Among 664 patients enrolled, 13 (1.9%) did not start the medications to which they were initially randomized and hence were excluded from the present analysis. Of 651 patients with evaluable data, 191 (29.3%) switched antipsychotics, while the remaining 460 (70.7%) continued with their initial medications (Figure ). A total of 155 individuals who discontinued the study early (prior to completing 1-year) without switching to a different medication were included in the continuer group. When excluding patients with acute-care service use at baseline, 156 patients switched antipsychotics, and 376 continued with their original medications. On average, the duration of treatment with antipsychotic medications before switching was 100 days, while the average duration of treatment after switching was 265 days. Of the switchers, 70 switched from conventional antipsychotics to olanzapine, 41 from risperidone to olanzapine, 15 from olanzapine to conventional, 11 from olanzapine to risperidone, with the rest being other combinations. Following the medication switch, 129 patients completed the study on their new medication and 12 required an additional medication switch during the study. Unfortunately, reasons for medication switching were not obtained for 28.2% of switchers, with the rest evenly divided between patient request (26.3%), lack of efficacy (23.1%), and adverse events (22.4%).
Clinical and demographic characteristics at baseline were fairly similar between switchers and continuers in the analysis sample (Table ), although a significantly lower proportion of switchers (vs. continuers) were male and the switchers had a lower mean PANSS total score. Switchers also tended to be prescribed lower mean dosages than continuers. For instance, among patients randomized to risperidone, the mean starting doses (2.5 mg/day for switchers vs. 3.0 mg/day for continuers), the modal (4.2 vs. 4.9), and maximum (5.3 vs. 6.3) doses were numerically lower for switchers compared to continuers. For patients randomized to olanzapine, the mean starting doses were similar (8.0 mg/day for switchers vs. 8.2 mg/day for continuers), but the modal (10.0 vs. 13.0) and maximum (13.3 vs. 17.8) doses were lower for switchers.
Patient baseline characteristics for switchers and continuers of the initial antipsychotic - analysis sample
Individuals who switched antipsychotics were significantly more likely to use acute-care services compared with their counterparts continuing with their initial medications. For example, as shown in Table , the proportion of patients using any acute-care service and the rate of admission to facilities providing such services were higher among switchers compared with continuers (p < .001). Differences in acute-care service use were driven primarily by differences in hospitalizations and crisis service use as opposed to partial hospitalizations. The proportion of patients hospitalized and the rate of hospitalizations were statistically significantly higher in those who switched. Similar differences were found for crisis service use. There were no statistically significant differences in partial hospitalizations between the two groups.
Annual acute-care service usea for switchers and continuers on the initial antipsychotic
As shown in Figure , the risk of new acute-care service use (rate of admissions) was significantly higher among individuals switching antipsychotics (vs continuers) for any acute-care service (p < .001), hospitalization (p = .013), and crisis services (p = .011) but not partial hospitalization. Not only did switchers have higher risks of using new acute-care services; they also used such services significantly earlier than continuers (p = .006; Figure ).
Figure 2 Risk of new acute-care service use (admissions).* *The percentage of days with new service use is calculated by the number of days with a new service (hospital admission, new partial hospitalization treatment, or ER admission) divided by the number of (more ...)
Time to first acute-care service event.
On average, the total annual health care cost of treatment per patient was $14,954 among patients who switched antipsychotics compared with $11,918 for those who continued with their initial treatment (p = .107), translating to an excess of just over $3,000 (or 25%, Figure ). More than half of this cost difference was attributed to higher expenses for acute- or intensive-care services.
Average 1-year total health care costs and cost components for switchers and continuers.
Results of the sensitivity analysis which included only patients who completed the 1-year study were consistent with the above findings. In addition, a total of 68 patients in the analysis dataset had been randomized to the medication they had been taking immediately prior to the trial. The majority of these patients were previously treated with (and were randomized to) conventional antipsychotics (N = 43), with smaller numbers for olanzapine (n = 14) and risperidone (n = 11). The sensitivity analysis excluding such patients did not change the direction nor the statistical significance of any of the findings. For instance, total 1-year costs for switchers and non-switchers after excluding these patients were (N = $12,157 vs. $14,729 for non-switchers vs. switchers). Patients randomized to the same medication that was being taken prior to the trial were found to have a numerically lower, but not significantly different, risk of switching/discontinuation of the randomized medication (HR = 0.89, p = .506).