Hospital utilization rates following antipsychotic dose reductions: implications for tardive dyskinesia

Antipsychotic medication is the mainstay of treatment for schizophrenia [1]. Agreement on the therapeutic doses of antipsychotics has been controversial; doses are determined empirically based on assessment of efficacy and tolerability for individual patients [1]. When side effects emerge, management options include discontinuation of the current antipsychotic, switching to a different drug, or lowering the dose [2‐4].

For example, tardive dyskinesia (TD), which occurs in up to 30% of patients receiving antipsychotics, is a serious side effect of antipsychotics for which dose reduction has been proposed [2‐4]. However, controlled evidence on the effect of dose reduction for the management of TD remains limited. [5]. Although novel drugs for treatment of TD are now available, management begins with consideration of antipsychotic prescribing decisions. In non-psychotic patients with TD, antipsychotic drug discontinuation may be considered and drugs may be tapered off if clinically appropriate [6]. Early diagnosis and drug cessation may facilitate remission, but evidence is limited on whether or how often TD resolves, and how long it takes to do so, after discontinuation of antipsychotics [7, 8]. In psychotic patients, maintenance antipsychotic treatment is essential to prevent deterioration and relapse [9]. Approximately 50% of patients taking antipsychotics are being treated for schizophrenia [10], which has a relapse rate of 53% within 9 months after antipsychotic discontinuation [11]. Antipsychotic discontinuation, which is associated with increased risk of violence, incarceration, hospitalization, increased healthcare resource utilization, and interruption of rehabilitation efforts, may be impractical in most patients with schizophrenia [12, 13].

Although some published evidence supports increasing doses or switching to other antipsychotics to suppress TD symptoms [14‐16], these changes may raise concerns about increasing acute extrapyramidal symptoms, limiting chances of TD reversibility, and possibly destabilizing psychiatric status. Dose reduction has also been recommended in past reviews of treatment options [5, 6, 17]. However, published meta-analyses have concluded that available data are insufficient to either support or refute treatment of TD by antipsychotic dose reduction [5, 18]. Furthermore, published studies of dose-reduction strategies to minimize the risk of developing extrapyramidal symptoms including TD, involving dose reductions of 50% or more, have been inconsistent but generally reported significant risks of psychotic exacerbation and relapse [9, 19‐23].

To further address the risks of dose reduction as a recommended treatment intervention for side effects of antipsychotic treatment, including TD, we conducted a retrospective cohort study to analyze the healthcare burden in terms of utilization of hospital-based resources resulting from ≥10% and ≥ 30% reductions in antipsychotic doses for patients with schizophrenia.

Patients with antipsychotic dose reductions showed significant increases in both all-cause and mental health–related admissions and ER visits. Absolute percentage differences in first-year hospital event rates were small (1–6%), but were statistically significant given the large sample size. The clinical meaningfulness of the observed differences should be considered by healthcare decision makers, because even a 1% difference in event rates reflected an additional 196 patients using hospital services during the study period. In addition, the shorter duration of follow-up observed for cases compared with controls suggests that dose reduction not only may increase the risk of hospital utilization, but also may result in shorter periods of treatment stability.

Hospital re-admissions for schizophrenia may be an indicator of an adverse outcome or failure of antipsychotic treatment after dose reductions. The chronic nature and severity of schizophrenia contribute to its large overall healthcare burden; direct healthcare costs for schizophrenia were estimated at 38 billion US dollars in 2013 [24]. Inpatient admissions represent one of the largest portions of the healthcare burden for schizophrenia, estimated in one study as 60–70% of direct costs [25]. The cost for schizophrenia patients who relapse is estimated to be two to five times higher than that for non-relapsed patients [26].

However, antipsychotic dose reduction in patients with schizophrenia may be necessary and appropriate for a variety of reasons; for example, patients in the dose-reduction cohort had higher mean doses to start with, as well as more cases of diabetes, which may have prompted reductions of antipsychotic doses. Although dose reduction has also been recommended to manage TD in previous practice guidelines for patients with schizophrenia who require maintenance antipsychotic treatment [5, 6], our results suggest that even relatively modest reductions in antipsychotic dosing, regardless of the rationale, may have significant adverse effects on outcomes for some patients. Therefore, prescribing decisions on maintenance antipsychotic treatment for preventing or treating TD should be carefully considered on an individualized basis for each patient. Apart from antipsychotic prescribing decisions, other measures, e.g., discontinuation or modification of anticholinergic drugs or treatment with vesicular monoamine transporter-2 inhibitors, may be considered in the management of TD.

The current study also attempted to evaluate antipsychotic dose reduction with respect to the diagnosis of TD and found no significant difference between stable dosing and dose reduction groups in either the number of new claims for TD during the study period based solely on the ICD-9 code of 333.85 and the ICD-10 code of G24.01 or in the persistence of a TD diagnosis following dose reduction. However, the duration of follow-up was significantly shorter in the dose-reduction group, and there were too few claims for TD in the databases compared with an expected prevalence of up to 30% in some previous studies [27]. This severely limits any interpretation of the relationship between dose reduction and the treatment of TD that was present at baseline, as well as the risk of new cases of TD developing in the total patient population. We suspect that TD is vastly underreported in claims databases [28], possibly due to complacency or a lack of clinical awareness, screening, approved treatments, standardization of the diagnosis or clarity regarding ICD codes. This study therefore highlights the need to document TD in a more systematic fashion. In addition, further study is needed to evaluate whether antipsychotic dose reduction is an effective treatment strategy for established cases of TD.

Strengths of the current study include the use of large patient cohorts with a 2-year follow-up period. In addition, claims data were analyzed from multiple states, with closely matched case and control groups. Study limitations include the possibility that the dose-reduction group had more hospitalizations because it represented a more-severe or chronically ill population compared with controls as suggested by the slightly longer duration of underlying schizophrenia among the case patients, the higher mean doses of antipsychotics they received, and the higher number of patients receiving clozapine at baseline. However, patients with more-severe disease would have been less likely to have doses reduced by prescribers and qualify as cases. In addition, more-severely ill patients were excluded from the analysis if they received multiple antipsychotics during the study period, had recent increases in doses (dose-reduction group), or were receiving long-acting injectable antipsychotics, all reflective of more-resistant disease. Comorbid psychiatric diagnoses that contribute to severity and hospitalizations, especially substance use, were actually more common among controls and may have biased results against our findings, resulting in worse outcomes among controls. Finally, outcome measures were controlled for numerous covariates in the Cox model that may reflect severity of illness, including psychiatric, substance use and medical co-morbidities, illness duration, and concomitant medication use. Nonadherence to treatment is another major contributor to relapse and hospitalization that may have affected differences in outcomes. However, enrollment criteria screened against including recognized nonadherent patients; patients were excluded if they received long-acting injectable antipsychotics or if they had fewer than 90 days of stable prescription monotherapy with antipsychotics prior to the index study date. Even if the dose reduction group showed greater rates of hospitalization because of nonadherence after the index date, that would still be clinically meaningful evidence suggesting untoward treatment effects of dose reduction. The high attrition during the 2 years of the study and the difference in attrition rates between cohorts were due to the cohort definitions, the differences in outcomes, and possibly nonadherence. Specifically, patients in the dose-reduction and control groups were censored at any change in antipsychotic use outside of the original inclusion criteria or after occurrence of an outcome event. The attrition and censoring of patients were accounted for in the Kaplan–Meier analyses and in the Cox models, which are statistical tools designed to handle censored or truncated time-to-event data [29]. Finally, given the retrospective nature of this claims-based study, we could not confirm the reasons for antipsychotic dose reductions and whether dose reduction affected outcomes positively in relation to other clinical domains.

Results from this study show statistically significant differences in hospital utilization event rates, and suggest that dose reductions of antipsychotics may destabilize psychiatric status and lead to increased hospitalization rates in some patients with schizophrenia, resulting in damaging effects on recovery and adding to overall healthcare costs. Dose reductions had no effect on claims for TD, but this finding was confounded by the limited duration of follow-up and underreporting of the diagnosis. Therefore, decisions on dose reduction of antipsychotics in schizophrenia patients who require maintenance antipsychotic treatment, whether to prevent or reduce TD symptoms or for other reasons, should be carefully considered on an individualized basis. These results may highlight the need for alternative management strategies apart from dose reduction to reduce distressing side effects that would allow physicians to maintain effective maintenance treatment for patients with schizophrenia.