Trends Over Time in Antipsychotic Initiation at a Large Children's Health Care System

Introduction

The use of antipsychotic drugs prescribed for children and adolescents increased dramatically starting in 1990 (Cooper et al. 2006; Olfson et al. 2012, 2015), raising concerns among parents, clinicians, advocates, and policymakers because of their high cost, unknown effects on the developing brain, and adverse side effects (Daviss et al. 2016; Galling et al. 2016; Cohen et al. 2017; Minjon et al. 2019). Moreover, prescriptions were frequently prescribed for disruptive behaviors, rather than for documented psychotic disorders or other approved indications, further exacerbating concerns (Cooper et al. 2006; Olfson et al. 2012, 2015; Correll and Blader 2015). Antipsychotic medication use is associated with significant increases in weight and can cause metabolic disorders (Bobo et al. 2012; Nicol et al. 2018). The low rates of weight and metabolic monitoring for pediatric patients prescribed antipsychotics underscored the lack of adequate oversight in use of antipsychotic medications with potential for long-term health consequences, when other, safer, treatment options existed for many patients (De Hert and Detraux 2018; Nicol et al. 2018).

As a result of the growing concern regarding use of antipsychotics in children, a variety of local, state and federal governmental policies and health care pharmaceutical control practices were put into place over the past two decades (Constantine et al. 2012; Barclay et al. 2017; Melvin et al. 2017; Zito et al. 2018). Some evidence suggests that these efforts, which have gained momentum since 2010, may have been associated with modest declines in antipsychotic prescribing (Stein et al. 2014; Crystal et al. 2016; Akincigil et al. 2020; Mackie et al. 2021). However, most studies in the United States have relied on Medicaid claims data, with little to no research on commercially insured children or of all-payer prescribing data on children and adolescents, such as in hospital-based health care settings (Saldana et al. 2014; Spence et al. 2020). Moreover, most studies have examined trends in overall prevalence of antipsychotic prescribing (Edelsohn et al. 2017; Kalverdijk et al. 2017), rather than prescribing among youth who do not have psychotic disorders or other indicated conditions (Olfson et al. 2012), for whom safer alternatives to antipsychotics may be available (e.g., first-line psychosocial care or other medications). Finally, studies to date have largely examined filled prescriptions based on claims data, which may be affected by patient-level factors and may differ from trends based on antipsychotic orders by clinicians (i.e., when prescriptions go unfilled). Thus, the current literature leaves an incomplete picture of the use of antipsychotics in U.S. children and changes over time.

The present study addresses the limitations of prior studies. We report on the trends observed between 2013 and 2017 in antipsychotic prescribing for patients at a large hospital-based pediatric health care system, Nationwide Children's Hospital (NCH). We make use of data on antipsychotic medication orders from electronic medical records (EMRs), which included children and adolescents prescribed antipsychotics, regardless of insurance type or whether patients/families filled prescriptions. NCH was part of a large multisite pragmatic trial, the “Safer Use of Antipsychotics in Youth” (SUAY) study (ClinicalTrials.gov 2020; Penfold et al. 2020), which was designed to address off-label antipsychotic prescribing, and we report on the baseline trends before beginning recruitment. We describe the overall trends in prescribing, as well as differences in trends observed based on key patient-level factors shown to be predictors of antipsychotic prescriptions in prior research, including age, gender, insurance coverage, and race (white or non-white).

Methods

Measures

Results

The average eligible patient population size per calendar quarter was 8758 and the average number of antipsychotic medication orders was 41.5 per 10,000 person months. There was an overall declining trend over time in rate of index antipsychotic prescriptions for eligible patients, with a rate of 54.9 prescriptions per 10,000 person months in the first quarter of 2013 and 34.1 per 10,000 person months in the last quarter of 2017 (Fig. 1). Although the rates of antipsychotic prescribing declined for all groups, they were significantly higher at nearly all time points for adolescents compared with children, patients who were commercially insured compared with Medicaid insured, and at most time points for white compared with non-white children and adolescents compared with children (Fig. 2). However, there were no significant differences in rates of antipsychotic prescribing based on gender.

OPEN IN VIEWER

FIG. 1.

Overall trend over time in rate of antipsychotic medication ordering.

OPEN IN VIEWER

FIG. 2.

Differences in rates of antipsychotic medication ordering by patient characteristics.

Discussion

This study is the first to report on changes over time in initiation of antipsychotic prescribing in a large hospital-based pediatric health care system and among children and adolescents who did not have documented psychotic disorders, mania associated with bipolar disorder, autism spectrum disorder, or intellectual disability. With concern over potential overuse of antipsychotics in children, it is important to look separately at pediatric patients for whom safer alternatives may be available (e.g., antidepressants, stimulants for disruptive behaviors, or first-line psychosocial therapy). In addition, this is one of few studies to include pediatric patients with multiple types of insurance in the same sample, both those commercially and publicly insured. We found a marked decrease over time in the rates of new antipsychotic prescribing across all subgroups of patients. However, overall prescribing was consistently higher among both commercially insured patients (relative to Medicaid) and White patients (relative to non-White).

The setting and data source of the present study differ from many prior studies, but results appear to confirm recent declining trends in antipsychotic use among children and adolescents (Edelsohn et al. 2017). Although prior studies that relied on prescription fills data identified declining trends, a limitation of these studies was that they could not rule out the possibility that observed trends were the result of changes in patient/family attitudes toward antipsychotic medications, rather than changes in providers' prescribing of antipsychotics. For example, it could indicate that strategies used thus far to target antipsychotic prescribing, many of which are costly and burdensome for prescribers, are not worth the investment. However, the fact that the present study confirmed prior observed trends using data that more immediately captured medication orders offers further evidence that strategies targeting prescribing behaviors are having the intended effects. This has important implications for how we proceed regarding quality improvement of medications for behavioral health conditions as well as other medications that may be of concern, such as opioid analgesics.

There were also some notable differences between the findings of the present study and those of previous studies. We observed lower rates of antipsychotic prescribing for Medicaid relative to commercially insured patients, which differs from most prior research that highlighted the urgent need to reduce antipsychotic prescribing in Medicaid patients, particularly those children in state foster care systems (Leckman-Westin et al. 2018). Lower prescribing rates among Medicaid patients could reflect the greater progress in reducing antipsychotic prescribing through focused Medicaid initiatives, such as prior authorization programs or peer review programs (Akincigil et al. 2020; Mackie et al. 2021). In Ohio, a state-wide behavioral health quality improvement initiative through Ohio's Medicaid program, “Ohio Minds Matter (OMM),” was credited with reducing antipsychotic prescribing for children and adolescents from 2013 through 2015 (Thackeray et al. 2018). However, Medicaid initiatives aimed at reducing antipsychotic prescribing have not been shown to have spillover effects for reducing antipsychotic prescribing among commercially insured patients (Spence et al. 2020; Mackie et al. 2021). This is the first study to compare pediatric prescribing practices based on insurance status within the same large health care system. Moreover, with the present study's use of prescribing data we can rule out the possibility that the observed differences based on insurance result from a lower likelihood of filling prescriptions among caregivers of children enrolled in Medicaid relative to those whose children are enrolled in commercial insurance (Finkelstein et al. 2000; Wang et al. 2006).

Contrary to prior studies, we did not observe a markedly higher rate of antipsychotic prescribing for male compared with female patients or for minority compared with white patients, as has been found in multiple prior studies (Patel et al. 2002, 2005; Lohr et al. 2015; Goddard et al. 2016; Edelsohn et al. 2017; Fontanella et al. 2018; Substance Abuse and Mental Health Services Administration 2019). One important distinction between our study and prior studies is that we excluded patients with a prior medication order for antipsychotics in the previous 180 days or who had specific diagnoses. Thus, it is possible that when comparing the prevalence of antipsychotics use for a broader group of patients, including both incident and prevalent use and youth with other diagnoses, higher rates of use for males and non-white patients would be observed. The differences in prescribing by age, gender, and race in our analysis could be related to parent/guardian preferences for starting new medications versus alternative treatments. Further research is needed to explore the factors parents and guardians consider at the time of treatment initiation.

There are potential limitations to the present study. First, because the data source was limited to the EMR, we do not have information from pharmacy claims regarding whether a prescription was filled and cannot report rates of prescription filling or compare them to prescribing rates. Still, the focus on provider behavior through use of data on antipsychotic prescribing in the EMR is also an advantage of this study. Second, the study is limited to specialty departments within a pediatric hospital-based health care system. Therefore, we do not know whether our findings would generalize to other health systems in other locations. Nevertheless, this is a setting that has been seldom included in the study of antipsychotic medication trends and offers a unique view.

Conclusions

Prescribing of antipsychotics in children and adolescents is decreasing overall in a hospital-based pediatric health care system, but differences in rates are observed between children and adolescents, white versus non-white patients, and Medicaid versus commercially insured patients, but not by gender. These overall trends likely reflect the combined progress of policies, legislation, guidelines, and increased knowledge of risks associated with antipsychotics in children.

Clinical Significance

This study is the first to report trends in antipsychotic medication orders among patients who do not have a documented psychotic disorder, mania, autism spectrum disorder, or intellectual disability. The reported declining trends in prescribing suggest that efforts to reduce prescribing, particularly for pediatric patients for whom there are preferred first-line treatment alternatives, are having an impact on care delivery. However, other findings suggest these gains may be limited to patients in specific subgroups. In particular, a higher rate of new antipsychotic prescribing was observed among patients with commercial insurance, which suggests initiatives to reduce potentially avoidable antipsychotic prescribing, similar to those implemented in state Medicaid programs, may be successful among the commercially insured.