Comparison of Discharge Rates and Drug Costs for Patients With Schizophrenia Treated With Risperidone or Olanzapine

The database used for our analysis was designed to prospectively evaluate use and dosage of atypical antipsychotics in Maryland state psychiatric inpatient facilities. All patients who started therapy with risperidone or olanzapine in six facilities were included. To classify appropriate diagnoses, chart reviews were performed by two members of the research team to verify the most recent diagnoses according to DSM-IV criteria.

The period for analysis was July 1, 1997, through June 30, 1998. Risperidone has been used in the state system since April 1994 and olanzapine since October 1996. Medication costs were determined from the actual prescribed regimens for each patient. Prices are negotiated by a large pharmaceutical buying group that represented 31 states and 1,850 facilities during the study period. These prices are believed to be more representative of actual expenditures by institutions than is the average wholesale price.

Student's t tests were used to analyze mean dosages, costs, and demographic variables between the two groups. To determine the discharge rate 30 days after the start of treatment, time to discharge was measured by product-limit (Kaplan-Meier) survival analysis. Statistical significance was measured by the log-rank chi square test. A log-normal regression model was used to determine variables highly predictive of discharge, such as age, sex, race, number of previous hospitalizations, and dosage. All tests were two-tailed, and the significance level was set at .05 or less.

Between July 1, 1997, and June 30, 1998, a total of 398 patients started treatment with olanzapine and 391 patients started treatment with risperidone. A total of 377 of the patients treated with olanzapine and 367 of those treated with risperidone were evaluable for race comparisons—that is, there was sufficient information in the patient's chart to enable such comparisons. Of these, 239, or 63 percent, of those taking olanzapine and 206, or 56 percent, of those taking risperidone were Caucasian (χ²=4.08, df=1, p=.043). The mean± SD age of the patients in the olanzapine group was 42.96±13.03 years, compared with 41.79±13.32 years for those in the risperidone group; the difference was not significant. A total of 393 patients taking olanzapine and 379 patients taking risperidone were evaluable for gender comparisons. Of these, 226, or 58 percent, of those taking olanzapine and 243, or 64 percent, of those taking risperidone were men (χ²=3.54, df=1, p=.06).

For comparisons of patients in each drug group who had had a previous trial with the other drug, 193 patients who started treatment with olanzapine and 175 patients who started treatment with risperidone were evaluable—that is, information on previous trials was available for these patients. Of these patients, 42 in the olanzapine group, or 22 percent, had had a previous trial of risperidone and 30 in the risperidone group, or 17 percent, had had prior olanzapine treatment. The difference between the groups was not significant. The mean number of previous hospitalizations since the availability of atypical antipsychotics was 1.25±.50 for the patients in the olanzapine group and 1.21±.51 for the patients in the risperidone group. The mean dosages were 4.8±2.7 mg a day for risperidone and 17.5±7.3 mg a day for olanzapine. The mean daily cost of the drugs was $6.42 for risperidone and $12.29 for olanzapine (t=18.57, df=701, p<.001).

Discharge rates 30 days after the start of treatment were significantly different between the patients treated with risperidone and those treated with olanzapine (log-rank χ²=10.82, df=1, p=.001) (Figure 1). Discharge rates were 45 percent (95 percent confidence interval=39 to 52 percent) in the risperidone group and 32 percent (CI=26 to 38 percent) in the olanzapine group 30 days after the initiation of treatment. Younger age was predictive of discharge in both the olanzapine group (χ²=6.47, df=1, p=.011) and the risperidone group (χ²=29.61, df=1, p<.001). Lower dosages of risperidone were associated with a greater likelihood of discharge (χ²=10.98, df=1, p<.001). No significant association between discharge and dosage was seen for olanzapine. There was no association between discharge and race or sex for either medication.

We found that risperidone was associated with higher 30-day discharge rates than olanzapine over a one-year period. Age, sex, race, number of previous hospitalizations, and whether the patient had had previous trials with atypical drugs were largely comparable between groups and did not account for the difference in discharge rates by regression analysis. Additionally, the average drug costs of risperidone were about half those of olanzapine, a cost difference of more than $2,100 per patient annually.

Other researchers have found similar medication cost savings and favorable outcomes with risperidone compared with olanzapine. Nasrallah and colleagues (8) found similar lengths of stay and response rates with the two drugs, but lower drug costs with risperidone. Likewise, Procyshyn and Zerjav (9) found that daily drug costs were lower with risperidone, whereas discharge and response rates were higher.

The differences in discharge rates in our study could have been explained if one group had consisted of more chronic patients than the other; however, this was not the case. Although measures of symptom severity were not available, proxy measures of chronicity, such as number of previous hospitalizations and whether the patient had had previous trials of atypical agents, were similar between the two groups. Younger age was associated with better outcomes in both groups, yet there were no significant differences in age, which would have contributed to differences in discharge rates.

Lower risperidone dosages were associated with higher discharge rates, whereas olanzapine dosage was not associated with outcome. This finding was not unexpected—there is much evidence that lower dosages of risperidone are more effective than higher dosages (7,10). This phenomenon does not occur with olanzapine, for which dosages have gravitated upward since the drug's introduction. Moreover, no dosage differences among inpatient facilities contributed to the differences in discharge rates. Because the most effective dosage of olanzapine has yet to be determined, drug trials may be more difficult to optimize with olanzapine than with risperidone.

Although this study lacked some of the benefits of a prospective, double-blind clinical trial, the naturalistic design had its own advantages. The results of the study reflect the real-world use of the two drugs. Practitioners prescribed either risperidone or olanzapine and discharged patients on the basis of their clinical judgment. These practitioners were unaware of the study and were not subject to inherent biases. However, other costs—such as outpatient services, emergency department visits, concomitant therapies, and rehospitalization—were not addressed, which was a limitation of our study.

Our data suggest that risperidone offers a higher discharge rate at a lower drug cost compared with olanzapine among inpatients with schizophrenia. Olanzapine's higher cost could be justified if it was associated with higher discharge rates, but this was not the case in our population. Nevertheless, other studies are needed to verify our findings, to compare the naturalistic use of these medications in first-break patients or outpatients, and to rule out regional differences. Additionally, rigorous pharmacoeconomic analyses should be undertaken to compare other contributing factors to overall costs of care.

The authors thank Carol Younkins, R.Ph., for her contributions to the collection and maintenance of the data. This study was supported in part by the Theodore and Vada Stanley Foundation and National Institute of Mental Health Interim Research Center grant MH-40279.

Dr. Kelly, Ms. Yu, and Dr. Conley are affiliated with the Maryland Psychiatric Research Center of the University of Maryland School of Medicine in Baltimore. Send correspondence to Dr. Kelly at the University of Maryland School of Medicine, Box 21247, Baltimore, Maryland 21228 (e-mail, [email protected] land.edu). Dr. Nelson is affiliated with Perry Point Veterans Affairs Medical Center in Perrysville, Maryland. Dr. Love is affiliated with the University of Maryland School of Pharmacy in Baltimore. Earlier versions of this paper were presented at the International Congress for Schizophrenia Research held April 17-21, 1999, in Santa Fe, New Mexico; the annual meeting of the American Psychiatric Association held May 15-20, 1999, in Washington, D.C., and the annual meeting of the New Clinical Drug Evaluation Unit held June 1-4, 1999, in Boca Raton, Florida.