Economic Grand Rounds: The Mismanagement of Medication Management

Advances in psychopharmacology have created situations in which conditions formerly thought to be manageable by psychotherapy alone are now treated with psychotherapy augmented by medication. Medication augmentation pertains not only to various types of depressive reactions but also to obsessive-compulsive disorders and anxiety states.

Certain other illnesses, such as severe schizophrenic reactions, are traditionally thought to be primarily treatable with psychopharmacology. In the past the treatment of many of these illnesses was rather straightforward, but increasingly complex algorithms that address the treatment issues are being developed. The algorithm developed by the Texas Schizophrenia Medication Algorithm Project mandates that all patients with schizophrenia be converted from conventional antipsychotics to atypical antipsychotics, even those who are doing well on conventional antipsychotic therapy (2).

Supporters of this algorithm suggest that it be used as a prototype for medication decision making in the rest of the country. If the algorithm is widely adopted, code 90862 would be used more frequently because of the necessity of following patients closely during the changeover period. During this interval, patients would be continued on maintenance dosages of conventional antipsychotic medications while atypical antipsychotic medications are titrated up to maintenance dosages. Then patients would have to be followed equally closely during the subsequent downward titration of conventional antipsychotics. The surge in the use of code 90862 may be temporary but could well continue for years, because the changeover would involve 1 percent of the population, or approximately 2.5 million people.

Algorithms of even greater complexity and specificity that require a higher level of diagnostic sophistication have been developed for schizoaffective and bipolar treatment. A recently published algorithm addresses the treatment of five different types of acute manic presentations and encompasses 15 steps (3), involving lithium or divalproex used either alone or in various combinations with three other types of anticonvulsants, calcium channel blockers, clozapine, and thyroid preparations. The new algorithms contribute to the complexity, and the efficacy, of psychopharmacology.

Even when some of the medications listed above are used alone, increased psychopharmacological oversight is required because of complications associated with their administration. For example, lithium has a narrow therapeutic index, and clozapine carries the risk of agranulocytosis. With clozapine, white blood cell counts have to be evaluated either biweekly or weekly, and under certain circumstances twice a week. Serum lithium evaluations have to be done periodically, but equally important, the patient must undergo frequent clinical evaluations because of the possibility of toxic reactions in the intervals between blood tests.

To make matters more complicated, recent advances in the understanding of pharmacodynamic interactions involving the cytochrome enzyme system have added an exponential complexity to the practice of psychopharmacology. We now appreciate that not only can psychiatric medications influence the metabolism of other concurrently administered psychiatric medications, but also such influences can be exerted by a variety of other drugs. This increase in interactive complexity is another factor that will necessitate closer psychopharmacological surveillance and contribute to more frequent use of medication management codes.

The fact is that more and more chronically mentally ill patients are being actively treated with new medications for other chronic medical illnesses. I recently admitted a new patient to a board-and-care home who was on 18 different medications for the treatment of asthma, chronic obstructive pulmonary disease, hypertension, arthritis, and diabetes.

There is an increasing capitulation, rightly or wrongly, to the philosophy that psychopharmacological approaches may be preferable to the customary psychotherapeutic approaches of the past. This changing view has several sources. One is managed care, in which medication management is regarded as more efficient than psychotherapy. Also, because of the high cost of hospitalization, increased pressure for early discharge is exerted by Medicare and other third-party payers. Thus more and more severely disturbed psychotic patients are being released into the community after only brief stays in the hospital. Such patients are most effectively treated psychopharmacologically, but they may require more visits, which is another reason why code 90862 services are being used—and will continue to be used—more frequently.

One advantage of more frequent visits is that, besides addressing psychopharmacological problems, they enhance the psychotherapeutic relationship. The importance of this relationship is sometimes overlooked in the treatment of severely mentally ill persons and may be instrumental in preventing future decompensation.

A high volume of 90862 services may be generated in several different types of settings besides traditional office practices and hospitals. These settings include specialized practices that render on-site services to residents of board-and-care homes, partial hospitalization settings, locked facilities, specialized clinics such as lithium or clozapine clinics, and managed care settings, especially those in which clinicians split responsibilities, with a nonmedical clinician providing psychotherapy and a psychiatrist providing pharmacotherapy.

High-volume practices immediately generate suspicion because of an aberrant procedure code profile. A physician in such a practice may be automatically targeted by a computer for a Medicare audit simply because the profile is aberrant, and for no other reason. What looks like poor quality of care may only appear that way, because high-volume providers with specialized foci may be able to work much more rapidly than generalists, particularly when the high-volume provider sees the same patients recurrently. Such psychiatrists can work very efficiently if they know the patients and their problems and have organized the patient's drug history on the basis of their own progress notes as well as information from hospital summaries.

Quality of care should not be inferred from numbers alone but can be inferred from several other factors such as usefulness of the progress notes (a quality not necessarily related to length), patients' complaints, evaluations of coworkers and house staff, and, under certain circumstances, patients' criminality rates, rehospitalization rates, and suicide rates. (Approximately 10 percent of deaths among patients with schizophrenia are suicides, according to some estimates).

Unfortunately, the increased use of code 90862 has been perceived by some Medicare administrators as abusive or fraudulent, and recent efforts have been made to control its use by mandating time intervals, such as requiring 20 minutes per session; by lowering allowed charges at a time when psychopharmacology is more complex and more effective than ever before; and by attempting to implement a two-tier system in which 90862 would be used only for patients who are "ineffectively controlled" and M0064 would be used for patients who are "stable from a psychopharmacological point of view" (4).

While such patients may be easy to identify at the extremes of a normal distribution, most patients fall within the extremes, and the ambiguity creates an inevitable dispute between psychiatrists and Medicare auditors about the use of the correct procedure code. "Ineffective control" appears to be a concept borrowed from the treatment of hypertension or diabetes, conditions in which control can be quantitatively determined. Psychiatry lacks such precision.

These coding problems would appear to be a trivial matter except that a $10,000 penalty is imposed per occurrence if an auditor determines that a physician is guilty of upcoding, or deliberately using a code that has a higher reimbursement than the code that should be used. It is the auditor who determines whether violations have occurred as a result of "deliberate ignorance." In most audit situations, once the allegation of deliberate ignorance is made, it will be up to the physicians to prove that they were ignorant. Ignorance, unlike knowledge, is difficult to prove, and it is even more difficult to prove that such ignorance did not deliberately occur.

Such proofs will be rendered even more challenging because programmatic incentives now exist for auditors and prosecutors to recover money. These incentives are included in the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Moreover, levying the $10,000 penalty does not necessarily depend on the discovery of a physician's fraudulent intent. HIPAA expressly states that "no specific intent to defraud is required" (5). The act specifically identifies "deliberate ignorance" as a sufficient condition for the implementation of these penalties.

HIPAA requires a similar $10,000 penalty per occurrence for rendering "unnecessary medical services." In this case, the burden of proof is on physicians to show that their services were necessary. In this regard, the proposed two-tier system creates another problem in that the literature contains no clear definition of an acceptable frequency of service for patients who are "stable from a psychopharmacological point of view." In an adversarial audit, the psychiatrist may be hard put to defend a given visit as "medically necessary" for a "stable" patient when the audit is conducted in isolation from other factors. The fact is that the chronic psychotic patient is unpredictable, and when the unpredictable happens, such as suicide or arson, and no recent progress note exists, it can be a legal, ethical, economic, and professional disaster for the treating physician. Even if we did not consider the ethical implications of benign neglect, the litigious climate in which American medicine is practiced does not permit clinicians this option.

In the above instances, in which civil penalties are generated in a Medicare audit, it is the responsibility of physicians to prove their innocence, unlike criminal proceedings, in which one is presumed innocent until proven guilty. These facts are not generally appreciated by most psychiatrists or other physicians.

Psychiatrists who assume responsibility for the most difficult cases in which medication management codes are used more frequently will generate procedure code profiles that are aberrant. The profiles will automatically trigger computer-driven audit selections. Psychiatrists and other physicians selected for audits will find that in such circumstances their guilt will be presumed, and they will have to prove their innocence. Some physicians will pay a heavy price because of the draconian financial penalties associated with upcoding or rendering unnecessary medical services. These charges will frequently be difficult to defend against because of ambiguously worded procedure code descriptors or poorly understood concepts of medical necessity, especially in the proposed two-tier system of medication management.

The net effect of these policies is that psychiatrists will opt out of the outpatient treatment of severely mentally ill patients, which will result in increased hospital readmission rates and higher overall Medicare costs. It behooves the national Medicare administration and regional carriers to work with respected and experienced practitioners in this field to obtain a more realistic and clinically appropriate concept of how medication management should be implemented.

Dr. Fleishman is a staff psychiatrist at St. Mary's Medical Center at Golden Gate Park, 2305 Van Ness Avenue, Suite F, San Francisco, California 94109 (e-mail, [email protected]). Steven S. Sharfstein, M.D., is editor of this column.