Comparison of hospitalization risk before and after changing from risperidone long‐acting injection to another long‐acting injection or oral antipsychotic in patients with schizophrenia: Mirror‐image study

Abstract

In a previous meta-analysis of mirror-image studies, long-acting injection (LAI) significantly reduced the length of hospital stay compared to the reduction by an oral antipsychotic.1 On the other hand, to our knowledge, reports that have evaluated the comparison between hospitalization risk before and after risperidone LAI (RLAI) discontinuation are scarce. Therefore, we examined the comparison of hospitalization risk for the 6 months before and after changing from RLAI to another long-acting injection or oral antipsychotic in a clinical setting by using a mirror-image analysis. This was a mirror-image design involving patients with schizophrenia basis who were given RLAI between July 2009 and November 2015 at the psychiatry department of Fukui Kinen Hospital. This study was approved by the ethics committee of Fukui Kinen Hospital. The primary outcome measures were the rates of hospitalization within 6 months before and after RLAI discontinuation. Testing of independence was done using Fisher's exact test. The significance level was P < 0.05. A total of 88 patients discontinued RLAI after having continued it for 6 months or more. However, 31 patients who were hospitalized throughout the investigation period were excluded. Therefore, we analyzed 57 patients (male/female, 28/29, 49.1%/50.9%). The mean ± SD age was 41.6 ± 14.8 years and duration of illness was 16.0 ± 11.1 years. The mean Global Assessment of Functioning (GAF) score was 45.3 ± 11.0. The mean number of hospitalizations before the study enrollment was 0.4 ± 0.7. The hospitalization rate during RLAI continuation was 29.8% (17/57). Within 6 months of discontinuing RLAI, three patients discontinued their hospital visits, and two patients died. After discontinuing RLAI, 25 patients switched to another LAI, and 27 patients switched to oral antipsychotics. Of these, three patients restarted RLAI within 6 months of switching to oral antipsychotics. The rates of hospitalization within 6 months of switching to another LAI or oral antipsychotic following RLAI discontinuation were 28% (7/25) and 25% (6/24), and no significant difference was seen in hospitalization risk for the 6 months after RLAI discontinuation (risk ratio = 0.89; 95% confidence interval, 0.35–2.28; P = 0.82). The limitations of a mirror-image study are the expectancy effect, natural course of disease, and the possibility of influence of time as a bias. However, the mirror-image study performed is considered to reflect the true state of treatment given in clinical practice better than a randomized controlled trial, which is a gold standard in clinical studies for comparison of the hospitalization risk for the period of 6 months after RLAI discontinuation in the same patients. In the present study, no significant difference was seen in hospitalization risk for the 6 months after RLAI discontinuation; this finding was similar to the findings obtained in previous studies in which a mirror-image analysis was performed.2 Therefore, no difference was found between the hospitalization risk before RLAI discontinuation and that after RLAI discontinuation following switching to another LAI or oral antipsychotic. Dr Suzuki has received honoraria from Janssen, Otsuka, Dainippon Sumitomo, Shionogi, and Yoshitomiyakuhin. Dr Hibino has received honoraria from Janssen, Lilly, Otsuka, and Glaxo-SmithKline. Dr Inoue has received honoraria from Eisai and Novartis. Dr Takaya has received payment from Otsuka and Yoshitomiyakuhin.