National Trends in Second Generation Antipsychotic Augmentation for Non-Psychotic Depression

Abstract

Objective:

This study estimates national trends and patterns in use of second-generation antipsychotics (SGAs) for adjunctive treatment of non-psychotic adult depression in office-based practice.

Method:

12 consecutive years (1999–2010) of the National Ambulatory Medical Care Survey were analyzed to estimate trends and patterns of adjunctive SGA treatment for adult (≥18 years) non-psychotic depression in office-based visits. Adjunctive SGA use was examined among all office visits in which depression was diagnosed (n= 7,767), excluding visits with diagnoses for alternative SGA indications (schizophrenia, bipolar disorder, pervasive development disorder, psychotic depression, dementia) and those without an active antidepressant prescription.

Results:

8.6% of adult depression visits from 1999–2010 included an SGA. SGA use rates increased from 4.6% in 1999–2000 to 12.5% in 2009–2010, with an adjusted odds ratio (AOR) for time trend of 2.78 (95%CI, 1.84–4.20). The increase in SGA augmentation was broad based with no significant differences in time trends between demographic and clinical subgroups. For the most recent survey years (2005–2010, SGA use rates were higher in visits to psychiatrists than to other physicians (AOR 5.08; 95%CI, 2.96–8.73), visits covered by public than private insurance (AOR 3.20; 95%CI, 2.25–4.54), visits with diagnosed major depressive disorder than other depressive disorders (AOR 1.49; 95%CI, 1.08–2.06), and visits with diabetes, hyperlipidemia, or cardiovascular disease (AOR 2.13; 95%CI, 1.12–4.03), and lower in visits by patients >65 years than 18–44 years (AOR 0.51; 95%CI, 0.32–0.82) and visits that included psychotherapy (AOR 0.68; 95%CI, 0.47–0.96).

Conclusions:

Between 1999 and 2010, SGAs were increasingly accepted in the outpatient treatment of adult non-psychotic depression.

Antidepressant medications, particularly the newer selective serotonin reuptake inhibitors and serotonin-noradrenaline reuptake inhibitors, are the recommended first-line treatment for major depressive disorder (MDD).¹ However, only about half of patients suffering from depression are likely to respond to an initial antidepressant trial and only about 30–40% of patients achieve full remission.^{2, 3} Pharmacological strategies for patients with incomplete response to a trial of antidepressant monotherapy include switching to another antidepressant (within or between antidepressant drug classes),⁴ augmentation with a second antidepressant agent,⁵ and augmentation with non-antidepressants, including lithium, triiodothyronine, and, more recently, second-generation antipsychotics (SGAs). ^{1, 6} Several non-pharmacological approaches, including various psychotherapies, electroconvulsive treatment, phototherapy, sleep deprivation, repetitive trans-cranial magnetic stimulation, and acupuncture, have also been used to treat depression following incomplete response to antidepressant monotherapy.⁷

Over the last decade, there has been an accumulation of evidence describing the effectiveness of SGAs as an adjunctive treatment for depression, particularly when patients fail to respond to first-line treatment.^8–22 Currently, augmentation with SGAs is the only FDA approved pharmacological treatment alternative for patients with incomplete response to antidepressant monotherapy.^{6, 23} Meta-analyses based on sixteen trials comparing adjunctive SGAs to placebo in patients with treatment-resistant or inadequately responding MDD (n=3,580) have generated estimates of efficacy of an odds ratio (OR)=1.69; 95% confidence interval (CI), 1.46–1.95 and number-needed-to-treat (NNT)=9 for response and OR=2.00; 95%CI,1.58–2.72, NNT=9 for remission.⁶ Since 2007, the FDA has approved aripiprazole (2007), olanzapine (in fixed combination with fluoxetine, 2009), and quetiapine (2009) as adjunctive therapy to antidepressants for the treatment of MDD in adults who have an inadequate response to antidepressant treatment alone.

Despite having the strongest evidence base for adjunctive use in MDD patients with inadequate response to standard antidepressant therapy, SGAs are associated with increased risks for extrapyramidal side effects, tardive dyskinesia, weight gain, diabetes, and dyslipidemia.^24–29 The metabolic risks may be increased in antipsychotic-naïve patients,^{28, 30} a patient group that is likely overrepresented in individuals with non-psychotic depressive disorders. Although the risk for cardiometabolic adverse effects associated with SGA treatment can be mitigated by adequate monitoring and management,³¹ metabolic monitoring in SGA treated patients remains suboptimal in clinical practice.^{32, 33} Moreover, SGAs have also been associated with rare, but serious, adverse events in other clinical populations, including death, acute myocardial infarction, and stroke.^34–37

Surprisingly little is currently known about the prevalence and patterns of antipsychotic use in outpatients treated for depression. One recent study estimated a 1-year prevalence of SGAs of 20.6% in Veterans with MDD uncomplicated by schizophrenia, schizoaffective disorder, or bipolar disorder.³⁸ In another study, approximately one-third (32.9%) of adult Medicaid patients in one state who were diagnosed with MDD but not psychotic disorders received antipsychotics.³⁹ To date, however, no national estimates are available.

The present study aimed to determine nationally representative estimates of prevalence, patterns, and trends in the office-based use of SGAs for adjunctive treatment of adult, non-psychotic depression. A greater understanding of this emerging clinical practice may help to identify patient populations with high and low likelihoods of receiving SGA adjunctive treatment and put into perspective concerns over the safety profiles of SGAs for treatment-resistant depression.

METHOD

Data Source and Sample

We analyzed twelve years (1999–2010) of data from the National Ambulatory Medical Care Survey (NAMCS). NAMCS is an annual survey conducted by the National Center for Health Statistics that yields nationally representative estimates of visits to US physicians in office-based practice. Each year, the survey samples approximately 3,000 non-federally employed office-based physicians who are primarily engaged in direct patient care. The treating physician or a member of the physician’s staff provides information about patient demographics, current diagnoses, and prescriptions. The unit of observation is the physician-patient encounter or visit. Further details on survey and sampling design are available from the National Center for Health Statistics.⁴⁰

For the present study, we examined visits with a diagnosis for depression (ICD-9-CM 296.2, 296.3, 300.4, 311). Because the scope of the present study is limited to adults, visits of patients younger than age 18 were excluded. To restrict the study population to those most likely to receive antipsychotics for treatment-resistant depression rather than for another indication, we excluded all visits with a diagnoses for schizophrenia (ICD-9-CM 295), bipolar disorder (ICD-9-CM 296.00–296.16, 296.4–296.81, 296.89), pervasive development disorder (ICD-9-CM 299), major depression with psychotic features (ICD-9-CM 296.24, 296.34), or dementia (ICD-9-CM 290, 294.1, 331.0–2, 331.82, 331.9). Because the FDA has approved antipsychotics in combination with antidepressant treatment and to ensure that patients are actively treated for depression at the time of the visit, visits without a prescription for an antidepressant were also excluded from the primary analysis sample (Figure 1).

Figure 1: Study Sample and Details of Exclusions^a.

^aData from NAMCS.

Psychotropic Medications

We created a binary variable indicating whether a SGA was prescribed during the visit. In addition, we also recorded individual SGAs, specifically aripiprazole, olanzapine, quetiapine, risperidone, and a residual group of all other SGAs. NAMCS recorded up to 6 medications for each visit from 1999 to 2002. Starting in 2003, up to 8 medications were recorded. To make years comparable for our study, we limited the maximum number of medications to the first 6 listed in all years (all 8 medications were included in a sensitivity analysis to empirically assess the impact of the aforementioned restriction). First generation antipsychotics (FGAs) were included in a secondary analysis to examine to what extent increase in SGA utilization may reflect a substitution of FGAs in the study population.

Demographic and Clinical Characteristics

Demographic characteristics assessed included patient sex, age, ethnicity/race, and expected source of payment. The survey year was transformed by subtracting 1999 from the year and dividing the results by 11. Thus, the transformed value was 0 for the year 1999 and 1 for the year 2010.⁴¹

We also included several clinical variables associated with each visit, such as visit sequence (new or repeat visit), physician specialty (psychiatry vs. not); comorbid anxiety disorder (ICD-9 CM:309.81, 308.3, 300.01, 300.21, 300.22, 300.02, 300.3, 300.2, 300.23, 300.29, 293.84, 300.0, 300.09, 309.21, 313.0); type of depression diagnosis (MDD [296.2, 296.3] vs. non-MDD [300.4, 311]); presence of other mental disorders (ICD-9 CM 290–319 not listed above); comorbid diabetes, hyperlipidemia or cardiovascular disease (ICD-9 CM 250, 401–404, 410–416, 425–437, 440–447, 272.0–272.4, 272.7, 272.9); and whether psychotherapy was ordered or provided during the visit. All diagnoses were assigned based on any of the 3 diagnosis fields in the survey instrument.

Analysis

Initially, we estimated the proportion of visits with SGA augmentation among all qualifying depression visits with active antidepressant treatment over the 12 year study period (1999–2010). To examine time trends in SGA use, we built multivariate logistic regression models to estimate the association between the transformed survey year variable and SGA augmentation, controlling for all demographic and clinical characteristics described above. Estimated coefficients are presented as odds ratios for ease of interpretation. These models were fit for the study population overall and separately for each subpopulation to estimate overall and subgroup specific trends. Following the analyses of trends in SGA augmentation on the class level, we also examined trends in the use of individual SGAs. In secondary analyses, we also examined trends in use of 2 concurrent antidepressants, use of any antipsychotic (FGA or SGA), and SGA use in non-psychotic depression visits without active antidepressant use.

We then compared demographic and clinical characteristics of patients receiving and not receiving SGA augmentation and modeled the likelihood of receiving an antipsychotic prescription using similar multivariate logistic regression models. These analyses were limited to the six most recent survey years to assure that the estimates reflect recent practice patterns.

NAMCS is a multistage probability sample. Analyses were adjusted for visit weights, clustering, and stratification and reflect national estimates.⁴² Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC).

RESULTS

eTable 1 presents demographic and clinical characteristics of all qualifying depression visits with active antidepressant treatment during the 12 year study period (N=7,767). Overall, 8.6% (95%CI, 7.7%−9.6%) of these visits included a prescription for an adjunctive SGA. Weighted to the US population, this represents approximately 17.85 million office-based visits with an SGA prescription or an average of 1.49 million visits per year.

Trends in Antipsychotic Augmentation

The percentage of visits that included an adjunctive SGA increased from 4.6% (95%CI, 2.9%−6.3%) in 1999–2000 to 12.5% (95%CI, 9.7–15.3%) in 2009–2010, representing an annualized estimated 0.7 million visits in 1999–2000 and 2.2 million visits in 2009–2010. Adjusted for all covariates, this increase translates into a multivariate-adjusted odds ratio (AOR) for time-trend of 2.78; 95%CI, 1.84–4.20. Similar relative increases occurred in patients with and without MDD (Figure 2). Over the same period, treatment with two concurrent antidepressants increased from 15.4% in 1999–2000 to 21.4% in 2009–2010 (AOR 1.82; 95%CI, 1.42–2.30; data not shown). Overall antipsychotic use (FGA or SGA) increased from 6.3% (95%CI, 4.4%−8.1%) in 1999–2000 to 13.0% (95%CI, 10.2%−15.9%) in 2009–2010 with an adjusted odds ratio for time trend of 2.39 (95%CI, 1.60–3.58). The restriction to 6 medication fields for the period from 2003–2010 had minimal impact on the results. Inclusion of all 8 medication fields over the 8 year period increased the number of survey visits with adjunctive SGA use by 1.6% from 568 to 577 (unweighted counts).

Figure 2: Trends in Percentage of Antidepressant Treated Non-PsychoticDepression Visits with Antipsychotic Augmentation, United States, 1999 to 2010^a.

^aData from NAMCS. SGA denotes second generation antipsychotic medication.

The increase in SGA augmentation of antidepressant treated depression visits was broad based (Table 1). Significant increases occurred among visits by males and females, adult patients of all age groups, and patients treated by psychiatrists and non-psychiatrists. The only visit group which did not experience a substantial numerical increase in antipsychotic augmentation was patients of non-White or Hispanic backgrounds, who already had extensive use of antipsychotic augmentation at the start of the study period. The largest increases in antipsychotic augmentation were apparent in visits by patients with cardiovascular diagnoses, by patients >65 years, new patients, publicly insured patients, non-Hispanic Whites, and males. However, confidence intervals were wide and none of the differences across visit strata in time-trend for antipsychotic use reached statistical significance.

Table 1.

Antipsychotic Prescribing Trends in Office-Based Medical Visits of Antidepressant Treated Adult Non-Psychotic Depression, United States, 1999 to 2010

Variable	SGA Rx Rates			AOR for Time Trend (95% CI)a,b
	99-02	03-06	07-10
	n=2,506	n=2,686	n=2,575
Overall	6.1	8.4	11.2	2.78 (1.84 – 4.20)
Demographic Characteristics
Sex
Male	6.3	8.2	12.5	3.21 (1.78 – 5.83)
Female	6.0	8.6	10.6	2.60 (1.59 – 4.24)
Race/Ethnicity
White, non-Hispanic	5.4	8.0	11.1	3.22 (2.14 – 4.84)
Other	15.3*	13.4	12.3	1.12 (0.41 – 3.08)
Age (years)
18 – 44	5.9	7.3	10.1	3.06 (1.73 – 5.43)
45 – 64	7.0	10.5	10.5	2.17 (1.31 – 3.57)
> 65	4.6 *	6.4 *	12.2	4.60 (1.34 – 15.80)
Payment Source
Private Insurance	4.9	5.7	7.1	2.63 (1.31 – 5.25)
Public Insurance	8.6	14.7	18.4	3.44 (1.79 – 6.60)
Self-Pay, Otherc	7.0	9.9	9.4	2.82 (1.56 – 5.08)
Clinical Characteristics
Physician Patient Relationship
New Patient	6.1*	5.6*	12.3*	4.14 (0.88 – 19.38)
Established Patient	6.1	8.6	11.1	2.71 (1.77 – 4.17)
Physician Specialty
Other	2.3 *	4.6	4.7	2.85 (1.29 – 6.32)
Psychiatry	10.5	13.1	20.1	2.91 (1.87 – 4.55)
Major Depressive Disorder
No	3.8	5.7	7.1	3.16 (1.75 – 5.71)
Yes	11.0	14.2	20.0	2.67 (1.62 – 4.39)
Anxiety Disorder
No	5.4	7.5	10.3	2.97 (1.85 – 4.78)
Yes	11.3	15.3	17.7	2.23 (1.19 – 4.20)
Other Mental Health Condition
No	4.1	5.8	7.6	3.19 (1.76 – 5.77)
Yes	9.7	12.6	14.2	2.41 (1.48 – 3.91)
Diabetes, Hyperlipidemia or CVD
No	7.1	8.7	11.2	2.57 (1.71 – 3.88)
Yes	1.7 *	7.3 *	7.2 *	9.65 (1.76 – 53.07)
Psychotherapy Provided
No	5.1	7.4	8.8	3.07 (1.78 – 5.17)
Yes	9.6	11.0	15.5	2.52 (1.49 – 4.26)

Data from NAMCS. Visits with diagnoses for schizophrenia, bipolar disorder, pervasive development disorder, psychotic depression, or dementia were excluded

^aBased on pooled NAMCS data from 1999–2010.

^bEstimates from the logistic regression models in which the dependent variable is AP prescription. We report the coefficients of the transformed survey year variable, converted to odds ratios. The models were calculated separately for each subpopulation and control for all remaining covariates.

^cIncludes self pay, worker’s comp., no charge/charity and others. Rates are based on weighted calculations therefore reflect national estimates. SGA, denotes second generation antipsychotic medication.

^*Cells size is less than 30. SGA denotes second generation antipsychotic medication. CVD denotes cardiovascular disease

Marked changes in prescribing preferences for individual SGAs were observed over the study period (Figure 3). From 1999 to 2002, olanzapine (42%) was the most frequently used antipsychotic medication for augmentation, followed by risperidone (32%) and quetiapine (22%), with 0% representation of aripiprazole (which was first approved in the US in November 2002, at that time for schizophrenia). By 2007 to 2010, quetiapine (34%) and aripiprazole (30%) had largely displaced olanzapine (11%) and risperidone (18%). Changes in prescribing preference over the course of the study period were significant for aripiprazole (p<0.001), olanzapine (p<0.001), quetiapine (p<0.001), and risperidone (p=0.025).

Figure 3: Trends in Antipsychotic Augmentation by Agent in Office-Based Medical Visits for Antidepressant Treated Adult Non-Psychotic Depression, United States, 1999 to 2010^{a, b, c}.

^aData from NAMCS. ^bp-values reflect test of difference inproportions between the three time periods for each SGA. ^c2003–2006 vs. 2007–2010 only. SGA denotes second generation antipsychotic medication.

Of those diagnosed with depression, but not receiving current treatment with antidepressants (3,263 observations representing approximately 88 million visits), 6.1% received SGAs, an increase from 3.6% in 1999–2002 to 6.8% in 2007–2010 (AOR 3.36; 95%CI, 1.63–6.90; data not shown). Trends in prescribing preferences over the study period were similar to those observed in patients receiving SGAs as adjunctive treatment. From 1999 to 2002, olanzapine (44%) and risperidone (45%) were the dominant SGAs with quetiapine (11%) a distant third. In contrast, by 2007 to 2010, quetiapine (45%) was the most widely used SGA, followed by aripiprazole (20%), risperidone (19%), and olanzapine (11%) (data not shown).

Antipsychotic Augmentation Patterns

Among depression visits with active antidepressant treatment during 2006–2011 (n=3,920), SGA use rates were higher in visits to psychiatrists than other medical specialists, visits covered by public insurance or self-pay/other as compared with private insurance, and visits with a diagnosis of MDD, but lower in visits that provided or ordered psychotherapy, and patients >65 years (Table 2). Visits with comorbid cardiovascular diagnoses were more likely than those without these diagnoses to include a SGA. Visits with comorbid anxiety disorder or other comorbid mental disorder diagnoses were not significantly related to SGA use.

Table 2.

Patterns and Predictors of Antipsychotic Prescribing in Office-Based Medical Visits of Antidepressant Treated Adult Non-Psychotic Depression, United States, 2005 to 2010

Variable	SGA Augmentation n= 435	No SGA Augmentation n= 3,485	AOR for SGA Augmentation (95% CI)
Demographic Characteristics
Sex
Female	66.0	68.9	0.95 (0.73–1.25)
Male	34.0	31.1	1.0
Race/Ethnicity
White, non-Hispanic	87.5	91.3	0.68 (0.43–1.08)
Other	12.5	8.7	1.0
Age, years
18 – 44	40.6	38.6	1.0
45 – 64	45.9	45.3	0.87 (0.67–1.13)
> 65	13.6	16.8	0.51 (0.32–0.82)
Source of Payment
Private Insurance	32.7	49.6	1.0
Public Insurance	38.2	21.9	3.20 (2.25–4.54)
Self-Pay, Othera	29.1	28.5	1.57 (1.10–2.24)
Clinical Characteristics
Visit Type
New	5.8	6.2	1.0
Established	94.2	93.8	1.14 (0.61–2.12)
Physician Specialty
Other	25.3	60.3	1.0
Psychiatry	74.7	39.8	5.08 (2.96–8.73)
Major Depressive Disorder
No	42.4	70.7	1.0
Yes	57.6	29.3	1.49 (1.08–2.06)
Anxiety Disorder
No	77.1	88.4	1.0
Yes	22.9	11.7	1.21 (0.89–1.64)
Other Mental Health Condition
No	52.8	62.6	1.0
Yes	47.2	37.4	1.03 (0.79–1.35)
Diabetes, Hyperlipidemia or CVD
No	88.7	84.6	1.0
Yes	11.3	15.4	2.13 (1.12–4.03)
Psychotherapy Provided
No	61.5	72.4	1.0
Yes	38.5	27.6	0.68 (0.47–0.96)

Data from NAMCS. Visits with diagnoses for schizophrenia, bipolar disorder, pervasive development disorder, psychotic depression, or dementia were excluded

^aIncludes self pay, worker’s compensation, no charge/charity and others. SGA denotes second generation antipsychotic medication. CVD denotes cardiovascular disease.

DISCUSSION

Adjunctive SGA treatment for non-psychotic depression in US office-based medical practice increased more than 2.5-fold since publication of the first major positive trial of SGA augmentation for treatment-resistant depression in the early 2000s.²⁰ Increase in use was broadly similar across a range of demographic and clinical subgroups indicating a mainstreaming of SGA use for a large patient population that has historically had low rates of antipsychotic treatment. Treatment by a psychiatrist, public insurance, diagnosis of MDD and comorbid diabetes, hyperlipidemia or cardiovascular disease were among the strongest predictors of SGA augmentation. In more recent years (2007–2010), SGA augmentation rates approached or surpassed 20% in visits to psychiatrists, visits of patients diagnosed with MDD, and those covered by public insurance. It is somewhat reassuring that the highest use rates are observed in patients likely to have more severe depression, i.e., those with MDD and seen by a psychiatrist.

Elevated cardiovascular or metabolic risk was also associated with adjunctive SGA treatment. Given the well-established metabolic risks of SGAs,^26–30 and limited cardiometabolic monitoring in antipsychotic treated patients,³¹ this trend warrants closer scrutiny. It is possible that relatively high rates of SGA use in this population reflect a greater degree of depression severity and treatment resistance in patients with CV/metabolic conditions or associations between depression and insulin resistance, metabolic syndrome and cardiovascular illness.^43–50

A substantial increase in SGA use was also observed among visits by older depressed patients. Although older age was predictive of less antipsychotic augmentation compared to younger age, the substantial increase in SGA use in the older patient population should prompt a careful risk assessment, given the black box warning for increased all-cause mortality of SGAs in the elderly with dementia.⁵¹

Another potentially concerning observation was the considerable and growing use of SGAs for non-psychotic depression without concurrent antidepressant treatment, as approval for SGAs for this indication is limited to adjunctive treatment. We found no evidence that the increase in SGA augmentation was compensated by a corresponding decrease in antidepressant combination treatment, as antidepressant cotreatment also increased over time, although the evidence base for this combination treatment is weak.⁵ Despite the inability to distinguish between antidepressant combination therapy and cross-titration between two antidepressants using cross-sectional survey data, it is unlikely that the observed increase in visits with 2 concurrent antidepressants is compatible with a reduction in intended antidepressant combination treatment. If confirmed in longitudinal datasets, this finding suggests an overall intensification of pharmacological combination treatments in depression and may reflect greater clinical attention to incomplete treatment response. Our finding that inclusion of FGAs did not significantly alter the magnitude of the observed increase in antipsychotic use suggests that the increasing trend of SGA augmentation does not merely represent a substitution effect from first generation agents.

A substantially larger proportion of publicly than privately financed depression visits included a SGA. While the surveys lack the necessary clinical detail to evaluate the causes of this treatment pattern, it is possible that public insurance serve as a proxy for greater depression severity or greater treatment resistance. Differences in SGA utilization may also reflect disparities in how physicians approach depression treatment in these two patient populations or a greater degree of formulary restrictions or cost-sharing in private insurance programs.^{52, 53}

The observation that SGA augmentation was approximately 30% less likely in visits that included or ordered psychotherapy is of potential importance, as it may indicate that psychotherapy can reduce the need for SGA augmentation^{54, 55} and may provide a safer initial treatment alternative for patients who do not achieve remission of depressive symptoms with antidepressant monotherapy. Adjunctive cognitive behavioral therapy has recently been shown to potentially be an effective treatment alternative for reducing depressive symptoms in this population.⁵⁶ However, because the NAMCS does not sample mental health professionals, other than psychiatrists, it is not possible to assess the full extent to which patients receive combined psychotherapy and pharmacological treatments for depression. While an alternative explanation for this finding could be that visits with psychotherapy generally involve less severely ill patients or that physicians who provide psychotherapy are more reluctant to prescribe SGAs,⁵⁷ the fact that absolute SGA augmentation rates were higher in visits with psychotherapy militates against these explanations. Nonetheless, due to the aforementioned limitations, this finding should be viewed as strictly hypothesis generating.

In visits with SGA augmentation, we observed a clear trend towards treatment with aripiprazole and quetiapine, and away from olanzapine and, to a lesser degree, risperidone. This observation parallels trends in the broader use of SGAs in other clinical populations during this time period, particularly the declining use of olanzapine in response to increased awareness of its metabolic risks.⁵⁸ Risperidone, despite not having FDA approval for augmentation in depression, still remains relatively commonly used for non-psychotic depression.

Strengths and Limitations

Our study, the first to present nationally representative data on patterns and trends in SGA augmentation for non-psychotic depression, is subject to several limitations. First, because of the cross-sectional design and limited clinical detail of the survey data, it is not possible to examine the history of a patient’s depression treatment or follow patients over the course of their SGA trials. We thus cannot examine SGA augmentation rates in patients who meet criteria for treatment resistance, assess the appropriateness of SGA augmentation, characterize treatment dose or duration, or examine treatment outcomes. More detailed research is needed on dose and duration of SGA trials in usual practice to assess the potential metabolic risks at doses that are generally lower when given adjunctively for depression than when used in schizophrenia or bipolar mania.³⁸ Neither the RCTs for SGA augmentation in depression, nor clinical practice guidelines give much guidance to clinicians in terms of how long depressed patients should be maintained on SGAs if they respond to this adjuvant treatment.

Second, because the NAMCS samples visits rather than patients, some duplication of patients is possible. However, because each sampled physician practice contributes visits during a single, randomly assigned week in the survey year, such duplication is likely to have limited impact on the reported estimates. Third, diagnoses in the NAMCS are based on the independent judgment of the treating physician, rather than research diagnostic interviews.⁵⁹ Fourth, we restricted the number of medications per visit to 6 to assure that trend analyses would not be biased by changes in the structure of the NAMCS medication assessment. While this could potentially result in an under-ascertainment of SGA use during the 2003–2010 period, the impact of the change in the assessment structure on the number of visits with adjunctive SGA treatment was minimal as demonstrated in sensitivity analysis. Fifth, despite the relatively large sample sizes, the limited number of visits within certain patient groups resulted in wide confidence intervals for some estimates, particularly for subgroup-specific time trends. Sixth, several settings that provide ambulatory care for patients with depression, such as community mental health centers or hospital outpatient departments, are not within the scope of NAMCS.

Summary and Conclusion

Growing use of SGAs for non-psychotic depression should be seen in the context of the evidence base of adjunctive SGA treatment. Despite being the best-supported intervention for treatment-resistant depression, SGA augmentation has a moderate effect size with a number needed to treat of approximately 9 for both response and remission.⁶ Given the significant metabolic adverse effects associated with SGAs, careful examination of the long-term benefit-risk of adjunctive treatment of depression with SGAs is vital to guide clinical decision-making for the treatment of patients who have inadequate response to antidepressant treatment. Education and a focus on medical comorbidities in depressed patients are particularly relevant, as our data point towards frequent use of SGAs in higher cardiovascular risk subgroups.