Polypharmacy in psychiatry is becoming the norm rather than the exception. For antipsychotic polypharmacy, several justifications are offered: improving efficacy in refractory illness, achieving rapid control of symptoms, minimizing adverse effects of an initial drug, and supplementing an initial drug for which maximum dosages have been reached (1,2).
However, many instances of poly-pharmacy appear less justifiable, such as when medications are not reviewed on a regular basis, when a clinician does not want to "rock the boat" if a patient is doing well, when a clinician neglects to stop one drug as another is started, when symptoms are treated in isolation, in response to pressure from other staff, when a clinician is reluctant to prescribe clozapine, and when low doctor-patient ratios promote medicating as an alternative to intensive interpersonal interventions (3).
To the extent that increased prevalence of polypharmacy is a result of less justified practices, it may be feasible to decrease polypharmacy simply through systematic attention to prescribing practices. Several interventions across a range of clinical settings have been used to alter polypharmacy by psychiatrists, including drug consultations with a pharmacologist (4), educational articles and videos (5), peer reviews (6,7), and ongoing computer monitoring and feedback (8,9). Such interventions, introduced and studied primarily in the 1970s, have led to decreased polypharmacy of varying degrees.
Polypharmacy can be especially prominent in state institutions, where patients who have chronic and refractory mental illnesses and multiple comorbid conditions are treated. As part of a general performance improvement initiative to reduce polypharmacy in a state psychiatric hospital in a Northeastern state, we undertook a low-intensity intervention to heighten awareness and reduce unnecessary antipsychotic medication combinations. We assessed whether polypharmacy would decrease when it was called to the attention of each treating psychiatrist by the chief of psychiatry and when each physician was provided with data summarizing his or her prescribing practice compared with those of colleagues. We predicted that antipsychotic polypharmacy would decrease among all psychiatrists and especially among higher utilizers of polypharmacy who might attempt to move closer to the standard practice of peers.
The initiative was undertaken in a 570-bed hospital that served primarily local counties and that ran at or near capacity, with many admissions lasting several months to several years. Baseline prescribing practices were summarized for each psychiatrist on the basis of physicians' order forms for May 2001. The number of patients receiving one antipsychotic and the number receiving two or more were determined, as well as the number for whom psychotropic medications other than antipsychotic agents were prescribed. During the ensuing months, case discussions, consultations with psychiatrists, and psychopharmacology seminars were conducted. Topics addressed included general psychopharmacologic principles and practices, specific treatment of schizophrenia and refractory schizophrenia, augmentation strategies, switching antipsychotics, drug-drug interactions, and the pros and cons of polypharmacy.
However, data collected in November 2001 showed that the prevalence of antipsychotic polypharmacy had not declined, which suggested that the measures used resulted in stable assessments of standard practice and that a standard educational program had little impact on prescribing behavior. An additional intervention was then implemented by the new chief of psychiatry, who met individually with each psychiatrist to compare his or her performance data with data of anonymous peers. The chief also asked all psychiatrists to decrease their polypharmacy by at least 10 percent over the ensuing six months. The psychiatrists were assured that the results of this performance improvement initiative would not influence their performance evaluations. Follow-up reviews of physicians' orders were conducted in August 2002, and the data were compared with data collected in November 2001. Chi square tests and Pearson r statistics (all two-tailed) were used to conduct the analyses. The software used was SPSS version 12.0.1. Institutional review board approval was obtained for the analytic studies.
Prescribing practices of the 14 psychiatrists who were on the staff in both November 2001 and August 2002 were evaluated. Of the 492 patients treated by the 14 psychiatrists during November 2001, 28 (5.7 percent) received no antipsychotics; of 408 treated in August 2002, 22 (5.4 percent) received no antipsychotics. No psychiatrist prescribed more than three antipsychotics to a given patient during either of the months examined. In November 2001 the most commonly prescribed antipsychotic agents included olanzapine (for 159 patients, or 32 percent), risperidone (for 133 patients, or 27 percent), and haloperidol (for 122 patients, or 25 percent, 48 of whom received depot medication). In August 2002 the most commonly prescribed antipsychotic medications were olanzapine (129 patients, or 32 percent), risperidone (110 patients, or 27 percent), and haloperidol (99 patients, or 24 percent, 27 of whom received depot medication).
First-generation antipsychotics were prescribed for 261 patients (53 percent) in November 2001 and for 187 patients (46 percent) in August 2002, a decrease that was not statistically significant. The proportion of patients for whom second-generation agents were prescribed (75 percent) did not change.
Overall, antipsychotics were prescribed to 464 patients in November 2001 and to 386 patients in August 2002. Polypharmacy decreased significantly during this period. At baseline 197 patients (42 percent of those receiving any antipsychotic) received two or more antipsychotics, compared with 127 patients (31 percent) at follow-up (χ2=8.2, df=1, p<.004). It should be noted that the number of patients who received three antipsychotics in either month was small and did not change significantly (eight patients, or 1.6 percent, at baseline and five patients, or 1.2 percent, at follow-up).
The use of mixed polypharmacy—that is, use of at least one second-generation agent plus at least one first-generation agent—also decreased significantly. In November 2001 mixed polypharmacy was used with 165 patients (36 percent of patients receiving any antipsychotic), compared with 106 patients (27 percent) in August 2002 (χ2=6.4, df=1, p<.02). Combinations of first-generation agents were rare (nine patients, or 1.9 percent, at baseline and nine patients, or 2.3 percent, at follow-up), as were combinations of second-generation agents (31 patients, or 6.7 percent, and 17 patients, or 4.4 percent, respectively). Of the 14 psychiatrists, 13 (93 percent) decreased their use of polypharmacy, and eight (57percent) met the goal of decreasing their polypharmacy by at least 10 percent.
To assess whether higher utilizers of antipsychotic polypharmacy at baseline showed greater change than the other psychiatrists after the intervention, Pearson correlations between the percentage of each physician's patients who were receiving polypharmacy in November 2001 and the percentage reduction in polypharmacy were calculated. No significant association was found. Among the 14 psychiatrists who used antipsychotic polypharmacy in November 2001, the upper 50 percent of these prescribers continued to have a higher percentage of patients receiving polypharmacy in August 2002 (t=2.64, df=12, p<.025). The mean percentage reduction in polypharmacy among the upper 50 percent was not significantly different from the mean percentage reduction among the lower 50 percent of users of polypharmacy (a 24 percent reduction and a 19 percent reduction, respectively).
Less antipsychotic polypharmacy was not associated with more use of other psychotropic medications. The mean±SD number of nonantipsychotic psychotropics for the 14 psychiatrists was 1.9±.3 per patient in November 2001 and 1.8±.4 per patient in August 2002. Less polypharmacy was not correlated with altered rates of other (nonantipsychotic) psychotropic prescriptions (r=-.24, p=ns).
No significant reduction in poly-pharmacy was seen among the patients who received depot medication. At the time of the study, only the first-generation agents haloperidol and fluphenazine were available, which led to polypharmacy when depot medication was indicated for patients who were receiving second-generation agents. In November 2001 the rate of polypharmacy among patients who received depot medications was 63 percent, compared with a rate of 38 percent among those who did not receive depot medications. The corresponding rates for August 2002 were 61 percent and 30 percent. The reduction in polypharmacy was significant among patients who did not receive depot medication (χ2=5.9, df=1, p<.02) but not among those who did. As anticipated, in most cases of polypharmacy among patients who received depot medication, the additional antipsychotic was a second-generation agent; in both November 2001 and August 2002 only two patients who received a depot antipsychotic received another first-generation agent.
We found that a performance improvement initiative that used a low-intensity individualized supervisory intervention resulted in a significant, albeit moderate, reduction in antipsychotic polypharmacy—from 42 percent of patients treated with antipsychotics to 31 percent nine months later. An earlier initiative that involved consultation and education showed no effect. No other changes or interventions occurred at the hospital between November 2001 and August 2002 that would have accounted for the reduction; for example, there were no formulary changes or restrictions on antipsychotic prescription and no changes in nursing education or prompts by pharmacists.
The performance improvement strategy was noteworthy, because it did not include sanctions or compromise anonymity but relied solely on the chief of psychiatry's personal expression of expectations along with provision of individual data for comparison. The intervention also did not attempt to proscribe specific medication combinations or to link prescription practices to specific clinical parameters. Instead, it offered a generic perform-ance expectation that all psychiatrists, whether they were high or low baseline users of polypharmacy, would reduce polypharmacy by 10 percent. Thus physician authority and responsibility for change was preserved. The provision of peer data, expectation of follow-up, and motivation to achieve positive recognition from the newly appointed chief of psychiatry are likely to have contributed to behavioral change. Further study is required to determine whether both the personal directive from the chief of psychiatry and the benchmarking comparison with peers were necessary or whether one of these would have been sufficient.
The prescription of psychotropic medications other than antipsychotics did not increase during the study period, which suggests that physicians were not simply adjusting the classes of medications to accommodate the mandate of the director. Also, the proportion of patients prescribed polypharmacy among those who received depot medications did not change, which is consistent with physicians' examining the rationale for polypharmacy rather than making arbitrary changes in their prescribing behavior. Overall, the project was not associated with reported adverse events, suggesting that polypharmacy may be decreased without negative impact, as previously described (10).
The study had some limitations. Although the hospital's admission and discharge criteria did not change dramatically during the nine-month study, the possibility that the hospitalwide reduction in polypharmacy was related to changes in the patient population cannot be ruled out. Because patient data were aggregated, it also cannot be determined whether changes in physicians' practice were restricted to newly admitted patients. Patients who were being cross-tapered during either of the survey months and those who were transferred to new psychiatrists close to the survey months (which may have left insufficient time for a review of medications) would have been counted among the patients who received more than one antipsychotic, leading to an exaggerated rate of polypharmacy. We also did not review changes in dosages or in PRN medications that might have occurred in response to less use of polypharmacy. In addition, information on specific clinical outcomes was not available.
A performance improvement initiative that sought to reduce antipsychotic polypharmacy in a state institution by increasing physicians' awareness of this practice, expressing an expectation that a reduction would occur, and providing individualized data to psychiatrists resulted in less use of polypharmacy by most staff psychiatrists. Overall, the findings suggest that such initiatives, which involve the focused attention of leadership but only a modest investment of effort, can result in significant change in a state hospital setting. It will be important to examine prescribing practices at the hospital at a future date to determine whether the altered behavior is maintained either as a direct effect on individual physicians or through a more global change in institutional culture.
Dr. Patrick is clinical associate professor of psychiatry and Dr. Schleifer is professor of psychiatry at the University of Medicine and Dentistry of New Jersey (UMDNJ)-New Jersey Medical School in Newark. Dr. Nurenberg is lecturer in the department of psychiatry at Columbia University College of Physicians and Surgeons in New York City. Dr. Gill is professor and chair in the department of psychiatric rehabilitation and behavioral health care in the School of Health Related Professions at UMDNJ. Send correspondence to Dr. Patrick at 4873 S.E. 7th Avenue, Ocala, Florida 34480 (e-mail, firstname.lastname@example.org). William M. Glazer, M.D., is editor of this column.