Better benchmarks for E/M coding comparisons

Accurate coding is critical to a medical practice’s success. Each time a provider interacts with a patient, the encounter will be documented in the patient’s medical record along with the appropriate diagnostic code to describe the patient’s presenting problems and a procedural code describing the medical services provided to the patient. Since fee-for-service billing is automated, the insurer will use the procedure and diagnostic codes to determine the amount of payment the practice will receive.

The Current Procedural Terminology, 4th Edition (CPT-4), published by the American Medical Association, is the standard procedural code that describes medical and surgical procedures, office and hospital encounters, laboratory tests, imaging procedures and all other types of patient services.

Since accurate coding is essential for appropriately documenting patient services and ensuring correct payment, practice leaders need to know if their providers are coding correctly. For some specialties and services, the best process for assessing coding is to audit a sample of medical records comparing the documentation to the procedural and diagnostic codes on the patient bill.

However, for some services, especially the codes for office and hospital encounters (designated E/M codes), the process is more complex. E/M codes are organized by type and location of services, and within a category — established patient, new patient, office or other outpatient service, inpatient service, etc. — there will be different levels of care describing the relative complexity of the encounter.

The most common process in assessing E/M codes is to compare coding profiles of individual providers to outside benchmarks to determine if the pattern is similar or if there may be evidence that providers are coding differently than their peers. This process works well if the practice’s doctors have similar patients as the doctors in the comparison group. It also means the practice needs to find an appropriate benchmark.

Commonly, practices will try to benchmark E/M coding using information published by the Centers for Medicare & Medicaid Services (CMS) for Medicare patients. This information is readily available on government websites. Unfortunately, while the coding profiles are easily available, the type of services for Medicare beneficiaries may be very different than those provided to younger patients.

The CMS National Health Expenditures Fact Sheet describes how much Americans spend each year on healthcare, which shows that younger patients spend only a fraction of what is spent by older patients, and the difference in cost is directly related to the services each group receives. The most recent data shows that patients between 19 and 44 years of age spend an average of $4,458 per year, while patients 65 to 84 spend $16,872, and those 85 and older spend $32,411 per year. Therefore, practices using Medicare data as a benchmark need to select only Medicare patients for their comparisons or they may want to find another benchmark.

The 2019 MGMA DataDive Procedural Profile data potentially solves the problem as it is based on a cross section of all patients.


Examine the graph displaying the E/M code distribution for CPT codes 99211 through 99215, established patients, office or other outpatient service for internal medicine and family medicine physicians. Neither graph displays a bell-shaped curve where the most commonly performed procedure is in the center of the graph. While the number of level one (99211) and level five (99215) procedures are similar in internal medicine, more than half of the procedures are level four (99214). For family medicine the curve is slightly more bell shaped, but the curve favors more complex procedures.

Most importantly, the two graphs compare the Medicare coding profile for the specialty with the information from the 2019 MGMA DataDive Procedural Profile. The two profiles are relatively similar for internal medicine, since most patients who have an internist as their physician are older and are Medicare beneficiaries.

However, the coding profiles are very different for family medicine physicians. It comes as no surprise that family medicine will have many younger, less complex patients who therefore warrant a lower-level office visit.

The table displays the percentages used to build the internal medicine and family medicine graphs and shows two other specialties for further comparison, general orthopedic surgery and noninvasive cardiology. These specialties have very different coding profiles as we would expect based on the type of patient and the type of services these specialists provide. The coding profiles for Medicare and from the 2019 MGMA DataDive Procedural Profile are relatively similar for noninvasive cardiology, but not for general orthopedic surgery. Again, as with internal medicine, most cardiology patients are older and therefore likely to be Medicare beneficiaries, whereas general orthopedic surgery will have a broad mix of patients with a different profile of office visits.

While having the right benchmark is important, how you use the results matters most. Examining the coding profiles provides invaluable information. Are your physicians’ E/M coding profiles similar or different from the benchmark? If the profile is different, do you know the reason? Do your physicians have a different type of patient? Are they younger or older? Do they present with comorbidities? If you feel there is a possibility that an insurer could audit submitted claims, you can preempt potential problems by performing an internal audit and taking corrective action, if warranted.

Comparing E/M procedure profiles is not difficult. With the right benchmark, practice leaders can rest easier knowing that everyone is doing their job.