Antipsychotic Formulation and One-Year Rehospitalization of Schizophrenia Patients: A Population-Based Cohort Study

A population-based cohort study was conducted by analyzing data retrieved from the Taiwan National Health Research Institutes database (22.60 million of Taiwan’s 22.96 million people are enrolled in the insurance program; w3.nhri.org.tw/nhird/en/index.htm). In this cohort study, patients had a diagnosis of schizophrenia or schizoaffective disorders (ICD-9-CM code 295). Records of patients with schizophrenia who were being admitted frequently were retrieved for analysis. Frequent admission was defined as hospitalization for schizophrenia treatment at least twice in an acute psychiatric ward within two consecutive years. Patients were followed from January 1, 2003, to December 31, 2008. In total, 14,610 patients with schizophrenia were enrolled in the study cohort.

The day of discharge from a second hospitalization was defined as the index day. Patients were divided into two groups on the basis of the formulation of the antipsychotics used for treatment. This included the oral-form antipsychotics group and the long-acting injectable antipsychotics group.

The group receiving treatment via injection was further divided according to medication received—thus risperidone, haloperidol, and flupenthixol subgroups. Patients were assigned to the injection group if long-acting injectable antipsychotics had been administered after the index day. If the cumulative dosage of long-acting injectable risperidone, haloperidol, or flupenthixol was, respectively, <300 mg, <300 mg, or <60 mg in 12 months, the patients were excluded from our study. The group receiving oral medication was divided into those receiving risperidone, a second-generation antipsychotic; some other second-generation antipsychotic; or a first-generation antipsychotic. If patients had received a particular type of oral antipsychotic for at least 50% of their treatment (determined by days of treatment), they were assigned to the corresponding antipsychotic treatment subgroup.

The specific aim of this study was to compare the rehospitalization rate of frequently admitted patients among different treatment subgroups within one year after the index discharge. A readmission within 14 days after a previous discharge with the same principal diagnosis was defined as the same (one) hospitalization episode. The patients were divided into four 20-year age groups. We also measured the impact of additional outpatient care by dividing these patients into two groups solely on the basis of clinic outpatient care or additional outpatient care.

Odds ratios and 95% confidence intervals for the associations between the incidence of rehospitalization and covariates were reported, and p<.05 was considered statistically significant. All analyses were performed with SAS, version 9.01.

We analyzed data for 14,610 patients with schizophrenia. A majority of patients were ages 20–39 (52.0%, N=7,602) or 40–59 (37.9%, N=5,535). Male gender was mildly predominant in this study (of 14,567 patients: male, 54.8%, N=7,988; female, 45.2%, N=6,579). More patients lived in urban areas than in suburban and rural areas. Patients in northern and southern regions of Taiwan accounted for 75.5% of the study population. [A table summarizing clinical characteristics of the patients is available online as a data supplement to this report.] We excluded data for 4,058 patients because their antipsychotic dosage did not reach the criteria we defined, and a final 10,552 participants were categorized according to their prescriptions. In addition, the data showed that most patients received treatment with oral antipsychotics rather than long-acting injectable antipsychotics (oral group, 90.5%; injectable group, 9.5%). More patients in the oral group were being treated with second-generation antipsychotics (second-generation antipsychotics, 59.7%, N=6,302; first-generation antipsychotics, 30.8%, N=3,251). In the group receiving antipsychotics by injection, more patients were being treated with flupenthixol (flupenthixol, 5.3%, N=556; haloperidol, 2.2%, N=236; risperidone, 2.0%, N=207). [A table summarizing usage of antipsychotics among the patients is available online in the data supplement.]

The rehospitalization rates in the one-year follow-up after discharge showed that 27.3% of the study participants in both the oral medication and injected medication groups experienced a third psychiatric hospitalization within that period (Table 1). In the group receiving oral medication, patients with risperidone treatment had rehospitalization rates similar to those who received first-generation antipsychotics (risperidone, 26.3%; first-generation medication, 26.5%, but patients taking risperidone fared significantly better than those who received other second-generation antipsychotics (28.8%).

In the long-acting injection group, patients with long-acting injectable haloperidol treatment had a lower rehospitalization rate than those with long-acting injectable risperidone or flupenthixol treatment (haloperidol, 22.5%; risperidone, 27.1%; and flupenthixol, 29.5%). Patients receiving long-acting injectable haloperidol treatment had the lowest rehospitalization rate. Patients receiving long-acting injectable flupenthixol had the highest likelihood of rehospitalization (odds ratio [OR]=1.44), and patients taking second-generation antipsychotics other than risperidone had the second highest likelihood of rehospitalization (OR=1.39). Patients who were admitted a third time were readmitted within an average 200 days of the index discharge.

Significant effects of age, region, and insurance payment were noted (p<.05), but gender, comorbid general medical condition, and form of outpatient care (additional or standard services) showed no significant effect on the risk of relapse. We found that the risk of rehospitalization decreased with age, and patients with lower insurance payments were more vulnerable to rehospitalization. Patients with additional outpatient care appeared to be less likely to be readmitted, but this finding was nonsignificant. Patients with a comorbid general medical condition had a slightly but nonsignificantly increased rehospitalization rate. [A table summarizing Cox regression model results for rehospitalized patients with schizophrenia is available in the data supplement.]

To the best of our knowledge, this is the first study to use data retrieved from a national database in an Asian country to analyze over a one-year follow-up period the rehospitalization rates of frequently admitted patients with schizophrenia. This cohort study showed that, with respect to rehospitalization rate, long-acting injectable antipsychotics were not superior to oral antipsychotics in treating these patients. Our findings are in accordance with the results of a recent U.S. study of unstable patients with schizophrenia (6). Similarly, a previous study of stable patients with schizophrenia found no difference between injectable and oral antipsychotics in regard to rehospitalization rate (7). However, the issue of whether long-acting injectable or oral antipsychotics are more effective in preventing rehospitalization remains controversial (6–8). No significant differences in the rehospitalization rates between the groups receiving oral versus injectable medication in our study could be attributed to long-acting injectable flupenthixol, which had a higher rehospitalization rate. One review suggested that long-acting injectable flupenthixol was inferior to other long-acting injectable antipsychotics for the medium-term outcome of six to 12 months (9).

Although there were no significant differences in rehospitalization rates for patients receiving oral versus injected medication in this study, further analysis of the group receiving antipsychotics by injection indicated that haloperidol had a lower rehospitalization rate than occurred with the other two long-acting injectable antipsychotics (haloperidol, 22.5%; risperidone, 27.1%; and flupenthixol, 29.5%). A previous study showed that patients receiving long-acting injectable risperidone may have longer hospital stays than patients receiving long-acting injectable haloperidol (10). One report assumed that treatment with long-acting injectable haloperidol would result in better adherence than would occur with long-acting injectable risperidone because of haloperidol’s longer half-life (11). However, a previous study showed better effectiveness with long-acting injectable risperidone than with long-acting injectable haloperidol, which conflicts with the results of this study (12). The comparative efficacy between long-acting injectable antipsychotics remains in debate.

Analyzing the risk factors of rehospitalization, we found that rehospitalization rates decreased with patients’ age. Younger age was found to be a predictor of relapse in this study. Two previous studies had similar results (13,14).

Our study showed that patients with lower insurance payments lower had a higher rehospitalization rate. One large population study also found that patients with better economic status may have a lower rehospitalization rate (14). There are two reasons that may explain this phenomenon. First, better socioeconomic status may have a greater protective effect against rehospitalization, and second, patients with higher insurance payments may have less functional impairment and thus a tendency to have a lower rehospitalization rate.

In this study, patients in southern and eastern Taiwan had higher rehospitalization rates. Again, there are two possible explanations. First, there are more psychiatric beds to be filled in southern and eastern Taiwan, and after considering the occupancy rate, doctors may recommend admission for patients seeking voluntary commitment. Second, people living in these regions of Taiwan are thought to be generally more agreeable to doctors’ recommendations concerning hospital admittance.

Several limitations of our study should be noted. First, given the observational nature of our study, we were unable to demonstrate conclusively that there were no significant differences in rehospitalization rates between patients who received oral antipsychotics and those who received long-acting injectable antipsychotics. Because this study used a cohort study design, the patients were not randomly assigned to the treatment groups. However, observational studies may represent clinical practice in the real world.

Second, patients receiving one antipsychotic agent may have been poorly responsive to another. However, we did not analyze the prior usage of antipsychotics before the study period, thus raising the possibility of bias.

Third, patients recruited for this study were identified by diagnostic code in an administrative claims database, which may not always be completely accurate. However, the study patients had at least two psychiatric admissions, so the diagnosis should be reliable.

Fourth, patients were categorized to different groups on the basis of the antipsychotics prescribed, but we cannot be sure whether patients took the medication or not. Questions about adherence cannot be answered in this study. Use of the prescribed dose of antipsychotic in the injectable group may be more reliable than in the oral group.

Fifth, 4,058 patients were excluded from the analyses because their usage of antipsychotics did not reach the dose level we defined in our criteria, and they were lost to follow-up during the one-year follow-up period. These patients tended to have poor adherence, and their exclusion from the analyses may have biased the rehospitalization rate.

Sixth, there was no measurement of symptom severity in our data set, which is an important factor known to affect frequency of relapse.

In summary, this prospective cohort study showed that use of different formulations of antipsychotics for patients with schizophrenia and frequent hospitalizations did not appear to affect the likelihood of rehospitalization. However, patients treated with long-acting injectable haloperidol appeared to have a lower rate of rehospitalization in the one-year follow-up period. This study reminds psychiatrists that older antipsychotics may be just as good as newer antipsychotics.

This study is based, in part, on data obtained from the National Health Insurance Research Database. The data were provided by the Bureau of National Health Insurance of the Department of Health and managed by National Health Research Institutes (registration A101095). The statistical analysis was supported by the Biostatistics Task Force of Taichung Veterans General Hospital. The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, or National Health Research Institutes.

The authors report no competing interests.