Off-label antipsychotic use patterns among Texas Medicaid adults 2013–2016

Antipsychotics have been US FDA-approved for a range of psychiatric conditions including schizophrenia, bipolar disorder, major depressive disorder (MDD), irritability in autism and Tourette syndrome. Off-label use refers to FDA-approved medications that are prescribed for unapproved clinical indications, age groups or dosage forms [1]. Off-label use is widely adopted in clinical practice, especially in psychiatry [2,3] and antipsychotic medications have been prescribed frequently for patients without FDA-approved indications [4–8]. Although off-label use is common, antipsychotics have been associated with adverse effects such as extrapyramidal symptoms [9], cardiovascular and metabolic abnormalities [10,11] and increased risk of death among older people [12]. Thus, it is important to continually monitor trends regarding off-label antipsychotic use.

In the early 2000s, approximately over one-half (49.0%–63.6%) of antipsychotics prescribed were for off-label use, which has shown a decline over time for Medicaid populations. Antipsychotic off-label use in Georgia Medicaid was 63.6% in 2001 [5], while a 42-state Medicaid study reported 49.0% of antipsychotic-treated adults received off-label prescriptions in 2003 [6]. Similar findings were reported in a study using 2001–2002 commercial claims where 58.6% of antipsychotics were prescribed off-label during office visits [13]. However, a remarkably lower rate of off-label use was observed in studies using data after 2009, which could be attributed to more FDA-approved indications for antipsychotics since 2006. A study using national commercial data from 2009 to 2010 reported that 13.7% of second-generation antipsychotics (SGAs) were prescribed off-label [14]. Similarly, a nationally representative survey (i.e. National Ambulatory Medical Care Survey [NAMCS]) using 2012–2013 data reported that 17.4% of office visits associated with antipsychotic prescriptions were off-label [3].

Some studies examined antipsychotic-treated patients with no documented mental health diagnosis and prevalence varied widely (5.9%–67.1%) [4,6,15–17]. Of those with lower prevalence, one was conducted using 2009–2010 national commercial data where 5.9% of adults who received SGAs had no diagnosis [14]. Another study was conducted using 42-state 2003 Medicaid data where 10.8% of antipsychotic-treated patients (including children and elderly) had no mental health diagnosis [6]. Several studies were conducted using nationally representative survey (i.e., NAMCS) data over three time periods with varying results regarding antipsychotic prescribing for adults with no documented mental health diagnosis: 2005–2009: 34.2% [16]; 2012–2013: ∼48.0% [15]; 2014–2015: ∼31.0% [15]. The highest prevalence of no mental health diagnosis occurred in studies using 2009 IMS claims data, where over three-quarters (76.8%) of antipsychotic-treated adults had no mental health diagnoses [4] and similarly, most (67.1%) young adults (19–24 years) who received antipsychotic medications did not have any mental health diagnoses [17]. Some of the no mental health diagnosis prevalence estimates could be explained by provider efforts and/or patient requests to not document mental health conditions due to potential stigma and disparate treatment [16]. In addition, antipsychotic medications might be prescribed to patients who experience insomnia or behavioral symptoms but not meet the diagnostic criteria for mental health disorders [16,18]. Additionally, in some studies, occasions of no diagnosis may be caused by database documentation limitations where only a subset had clinical diagnoses recorded and/or where a maximum number of diagnoses (i.e., ≤3 diagnoses) could be listed per visit [4,15–17]. As mentioned previously, since 2006, more expanded FDA indications have been approved for antipsychotics. Specifically, risperidone was the first approved SGA for treating irritability in autism for patients aged ≥5 years (2006). Several SGAs (i.e., aripiprazole in 2007, olanzapine in 2009 and quetiapine in 2009) have added indications such as augmentation combined with antidepressants for MDD.

A comparative effectiveness report by the Agency for Healthcare Research and Quality was updated in 2011 and new evidence regarding off-label antipsychotic use was added [19]. However, while most antipsychotic utilization pattern studies focused on elderly or young populations as the Agency for Healthcare Research and Quality guideline mentioned [19], studies in adult populations were limited. Early Medicaid database research (i.e., before 2003) does not provide up-to-date information for policymakers, payers and providers regarding antipsychotic off-label use [5,6]. Additionally, because most studies examined only 1 year of data, the literature examining off-label antipsychotic use trends is limited [5,6,14], and little is known regarding these issues in Texas Medicaid. Therefore, this study was conducted to describe the proportion of antipsychotic off-label use over time and to examine factors associated with off-label use among adults enrolled in Texas Medicaid who were prescribed antipsychotics.

From 2013 to 2016, 51,257, 54,523, 49,843 and 39,151 (January–August data only available in 2016) patients with prevalent antipsychotic use were identified from Texas Medicaid data, respectively (Table 2). The proportion of antipsychotics prescribed among each age group was stable during the study period. Regarding age groups, the proportion of Medicaid patients who received antipsychotics were ∼16% (18–24 years), ∼20% (25–34 years), ∼19% (35–44 years), ∼25% (45–55 years) and ∼20% (55–63 years). More than half (54.0%–57.5%) of those prescribed antipsychotics were female and approximately 90% were on SGAs. In 2015, the mean age for the study sample was 41.1 ± 13.5 years.

The proportions of off-label antipsychotic users decreased from 22.5% in 2013 to 17.4% in 2016. When employing the sensitivity analysis (i.e., diagnostic groups based on drug class level vs individual drug), the proportion of off-label use was lower (∼6–7 percentage points) and ranged from 16.1% in 2013 to 10.3% in 2016 (data not shown in Table 2). Less than 10% (7.5%, 8.1%, 7.3% and 9.4%, from 2013 to 2016, respectively) of patients receiving antipsychotic prescriptions had no diagnosis (Table 2).

As shown in Table 3, during 2015, among those with off-label antipsychotic use, prevalence was highest in young adults aged 18–24 years (25.5%), followed by the 55–63 age group (20.6%). Of note is that the middle-aged patients (35–44 years and 45–54 years) had the lowest proportion of off-label use (17.6%). Among young adults (18–24 years) with an off-label diagnosis, pervasive developmental disorder (29.1%) and attention-deficit/hyperactivity disorder (ADHD; 22.6%) were most common. Among adults aged 25–34 years, pervasive developmental disorder (25.8%) and depression (20.3%) were most prevalent. Among age groups ≥35 years old, depression was the most common off-label diagnosis (range: 29.5%–33.5%) followed by anxiety (range: 21.3%–23.0%).

As shown in Table 4, the proportion of off-label use/no diagnosis (combined) across each age group was significantly (p < 0.0001) different. Chi-square analysis also showed males had a significantly (p < 0.0001) higher proportion of off-label use/no diagnosis compared with females (28.3% vs 26.3%, respectively). Last of all, FGA-treated patients had a significantly (p < 0.0001) higher proportion of off-label use/no diagnosis than SGA-treated patients (49.8% vs 25.0%, respectively).

When controlling for covariates, age group and antipsychotic class were statistically significantly associated with off-label use/no diagnosis (Table 5). Young adults aged 18–24 years and those prescribed FGAs were more likely to receive antipsychotics off-label/no diagnosis. Compared with young adults aged 18–24 years, patients in all older age groups were significantly (p < 0.0001) less likely to be prescribed antipsychotics off-label/no diagnosis. Compared with young adults aged 18–24 years, the odds of receiving antipsychotics off-label/no diagnosis were 32% lower for adults aged 25–34 years (p < 0.0001; OR: 0.68; 95% CI: 0.64–0.72), 45% lower for adults aged 35–44 years (p < 0.0001; OR: 0.55; 95% CI: 0.52–0.59), 47% lower for adults aged 45–54 years (p < 0.0001; OR: 0.53; 95% CI: 0.49–0.56) and 35% lower for adults aged 55–63 years (p < 0.0001; OR: 0.65; 95% CI: 0.61–0.70). SGA recipients had a 68% lower likelihood of off-label use/no diagnosis antipsychotic use compared with FGA recipients (p < 0.0001; OR: 0.32; 95% CI: 0.30–0.34).

From 2013–2016, the proportion of off-label use/no diagnosis decreased from 30.0% to 26.8%, which represented a 10.7% decline. This modest decrease was robust when using sensitivity analyses that defined diagnostic status (i.e., on-label, off-label and no diagnosis) based on drug class level (vs individual drug level). When using the less conservative sensitivity analysis of class versus individual level, our results ranged 10.3%–16.1% for off-label, and 7.3%–9.4% for no diagnosis, which was comparable to a study using 2009–2010 national commercial data, which reported 13.7% for off-label use and 5.9% for no diagnosis [14]. In our study, 19.9% of antipsychotics were prescribed off-label and 7.3% had no mental health diagnosis in 2015. However, these results were notably lower when compared with previous Medicaid database studies (49.0%–63.6%), which were conducted using data from 2001–2003 [5,22]. It might be reasonable that early Medicaid studies had a higher proportion of off-label use due to several factors including: limited FDA-approved indications, different definitions for off-label use, and different inclusion criteria. Regarding limited FDA-approved indications, since 2006, antipsychotics have been approved for several additional indications including irritability in autistic disorders and MDD. The difference between Georgia Medicaid study and our study was likely due to more approved indications after 2006, since both our study and the Georgia Medicaid determined diagnostic status at the individual drug level [5]. Despite the approved indication additions, the difference between the 42-state Medicaid study and our results can be explained by different definitions for off-label use. In the 42-state Medicaid study by Leslie and Rosenheck, off-label use was determined based on drug class level and any diagnosis with the exception of schizophrenia or bipolar disorder were off-label [6]. Additionally, inclusion of the elderly population in the Georgia Medicaid study may account for a higher prevalence of off-label use when compared with our study, which only included nonelderly adults aged 18–63 years. The Georgia Medicaid study revealed that patients ≥65 years were 5 times more likely to receive antipsychotics off-label than patients younger than 65 years [5]. Olfson et al. also indicated that the prevalence of dementia, the most common off-label diagnosis among elderly, increased as advancing age among antipsychotic-treated patients [4]. This exclusion of patients ≥64 in our study may contribute to a lower prevalence of no mental health diagnosis when compared with the Leslie and Rosenheck [6]. This is also supported by a recent NAMCS study (2019) that found the odds of no mental health diagnosis were two-fold higher for elderly aged ≥65 years when compared with 18–44 years [15].

As expected, bipolar disorder and schizophrenia were the most prevalent on-label indications. Our study results were similar to another study’s findings, regarding depression and anxiety as the most prevalent off-label indications in adults [16], which is congruent with guidelines [23–25]. Practice guidelines suggested SGAs improve the response or remission rate of depressive symptoms even in patients without any psychotic symptoms [23,24]. Although the guidelines acknowledge that limited evidence exists to support this claim and that adverse effects exist, quetiapine was recommended as second-line therapy for anxiety [25]. Antipsychotic use in pervasive developmental disorder, which was the third most commonly diagnosed off-label indication among adults, was also supported by guideline. Guideline supported the efficacy of aripiprazole and risperidone in ‘reducing repetitive movements, self-injury and severe disruptive behavior’ for patients with pervasive developmental disorder [26].

From the logistic regression, patients in age groups ≥25 years and receiving SGAs were associated with a lower likelihood of off-label antipsychotic use. The age pattern of off-label antipsychotics use can be linked to prevalent diagnosis of ADHD and developmental disorder diagnosis among young adults (18–24 years) [27]. The higher likelihood for FGA users receiving off-label prescribing compared with SGA users was also observed in the Georgia Medicaid study [5]. Most FGA medications have limited approved indications such as schizophrenia and psychotic disorders. Only chlorpromazine and loxapine have been approved for bipolar disorder, which was the most common on-label indication in our study. Although several studies indicated that males were associated with a lower likelihood of off-label use [5,6], our study did not indicate gender as a strong predictor of off-label use.

There were some limitations in this study. First, a psychiatric diagnosis associated with the index antipsychotic might have been misclassified. In addition, accurate diagnoses of mental health conditions may take years and involve trial and error. Patients might have received other diagnoses during the year and/or the previous year that were associated with the index prescription. It is difficult to accurately assess underlying reasons for off-label prescribing when claims data are used. However, our study utilized similar methods as other studies examining this issue [21,28]. Second, data such as race/ethnicity, community, metropolitan status area, and healthcare provider type were not available, which limited our capability to employ a conceptual framework (i.e., Anderson Behavioral Model) to identify other predictors and to assess their association with off-label antipsychotic use. There was no consensus regarding the association between provider type and likelihood of off-label prescribing. Some studies have shown that psychiatrists were less likely to prescribe antipsychotics for patients without an approved diagnosis (i.e. off-label prescribing/prescribing with no diagnosis) [15,16], while other studies found opposite results [3,5]. Third, this decrease in off-label use might not merely reflect the change in prescribing behaviors, but can also be explained by an increase in documentation of mental health disorders in Medicaid. Finally, our findings regarding off-label use patterns of antipsychotics were specific to the Texas Medicaid adult population, which might not be applicable to other populations.

In our analysis, off-label use of antipsychotics and antipsychotic use with no concurrent psychiatric diagnosis in Texas Medicaid declined from 2013 to 2016. Overall, depression and anxiety were the most common off-label diagnoses among adults, but ADHD was the most common diagnosis among young adults. Young adults and FGA recipients were more likely to have an off-label antipsychotic prescription. As adolescent patients transition to young adults, use of antipsychotics for ADHD may need to be closely monitored.

Future research should continue to examine off-label antipsychotic prescribing practices, especially for depression and anxiety diagnoses. In addition, additional research should examine if off-label use with these mental health conditions result in optimal patient outcomes.

All authors had substantial contributions to the conception or design of the work, involved in data interpretation, provided final approval of the version to be publish and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. S Chen analyzed and interpreted the data and drafted the manuscript; JC Barner and E Cho revised the manuscript critically.

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

This study was approved by the Institutional Review Board of the University of Texas at Austin.