Factors Associated With Off-Label Utilization of Second-Generation Antipsychotics Among Publicly Insured Adults

Using a conservative approach to examine patterns of off-label utilization of second-generation antipsychotic medications among publicly insured adults in four states and in fiscal years 2009–2013, we found that utilization was always off-label for a large fraction of users, roughly two in five, many of whom (26%–36%) had no mental illness diagnosis. Key findings were that although fee-for-service payer did not modify the relationship between race-ethnicity and off-label utilization, race-ethnicity and payer were generally associated with off-label utilization (lower for users belonging to racial-ethnic minority groups, especially Blacks relative to Whites, and higher for those having Medicaid and to a lesser extent Medicare as payers relative to dually eligible beneficiaries). We also found that although off-label utilization of second-generation antipsychotics increased in all states relative to the baseline year, a downward trend followed in three of the four states. Other findings were that the doses used for off-label indications were up to four-fifths the dose used for on-label indications, and average costs were at best two-thirds of those associated with on-label utilization.

Although off-label prescribing has a long tradition in medicine, and may be appropriate in some cases, the decision to prescribe medications off-label should be made with careful consideration of the potential risks when the evidence of benefit is minimal or mixed (16). A 2011 review of the evidence on off-label utilization of second-generation antipsychotics concluded that selected drugs may be associated with small benefits in the treatment of patients with generalized anxiety disorder, obsessive-compulsive disorder, or dementia (6). However, adverse events, including cardiovascular morbidity, were common, despite observation periods that were far shorter than the longer exposures typical of routine care (6). A recent evidence review concluded that off-label prescribing of second-generation antipsychotics is inappropriate because the risks of such prescribing outweigh potential benefits that are uncertain and at most minimal (17). The practice has a particularly unfavorable risk-benefit profile for individuals who are Black or Latino. Not only do Black adults have a higher risk for cardiometabolic disorders than Whites (18, 19), which, with some exceptions, is also the case for Latinos (20, 21), but Black race appears to confer a higher risk for weight gain related to use of second-generation antipsychotics (22). The evidence is less clear for other cardiometabolic effects (23).

Off-label utilization has economic implications for the payers. Until well into our study period, antipsychotics were the costliest drug class for Medicaid; by 2009, these costs were almost double those of the second costliest class among fee-for-service beneficiaries (24). The Medicare prescription drug plans that manage Medicare’s Part D benefit on behalf of the Centers for Medicare and Medicaid Services (25) are also large purchasers of second-generation antipsychotics (26). Although costs of second-generation antipsychotics have dropped after generic products entered the market, they remain expensive to payers (27). Thus, public payers bear a large portion of the economic consequences of off-label utilization of second-generation antipsychotics, and they are also likely to be the payers for care associated with cardiometabolic morbidity related to second-generation antipsychotics use.

In our study, Blacks in all states and Latinos in two states, including California, where they made up one-fifth of the study population, had lower off-label utilization of second-generation antipsychotics than did Whites. Although our finding might stem from prescribers’ efforts to minimize risks, alternatively, it may have resulted from provider and system factors that underlie widely documented racial-ethnic disparities in health care (15, 28). This explanation is supported by evidence that prescribers have limited knowledge of approved indications, including approved indications for second-generation antipsychotics (29).

Our finding of differences among payers in off-label utilization is novel, because little evidence for such differences currently exists. Medicaid and Medicare employ utilization management strategies such as previous authorization to steer care toward lower-cost drugs (30, 31). However, the programs vary in their use of these strategies to shape utilization; they also vary in their approaches to negotiate lower prices and the size of the rebates (32) and in their adoption of value-based payment models over the past decade (33). The absence of substantive differences between data for Medicare beneficiaries and dually eligible beneficiaries may be due to the federal government’s regulation of the pharmacy benefit for both. Prescription drug plans have latitude in their implementation of utilization management (30) and tend to use previous authorization to manage utilization of second-generation antipsychotics. Medicaid beneficiaries’ higher off-label utilization may stem from Medicaid programs’ preference for not employing aggressive antipsychotic utilization management for adults (30, 34). Specifically, the Medicaid states in our study (California and Oklahoma) have implemented only modest diagnosis-related restrictions to adults’ access to second-generation antipsychotics (35, 36).

The drivers of the general downward temporal trend that followed an initial surge in the likelihood of off-label utilization in three of the four study states are likely multifactorial but may include concerns about the health or economic implications of the practice. Our dose findings are noteworthy because no systematic evidence exists to guide dosing for off-label indications. Moreover, because of limitations in the evidence on dose dependence of cardiometabolic effects of second-generation antipsychotics among people with serious mental illnesses (37, 38), the use of lower doses for off-label users may be less protective than prescribers might believe.

Our finding of substantial cost reductions in fiscal year 2013 for most states and payers was not the result of reductions in doses of second-generation antipsychotics; our analyses showed that doses were quite stable throughout the study period (results not shown). A potential explanation is savings resulting from the entry in late 2011 of generic olanzapine and in early 2012 of quetiapine and ziprasidone, all three of which are widely used second-generation antipsychotics.

We cannot draw firm comparisons between our results and those in the available literature. Published rates of off-label utilization of second-generation antipsychotics among adults have varied widely, ranging between 17% and 85% (4, 17, 39–42), probably as a result of differences in populations, study years, and study methods. The limited evidence on the association between race-ethnicity and off-label utilization is mixed, and no study has assessed the modifying effect of payer (40, 43). We are aware of only two survey-based studies that reported on variation of off-label utilization of second-generation antipsychotics and associated costs by payer (42, 44), but the only study conducted after Medicare Part D was implemented based its conclusions on analyses of aggregate visits (42). Few studies have reported on temporal trends (43), and we are not aware of recent evidence. Our finding that off-label doses are modestly smaller than on-label doses is consistent with limited previous research (41).

Our study had some limitations. First, because treatment-resistant major depressive disorder requires an assessment of failure of antidepressant monotherapy, a complex undertaking when using administrative data, we required only evidence of illness severity. This decision and our assumption of class effects may have led to underestimation of the true prevalence of off-label utilization of second-generation antipsychotics. Our decision to restrict our analysis to two diagnostic codes for person-months covered by Medicare or dual payers to avoid ascertainment bias may have had the opposite effect, although the decrease in always off-label utilization was minor. Underdocumentation of serious mental illness diagnoses might also contribute to overestimating off-label utilization. However, there is substantial agreement of claims-based serious mental illness diagnoses with diagnoses based on more clinically rich sources (45, 46), and it is unlikely that underdocumentation differed systematically among payers. Second, we may not have fully controlled for differences in case mix among the payers. However, if we assume that all off-label uses of second-generation antipsychotics are of equal low value, such differences should not affect our conclusions.

Third, our study may not be generalizable to other states, nor is it generalizable to off-label utilization of second-generation antipsychotics under managed-care arrangements, which might make greater use of utilization management strategies. Fourth, because we lacked information on prescriber discipline or specialty, we were unable to assess its association with off-label prescribing, particularly in the absence of a mental illness diagnosis. However, some evidence suggests that although psychiatrists may be more likely to prescribe antipsychotics off-label (42), nonpsychiatrists appear to be far more likely to prescribe antipsychotics without documenting a mental illness diagnosis (5). Fifth, given the years covered by our analyses, they may not reflect current patterns of off-label utilization among publicly insured fee-for-service beneficiaries that may have been influenced by changes in costs of second-generation antipsychotics or quality measurement initiatives launched as part of the Affordable Care Act and the move toward value-based payment models. Although additional generic drugs have entered the market in recent years, entry of generic drugs does not necessarily translate to marked cost reductions for payers, and any cost reductions may have been offset by the entry in 2015 of two new second-generation antipsychotics. We are unaware of any performance measurement system broadly used by public payers that monitors off-label utilization of second-generation antipsychotics. Last, we overestimated the costs to the payers, more so for Medicaid, because like others (47), we did not account for rebates and discounts. Nevertheless, our cost findings are meant to provide only a signal of the significance of off-label utilization for the payers.

Off-label utilization of second-generation antipsychotics was common among nonelderly Medicaid, Medicare, and dually eligible adult beneficiaries in four states for whom second-generation antipsychotics were prescribed. Members of racial-ethnic minority groups, especially Black individuals, had a lower likelihood of off-label utilization regardless of payer, an encouraging finding given racial-ethnic minority groups’ higher baseline risk for cardiometabolic morbidity, even if the difference was potentially driven by factors associated with health care disparities. Our finding that off-label utilization of second-generation antipsychotics varies by payer expands a small body of evidence on this topic and suggests that payer policies such as previous authorization or financial incentives tied to quality metrics can affect off-label utilization.

Preliminary results of earlier analyses of the data in this article were presented at the National Institutes of Mental Health 24th Mental Health Services Research Conference, August 1, 2018, Bethesda, Maryland.