management.