Charting Sea Changes in Outpatient Pharmacotherapy of Bipolar Disorder

In this issue, Rhee and colleagues (1) draw on data collected as part of the National Ambulatory Medical Care Surveys (NAMCS) to provide an up-to-date and nationally representative description of recent trends in the pharmacotherapy of bipolar disorder. Extending from 1997 through 2016 and grouped into five 4-year epochs, the NAMCS survey data document several noteworthy changes in the outpatient care provided by psychiatrists in the United States, including trends in the medications they prescribe.

These data are important for several reasons. The condition that Kraepelin called “manic-depressive insanity” in the late 19th century (2) continues to be one of the several classical forms of severe psychiatric illness that have helped to define modern psychiatric practice. As the 100th anniversary of publication of the final edition of Kraepelin’s text nears, the conditions now grouped together as bipolar disorders remain among the world’s most important public health problems because the illness often begins in adolescence or young adulthood; is characterized by episodic recurrences and, not infrequently, runs a chronic or rapid cycling course; is associated with high rates of vocational impairment and disability; and results in a significant reduction in life expectancy, both from an increased risk of suicide and from cardiometabolic comorbidities (3).

Effective treatment can mitigate the public health impact of bipolar disorder, and effective antimanic medications have been available in the United States for 60 years; 2020 marks the 50th anniversary of the U.S. Food and Drug Administration (FDA) approval of the first lithium salts for prophylactic therapy of bipolar disorder. Psychiatrists continue to prescribe most of the medications received by people with bipolar disorder, and skillful use of the full range of treatments represents the best means to reduce the deleterious impact that this illness has on the patient’s mental and physical health and quality of life, as well as the effects it has on family, social, and vocational functioning (4). Despite a number of developments in the pharmacotherapy of bipolar disorder, there is still a great unmet need for more effective or better tolerated treatments; finding the right treatment regimen for patients who do not respond to standard therapies is one of the biggest challenges faced by psychiatrists in contemporary practice.

The new data on temporal trends in the therapeutics of bipolar disorder revealed in the report by Rhee and colleagues (1) provide an objective way of taking stock of how our treatment choices have evolved since the late 1990s. Some changes in prescription patterns, such as the brief uptick in off-label prescription of gabapentin that occurred between 2001 and 2004, were short-lived. Other changes have occurred more gradually, as illustrated by the slow decline in use of first-generation antipsychotics. Yet others reflect valuable additions to the options available for psychiatrists to treat their patients. Hopefully, learning from the past will help psychiatrists be more mindful about the possibility that what appears to be an exciting new development is actually a fad and, on occasion, about recognizing that there may still be a role for strategies that appear to have fallen out of favor.

One striking finding in the NAMCS data actually has nothing to do with the particular treatments prescribed by psychiatrists. At the outset of the survey period (the 1997–2000 epoch), the amount of care provided by psychiatrists to patients with bipolar disorder was estimated at 467,000 visits, which was only about 20% greater than the amount of care provided to patients with schizophrenia (estimated at 380,000 visits). The estimated treated prevalence of these two conditions circa 2000 was thus remarkably similar to the actual prevalences for bipolar disorder and schizophrenia reported in DSM-IV (5). Over the years that followed, the NAMCS survey documented a significant change: whereas the number of estimated visits was almost constant for schizophrenia, the number of visits for bipolar disorder more than doubled. By the final epoch of the survey, the care provided by psychiatrists amounted to three times the number of visits for patients with bipolar disorder than for patients with schizophrenia.

One limitation of this type of survey data is that we can only speculate about where all those “new” bipolar patients came. Arguably, the most likely explanation is the growing popularity of Akiskal’s concept of a bipolar spectrum (6) largely populated by patients who were formerly thought to have unipolar depressions. Indeed, it has been shown that relatively small changes in the thresholds used to diagnose hypomania, such as requiring fewer symptoms or fewer symptomatic days to meet criteria for an episode of illness, can have a large impact on the apparent prevalence of bipolar disorder (7).

Two other temporal trends in the NAMCS data set revealed that although psychiatrists were treating more patients with the diagnosis of bipolar disorder, the kind of care they were providing was evolving in way that might adversely affect outcomes. The first trend reflected a decrease in the percentage of patients receiving concomitant psychotherapy, which dropped from about 50% at the outset of the survey to only 35.7% during the final epoch. The second pertained to a reduction in the length of visits, which is likely a proxy for the provision of psychotherapy by the treating psychiatrist. During the first epoch of the survey, 44.5% of the psychiatrist visits were longer than 30 minutes; by the final epoch, only 29.3% of the visits were longer than 30 minutes. It is an unfortunate irony that as the evidence documenting the value of cognitive-behavioral, interpersonal, and family-focused therapies for patients with bipolar disorder grew progressively stronger (8), the actual use of psychotherapy declined.

With respect to the specific types of medications prescribed by U.S. psychiatrists for outpatients with bipolar disorder, the most dramatic changes involved the use of antipsychotic medications and mood stabilizers. During the first epoch (1997–2000), an antipsychotic medication was prescribed at only about one in five patient visits (19%). By the final two epochs (2009–2016), antipsychotic medications were prescribed at more than half of visits (53%). This large shift was driven entirely by the growing prescription of second-generation antipsychotics, increasing from about 12% in the 1997–2000 epoch to about 50% in the 2005–2008 epoch and remaining relatively constant thereafter. The older antipsychotics—already declining in use in the 1997–2000 epoch—were seldom prescribed by the end of the study period, as the frequency of use decreased from 7% of visits to less than 2%.

Greater use of second-generation antipsychotics was closely mirrored by decreasing prescription of mood stabilizers. For example, the percentage of visits at which lithium was prescribed decreased from 30% to 18% across the two decades of survey data that were analyzed. An even larger reduction was observed in the use of the “anticonvulsant mood stabilizers” (a pooled grouping of divalproex and carbamazepine): their use decreased from 35% of visits at the beginning of the survey to only about 5% of visits by the final epoch—an 86% reduction in use! The introduction of a third anticonvulsant mood stabilizer, lamotrigine, which was approved by the FDA in 2003, had little impact on this sea change in therapeutics, as lamotrigine was prescribed at only about 5% of visits for bipolar disorder across the final three epochs of the survey.

It would appear that by 2010, U.S. psychiatrists were practicing as if the second-generation antipsychotics constituted a “new” and preferred subcategory of mood stabilizers. How did this happen, and did the state of the evidence justify such paradigm shift? Rhee and colleagues (1) correctly point to the influence of marketing on physicians’ decisions about prescribing newer drugs. During the early years of the NAMCS survey, the second-generation antipsychotics were available only in more expensive branded formulations, and, as a result, there is no doubt that this particular prescribing trend increased the costs of care. Moreover, any anticipation of offsets in costs resulting from less need for therapeutic drug monitoring or periodic laboratory testing to ensure safety during preventive therapy (e.g., annual assessments of thyroid function or creatinine clearance) was relatively quickly dispelled by the growing recognition that patients taking second-generation antipsychotics needed to be monitored to ensure that therapy was not adversely affecting lipid or glucose metabolism (9–11).

Although the impact of marketing cannot be discounted, it is noteworthy that equally capable promotional efforts did not have the same impact on psychiatrists’ prescriptions of anticonvulsant mood stabilizers. The small spike in lamotrigine prescription that accompanied FDA approval was short-lived, and the introduction of newer, branded formulations of divalproex and carbamazepine appeared to have little impact on the use of these medications.

Another potential reason for the growing use of second-generation antipsychotics is continuity with inpatient treatment: even in the early years of the survey, the second-generation antipsychotics were widely perceived to be very useful options for treatment of acute mania, in terms of both effectiveness and ease of initiating therapy (12). That said, continuity of medications following hospitalization is not likely to be the major factor driving the use of second-generation antipsychotics, specifically because the increase in visits for treatment of bipolar disorder appears to have come largely from the “softer” side of the bipolar spectrum, that is, patients who have never experienced a manic episode.

Conversely, one might speculate that greater use of second-generation antipsychotics was at least partly the result of growing evidence that some of these drugs have meaningful antidepressant efficacy without a significant risk of provoking treatment-emergent affective shifts (9, 10). However, if psychiatrists were prescribing second-generation antipsychotics to treat bipolar depression, why did the use of conventional antidepressants not decrease? In fact, use of antidepressants actually increased significantly across the 20 years of the NAMCS survey, from 47% in 1997–2000 to 57.5% in 2013–2016. As the authors note, the predilection of psychiatrists for prescribing antidepressants to patients with bipolar depression remains a curiosity given the inconsistent evidence of efficacy in controlled studies (13) and low ratings of confidence in practice guidelines (10, 12).

Perhaps it is simply true that by about 2005 many psychiatrists had recognized that at least some of the second-generation antipsychotics have a constellation of clinical effects that are fully consistent with the definition that was used to define the conventional mood stabilizers: antimanic efficacy, antidepressant efficacy, low risk of treatment-emergent affective shifts or rapid cycling, and prophylactic efficacy (14). Maybe it is time that the second-generation antipsychotics that show these effects—olanzapine, quetiapine, and possibly others—are classified as mood stabilizers in addition to being antipsychotics.

According to the NAMCS survey, the second-generation antipsychotics have served as the cornerstone of therapeutics of bipolar disorder for the past 15 years. It is not clear how long this trend will last; promising alternatives do not appear to be on the immediate horizon. Lithium was in a class by itself for more 25 years before it was supplanted by divalproex in the late 1990s. According to the NAMCS survey results, divalproex had a shorter time as the most preferred option, being passed by the second-generation antipsychotics at some point before 2004. As generic formulations of the more widely prescribed members of the class continue to proliferate, concerns about acquisition cost and cost-effectiveness have naturally faded. However, there are good reasons for ongoing concern about other “costs” associated with such widespread use of the second-generation antipsychotics, including the risks of weight gain and the aforementioned metabolic consequences (9–11), as well as the uncommon but real specter of tardive dyskinesia during more extended courses of therapy (15).

Because of the necessity of prophylactic therapy, an ideal mood stabilizer must be both well tolerated and safe across months or even years of therapy. The psychiatrists who care for patients with bipolar disorder thus have the clear responsibility to ensure that their patients are well-informed about the risks of their medications. Likewise, there is the obligation for prescribers to monitor their patients who are on longer-term pharmacotherapy to ensure that the incidence of treatment-emergent side effects, including weight gain, dyslipidemia, diabetes, movement disorders, and other complications, are minimized. There is a similar need to ensure that when adverse effects are recognized, appropriate actions are taken in collaboration with the patient’s primary care provider.