Variation in Psychotropic Medication Treatment of Adults Diagnosed with Schizophrenia in the United States

Abstract

Objective

The authors examined practice variation in prescription of psychotropic medications common in treatment of schizophrenia-spectrum disorders in a national U.S. sample.

Methods

Data from the 2011–2012 Medicaid Analytic eXtract was examined for state variation and demographic predictors of psychotropic prescription among beneficiaries diagnosed with schizophrenia-spectrum disorders (N=357,914). Adjusted proportions of percentages of beneficiaries in each state filling prescriptions of at least 15 days of any antipsychotic, clozapine, antidepressant, benzodiazepine, mood stabilizer, and any long-acting injectable antipsychotic (LAI) were determined, adjusting for demographic and clinical covariates. Multivariate regression models of clinical and demographic factors predicting prescription were also conducted.

Results

Prescribing patterns for all types of psychotropics varied significantly between states. Clozapine and LAIs had the most dramatic differences between states and by demographic characteristics. LAIs had a range of adjusted proportions from 4% (Colorado) to 22% (Rhode Island), a 5-fold difference, and clozapine ranged from 1–11%, more than a 10-fold difference. Non-Hispanic Blacks and people of other races were more likely than Non-Hispanic Whites to fill prescriptions for LAIs, but Non-Hispanic Whites were more likely than other races and ethnicities to fill prescriptions for clozapine.

Conclusion

Considerable variation in prescribing patterns of LAIs and clozapine by race and ethnicity, as well as across states, suggest uneven quality of care for individuals diagnosed with schizophrenia-spectrum disorders in the U.S. A better understanding of these variations and possible policy changes could support efforts to improve treatment for this vulnerable population.

Several classes of psychotropic medications are commonly prescribed to people with schizophrenia, but evidence for effectiveness is strong only for antipsychotics.^{1, 2} Among antipsychotics, clozapine and long-acting injectable formulations have unique roles. Clozapine is the established medication of choice for treatment-resistant schizophrenia and is the only drug approved for this indication.³ Long-acting injectable antipsychotics (LAIs) are important options to help ensure adherence,⁴ but there is controversy about their benefits and how they are used.^{5, 6} There is increasing evidence for an important role for antidepressants,^{1, 7} while mood stabilizers and benzodiazepines have limited evidence for effectiveness in schizophrenia.^8–12

Treatment variation refers to differences in treatment for persons with the same condition that are not attributable to differences in clinical presentation. Variation in treatment patterns is thought to reflect professional uncertainty—a lack of consensus or evidence regarding the best treatment choice for a condition.¹³ Other factors contributing to variation are uncertainty of sequential timing of treatments; differences in availability of providers and services; and provider preferences.^{13, 14} Regulations, variations in health insurance benefits, pharmacy benefit management, and consumer preferences are other possible contributors.^{13, 14} In some cases, treatment variations may reflect inequities based on socioeconomic status, race, or ethnicity.

In an extant review of the literature, variations in prescription of psychotropic medications for schizophrenia have been documented despite treatment guidelines.¹⁵ These studies provide evidence of variation related to geographic location,¹⁶ specific states,^17–19 specific medications (clozapine versus other antipsychotics),²⁰ medication classes (antidepressants versus anxiolytics), or demographic characteristics (African Americans versus non-Hispanic whites).²¹ Prior studies, however, have not provided a national portrait of variation across the major psychotropic medication classes used to treat schizophrenia.

This national study of variations in psychotropic medication prescriptions for adults diagnosed with schizophrenia-spectrum disorders extends these prior investigations. In addition to antipsychotic medications, we included antidepressants, mood stabilizers and benzodiazepines because of clinical interest regarding the appropriate role of these classes in the management of schizophrenia. ^{2, 7, 22, 23} We analyzed Medicaid claims data from 47 states and the District of Columbia with two complementary goals: 1) to identify which beneficiary-specific factors predict prescription of these medications and 2) to identify whether practice variation in psychotropic medication prescription rates between states remain after controlling for beneficiary clinical, demographic, and service use factors.

Methods

Source of Data

Data come from the 2011–2012 national Medicaid Analytic eXtract (MAX) Inpatient, Other Services, Prescription Drug and Personal Summary datasets, excluding Hawaii, Idaho and Maine due to missing data. Dual-covered individuals (both Medicaid and Medicare) were excluded, but we included individuals covered by Medicaid fee-for-service (FFS) and managed care (MC) plans. The New York State Psychiatric Institute Institutional Review Board approved this study.

Cohort Construction

The final analytic sample was limited to beneficiaries with one inpatient or two outpatient diagnoses of schizophrenia spectrum disorders (ICD-9-CM 295.xx), hereafter schizophrenia, using the principal and second listed diagnosis codes in the 2011 Inpatient and Other Services datasets, a previously validated method.²⁴ To improve interpretation of mood stabilizer prescription, often used as anti-seizure medications, beneficiaries with 2011 principal and second listed diagnoses of epileptic disorders (ICD-9-CM 345.xx) were excluded. The sample was limited to adults aged 18 to 64 years, who filled at least a 15-day prescription of at least one oral psychotropic medication or prescription of an LAI medication, and were eligible for Medicaid for all months of 2012 (N=357,914).

Variable Construction

The outcomes of interest were prescribing patterns for six types of psychotropic medications (Online Supplement Table 1) by state in 2012: any antipsychotics, clozapine, LAIs, antidepressants, mood stabilizers, and benzodiazepines. Filled prescriptions were defined as at least a 15-day supply of any oral antipsychotic, antidepressant, benzodiazepine, or mood stabilizer; or any LAI. To examine practice variation in psychotropic medication prescription by state, indicator variables of each state were the primary predictor of interest in these analyses.

Demographic, clinical, and service use covariates

Our sample included Medicaid FFS and MC plans that could affect prescribing patterns, so we included a covariate indicating Financing Arrangement (FFS; MC; Other). Demographic and clinical covariates included age, sex, race/ethnicity (Non-Hispanic Black; Non-Hispanic White; Hispanic; an Other Race category that includes: American Indian/Alaska Native, Asian/Pacific Islander, Native Hawaiian/Other Pacific Islander, More than one race, and unknown), prior year (2011) schizoaffective (defined as over 50% of 295 diagnosis codes as 295.7x) or schizophreniform (defined as over 50% of 295 diagnosis codes as 295.4x) subtypes, prior year (2011) psychiatric comorbidities (mental retardation, substance use disorders, depression, anxiety disorders, self-harm; Online Supplement Table 2), prior year (2011) selected physical comorbidities that may affect antipsychotic or other treatment choices (diabetes, cardiovascular disease (excluding essential hypertension diagnosis code 401.xx),^25–27 white blood cell disease, intracranial injury,²⁸ hyperlipidemia, drug induced dyskinesia,²⁹ HIV³⁰; Online Supplement Table 2), and prior year (2011) service use (count of psychiatric outpatient visits, any mental health emergency department visits, count of inpatient psychiatric admissions; Online Supplement Table 2).

Analytic Strategy

Medicaid programs differ between states in demographic characteristics of their covered populations and psychiatric and other medical services they cover. We examined practice variation in psychotropic medication prescription by state, controlling for individual level demographic, clinical, service use characteristics, and financing arrangement. Logistic regressions with filled prescriptions of each psychotropic class (yes/no) as the outcome were predicted by dummy variables for each state and all other covariates. We included a Bonferroni adjustment for multiple comparisons of these 6 models with a resulting p-value of <0.008 (0.05/6), or a confidence interval of 99.2% (hereafter rounded to 99%). SAS 9.4 software³¹ was used to conduct these statistical analyses. Odds ratios and 99% confidence intervals assessed the association of each demographic and clinical covariate. Adjusted proportions of prescriptions for each medication were calculated for each state from the logistic regressions (using the SAS Proc GENMOD with a binomial distribution and logit link) as the overall population mean values adjusting for all other covariates. Tests of state variation in filled prescriptions was done using the χ2 likelihood ratio test from the fully adjusted logistic regression models.

Results

Sample characteristics

Table 1 shows characteristics of the study population providing context for our analyses of state variation in prescribing patterns. The sample was 49% female, 35% Non-Hispanic Black, 39% Non-Hispanic White, and 12% Hispanic. Beneficiaries with schizoaffective disorder made up 33% of the sample while only 0.6% of beneficiaries had schizophreniform disorder. Past year depression (33%) was the most common psychiatric comorbidity Cardiovascular disease (35%), hyperlipidemia (24%), and diabetes mellitus (21%) were the most common physical comorbidities. Almost 61% of beneficiaries had 9 or fewer outpatient visits during 2011, 26% had a mental health emergency department visit, and 81% had no past-year psychiatric inpatient admissions. Almost 70% of beneficiaries had fee for service and 30% had managed care financing arrangements. Only 30% of beneficiaries were prescribed psychotropic monotherapy, while the highest rates of polypharmacy prescription were two types of medications (38%), supporting the need to study other psychotropics in addition to antipsychotic medications commonly prescribed for schizophrenia treatment.

Table 1.

Demographic and service use characteristics of national sample of persons with schizophrenia-spectrum diagnoses.^*

	Full Sample
Demographics	Mean	SD
Age (M, SD)	43.21	12.61
	N	%
N	357,914
Female (N, %)	174,272	48.69
Race/Ethnicity (N, %)
White, Non-Hispanic	141,261	39.47
Black, Non-Hispanic	125,256	35.00
Hispanic/Latino	41,262	11.53
Other	50,135	14.00
Schizophrenia Subtypes
Schizoaffective (N, %)	118,930	33.23
Schizophreniform (N, %)	2,007	0.56
Psychiatric Comorbidities (2011)
Substance Use Disorder (N, %)	54,895	15.34
Depression (N, %)	119,738	33.45
Anxiety (N, %)	55,468	15.50
Deliberate Self Harm (N, %)	2,796	0.78
Physical Comorbidities (2011)
Diabetes (N, %)	75,833	21.19
Serious Cardiovascular Disease (N, %)	126,636	35.38
Hyperlipidemia (N, %)	86,153	24.07
Financing Arrangement (N, %)
Fee For Service (FFS)	246,640	68.91
Managed Care (MC)	109,017	30.46
Other	2,257	0.63
Outpatient Visits for Schizophrenia (2011) (N, %)
0–9 visits	217,252	60.70
10–29 visits	64,774	18.10
30–49 visits	24,852	6.94
50+ visits	51,036	14.26
Mental Health Emergency Dept. Visit (2011) (N, %)	92,552	25.86
Inpatient Admission for Schizophrenia (2011) (N, %)
0 visits	288,876	80.71
1 visit	34,453	9.63
2 visits	15,819	4.42
3 visits	6,297	1.76
4+ visits	12,469	3.48
Number of psychotropic types taken in year (N, %)
Monotherapy	107,538	30.05
Two types of psychotropics	137,178	38.33
Three types of psychotropics	87,440	24.43
Four types of psychotropics	24,420	6.82
Five types of psychotropics	1,338	0.37
Psychotropic Medication Type**
Any Antipsychotics (N, %)	306,103	85.52
LAI Antipsychotics (N, %)	51,427	14.37
Clozapine (N, %)	15,679	4.38
Antidepressants (N, %)	207,531	57.98
Mood Stabilizers (N, %)	99,758	27.87
Benzodiazepines (N, %)	97,468	27.23

^*Data from Medicaid Analytic eXtract (MAX) years 2011–2012. Missing data from Hawaii, Idaho, and Maine

^**Not mutually exclusive categories

LAI: Long-Acting Injectable; Other Race category includes: American Indian/Alaska Native, Asian/Pacific Islander, Native Hawaiian/Other Pacific Islander, More than one race, Unknown; Other Financing Arrangement includes: Capitation Plan, Lump sum payment to provider, Supplemental payment above capitation fee, Unknown, Missing

Demographic, clinical and service use predictors of psychotropic medication prescriptions

Table 2 presents results from the logistic regression models predicting filled prescriptions of psychotropic classes and formulations, mutually adjusting for all covariates including state.

Table 2.

Adjusted Odds of prescription of psychotropic medications for persons with schizophrenia-spectrum diagnoses. Selected results from the full model, adjusted for all covariates.

	Antipsychotics		LAI		Clozapine		Antidepressants		Benzodiazepines		Mood Stabilizers
Demographics	AOR	99% CI**	AOR	99% CI	AOR	99% CI	AOR	99% CI	AOR	99% CI	AOR	99% CI
Age (continuous)	0.99*	0.99–0.99	0.99*	0.99–0.99	0.98*	0.98–0.98	1.01*	1.00–1.01	1.01*	1.01–1.01	0.98*	0.98–0.98
Sex (ref=Male)	0.81*	0.79–0.83	0.78*	0.76–0.80	0.82*	0.78–0.86	1.47*	1.44–1.50	1.38*	1.35–1.41	1.00	0.98–1.02
Race (ref=White Non-Hispanic)
Black, Non-Hispanic	1.16*	1.12–1.19	1.39*	1.35–1.44	0.40*	0.38–0.42	0.71*	0.69–0.73	0.51*	0.49–0.52	0.69*	0.67–0.71
Asian/Pacific Islander	1.26*	1.13–1.41	1.10	0.99–1.23	0.83*	0.72–0.96	0.60*	0.56–0.65	0.63*	0.58–0.69	0.72*	0.66–0.78
Hispanic	1.10*	1.03–1.17	1.26*	1.19–1.34	0.63*	0.57–0.69	0.95*	0.92–0.99	1.00	0.95–1.04	0.74*	0.71–0.78
Native Hawaiian/Other Pacific Islander	1.55*	1.30–1.84	1.26*	1.11–1.44	0.80*	0.66–0.97	0.69*	0.63–0.75	0.63*	0.57–0.70	0.72*	0.65–0.80
Schizophrenia Subtypes (ref=Schizophrenia)
Schizoaffective	1.00	0.98–1.03	0.77*	0.75–0.80	0.91*	0.87–0.96	1.35*	1.32–1.38	1.14*	1.11–1.16	1.60*	1.57–1.64
Schizophreniform	1.15	0.96–1.37	0.91	0.76–1.09	0.77	0.56–1.07	0.93	0.82–1.05	1.04	0.90–1.20	1.00	0.87–1.15
Psychiatric Comorbidities
Substance Use Disorder (ref=No)	0.82*	0.79–0.85	0.83*	0.80–0.87	0.45*	0.41–0.50	1.41*	1.37–1.45	1.09*	1.06–1.13	0.88*	0.86–0.91
Depression (ref=No)	0.72*	0.70–0.74	0.65*	0.63–0.67	0.56*	0.53–0.60	2.06*	2.01–2.10	1.17*	1.14–1.20	0.92*	0.90–0.94
Anxiety (ref=No)	0.69*	0.67–0.72	0.66*	0.63–0.69	0.65*	0.60–0.70	1.41*	1.37–1.45	2.33*	2.27–2.40	0.88*	0.86–0.91
Deliberate Self Harm (ref=No)	1.03	0.89–1.19	1.02	0.88–1.19	0.92	0.66–1.29	1.46*	1.29–1.66	1.07	0.95–1.19	1.05	0.94–1.17
Physical Comorbidities
Diabetes (ref=No)	1.08*	1.04–1.12	1.06*	1.03–1.10	1.48*	1.40–1.57	1.10*	1.07–1.13	1.08*	1.05–1.11	1.12*	1.09–1.15
Serious Cardiovascular Disease (ref=No)	0.86*	0.83–0.89	0.84*	0.81–0.86	0.98*	0.93–1.03	1.19*	1.16–1.21	1.30*	1.27–1.33	1.06*	1.04–1.09
Hyperlipidemia (ref=No)	1.13*	1.09–1.17	0.91*	0.88–0.94	1.44*	1.37–1.52	1.07*	1.05–1.10	1.04*	1.01–1.07	1.06*	1.03–1.08
Financing Arrangement (ref=FFS)
Managed Care (MC)	0.46*	0.44–0.48	0.81*	0.77–0.84	0.77*	0.72–0.83	1.24*	1.20–1.27	0.99	0.96–1.02	0.89*	0.86–0.92
Other	0.23*	0.20–0.26	0.61*	0.50–0.74	0.16*	0.08–0.31	0.24*	0.21–0.27	0.19*	0.14–0.24	0.20*	0.16–0.25
Psychiatric Service Use
Outpatient Schizophrenia Visits (ref=0–9)
10–29 visits	1.94*	1.86–2.02	2.65*	2.55–2.74	2.36*	2.22–2.51	0.78*	0.76–0.80	0.86*	0.84–0.89	0.89*	0.86–0.92
30–49 visits	2.27	2.13–2.42	4.25*	4.06–4.45	3.10*	2.86–3.36	0.75*	0.72–0.78	0.83*	0.79–0.87	0.97	0.93–1.01
50+ visits	2.75	2.61–2.88	5.32*	5.13–5.52	4.88*	4.59–5.18	0.73*	0.71–0.75	0.82*	0.79–0.85	1.15*	1.12–1.19
Mental Health ED Visit (ref=No)	1.18*	1.14–1.22	1.44*	1.39–1.49	0.77*	0.72–0.82	0.95*	0.92–0.97	1.12*	1.09–1.15	1.36*	1.32–1.40
Inpatient Schizophrenia Admission (ref=0)
1 visit	1.13*	1.08–1.19	1.36*	1.30–1.43	0.92	0.84–1.00	0.91*	0.88–0.95	1.04*	1.00–1.08	1.30*	1.26–1.35
2 visits	1.32*	1.23–1.42	1.58*	1.48–1.69	1.01	0.90–1.15	0.83*	0.79–0.87	1.04	0.99–1.10	1.45*	1.38–1.52
3 visits	1.53*	1.36–1.71	1.80*	1.64–1.98	1.12	0.93–1.34	0.83*	0.77–0.90	1.01	0.93–1.09	1.71*	1.59–1.84
4+ visits	1.51*	1.39–1.64	1.80*	1.67–1.94	1.22*	1.06–1.41	0.90*	0.85–0.96	1.08*	1.01–1.15	1.75*	1.65–1.85

^*p<0.008

^**Confidence Interval is 99.2% to adjust for multiple comparisons.

ED: Emergency Department; FFS: Fee For Service, Other Financing Arrangement includes: Capitation Plan, Lump sum payment to provider, Supplemental payment above capitation fee, Unknown, Missing; LAI: Long-acting injectable antipsychotic; All covariates included State, Age, Sex, Race, Substance Use Disorder, Depression, Anxiety, Self Harm, HIV, Financing Arrangement, Schizophrenia spectrum disorder, Diabetes, Cardiovascular disease, White Blood Cell disease, Mental Retardation, Intracranial Injury, Hyperlipidemia, Drug Induced Dyskinesia, Outpatient visit count, Mental Health Emergency Department visit, Inpatient admission count

The most notable differences were seen by race and sex. Any antipsychotics (AP) and LAI prescriptions were more likely to be filled for Black Non-Hispanics (adjusted odds ratio (AOR) for AP=1.16 (99% CI: 1.12–1.19); LAI=1.39 (99% CI: 1.35–1.44)), Hispanics (AOR AP=1.10 (99% CI: 1.03–1.17); LAI=1.26 (99% CI: 1.19–1.34)), and Native Hawaiian/Other Pacific Islanders (AOR AP=1.55 (99% CI: 1.30–1.84); LAI=1.26 (99% CI: 1.11–1.44)) than White Non-Hispanics. However, White Non-Hispanics were more likely than all other races and ethnicities to fill prescriptions for clozapine (AOR range of 1.17–1.60), antidepressants (AOR range of 1.05–1.40), benzodiazepines (AOR range of 1.37–1.49) and mood stabilizers (AOR range from 1.26–1.31).

Women were less likely to fill prescriptions for any antipsychotics (AOR AP=0.81 (99% CI: 0.79–0.83); LAI=0.78 (99% CI: 0.76–0.80); clozapine=0.82 (99% CI: 0.78–0.86)), but more likely to fill prescriptions for antidepressants (AOR 1.47 (99% CI: 1.44–1.50)) and benzodiazepines (AOR 1.38 (99% CI: 1.35–1.41)) than men.

Beneficiaries with schizoaffective disorder were less likely to fill prescriptions of LAIs or clozapine, but more likely to fill prescriptions for antidepressants, benzodiazepines, and mood stabilizers than those with other schizophrenia-spectrum disorders. Those with comorbid substance use disorder, depression, or anxiety were less likely to fill prescriptions for antipsychotics and mood stabilizers, but more likely for antidepressants (AOR range of 1.41–2.06) and benzodiazepines (AOR range of 1.09–2.33) than beneficiaries without these comorbidities. Deliberate self-harm was associated with increased odds of prescription of antidepressants (AOR 1.46 (99% CI: 1.29–1.66)), but not other classes. All physical comorbidities were associated with increased likelihood of benzodiazepine and mood stabilizer prescription. Those under managed care were more likely to be prescribed antidepressants (AOR 1.24 (99% CI: 1.20–1.27) and less likely to be prescribed all other medication types than those under fee for service financing arrangements.

State variation in the prescription of psychotropic medications

Adjusted proportions of beneficiaries in each state who filled prescriptions for LAIs, clozapine and benzodiazepines are shown in Figures 1–3 and for each psychotropic in the Online Supplement Table 4, 4a and Figures 2–3. State differences in prescribing for each medication revealed significant variation (LAIs: χ2=4892.99, df=47, p<0.001; clozapine: χ2=2709.06, df=47, p<0.001; antidepressants: χ2=4092.14, df=47, p<0.001; benzodiazepines: χ2=8186.17, df=47, p<0.001; and mood stabilizers: χ2=858.51, df=47, p<0.001).

Figure 1.

Adjusted proportions of LAI prescription by state (Difference: χ2=4892.99, df=47, p<0.001)

Figure 3.

Adjusted proportions of benzodiazepine prescription by state (Difference: χ2=8186.17, df=47, p<0.001)

Figure 2.

Adjusted proportions of clozapine prescription by state (Difference: χ2=2709.06, df=47, p<0.001).

The range of proportions for LAIs was 4% (Colorado) to 22% (Rhode Island), an adjusted 5-fold difference and interquartile range (IQR) of 4%. For clozapine, some states with few beneficiaries were outliers but most states ranged between 1–11%, a 10% spread representing an adjusted 10-fold difference and IQR of 2%. Adjusted proportions for antidepressants ranged from 47% (Vermont) to 75% (Missouri), a 25% spread representing a 55% difference and IQR of 8%. Benzodiazepines had two outlier states, but the rest were in the range of 20% to 35%, a 15% spread representing a 75% difference and IQR of 8%. Mood stabilizers ranged from 19% (Montana) to 33% (Pennsylvania), a 14% spread or 73% difference and an IQR of 5%.

Post-hoc analyses

By adjusting for all factors in the Medicaid data to identify whether heterogeneity in prescribing patterns persisted after accounting for this variation, we limited identification of statewide differences in these factors and their implications. To probe whether patient race, ethnicity and other characteristics helped account for interstate variation in prescribing patterns, we partitioned states by tertiles of LAI and clozapine prescription (i.e., one record per state for highest, middle, and lowest levels of LAI and clozapine prescription). We then examined the composition of characteristics of beneficiaries in these state-level tertiles of LAI and clozapine prescription (Online Supplement Table 3).

Discussion

For people diagnosed with schizophrenia, this study identified practice variation among states in prescriptions for important psychotropic medications. The most dramatic differences were in prescription of clozapine and LAIs, which hold distinctive places in medication management—clozapine for its unique efficacy, reducing suicidal behaviors, and underuse;^{15, 32–39} and LAIs for addressing non-adherence, invasiveness, and risk of coercion.^{5, 40} Significant variation across states held for all psychotropics studied even after adjusting for all covariates in the models. This national analysis expands prior investigations of specific city,⁴¹ state ⁴² and regional ⁴³ variations in prescriptions for individuals diagnosed with schizophrenia by including a range of psychotropic medications.

Identification of individual-level predictors of psychotropic medication prescription aids our understanding of state-level variation in these prescriptions. Non-Hispanic Whites were more likely to receive clozapine and less likely to receive LAI prescriptions than Non-Hispanic Blacks, Hispanics, and people of all other races. That these findings held in a national study further supports the need to understand drivers of these disparities and variation in state-level prescription of these medications. One hypothesis is that availability of providers to prescribe clozapine, including regular monitoring of neutrophil counts, may be unevenly distributed in states where different race and ethnic groups receive services. Providers may better recognize treatment-resistant schizophrenia and other indications for clozapine in Non-Hispanic Whites. Conversely, patterns of LAI prescription may be due to type of insurance, beneficiary clinical and demographic factors, or psychiatrist demographics between states.⁴⁴ Providers may better recognize indications for LAIs—non-adherence or its risk factors such as substance abuse, poor insight, or risk for relapse associated with aggression, violence, or self-harm⁴⁵—among Non-Hispanic Blacks, Hispanics, and other Non-White groups. Clinician enthusiasm for managing non-adherence,⁴⁴ and whether LAIs are stigmatizing from the person’s perspective⁴⁶ may influence individuals qualified to administer injections.

That race and ethnic minority groups are consistently less likely to receive prescriptions for antidepressants and mood stabilizers than non-Hispanic Whites may reflect receipt of care in settings with fewer resources and receipt of narrower evaluations leading to poorer recognition of mood and anxiety problems. State regulation of benzodiazepines affected their use— Tennessee’s Medicaid program had strict limits on benzodiazepine prescriptions.⁴⁷ Concern regarding benzodiazepine abuse liability that resulted in changes in FDA labeling,⁴⁸ may be associated with disproportionately lower numbers of prescriptions of these medications for race and ethnic minorities compared to non-Hispanic Whites.⁴⁹ These practice patterns bring into question the equity and quality of care provided to these different groups, and if rooted in implicit biases or systemic racism, must be addressed ensuring fair access to evidence-based treatments for everyone.

The observed variability in prescribing patterns of psychotropic medications across states should be viewed in the context of several limitations. Because this is an observational study, we could not account for unmeasured variation in several comparisons. The data reflect filled prescriptions, not medication ingestion. We are limited by using clinical rather than research diagnoses, although they are appropriate for describing real-world community practice patterns. This study used the latest year that included data from almost all states; practice patterns may have changed since 2012 including expansion of Medicaid coverage through the Affordable Care Act. Our analysis is limited to individuals diagnosed with schizophrenia in 2011 and with 2012 data, therefore we cannot account for length of schizophrenia diagnosis that could influence initiation and use of LAIs after failure on multiple oral antipsychotics. The data did not include dual-eligible (Medicaid and Medicare) beneficiaries, or those privately insured, insured by the VA, or uninsured adults. Previous studies have found that type of practitioner (e.g., psychiatrist versus general practitioner) is related to selection of psychotropic medications, but Medicaid data does not include a consistent provider type for prescriptions.⁵⁰

Results from the post-hoc analyses revealed higher percentages of non-Hispanic Whites (44%) and Blacks (43%) in states with highest rates of LAI prescription, and higher percentages of non-Hispanic Blacks (42%) in states with lowest rates of clozapine prescription. More Hispanics were in states with lowest rates of LAI prescription (16%) and states with highest rates of clozapine prescription (19%). As a measure of access to psychiatrists, we calculated psychiatrists per capita using Census Bureau data from 2012 for states in our study. States with the highest rates of clozapine prescription had more psychiatrists per capita (11/100,000) than those with the lowest rates (5/100,000).

Differences in state policies and Medicaid plans operating within states are not specified in Medicaid data. Eligibility requirements for Medicaid are different between states and can change yearly, as can percentages of costs covered by a state’s plan. The most recent state profiles for Medicaid programs from 2020 can help provide context for how states differed in 2012 ⁵¹. As proxies for state prescribing policies, we reviewed preferred drug lists (PDL) for some outlier states with the highest and lowest rates of prescription of LAIs and clozapine. Significant variation in current PDLs can provide context regarding these policies for 2012. Generic clozapine prescription did not require prior authorization (PA) in these states, however there was substantial variability in whether and how many types of LAIs and oral antipsychotics required PA. In Pennsylvania’s PDL PA is not required for the ten types of LAIs or eleven oral antipsychotics, PAs are approved after failure of one preferred drug, and the PDL includes a comprehensive list of generic and newly FDA approved branded oral antipsychotics.⁵² In Rhode Island only atypical antipsychotics are listed, three LAIs require PA with no PA required for four other LAIs and eight oral antipsychotics, PAs are approved after failure of a preferred drug in the past 90 days, and the PDL includes a comprehensive list of generic and FDA approved branded atypical antipsychotics.⁵³ In Louisiana only two branded LAIs require PA, but their generic forms and seven other LAIs do not, PA requests must show prior use of another preferred drug in the previous 30 days, only three oral antipsychotics require PA for both the brand and generic, and 18 other oral antipsychotics do not require PA.⁵⁴ In Colorado only atypical antipsychotics are listed, all seven LAIs on their PDL require PA, seven oral antipsychotics do not require PA, but their branded versions, three branded and their generic versions, and four newly FDA approved oral antipsychotics require PA showing prior failure of three preferred drugs.⁵⁵

Conclusions

Significant practice variation was found among states in prescription of six types of psychotropic medications despite adjusting for multiple beneficiary factors. Variations in state policies could affect these differences.

We found significant associations between financing arrangement and all medication types, identifying possible policies affecting treatment choice despite evidence-based practices and clinical experience. The exception for benzodiazepines, with known policies governing prescription practices, suggests that policy changes can affect prescribing patterns. The use of standardized PDLs across states where fewer LAIs require PA and more LAIs and oral antipsychotics are listed may be a concrete way to influence new prescribing patterns. Clinicians may favor one medication over others due to their listing in a states’ PDL. However, these policies need to consider variations in fiscal constraints across states given the state’s share of Medicaid financing.

A better understanding of the causes of wide variation in prescription of LAIs and clozapine is needed so efforts to improve access to these important treatment options can be targeted. Variations in Medicaid pharmacy benefit policies may be particularly influential for LAIs while differences in psychiatry training may be important for both LAIs and clozapine.

Highlights:

• This is a nationwide study reporting prescribing variation among major psychotropic medication classes used to treat schizophrenia.

• Significant practice variation was found among states in prescription of each psychotropic class with 10- and 5-fold variation in clozapine and long acting injectables (LAIs), respectively.

• All races and ethnicities were more likely to fill prescriptions of any oral and LAI antipsychotics than White Non-Hispanics, were more likely to fill prescriptions for clozapine and all other medications.

• A better understanding of reasons for divergent findings of clozapine and LAI prescriptions by race and ethnicity is needed to identify sources of inequities.