Antipsychotic Medication Treatment Patterns in Adult Depression

Abstract

Objective:

To characterize the role of antipsychotic medications in the community treatment of adult depression.

Method:

We identified adults (18-64 years) with new episodes of depression treatment (ICD-9CM 296.2, 296.3, 300.4, or 311) in US national Medicaid data (2001-2010). Patients with alternative antipsychotic indications, such as schizophrenia or bipolar disorder, were excluded. Each patient was followed for at least one year to characterize antipsychotic and antidepressant treatment and emerging alternative antipsychotic indications. For patients without alternative indications through day 45 following start of antipsychotic treatment, antipsychotics were considered to be intended for treatment of depression. Among this group, we determined whether antipsychotic initiation was preceded by minimally adequate treatment with antidepressants, defined as active antidepressant treatment for ≥31 days prior to and including the day of antipsychotic initiation.

Results:

Within one year following onset, 14.0% of patients started an antipsychotic medication. 41.3% of antipsychotic initiators developed an antipsychotic indication other than depression through day 45 following antipsychotic initiation, most often bipolar disorder or depression with psychotic features. The remaining 58.7% of antipsychotic initiators presumably started antipsychotics for nonpsychotic depression. 71.3% of these patients did not have minimally adequate antidepressant treatment prior to starting the antipsychotic medication.

Conclusion:

Antipsychotic medications are used in approximately one in seven patients with a new episode of depression. For one in twelve patients, the antipsychotic was considered to be intended for nonpsychotic depression. Almost three-quarters of these patients did not receive minimally adequate treatment with antidepressants prior to antipsychotic initiation. This pattern suggests potentially inappropriate and premature initiation of a drug class with substantial adverse effects and medical risks.

Depressive disorders affect more than 15 million US adults and cause substantial emotional, physical, and economic burden on the affected individuals, their families, and their communities.^1–3 Approximately two thirds of depressed patients fail to achieve remission from a first antidepressant trial.⁴ For those who do not achieve remission, switching to another antidepressant is recommended, followed by a variety of augmentation strategies.⁵ Augmentation with second-generation antipsychotics (SGAs) is arguably the best supported and the only FDA-approved pharmacologic alternative for patients with major depressive disorder who have failed two or more trials of antidepressant monotherapy.^6,7 Since 2007, the FDA has approved olanzapine with fluoxetine, aripiprazole, quetiapine and brexpiprazole as adjunctive treatments to antidepressants for adults with major depressive disorder. Each year in the United States, roughly two million outpatient visits for adult depression include an SGA.⁸

The role of antipsychotics in the community treatment of depression remains poorly understood. SGAs are associated with increased risks for extrapyramidal side effects, tardive dyskinesia, weight gain, diabetes, dyslipidemia, and increased mortality.^9–14 A study of the US Veterans Affairs system reported that 43% of patients treated with SGAs for major depressive disorder received antipsychotics at doses at or above those recommended for schizophrenia, rather than at the lower doses recommended for depression.¹⁵ Because antipsychotics have significant adverse effects, their inappropriate use, including as monotherapy rather than as augmenting agents, their initiation before failure of response to antidepressant medications could be reasonably established, or their use above the dosing range indicated for treatment of depression, could pose quality of care concerns. However, none of these issues has been examined in community practice. The present study is the first detailed characterization of antipsychotic use patterns in relationship to antidepressant treatment in the community management of adult depression.

METHOD

We conducted a retrospective cohort study using US Medicaid Analytic Extract (MAX) data from 44 states from January 1, 2001 through December 31, 2010. All adults (aged 18-64 years) with a new episode of depression were included. A new episode of depression was defined by a diagnosis claim for depression (single in- or outpatient claim of ICD-9CM 296.2, 296.3, 300.4, or 311 in any position) following at least 180 days of uninterrupted Medicaid eligibility without (1) any diagnostic claims for depression, (2) any claims for antidepressant medications (eAppendix 1), and (3) any claims for exclusion diagnoses including schizophrenia [ICD-9-CM 295], bipolar disorder [296.00–296.16, 296.4–296.81, 296.89], major depressive disorder with psychotic features [296.24, 296.34], pervasive development disorder [299], dementia [290, 294.1, 331.0–2, 331.82, 331.9], delusional disorders [297], depressive-type psychosis [298.0], or cyclothymic disorder [301.13].

To allow for antidepressant medication initiations just prior to documentation of a depression diagnosis, we allowed for antidepressant claims up to 30 days prior to a new depression diagnosis, if the 180 day period prior to the first antidepressant claim met criteria (1)-(3). Patients were excluded if they met any of the exclusion conditions during the 180 day pre-index period, including any prescription claims for antipsychotics, any of the exclusion diagnoses, or eligibility for Medicare.

The start of follow-up (index date) was the initial diagnosis date or antidepressant claim, whichever came first. Only episodes with at least one year of continuous eligibility after the index date were included. For patients with more than one new depressive episode, only the first episode was considered for analysis. Baseline characteristics were ascertained over the 180 day pre-index period.

Each patient was followed for at least one year to characterize initiation and patterns of antidepressant and antipsychotic medication treatment. For initiators of antipsychotics (eAppendix 1), we examined whether antipsychotic indications other than depression developed between the index date through day 45 following antipsychotic initiation. Those without alternative indications were considered to have initiated the antipsychotic for the treatment of depression. Characteristics of the depression episode (presence of major depressive disorder and whether or not depression was recorded as a primary diagnosis) were ascertained from all diagnostic claims starting at the index date and including the following 30 days.

We determined the total duration of antipsychotic treatment by calculating the sum of days supply for antipsychotic medications during the year following antipsychotic initiation. We also determined the single prescription fill with the highest daily dose. Prescription claims with missing information for the days supply variable (present in 0.15% of patients who initiated antipsychotic treatment) were excluded from analysis.

To prevent undue influence of administrative errors, we further excluded patients with prescription claims with daily doses outside of a clinically reasonable range from the dose analyses (eAppendix 2; approximately 1.8% of antipsychotic treatment initiators). Among those considered to have initiated antipsychotic medication treatment for depression, we determined whether treatment initiation was preceded by minimally adequate treatment with antidepressants. Minimally adequate antidepressant treatment was defined conservatively as at least 31 days of antidepressant supply prior to antipsychotic medication initiation (to examine potentially premature antipsychotic initiation) and active antidepressant treatment on the day of antipsychotic initiation (to examine potential antipsychotic monotherapy). Among those who initiated an antipsychotic for depression, we used multivariate logistic regression to identify risk factors for antipsychotic initiation without minimally adequate prior antidepressant treatment. Further, we implemented sensitivity analyses limited to depressive episodes occurring after January 1, 2008 to examine whether FDA approval and labeling of aripiprazole for adjunctive treatment of depression in late 2007 impacted antipsychotic use patterns in patients with depression.

The study used de-identified data and was approved by the Rutgers Institutional Review Board. Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, North Carolina).

RESULTS

Between 2001 and 2009, 1,594,180 patients met the criteria for a new episode of depression. Mean age at onset was 36 years. Women accounted for more than three-quarters of patients with new depressive episodes (Table 1). Within one year of onset, antipsychotic medication was initiated in 14.0% of these patients. Antipsychotic initiators were more likely to be male, African-American, eligible for Medicaid through disability, and less likely to be in managed care. Initiators were also more likely to be diagnosed with major depressive disorder. In addition, they were more likely to have a primary diagnosis of depression, any diagnosis of substance use disorder, and greater recent use of emergency department visits, hospitalizations, and outpatient mental health visits. Altogether, 96.8% of antipsychotic initiations were for a second-generation antipsychotic. Mean time from the start of the depressive episode to antipsychotic initiation was 98.8 days (SD 106.0). Between the start of the depressive episode through day 45 of antipsychotic medication initiation, 41.3% of patients who initiated an antipsychotic had one or more new alternative indications for an antipsychotic recorded (Table 2). The most common alternative indications were major depressive disorder with psychotic features, bipolar disorder, and schizophrenia. The remaining patients who initiated antipsychotic treatment but did not develop an alternative antipsychotic indication, representing 58.7% of antipsychotic initiators or 8.2% of all patients with new depressive episodes, were presumed to have initiated antipsychotic treatment for nonpsychotic depression. No time trend was apparent across the study period years with initiation rates ranging from a low of 8.0% in 2001 to a high of 9.3% in 2004 (data not shown).

Table 1:

Characteristics of adult Medicaid beneficiaries with new depressive episodes by treatment with oral antipsychotic medications

	Total Episodes N=1,594,180 (100%)	APM Initiators N=223,430 (14.0%)		Non APM Initiators N=1,370,750 (86.0%)		P(Chi-Square)
		N	%	N	%
DEMOGRAPHICa
Sex
Male	351,141	70,172	31.4	280,969	20.5	<.0001
Female	1,243,018	153,257	68.6	1,089,761	79.5
Age (mean, SD)	35.9 (12.5)	36.8	12.0	35.8	12.5
Age						<.0001
18-24	366,984	43,452	19.5	323,532	23.6
25-34	436,432	56,920	25.5	379,512	27.7
35-44	363,539	58,826	26.3	304,713	22.2
45-54	271,765	45,373	20.3	226,392	16.5
55-64	155,460	18,859	8.4	136,601	10.0
Race Ethnicity						<.0001
White, non-Hispanic	847,429	109,183	48.9	738,246	53.9
African American, non-Hispanic	364,213	62,815	28.1	301,398	22.0
Hispanic	229,011	29,452	13.2	199,559	14.6
other	153,527	21,980	9.8	131,547	9.6
Medicaid Eligibility						<.0001
Disability	602,389	114,774	51.4	487,615	35.6
Low income	644,988	67,520	30.2	577,468	42.1
Other	346,803	41,136	18.4	305,667	22.3
Managed care (not FFS)	608,939	73,106	32.7	535,833	39.1	<.0001
CLINICAL&UTILIZATIONb
Primary Depression dxc	1,151,853	182,702	81.8	969,151	70.7	<.0001
Major Depressive Disorderc	515,680	116,569	52.2	399,111	29.1	<.0001
Anxiety	129,006	20,845	9.3	108,161	7.9	<.0001
Substance use disorder	179,800	36,817	16.5	142,983	10.4	<.0001
Personality disorders	5,827	1,552	0.7	4,275	0.3	<.0001
Self-harm	2,282	496	0.2	1,786	0.1	<.0001
Diabetes	130,165	17,016	7.6	113,149	8.3	<.0001
Cardiovascular Disease	276,154	38,711	17.3	237,443	17.3	.9662
Cerebrovascular Disease	26,636	3,981	1.8	22,655	1.7	.0001
Psychotherapy	42,795	6,942	3.1	35,853	2.2	<.0001
MH ER visitd						<.0001
No visit	1,544,285	210,808	94.4	1,333,477	97.3
1 to 2 visits	46,140	11,657	5.2	34,483	2.5
3 or more visits	3,755	965	0.4	2,790	0.2
MH hospital admissionsd	15,888	5,087	2.3	10,801	0.8	<.0001
MH outpatient visitsd						<.0001
No visit	1,312,994	167,919	75.2	1,145,075	83.5
1 to 2 visits	164266	30543	13.7	133723	133723	9.8
3 or more visits	116920	24968	11.2	91952	91952	6.7

^aat index date;

^bduring 180 baseline period including index date;

^cwithin Index date and index date+30 days;

^dprimary diagnosis.

APM: antipsychotic medication; FFS: fee-for-service; MH: mental health; ER: emergency room

Table 2:

Antipsychotic use in the year following a new depressive episode by antipsychotic indications diagnosed in the period from the new depressive episode to 45 days following the antipsychotic medication initiation

	Antipsychotic Medication Initiators N=223,430
	N	%
No alternative APM Indicationa	131,176	58.7
At least 1 alternative APM Indicationa	92,254	41.3
Schizophrenia	23,829	10.7
Bipolar Disorder	40,214	18.0
Depressedb	10,995	4.9
Manicc	6,289	2.8
Mixedd	6,337	2.8
Othere	26,994	12.1
MDD with psychotic features	41,333	18.5
Dementia	2,348	1.1
Otherf	3,613	1.6

^afrom the index date to 45 days following antipsychotic initiation;

^bICD-9-CM 296.5, 296.50-296.56;

^cICD9-CM 296.01-296.06, 296.4, 296.40-46;

^dICD-9CM 296.6, 296.61-296.66;

^eICD-9-CM 296.7, 296.80, 296.89;

^fpervasive development disorder, delusional disorder, depressive type psychosis, cyclothymic disorder;

APM: antipsychotic medication

Patients who initiated an antipsychotic for depression had a mean age of 37 years, were predominantly female, and were most commonly eligible for Medicaid due to disability. Within the first 30 days of onset, 80% had a primary depression diagnosis and slightly fewer than half were diagnosed with major depressive disorder. Patients who initiated an antipsychotic for depression generally resembled patients who initiated an antipsychotic for alternative indications (Table 3). Those who initiated antipsychotics for depression included a smaller proportion of African-Americans, a lower rate of Medicaid eligibility through disability, and a lower rate of major depressive disorder.

Table 3:

Characteristics of initiators of antipsychotic medications by presence of antipsychotic indication during the period from a new depressive episode to 45 days following an antipsychotic medication initiation

	No alternative APM indication over follow-up N=131,176		At least 1 alternative APM indication over follow-up N=92,254		P(Chi-Square)
	N	%	N	%
DEMOGRAPHICa
Sex					<.0001
Male	40,172	30.6	30,000	32.5
Female	91,004	69.4	62,253	67.5
Age (mean, SD)	37.0	12.0	36.7	12.0
Age					<.0001
18-24	24,984	19.1	18,468	20.0
25-34	33,554	25.6	23,366	25.3
35-44	34,644	26.4	24,182	26.2
45-54	26,620	20.3	18,753	20.3
55-64	11,374	8.7	7,485	8.1
Race Ethnicity					<.0001
White, non-Hispanic	68,150	52.0	41,033	44.5
African American, non-Hispanic	32,433	24.7	30,382	32.9
Hispanic	17,024	13.0	12,428	13.5
other	13,569	10.3	8,411	9.1
Medicaid Eligibility					<.0001
Disability	63,308	48.3	51,466	55.8
Low income	42,209	32.2	25,311	27.4
Other	25,659	19.6	15,477	16.8
Managed care (not FFS)	44,239	33.7	28,867	31.3	<.0001
CLINICAL&UTILIZATIONb
Primary Depression dxc	105,462	80.4	77,240	83.7	<.0001
Major Depressive Disorderc	60,073	45.8	56,496	61.2	<.0001
Anxiety	13,063	10.0	7,782	8.4	<.0001
Substance use disorder	22,044	16.8	14,773	16.0	<.0001
Personality disorders	882	0.7	670	0.7	.1311
Self-harm	245	0.2	251	0.3	<.0001
Diabetes	9,913	7.6	7,103	7.7	.2116
Cardiovascular Disease	23,021	17.6	15,690	17.0	.0009
Cerebrovascular Disease	2,399	1.8	1,582	1.7	.0449
Psychotherapy	4,279	3.3	2,663	2.9	<.0001
MH ER visitd
No visit	124,453	94.9	86,355	93.6	<.0001
1 to 2 visits	6,198	4.7	5,459	5.9
3 or more visits	525	0.4	440	0.5
MH hospital admissionsd					.0003
No admission	128,255	97.8	90,088	97.7
1+ admissions	2,921	2.2	2,166	2.4
MH outpatient visitsd					<.0001
No visit	98,169	74.8	69,750	75.6
1 to 2 visits	17,498	13.3	13,045	14.1
3 or more visits	15,509	11.8	9,459	10.3

^aat index date;

^bduring 180 baseline period including index date;

^cwithin Index date and index date+30 days;

^dprimary diagnosis.

APM: antipsychotic medication; FFS: fee-for-service; MH: mental health; ER: emergency room

Mean duration of antipsychotic medication treatment ranged from 96.8 days (ziprasidone) to 121.6 days (quetiapine) (Table 4). The proportion of antipsychotic-treated patients who received >30 days of treatment ranged from 60.5% (ziprasidone) to 70.6% (quetiapine) (eAppendix 3). The means of the highest observed doses for each of the individual antipsychotics are shown in Table 4. Doses were consistently lower in patients without alternative indications for antipsychotics than in patients with alternative indications. eAppendix 3 depicts respective doses at the 95^th percentile of the dose distribution. The highest approved antipsychotic dose for depression (eAppendix 4) was exceeded by 14.7% (quetiapine, 300mg), 26.6% (olanzapine, 12mg), and 17.3% (aripiprazole, 15mg) of patients without alternative antipsychotic indications, respectively (data not shown).

Table 4:

Duration and dose of antipsychotic treatment among patients with new depressive episodes with and without alternative antipsychotic indications

	Without alternative Indication		With alternative Indication
Durationa
	N	Days, mean (SD)	N	Days, mean (SD)
Any APM	131,176	134.0 (108.4)	92,254	147.7 (110.0)
Quetiapine	63,001	121.6 (103.8)	39,310	122.9 (102.8)
Olanzapine	28,818	108.1 (99.0)	21,841	110.6 (99.4)
Risperidone	33,897	108.7 (99.0)	30,979	110.8 (98.9)
Aripiprazole	20,612	103.8 (92.1)	19,178	105.1 (92.7)
Ziprasidone	9,423	96.8 (93.2)	10,972	96.5 (92.1)
Doseb
	N	mg; mean, (SD)	N	mg; mean, (SD)
Quetiapine	62,136	186.4 (184.1)	38,745	263.4 (223.8)
Olanzapine	28,332	10.3 (7.3)	21,371	13.3 (8.1)
Risperidone	33,298	2.0 (1.5)	30,295	2.7 (1.7)
Aripiprazole	20,245	11.5 (7.5)	18,779	14.0 (7.9)
Ziprasidone	9,249	95.3 (50.6)	10,723	109.1 (51.6)

^atotal days of medication possession based on days supply dispensed;

^bin mg, based on the prescription fill with the highest dose dispensed; calculated as [(#dispensed*mg/dose)/days supply]; dose analyses excluded 4,858 episodes [1.8% of all initiations; 1.6% of those without indication; 2.0% of those with indication] where the highest calculated single dose was outside a clinically reasonable dosing range (eAppendix 2).

APM: antipsychotic medication

Altogether, 71.3% of patients initiating antipsychotics for depression did not have minimally adequate antidepressant treatment prior to antipsychotic initiation. Of those, 53.1% did not have at least 31 days of prior antidepressant use, 7.2% did not have active antidepressant supply on the day of the antipsychotic initiation, and 39.7% did not have either. The risk for not having received minimally adequate antidepressant treatment prior to initiating an antipsychotic was particularly high for African-Americans, younger individuals, men, and those eligible for Medicaid due to disability. (Table 5). Results from sensitivity analyses limited to new depressive episodes on or after January 1, 2008 were close to identical to the results obtained from the full cohort (eAppendices 5–10).

Table 5:

Predictors of antipsychotic use without minimal antidepressant treatment among adults with new depressive episodes and no observed alternative clinical indications for antipsychotic treatment

	Minimal antidepressant treatmenta		No minimal antidepressant treatment		Adjusted Odds Ratiob (95% CI)
	N	%	N	%
TOTAL	37,591	28.7	93,585	71.3
DEMOGRAPHICc
Sex
Male	9,369	24.9	30,803	32.9	1.32 (1.28-1.36)
Female	28,222	75.1	62,782	67.1	Reference
Age
18-24	6,431	17.1	18,553	19.8	1.27 (1.20-1.34)
25-34	10,465	27.8	23,089	24.7	1.03 (0.97-1.08)
35-44	10,283	27.4	24,361	26.0	0.97 (0.92-1.02)
45-54	7,485	19.9	19,135	20.5	0.92 (0.88-0.97)
55-64	2,927	7.8	8,447	9.0	Reference
Race Ethnicity
White, non-Hispanic	21,693	57.7	46,457	49.6	Reference
African American, non-Hispanic	7,463	19.9	24,970	26.7	1.59 (1.54-1.64)
Hispanic	4,832	12.9	12,192	13.0	1.24 (1.19-1.28)
Other	3,603	9.6	9,966	10.7	1.25 (1.19-1.30)
Medicaid Eligibility
Disability	14,634	38.9	48,674	52.0	Reference
Low income	14,947	39.8	27,262	29.1	0.56 (0.54-0.57)
Other	8,010	21.3	17,649	18.9	0.62 (0.60-0.64)
Managed care (not FFS)	13,218	35.2	31,021	33.2	1.05 (1.02-1.08)
CLINICAL&UTILIZATIONd
Primary Depression Diagnosise	29,883	79.5	75,579	80.8	1.04 (1.01-1.08)
Major Depressive Disordere	16,589	44.1	43,484	46.5	1.08 (1.06-1.11)
Anxiety	3,887	10.3	9,176	9.8	0.96 (0.92-1.01)
Substance use disorder	5,754	15.3	16,290	17.4	1.06 (1.02-1.11)
Personality disorders	178	0.5	704	0.8	1.33 (1.12-1.58)
Self-harm	62	0.2	183	0.2	1.10 (0.82-1.48)
Diabetes	2,850	7.6	7,063	7.6	0.95 (0.91-1.00)
Cardiovascular Disease	6,726	17.9	16,295	17.4	0.87 (0.84-0.90)
Cerebrovascular Disease	663	1.8	1,736	1.9	0.95 (0.86-1.04)
Psychotherapy	1,206	3.2	3,073	3.3	0.97 (0.90-1.04)
MH ER visitf
No visit	36,094	96.0	88,359	94.4	Reference
1 to 2 visits	1,388	3.7	4,810	5.1	1.25 (1.16-1.34)
3 or more visits	109	0.3	416	0.4	1.26 (1.02-1.57)
MH hospital admissionsf
No admission	37,005	98.4	91,250	97.5	Reference
1+ admissions	586	1.6	2,335	2.5	1.31 (1.19-1.44)
MH outpatient visitsf
No visit	28,900	76.9	69,269	74.0	Reference
1 to 2 visits	4,731	12.6	12,767	13.6	1.04 (1.00-1.09)
3 or more visits	3,960	10.5	11,549	12.3	1.05 (1.00-1.11)

^adefined as AP initiation preceded by ≥31 days of AD supply AND active AD supply on the AP index date;

^bin addition to all variables shown in Table 5, the model also adjusted for index year;

^cat index date;

^dduring the 180 day baseline period including index date;

^ewithin Index date and index date+30 days;

^fprimary diagnosis;.

APM: antipsychotic medication; FFS: fee-for-service; MH: mental health; ER: emergency room

DISCUSSION

Antipsychotics are used in the treatment of approximately one in seven adult patients who develop new episodes of depression that are not complicated by other psychiatric conditions for which antipsychotic medications are indicated. Although antipsychotic dosing was generally in line with clinical recommendations, ¹⁶ nearly three-quarters of antipsychotic-treated patients did not receive even minimally adequate antidepressant treatment prior to antipsychotic initiation. This pattern raises the possibility of premature or inappropriate initiation of antipsychotic medications.

Approximately 14% of patients with new depressive episodes initiated an antipsychotic medication within one year of onset. In over one-third of these patients, new diagnoses of alternative conditions with indications for antipsychotics were recorded after onset of the depressive episode, suggesting that for these patients, depression was not the likely primary target for antipsychotic treatment. In the remaining patients, approximately 8% of those with new episodes of depression, the antipsychotic was considered to be intended for the treatment of depression.

One previous study has reported a somewhat higher rate of antipsychotic treatment by patients with depression. In a study of veterans with major depressive disorder without comorbid schizophrenia, schizoaffective disorder or bipolar disorder, approximately one in five (20.6%) patients was treated with an antipsychotic medication.¹⁵ The male predominance of the veteran patient population may help to account for their relatively higher rate of antipsychotic treatment. In addition, over 50% of antipsychotic medication use in the VA study was for patients with comorbid PTSD, a condition for which antipsychotic treatment, despite limited evidence of effectiveness, is not uncommon.^17,18 Demographic and clinical predictors of antipsychotic use were largely consistent between the present study and the VA study, with younger patients, men, patients with more psychiatric comorbidities, and patients with more intensive mental health services use being more likely to receive antipsychotic medications.

The present estimate of patients with a new episode of depression without evidence of alternative antipsychotic indications who initiated an antipsychotic (8.2%) is similar to a report from a national survey of office-based practice that 8.6% of adult depression visits without alternative indications included a SGA.⁸ In the national survey, publicly insured depressed patients, such as those in the present study, were significantly more likely than their privately insured counterparts to receive an antipsychotic, and antipsychotic use rates significantly increased from 1999-2002 to 2007-2010. However, due to its sampling frame and cross-sectional single visit design, the office-based survey did not provide a rigorous means of excluding patients who had alternative psychiatric indications for receiving antipsychotic treatment.

The pattern of antipsychotic treatment in adult depression reflects several markers of greater illness severity. Higher rates of major depressive disorder, a primary diagnosis for depression, and a history of greater mental health service use all point towards targeted use in patients with greater severity of illness. In this context, the average duration of antipsychotic treatment in patients with depression, approximately 3-4 months, was almost identical to the duration of antipsychotic treatment of depressed patients who presumably initiated an antipsychotic for other indications (Table 4, eAppendix 3).

Antipsychotic dosing was largely consistent with clinical guidelines for adjunctive treatment for major depressive disorder (eAppendix 4).^16,19 With the exception of quetiapine for which the mean dose fell significantly below the lowest recommended dose for schizophrenia, the mean of the highest prescribed antipsychotic dose among episodes was generally at the lower end of the recommended dosing range for schizophrenia. The highest labeled dose for depression was exceeded in only 15% (quetiapine) to 27% (olanzapine) of patients initiating an antipsychotic for depression and maximum doses almost never exceeded labeled schizophrenia doses. This finding is reassuring when compared with the higher degree of dosing above clinical recommendations reported in the VA study (ranging from 32% for quetiapine to 91% for ziprasidone).¹⁵

A clear majority of patients initiating an antipsychotic for depression failed to meet minimal requirements for appropriate prior antidepressant treatment, particularly regarding adequate prior duration of antidepressant therapy. This observation extends longstanding concerns over poor antidepressant adherence in community practice to adjunctive antipsychotic treatment that should be initiated only after insufficient response to antidepressant monotherapy.²⁰ Antipsychotic use that occurred without concurrent antidepressant use or before a >30 day antidepressant trial was particularly common among African-Americans, younger individuals, and men. In light of the risk of adverse effects associated with antipsychotics,^9–14 careful attention should be given to ensuring that antipsychotics are generally reserved for situations in which treatment failure has been established with first line antidepressant treatment. Despite limited evidence suggesting potential effectiveness of quetiapine monotherapy in patients with unipolar major depression,²¹ such data are lacking for any other antipsychotic, and antipsychotic monotherapy for major depressive disorder is neither approved by the FDA nor endorsed by any guidelines.^5,22 Of note, while we considered any duration of prior antidepressant use exceeding 31 days as minimally adequate and didn’t impose a requirement regarding dosing of the antidepressant, guidelines typically recommend at 6-12 weeks of antidepressant treatment titrated, as tolerated, to standard maximal doses.¹⁹ Similarly, our definition did not require trials of more than one antidepressant prior to augmentation with an antipsychotic though two trials are generally recommended.¹⁹ The operational definition of minimally appropriate treatment used for this study was thus quite conservative because a stricter definition of minimally adequate treatment would have been failed by an even larger proportion of individuals initiating an antipsychotic for depression.

Our study spanned the period from 2002 to 2010 because in clinical practice antipsychotics had been used for the treatment of depression prior to receiving FDA approval for this indication.^8,15 Replication of the analyses limited to depressive episodes occurring after January, 1 2008 resulted in findings almost identical to those obtained from the full study period (eAppendices 5–10). This suggests that use patterns were largely unaffected by FDA approval and labeling.

As a claims-based observational study, our work has several particular strengths and limitations. Using 10 years of near national Medicaid data, it is the first longitudinal study of a large and diverse real-world population that examines the role of antipsychotic medications for depression in clinical practice. A focus on patients with new episodes of depression facilitates examination of antipsychotic use patterns that is not possible in cross-sectional studies or studies of patients with ongoing depression. However, inferring treatment intent of antipsychotic initiation solely from observed diagnostic claims for depression and alternative conditions with indications for antipsychotic initiation may result in misclassification. Nevertheless, our exclusion of patients diagnosed with many conditions for which antipsychotics are frequently used in practice helps to narrow the focus to patients in which it is likely that antipsychotics were used to treat depression.

The claims data do not provide information on drug samples. This might have resulted in misclassification of the initiation dates for both antidepressant and antipsychotic treatment. For antidepressant treatment, this would have led to an artificial reduction in the measured duration of treatment and could therefore have led to an underestimation of the proportion of individuals who received minimally adequate duration of antidepressant therapy. However, as discussed above, our definition of minimally adequate antidepressant treatment (≥31 days) was quite conservative and would be expected to be robust to underestimation of the duration of antidepressant treatment due to samples. In addition, our analyses underestimate the duration of antipsychotic treatment within the year following initiation because the inclusion criteria only assure 365 days of Medicaid eligibility following the date of onset of the new depressive episode, not following antipsychotic treatment initiation. Sensitivity analysis limited to patients with 365 days of eligibility following antipsychotic initiation (including 89% of antipsychotic initiators) showed an increase in follow-up ranging from 2.2% (ziprasidone) to 3.0% (quetiapine) (data not shown). Lastly, the study was limited to Medicaid-insured patients, and the described patterns may present differently in other patient populations.

Although our study provides reassurance concerning the overall rate and dosing patterns of antipsychotic medications for depression, the finding that nearly three-quarters of antipsychotic-treated patients did not receive even minimally adequate treatment with antidepressants prior to antipsychotic initiation is of concern. Minimizing premature or inappropriate initiation of medications with substantial adverse effects and medical risks should be a priority for physicians who are engaged in the pharmacological management of adults with depression.