In this large prospective naturalistic study of patients with schizophrenia, antipsychotic polypharmacy was found to be highly prevalent and to be of prolonged duration. Overall, most patients (57.7%) had at least one period of antipsychotic polypharmacy longer than 60 consecutive days, and only a third (35.7%) of the patients were treated predominately with monotherapy. More specifically, two thirds (66%) of the patients were treated with another antipsychotic at the time of initiation on the atypical agent, a practice that could have signaled a medication change process. However, most of these patients continued on antipsychotic polypharmacy for a substantial duration and only a small proportion of those patients (30%) were deemed to have gone through medication changes as polypharmacy ceased within the first 60 days after medication initiation. Findings suggest that for the majority of patients, polypharmacy is a prolonged and deliberate treatment choice rather than an interim, brief, or unintentional practice.
A third (34%) of the patients were not receiving another antipsychotic at the time of initiation on the studied atypical medications (olanzapine, quetiapine, and risperidone). These monotherapy-initiated patients continued on monotherapy for only about a third of the year post initiation, and at the end of that year, more than 50% of them were no longer receiving monotherapy with the initiating antipsychotic. Further attesting to the pervasive practice of antipsychotic polypharmacy is the finding that over 40% of all patients had no days of monotherapy with the initiating atypical antipsychotics during the 1-year treatment period. On the average, patients were treated with monotherapy for 54% of the year, with polypharmacy for 43% of the year, and without antipsychotic therapy for 3% of the year.
While polypharmacy was found to be prevalent in this patient population, the question of generalizability remains. However, a goal in designing the US-SCAP study was to generate a sample of patients representative of those treated in usual care. Participants in this large prospective naturalistic study were treated for schizophrenia at large public health care delivery systems in the United States, and were enrolled from multiple sites across six states. The patients were randomly identified from active client rosters at each site and only then approached about enrollment in the study. In addition, there were few exclusion criteria that would restrict the patient population.
This study provided a 1-year longitudinal perspective on the rate and duration of polypharmacy following initiation on the index antipsychotic whereas most previous studies assessed polypharmacy during shorter time periods, such as a 2-month window [15
], or during inpatient hospitalization [16
]. Generally, the larger is the studied time window, the higher is the likelihood of finding polypharmacy. Another major difference is the age of the data because polypharmacy has increased over the years. Studies using data from the earlier years after the introduction of the atypical antipsychotics [10
] tend to report lower prevalence rates of antipsychotic polypharmacy, whereas studies using more recent data reported higher polypharmacy rates [8
]. The complexity involved in comparing findings across studies is further compounded by variations in the definition of polypharmacy. Although most studies defined polypharmacy as any time with more than one antipsychotic [15
], others have set specific time requirements, such as at least 14 days of concurrent antipsychotic use [8
]. Other differences in study methods can generate different results. While the current study followed patients after their initiation on certain antipsychotic medications, other studies used a cross sectional method, assessing the prevalence of polypharmacy at a given time window, without using time of initiation on the studied antipsychotic medications as the reference point. While cross sectional designs provide important information about prevailing practices at a given time window, the cross sectional method is not well suited for comparisons between antipsychotic treatment groups. When the date of initiation on the antipsychotic is not used as a starting point, treatment group differences may be obscured by differential duration on the medication prior to the studied time window, data that are not included in cross sectional designs.
This study also found significant differences in monotherapy and polypharmacy between the most commonly prescribed atypical antipsychotic medications. Patients initiated on olanzapine were significantly more likely to be on monotherapy compared to quetiapine (rate and duration) or risperidone-initiated patients (rate only) during the 1-year post treatment initiation. The current findings appear consistent with several previous studies in which olanzapine-treated patients were found to have significantly higher rate of monotherapy compared to risperidone [18
], and compared to quetiapine-treated patients [25
], with the quetiapine treatment group being the least likely to be treated monotherapy [8
]. Findings are particularly congruent with those reported in a study of medication patterns in the Michigan Medicaid database [28
], in which the proportion of schizophrenia patients treated with olanzapine monotherapy remained relatively steady over the first 3 months of treatment while the proportion of patients receiving risperidone monotherapy decreased over time.
The present study helps demonstrate the dynamic and complex nature of medication management of patients with schizophrenia in usual clinical practice [14
]. In addition to providing new information on the rate of monotherapy during treatment with various atypical agents, this study further contributes to the literature by providing a longitudinal perspective on the duration of monotherapy/polypharmacy following treatment initiation. The strengths of this study appear to lie in its large representative and diverse sample, the ability to provide comparative data on a number of commonly used atypical antipsychotics, the availability of comprehensive medication information about use of antipsychotics in depot formulation and about antipsychotics used during hospitalizations (types of data often absent in claims databases), and notably, the ability to generalize the findings to patients treated at large public systems of health care across the United States.
This study also has its limitations. First is the use of naturalistic observational data to compare antipsychotic treatment groups, because observed group differences in rates or duration of monotherapy could result from pre-existing differences between the treatment groups rather than differences in medication choice. Physicians tend to select treatments for different types of patients and illness profiles, a practice that may lead to lack of comparability between the treatment groups at initiation. Analyses that do not appropriately control for such differences can be biased, and one can never be certain that all important group differences have been controlled for statistically. Indeed, in this study differences were observed between groups at initiation. Although differences in several available pre-existing patient and treatment characteristics were controlled for in the analysis, it is possible that other pre-existing group differences were present. For instance, the design of this study did not include assessment of symptom severity or substance use at the time of initiation on the index antipsychotic. Thus, differences in symptom severity or substance use at initiation could not be controlled for.
Another study limitation is lack of information about reasons for initiating antipsychotic medication changes or the specific treatment indications. This type of information was not assessed in US-SCAP but could be valuable in discerning whether there is a link between treatment effectiveness and antipsychotic monotherapy (or polypharmacy). Treatment indication is also important because some antipsychotics, particularly quetiapine in low doses, may have been used to treat insomnia rather than core symptoms of schizophrenia. Although some have speculated that higher doses of quetiapine produce better outcome, the available data [35
] do not support this. Davis and Chen [35
] observed, however, "the sponsor of quetiapine has 2 studies of dose response, one nearing completion, and the other just beginning, so more information will be available in the future". They also noted "the fact that individuals do clinically improve at higher doses in open studies does not prove that the higher dose was responsible, as such improvement might reflect the passage of time, augmenting drugs, or other confounding factors."
Further, this study did not control for potential "sponsorship" bias that could have influenced clinicians' prescription rates and duration of monotherapy in favor of the sponsor's antipsychotic medication (in this case, olanzapine). Although we cannot completely rule out this possibility, sponsor bias is unlikely to have played a role in this study because the current findings are highly consistent with findings of other studies conducted by independent researchers who used non-industry-sponsored data such as Medicaid claims database [18
]. Additional support for the validity of our findings and their freedom from sponsorship bias comes from a recent double-blind randomized study conducted by the sponsors of risperidone [35
], comparing risperidone and quetiapine on the rate of antipsychotic polypharmacy. That short term trial found the relative risk (quetiapine vs. risperidone) of using antipsychotic polypharmacy to be higher for quetiapine compared to risperidone (odds ratio 1.90, 95% CI, 1.29–2.80, p = 0.001), a finding similar to that in our observational study, in which the odds ratio for antipsychotic polypharmacy of quetiapine vs. risperidone was 1.53 (95% CI 0.94–2.47, p = 0.085).
This study also did not present information about concomitant use of psychotropic medications other than antipsychotics, thus providing a partial understanding of the full range of psychotropic polypharmacy that takes place in usual practice. This is a complex phenomenone that will benefit from a separate and detailed study. We, however, cognizant of the potential role of prior concomitant psychotropic medications and used five types of psychotropics (antidepressants, anti-anxiety, antiparkinsonian, mood stabilizers, and sleep agents) in the analyses. Because prior use of concomitant psychotropics is highly correlated with continued use of these concomitant medications post initiation of the new antipsychotic regimen, we opted to include patients' prior concomitant medication as covariates in the analyses. The correlations between concomitant psychotropic medication use prior to initiation on the index drug and its concomitant use post initiation on the antipsychotic were consistently high, ranging from .65 to .81 (r = .81 for antidepressants, r = .80 for anti-anxiety agents, r = .73 for antiparkinsonians, r = .81 for mood stabilizers, and r = .65 for sleep agents). These correlations suggest that the bigger factor in the use of concomitant psychotropic medications post antipsychotic initiation is likely patients' prior medication history and not the antipsychotic monotherapy/polypharmacy status. Generally, the mean daily total number of concomitant psychotropic medications was higher for patients on antipsychotic polypharmacy (1.56) than for patients on antipsychotic monotherapy (1.24). Lastly, this study did not address the potential impact of polypharmacy on the total cost of antipsychotic treatment, or on patients' clinical outcomes, an area of investigation that will require further study.