If coding for evaluation and management seems like working a jigsaw puzzle, take heart: Even the professionals trained to use the Current Procedural Terminology (CPT) codes don’t always agree on how to do it.

A survey of certified professional coding specialists who were asked to assign evaluation and management (E&M) codes to hypothetical cases found that they agreed on the appropriate code only slightly more than half the time.

The survey results, published in the February edition of the Archives of Internal Medicine, lend some scientific credence to what clinicians everywhere have discovered in their practice.

"The average physician has minimal to no training in CPT coding," said Mitchell King, M.D., an assistant professor of family medicine at Northwestern University Medical School in Chicago and lead author of the report. "Anyone who has been out in practice for a number of years is learning to do it on the fly. It’s a very complicated system that coding specialists themselves cannot come to agreement on. I think it raises the question of whether anyone should be audited."

In the study, 300 certified professional coding specialists randomly selected from the active membership of the American Health Information Management Association were sent six hypothetical progress notes of office visits, and 136 surveys were returned.

Coders agreed on the appropriate code to use 58.7 percent of the time. Across individual cases, the level of agreement ranged from 50 percent to 70 percent. Undercoding, relative to the most common or consensus code among the respondents, occurred more commonly than overcoding and occurred significantly more often for established patients. In contrast, for new patients’ progress notes, overcoding relative to the consensus code was more common than undercoding.

King and colleagues suggested in their paper that coding criteria are stricter for new patients, requiring more documentation to establish the same service level. In addition, physicians and coding specialists may recognize that caring for new patients requires more effort and that there is more uncertainty in providing this care than for established patients.

For this reason, physicians and coders may feel that new patients are more difficult—and established patients less difficult—and coding levels may reflect this.

A demographic questionnaire, which accompanied the survey, found that coders averaged 10.9 years of coding experience, with an average of 8.3 years’ experience coding in physicians’ offices.

"It highlights the need for some serious thought about revising a system that is so complex that even people who are trained in school can’t apply it consistently," King said.

Chester Schmidt, M.D., chair of APA’s Committee on RBRVS, Codes, and Reimbursements, agreed. He said the study is relevant to psychiatrists, who use the E&M codes for inpatient and partial hospitalization, consultation, nursing home visits, and some office visits.

"Psychiatrists, like all physicians, have some difficulty using these codes and selecting the level of service, and they are uncertain about the elements of documentation needed to support it," he said.

Schmidt cited some weaknesses in the study, including the fact that the hypothetical cases themselves were not presented. And he said mail surveys are subject to selection bias, noting that the 136 coders who responded represented fewer than half of the total (300) who were actually surveyed.

Finally, Schmidt noted that some of the variation among the coders may have been due to the survey method itself. "When you mail something like this out, you don’t have everyone in one place, so that everyone hears the same thing," he said.

But he said there is no doubt that the rules for E&M coding and documentation are all but byzantine. Especially controversial, he said, are the guidelines for determining the level of "medical decision making"—one of the three E&M components, along with history taking and examination.

The guidelines for that component use a four-by-four grid designed to produce a numerical approximation of medical decision making—a system so complex "it bears no resemblance to the actual practice of medicine," Schmidt said.

Despite vocal protests from physicians, and promises from the AMA and the Centers for Medicare and Medicaid Services (CMS) to revise the medical decision-making guidelines, there has been no action, Schmidt said.

He said that the AMA, which has a copyright on the CPT codes, may have little room for maneuvering to make the codes less stringent. It was the AMA that collaborated with CMS’s predecessor, the Health Care Financing Administration, on the Resource-Based Relative Value Scale, and the coding system to go with it, in 1992—partly in response to criticism that the older coding system was insufficiently rigorous, Schmidt said.

"No system has been designed that has been scientifically tested for its validity and reliability," Schmidt noted. ▪