Patient, Physician and Organizational Influences on Variation in Antipsychotic Prescribing Behavior

Abstract

Background

Physicians face the choice of multiple ingredients when prescribing drugs in many therapeutic categories. For conditions with considerable patient heterogeneity in treatment response, customizing treatment to individual patient needs and preferences may improve outcomes.

Aims of the Study

To assess variation in the diversity of antipsychotic prescribing for mental health conditions, a necessary although not sufficient condition for personalizing treatment. To identify patient caseload, physician, and organizational factors associated with the diversity of antipsychotic prescribing.

Methods

Using 2011 data from Pennsylvania’s Medicaid program, IMS Health’s HCOS^™ database, and the AMA Masterfile, we identified 764 psychiatrists who prescribed antipsychotics to ≥10 patients. We constructed three physician-level measures of diversity/concentration of antipsychotic prescribing: number of ingredients prescribed, share of prescriptions for most preferred ingredient, and Herfindahl-Hirschman index (HHI). We used multiple membership linear mixed models to examine patient caseload, physician, and healthcare organizational predictors of physician concentration of antipsychotic prescribing.

Results

There was substantial variability in antipsychotic prescribing concentration among psychiatrists, with number of ingredients ranging from 2-17, share for most preferred ingredient from 16%-85%, and HHI from 1,088-7,270. On average, psychiatrist prescribing behavior was relatively diversified; however, 11% of psychiatrists wrote an average of 55% of their prescriptions for their most preferred ingredient. Female prescribers and those with smaller shares of disabled or serious mental illness patients had more concentrated prescribing behavior on average.

Discussion

Antipsychotic prescribing by individual psychiatrists in a large state Medicaid program varied substantially across psychiatrists. Our findings illustrate the importance of understanding physicians’ prescribing behavior and indicate that even among specialties regularly prescribing a therapeutic category, some physicians rely heavily on a small number of agents.

Implications for Health Policies, Health Care Provision and Use

Health systems may need to offer educational interventions to clinicians in order to improve their ability to tailor treatment decisions to the needs of individual patients.

Implications for Future Research

Future studies should examine the impact of the diversity of antipsychotic prescribing to determine whether more diversified prescribing improves patient adherence and outcomes.

Introduction

Physicians often face many choices when prescribing drugs in a therapeutic category. Appropriate prescribing is the result of a matching process in which providers prescribe those medications that best fits the patient’s clinical characteristics and preferences. Personalizing prescribing choices to each individual could lead to better clinical outcomes potentially through improving patient adherence.^1,2 Unfortunately, evidence suggests that many physicians tend to prescribe the same drug (or a limited subset of drugs) to all patients because customized prescribing entails a variety of costs (e.g., coordination and cognition costs) that can be challenging and laborious.³

There are now more than 20 molecules and their reformulations in the class of antipsychotic drugs.⁴ Antipsychotics have been approved by the US Food and Drug Administration (FDA) to treat several serious psychiatric conditions including schizophrenia, bipolar disorder, major depressive disorder, and autism. Off-label use of antipsychotics for other conditions is also common.^5,6 The widespread substitution of second-generation antipsychotics (SGAs) for first-generation antipsychotics (FGAs) in the past two decades has resulted in high expenditures for antipsychotics, which are mainly financed by Medicare and state Medicaid programs.⁷ Because there is considerable variability in treatment response and medication side effects across individual patients using antipsychotic drugs,^8,9 tailoring treatment to the needs and preferences of patients is essential to improving clinical outcomes.

Although there are many antipsychotic medications, previous studies^4,10,11 (which differ in their sampling frames and study periods) have found antipsychotic prescribing to be relatively concentrated. These findings are similar to those reported from several studies examining prescribing concentration in multiple disease conditions³ (acute vs. chronic diseases) and antidepressants,¹ while one study examining ten prevalent therapeutic classes¹² found relatively diversified prescribing behavior (all these studies differ in concentration measures and study populations). The studies examining antipsychotic prescribing concentration adjusted for physician characteristics^4,10,11 but did not have information on the patient case mix or on the setting in which the physicians practiced. The characteristics of the treated patient population may influence prescribing decisions because physicians who see sicker patients may be more likely to customize the treatment than physicians seeing patients with lower severity.¹ Furthermore, physician prescribing practices may be shaped by the organizations within which they practice through multiple mechanisms (e.g., guideline dissemination, quality improvement initiatives, normative influences, financial incentives).^13,14

This study aimed to examine variation in psychiatrists’ prescribing of antipsychotics and the influence of patient caseloads, physician, and organizational features on the diversity/concentration of antipsychotic prescribing.

Methods

Data Sources

Medicaid Data

We obtained 2011 data on patient caseload characteristics and physician antipsychotic prescribing from Pennsylvania Medicaid, the fourth largest Medicaid program in the U.S. in expenditures. The data contain enrollment information, 46 medical claims (inpatient, outpatient, professional), and pharmacy claims for all 2.2 million Medicaid beneficiaries who were enrolled in either fee-for-service or managed care programs in 2011. We used the enrollment file to obtain beneficiaries’ demographic and enrollment information such as age, sex, race/ethnicity, dual eligible status for Medicare, eligibility type, and Medicaid payer (fee-for-service vs. managed care). We used medical claims to capture patients’ diagnoses in the year. The pharmacy claims file has information for each prescription on the National Drug Code (NDC), date of fill, quantity, form, and prescribing provider identifier. We used the Medi-Span^® database to acquire antipsychotic drugs’ name, dose, and active ingredient by NDC.¹⁵ Finally, we obtained a provider file from Medicaid containing Medicaid provider identifier (linkable to prescribing provider identifier in pharmacy claims), National Provider Identifier (NPI), and practice ZIP code.

Physician Characteristics

We obtained information on physician characteristics (including demographic, specialty and medical education) for the prescribers in our sample from the AMA Masterfile which includes data on all physicians (both domestic and foreign graduates) practicing in the US.¹⁶

Organizational Affiliation and Characteristics

We obtained information on organizational affiliation from IMS Health’s Healthcare Organizational Services^™ (HCOS) database, which identifies physician affiliations with health care organizations (e.g., medical groups, hospitals, nursing homes), along with the type of affiliation with each organization (e.g., attending, affiliated, admitting, staff, consulting, treating). HCOS^™ also contains information on the specialty of the organization, for example, whether a medical group specializes in primary care or behavioral health, and the total number of providers from all specialties affiliated with that organization.¹⁷

Study Population

The unit of analysis was at the physician-level; however, we started with a sample of claims at the patient-level from which we identified psychiatrists treating these patients. First, we selected Medicaid beneficiaries with any antipsychotic use in the Pennsylvania’s Medicaid program and then limited them to those who were under the age of 65 and not dually eligible for Medicare because Medicaid lacks complete claims information for dual eligible, particularly for prescription drugs. By linking Medicaid data with IMS Health’s HCOS^™ database and AMA Masterfile, we then identified psychiatrists prescribing at least one antipsychotic drug to these Medicaid beneficiaries. We further limited our analyses to all psychiatrists who regularly prescribed antipsychotics to Medicaid beneficiaries, defined as having ≥10 unique Medicaid beneficiaries in 2011. We excluded low volume prescribers because they provide little information about the diversity of antipsychotic prescribing and account for a small proportion of antipsychotic prescriptions (<1%) (see Appendix, Table A1 for the study population flow chart).

Outcome Variables

We constructed three physician-level measures of concentration/diversity for antipsychotic prescribing. The first outcome measure was the number of unique antipsychotic ingredients ever prescribed in the year. Prescriptions for reformulations (e.g., Risperdal) were combined with the original ingredient (e.g., risperidone). The second outcome measure was the share of prescriptions accounted for by the psychiatrist’s most preferred antipsychotic ingredient, a measure of prescribing concentration used in previous research.^1-3 The third outcome measure was the Herfindahl-Hirschman index (HHI), a commonly accepted measure of market concentration of firms,^18-20 which has also been used to measure the concentration of product choice within physicians.^1,2,4,10,11 The HHI index incorporates information on both number and share of antipsychotic ingredients prescribed by the psychiatrist, with a larger value implying more concentrated and a smaller value indicating more diversified prescribing behavior. We calculated the HHI for a psychiatrist who prescribed N unique antipsychotic ingredients in 2011, by summing up the square of each antipsychotic ingredient’s prescription share. The value of the HHI index ranged from 10,000/N (if a physician prescribed each of N antipsychotic ingredients with equal share) to 10,000 (if a psychiatrist prescribed only one antipsychotic ingredient). An HHI greater than 1,800 generally reflects a highly concentrated market.²¹

Each of the three measures of diversity/concentration captures a different aspect of concentration. Specifically, HHI describes the concentration of shares across all the antipsychotic ingredients prescribed by the psychiatrist, including both heavily and rarely selected ones. In contrast, the share of prescriptions made up by the most preferred antipsychotic ingredient focuses only on the top drug prescribed by the psychiatrist. The number of unique antipsychotic ingredients reports on the range of drugs ever prescribed by the psychiatrist regardless of share. Together, the three variables explore psychiatrists’ concentrated vs. diversified prescribing behavior of antipsychotic drugs. A larger number of ingredients indicates less concentrated (more diversified) prescribing of antipsychotics, while higher share of prescriptions made up by the most preferred ingredient and a higher HHI imply a more concentrated (less diversified) prescribing behavior.

Explanatory Variables

We examined three types of predictors of antipsychotic prescribing: patient caseload characteristics, physician characteristics, and features of the health care organizations with which psychiatrists were affiliated.

Patient Caseload Characteristics

We used data on patient caseload characteristics to adjust for differences in the psychiatrists’ treated population. For each psychiatrist we calculated the share of his or her patients that were female, non-Hispanic white, under 18 years old, and 50 or older. We included a variable indicating the share of patients eligible for Medicaid through SSI, a measure of disability. To adjust for variability in prescribing due to formulary management tools, we included an indicator of the share of a psychiatrist’s Medicaid patients in fee-for-service vs. managed care programs; however, the only prior authorization policies in existence during our study were for children <18 years old. To adjust for patients’ health status and diagnosis, we included psychiatrist’s share of patients with serious mental illnesses for which antipsychotics are indicated [any diagnosis of schizophrenia (ICD-9 codes: 295), bipolar disorder (ICD-9 codes: 296.0, 296.1, 296.4, 296.5, 296.6, 296.7, 296.8), major depressive disorder (ICD-9 codes: 296.2, 296.3), or autism spectrum disorder (ICD-9 codes: 299.0, 299.1, 299.8)]. To capture patients’ medical comorbidities, we first constructed for each patient separate indicators of 25 non-mental illness conditions in the Elixhauser comorbidity index^22-24 based on patient’s medical claims. Then we created a variable for physician-level share of patients with ≥2 of these conditions, a cutoff chosen based on the distribution in our patient sample.

Physician Characteristics

The physician characteristics included psychiatrist’s sex (female vs. male), age categories (≥40, 40–49, 50–59, ≥60), educational background (whether the psychiatrist graduated from a top 20 medical school according to 2011 US News and World Report Rankings, and whether he/she received a degree from a US vs. a foreign medical school), prescription volume measured by total number of antipsychotic prescriptions in the year, and whether the psychiatrist practiced in an urban area. We measured each psychiatrist’s professional network by calculating the total number of other regular antipsychotic prescribers in all health care organizations with which the psychiatrist was affiliated. We hypothesized that psychiatrists practicing in larger organizations may have more diversified prescribing practices due to greater availability of information from peers.²⁵

Organizational Setting

To measure features of affiliated health care organizations, we included dummy variables for organizational affiliation type (outpatient practice only, inpatient only, or both outpatient and inpatient organizations), and an indicator of whether the psychiatrist was affiliated with any behavioral health organization (psychiatric hospital, or medical group with specialization in behavioral health/addiction medicine). We also calculated the weighted average of the provider numbers (from all specialties) across all the organizations to which the psychiatrist belonged, as a proxy for organization size. Weights were applied based on the weighting structure of the multiple membership modeling²⁶ as discussed in more detail below.

Data Analytic Procedures

Descriptive Analysis

We examined the distribution of the 3 outcome variables across psychiatrists in the study sample, reporting mean, standard deviation, and percentiles, and the coefficient of variation, a commonly used measure of variability.^11,27 Since we do not have a gold standard to assess “appropriate” concentration of prescribing, we looked at the psychiatrists who were in the 25^th percentile and below (number of ingredients) and 75^th percentile and above (share of most preferred ingredient and HHI) and assessed the extent to which they overlapped. We speculate that those psychiatrists who were in the “most concentrated prescribing” group for all three measures may be less likely to be attempting to match patient and treatment.

Multiple Membership Data Structure and Modeling

Traditional multilevel models are usually used to analyze data that have hierarchical structures (i.e., each observation at a lower level is nested within a single unit at a higher level).^28-31 However, sometimes the assumption of purely hierarchical data structure does not hold in practice.^32,33 One complex type of non-hierarchical data is the multiple membership structure, in which lower-level observations are not nested within only one higher-level unit; instead, they are members of multiple higher-level units simultaneously.^32,34 In our study, some psychiatrists (lower-level unit) were affiliated with more than one health care organization (higher-level unit). In analyses with a multiple membership data structure, it is assumed that there are known weights which could be used to quantify the degree of membership for a lower-level unit to the different higher-level units. The sum of weights across different clusters for each lower-level unit equals 1.

To tackle the complexity of the multiple membership data structure, we used the multiple membership linear mixed models with restricted maximum likelihood estimation (REML) for the evaluation of the 3 continuous outcome measures, which were approximately normally distributed in the study sample. We used REML rather than maximum likelihood estimation (ML) because REML not only provides unbiased estimates but also takes into account the loss of degrees of freedom due to the inclusion of covariates.³⁵ All regressions were performed at the physician-level. The regression models included fixed effects for all explanatory variables discussed above and health care organization-level random effects. The model can be expressed as:

$$y_i={\beta }_0+{\beta }_1{x}_{i,\mathit{pt}}+{\beta }_2{x}_{i,\mathit{phy}}+{\beta }_3{x}_{i,\mathit{org}}+{\beta }_4\sum _{j\in \mathit{org}\left(i\right)}{w}_{j,i}^2{\text{size}}_j^2++\sum _{j\in \mathit{org}\left(i\right)}{w}_{j,i}^2{\mu }_j^2+e_i$$

where y_i represents the outcome variable for psychiatrist i, β₀ is the intercept, x_i,pt is a vector of variables for patient characteristics for psychiatrist i, with corresponding vector of coefficients β₁. Similarly, x_i,phy is a vector of psychiatrist characteristics with coefficients β₂ for psychiatrist i; and x_i,org is a vector of organizational setting factors for psychiatrist i with corresponding coefficients β₃. $w_{j,i}^2$ measures the degree of membership for psychiatrist i (level 1) to organization j (level 2), summing to 1 for psychiatrist i. ${\text{size}}_j^2$ represents total number of providers for organization j (level 2). ${\sum }_{j\in \mathit{org}\left(i\right)}{w}_{j,i}^2{\text{size}}_j^2$ is the weighted average of the provider numbers across all the organizations (level 2) with which psychiatrist i was affiliated, β₄ stands for corresponding coefficient. The first 5 items on the right-hand side of the model are termed the fixed part of the model. ${\sum }_{j\in \mathit{org}\left(i\right)}{w}_{j,i}^2{\mu }_j^2$ represents the weighted sum of organizational-level random effect, and e_i is the residual error term. The last 2 terms stand for the random part of the equation. Appendix, Table A2 contains details of the weighting structure used in both main and sensitivity analyses to capture the extent of membership for a psychiatrist to each of his or her affiliated organizations.

In addition, we ran the regressions on each outcome without including any explanatory variable; and we report the reduction in variance between the null and above full models. Finally, to assess the degree of concentration in response to the severity of patient illness (since prescribing customization is expected to meet patients’ clinical needs), we predicted the 3 outcomes on antipsychotic prescribing by share of patients with serious mental illnesses and by share of SSI-eligible patients, adjusting for all other covariates.

Results

Physician and Patient Caseload Characteristics

In 2011, a total of 764 psychiatrists treating 65,256 patients in the Pennsylvania Medicaid program were included in our sample (the median number of treated patients was 68 for a psychiatrist). Of the 764 psychiatrists in the study, 33.3% were female, more than half (55.9%) were ≥50 years old, and 44.0% graduated from foreign schools. At the physician-level, the mean share of SSI-eligible patients was 69.6% and the mean share of a physician’s patients with schizophrenia, bipolar disorder, major depressive disorder or autism was 77.7% (Table 1).

Table 1.

Descriptive Characteristics of Regular Psychiatrist Prescribers

Characteristic	Mean (SD) or percent
N	764
Patient caseload characteristics
Demographic information
Female patients (%)	49.6 (15.1)
SSI-eligible patients (%)	69.6 (13.9)
Non-Hispanic whites (%)	62.8 (27.2)
Patients <18 years old (%)	23.4 (31.4)
Patients ≥50 years old (%)	18.3 (15.0)
Health status
Mean number of non-mental comorbidities	1.4 (0.7)
Patients with serious mental illnesses (%)	77.7 (16.5)
Health insurance
Patients enrolled in fee-for-service program (%)	22.1 (32.5)
Physician characteristics
Female	33.3%
Physician age (yrs)
<40	16.0%
40–49	28.1%
50–59	33.3%
≥60	22.6%
Type of medical school attended
US top 20	8.8%
Ranked ≥21	47.2%
Foreign schools	44.0%
Number of antipsychotic prescriptions
1st quartile (13-99)	55.8 (23.5)
2nd quartile (100-279)	182.1 (54.9)
3rd quartile (281-742)	483.7 (141.9)
4th quartile (746-8,234)	1,572.7 (1,057.2)
Number of other antipsychotic prescribers in the same organizations
0	7.6%
1–9	39.9%
≥10	52.5%
Practice in urban only	86.7%
Organizational setting
Number of affiliated organizations
1	49.5%
2	28.1%
≥3	22.4%
Any affiliation with a behavioral health organization	38.2%
Organizational affiliation types
Outpatient only	15.8%
Inpatient only	57.9%
Both inpatient and outpatient	26.3%
Organization size measured by number of providersa
1st quartile (1-35)	12.2 (8.3)
2nd quartile (36-208)	96.8 (48.6)
3rd quartile (210-464)	314.8 (60.1)
4th quartile (466-3,392)	972.4 (603.6)

Notes:

^aNumber of providers includes providers of all disciplines affiliated with an organization.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved. AMA Masterfile, 2011.

Organizational and Physician Affiliations

Among the study sample, about half (50.5%) of the psychiatrists were affiliated with 2 or more organizations in 2011 (mean number of organizations per physician: 1.9; range: 1–7), 38.2% had an affiliation with at least one behavioral health organization (psychiatric hospital, or medical group with specialization in behavioral health/addiction medicine) (Table 1). Of the 539 organizations with which the psychiatrists billing Pennsylvania Medicaid were affiliated, the majority of them (63.5%) were non-behavioral health organizations and most (63.6%) were inpatient organizations (Table 2).

Table 2.

Features of Affiliated Organizations of the Study Sample

Characteristic	Number (percent)
Total number of organizations	539
Number stratified by organization specialty
Behavioral health	197 (36.6%)
Non-behavioral health	342 (63.5%)
Number stratified by organization type
Outpatient	196 (36.4%)
Inpatient	343 (63.6%)
Acute care hospitals	222 (41.2%)
Psychiatric hospitals	36 (6.7%)
Nursing homes	81 (15.0%)
Rehabilitation hospitals	4 (0.7%)
Mean number of providers of all disciplines/organization [mean (SD)]a	249.6 (464.6)
Mean number of regular antipsychotic prescribers/organization [mean (SD)]	3.1 (5.1)

Notes:

^aNumber of providers includes providers of all disciplines affiliated with an organization.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved.

Variation in Physicians’ Antipsychotic Prescribing

There was substantial variability (as reflected in the coefficient of variation²⁷) in all three measures of concentration across psychiatrists, with number of ingredients ranging from 2 to 17, share of most preferred ingredient ranging from 16% to 85%, and HHI from 1,088 to 7,270 (Table 3). On average, antipsychotic prescribing in our sample was relatively diversified. The average physician used 8.8 ingredients, the share of the most preferred ingredient was 38%; however, the HHI was above the threshold considered concentrated by economists. About 11% of these psychiatrists (N = 84) wrote an average of more than half (55%) of their antipsychotic prescriptions for the most preferred ingredient, and their mean HHI was 4,022 (maximum value 10,000), which suggests that these psychiatrists relied heavily on only 1 or at most 2 ingredients.

Table 3.

Distributions of Number of Ingredients, Share of Most Preferred Ingredient, and HHI of the Study Sample

Variable	Number of ingredients	Share of most preferred ingredient (%)	HHI of ingredients
Overall sample (N = 764)
Mean (SD)	8.8 (2.9)	38 (11)	2,603 (847)
5th percentile	4	22.7	1,530
25th percentile	6	29.8	1,991
Median	9	36.1	2,483
75th percentile	11	43.8	3,018
95th percentile	14	58.4	4,073
Range	2-17	16-85	1,088-7,270
Ratio of 75th to 25th percentiles	1.83	1.47	1.52
Coefficient of variation	0.33	0.29	0.33
Psychiatrists with the most concentration in all 3 measures (N = 84)a
Mean (SD)	4.9 (0.9)	55 (9)	4,022 (859)
Median	5	53	3,731
Range	2-6	44-85	3,058-7,270

Notes:

^aThe subset included psychiatrists in the lowest 25th percentile of number of ingredients and highest 25th percentiles of share of most preferred ingredient and HHI.

Predictors of the Concentration of Physicians’ Antipsychotic Prescribing

Patient Caseload Characteristics

Table 4 reports results from the multiple membership linear mixed models. Of the 3 types of factors included in the regressions, several patient caseload characteristics were associated with psychiatrists’ antipsychotic prescribing although the effects were relatively small. After adjusting for physician characteristics and organizational features, psychiatrists with a 1 percent increase in share of SSI-eligible patients were associated with a 0.02 unit increase in number of ingredients (p<.05), a 0.16 percent decrease in share of the most preferred antipsychotic ingredient (p<.01) and a 13.2 unit decrease in HHI (p<.01) (less concentrated prescribing). Similarly, a 1 percent increase in psychiatrists’ share of patients with serious mental illnesses was associated with a 0.02 unit increase in number of ingredients, a 0.11 percent decrease in the share of antipsychotic prescriptions for most preferred ingredient, and a 9.0 unit decrease in HHI (all p<.01). A 1 percent increase in a psychiatrist’s share of patients <18 years old was associated with a 0.03 unit decrease in number of ingredients, a 0.07 percent increase in most preferred ingredient, and a 8.0 unit increase in HHI (all p<.01). Other patient caseload characteristics, including a larger share of older patients and higher share of non-Hispanic whites, were also significantly associated with more diversified antipsychotic prescribing.

Table 4.

Predictors of the Concentration of Psychiatrist Prescribing of Antipsychotics and Related Variance Reduction

Variables	Coefficients (standard errors)
	Number of ingredients	Share of most preferred ingredient	HHI of ingredients
Patient caseload characteristics
Female patients (%)	−0.01 (0.01)	0.02 (0.03)	3.80 (2.24)*
SSI-eligible patients (%)	0.02 (0.01)**	−0.16 (0.03)***	−13.15 (2.32)***
Non-Hispanic whites (%)	0.00 (0.00)	−0.03 (0.02)*	−2.11 (1.26)*
Patients <18 years old (%)	−0.03 (0.00)***	0.07 (0.02)***	8.01 (1.61)***
Patients ≥50 years old (%)	0.00 (0.01)	−0.08 (0.04)**	−5.59 (3.03)*
Patients with serious mental illnesses (%)	0.02 (0.01)***	−0.11 (0.03)***	−9.00 (2.26)***
Patients with 2+ non-mental comorbidities (%)	0.01 (0.01)	0.06 (0.04)	3.71 (3.08)
Patients enrolled in fee-for-services (%)	0.00 (0.00)	0.02 (0.01)	1.79 (1.09)*
Physician characteristics
Physician sex (ref = male)
Female	−0.29 (0.18)*	1.23 (0.8)	97.53 (58.88)*
Physician age (ref = <40)
40-49	0.17 (0.25)	1.05 (1.14)	62.46 (83.84)
50-59	−0.05 (0.25)	1.24 (1.13)	76.55 (82.87)
≥60	0.25 (0.27)	1.92 (1.22)	103.37 (89.74)
Type of medical school attended (ref = ranked ≥21)
US top 20	−0.26 (0.30)	1.90 (1.33)	131.32 (97.92)
Foreign schools	0.11 (0.17)	1.41 (0.76)*	62.29 (55.98)
Total number of antipsychotic prescriptions	0.002 (0.000)***	−0.003 (0.000)***	−0.22 (0.035)***
Number of other antipsychotic prescribers in the same organizations (ref = 0)
1-9	0.18 (0.33)	0.06 (1.46)	−66.56 (107.72)
≥10	0.34 (0.34)	1.26 (1.47)	−33.44 (109.88)
Practice location (ref = otherwise)
Urban only	−0.24 (0.27)	0.78 (1.21)	41.23 (88.81)
Organizational setting
Organization specialty (ref = otherwise)
Any affiliation with a behavioral health organization	0.79 (0.45)*	1.42 (1.96)	46.82 (145.66)
Organizational affiliation type (ref = outpatient only)
Inpatient only	0.36 (0.49)	2.74 (2.08)	135.78 (155.89)
Both inpatient and outpatient	0.06 (0.28)	1.50 (1.22)	101.45 (90.76)
Organization size	0.00 (0.00)	0.00 (0.00)	0.05 (0.07)
Intercept	5.09 (0.96)***	50.84 (4.24)***	3751.94 (312.75)***
Variance reduction by adding above explanatory variables
Total variation reduction	45.68%	21.05%	28.02%

Notes:

^*Significant at 10 %;

^**significant at 5 %;

^***significant at 1 %.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved. AMA Masterfile, 2011.

We predicted the marginal effects on prescribing concentration by 2 patient caseload variables of interest. By increasing a psychiatrist’s share of patients with serious mental illnesses from 20% to 100% (using the range observed in our study sample), the degree of concentration would decrease (the share of most preferred ingredient from 43.4% to 34.6% and the HHI from 3,102 to 2,382). Similar patterns were also found by share of SSI-eligible patients (Figure 1).

Figure 1.

Concentration of antipsychotics prescribing by psychiatrist’s share of patients with serious mental illnesses, share of SSI-eligible patients^a

^aAll measures were adjusted for patient caseload, physician, and organizational setting covariates listed in the regression.

Physician Characteristics

Of the several physician characteristics examined, only physician sex and prescribing volume were significantly associated with prescribing concentration (Table 4). Female psychiatrists prescribed 0.29 fewer antipsychotic ingredients than did male psychiatrists and had a HHI that was 97.5 units higher than that of their male counterparts although both associations were only significant at the p<.10 level.

Organizational Setting

Psychiatrists who had any affiliation with behavioral health organizations prescribed 0.8 more unique antipsychotic ingredients on average compared to those who did not have affiliation with any behavioral health organization, although this association was only significant at the p<.10 level (Table 4). The other two organizational characteristics (organizational affiliation type and size) were not significantly associated with the concentration of antipsychotic prescribing.

In total, our explanatory variables accounted for a 45.7% reduction of the total variance for number of ingredients, 21.1% for share of most preferred ingredient, and 28.0% for HHI (Table 4). Regression results of the 3 outcome variables for all sensitivity analyses were very similar to the main analysis (see Appendix, Tables A3-A5 for results of sensitivity analyses).

Discussion

Our study provides a comprehensive assessment of how the diversity of psychiatrists’ antipsychotic choice is shaped by patient caseload, physician, and organizational characteristics in a large Medicaid program. Psychiatrists were relatively diversified in the number of antipsychotics they prescribed – using an average of 8.8 different molecules. However, we found that the degree of concentration in antipsychotic prescribing varied substantially across psychiatrists who were regular prescribers of antipsychotics, and that some psychiatrists having very concentrated antipsychotic prescribing behavior as measured by all three outcomes. Several patient caseload and physician characteristics were significantly associated with psychiatrists’ diversity vs. concentration of antipsychotic prescribing although the effects were small. The few characteristics of organizations we were able to measure had little influence over psychiatrist prescribing behavior.

Our finding that there was substantial variability in the degree of concentration in antipsychotic prescribing across psychiatrists suggests that patients seeing psychiatrists who only prescribe a limited number of antipsychotics (e.g., with an average of 55% of antipsychotic prescriptions written for the most preferred ingredient) may have a different treatment experience than do patients whose doctors prescribe a wide range of antipsychotic products. Antipsychotics have been approved to treat several serious mental disorders (such as schizophrenia and as adjunctive treatment for major depressive disorder), and prescribers must sometimes try multiple drugs before finding the “best drug” for a particular patient because of the wide variability in individual treatment responses and medication side effects. For instance, the large Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) found that majority (74%) of patients with schizophrenia failed the first trial,³⁶ indicating that most patients with schizophrenia would need multiple trials.³⁷ For patients not responding adequately to initial therapy, doctors need to alter treatment either by dose adjustment or switching medications. Each of these changes in prescribing for a particular patient requires compliance with evidence-based guidelines and regular follow-up to customize prescribing choices to individual patients.^38,39 We created summary measures of concentration across all antipsychotic prescriptions written by a psychiatrist; however, the consequences of concentrated prescribing may vary based on the timing within episode. For example, consistent choice of the same first line agent may be appropriate whereas concentrated prescribing later in a treatment episode when a medication change is warranted may be inappropriate. Nevertheless, our finding that 11% of psychiatrists treating Medicaid enrollees with antipsychotics had very concentrated prescribing behavior (relying heavily on only one or two antipsychotic agents) indicates that they may not tailor treatment to individual patient needs and preferences. Future studies may improve upon our analysis by examining within-episode variation in concentration to determine whether more diversified prescribing improves patient adherence and outcomes.

Previous literature examining multiple therapeutic classes (which differ in concentration measures and study populations) yielded mixed results.^1,3,12 Three studies have examined physician antipsychotic prescribing previously; all used the IMS Health’s Xponent^™ database which has comprehensive information on physician prescribing but limited patient information.^4,10,11 Using Medicaid data from a more recent time period (2011), measuring concentration in three ways, and adjusting for a comprehensive set of factors, our study finds that antipsychotic prescribing varied substantially across regular psychiatrist prescribers and a subset of the prescribers had very concentrated prescribing behavior.

To our knowledge, this study is the first to examine the concentration of antipsychotic prescribing with all three types of information on physicians: characteristics of the physicians, their treated patients and practice settings. Previous research found that patient clinical factors played trivial role in medication switches.³ Our study found a significant relationship between patient caseload characteristics and the diversity of physicians’ prescribing of antipsychotics although effects were relatively small. Psychiatrists had more diversified prescribing using all three measures of concentration if they had a higher share of their patient population with a disability or with relatively severe mental illnesses. Of the several physician characteristics examined, female psychiatrists tended to have slightly more concentrated prescribing behavior than their male counterparts although the association was only marginally significant. Our organization-level factors were limited to organization type, specialty, and size. Psychiatrists who had any affiliation with a behavioral health organization (clinic or psychiatric hospital) prescribed more antipsychotic ingredients on average than those not affiliated with behavioral health organizations although it was only marginally significant. We did not have information on other organizational factors such as quality improvement initiatives, financial incentives, and guideline dissemination which may also play an important role. The fact that the explanatory variables included in the regressions accounted for a moderate reduction of the total variance for the 3 outcome measures (28.0%-45.7%) implies that other factors not included in our analyses also influence physicians’ antipsychotic prescribing behavior.

This study has several limitations. First, our study examined prescribing behavior in the Pennsylvania Medicaid program and thus our findings may not necessarily be generalizable to other states. Second, we lack information on non-Medicaid prescribing for the psychiatrists in our study. Third, the HCOS^™ database captures all hospitals and mid- to large-medical groups and clinics, but is less likely to capture small practices with only 1-2 providers. We may have misclassified some providers as practicing only in inpatient settings as a result. Furthermore, we could not adjust for other important factors, such as pharmaceutical manufacturer promotion on specific antipsychotic drugs, which may shape physician prescribing behavior.⁴ In addition, the patient caseload characteristics were aggregated at the physician level hence extrapolating to patient level inferences would result in ecological fallacy.⁴⁰ Finally, approximately 23% of the physician sample’s patient caseloads was <18. Psychiatrists prescribing primarily to children may be more concentrated in their prescribing because fewer antipsychotics are approved for use in children. However, given that two thirds of psychiatrists in our sample treated both adults and children we addressed this issue by including a variable for the share of patients <18 in all the regressions. This variable had very small effect on prescribing concentration.

In conclusion, antipsychotic prescribing by individual psychiatrists in a large state Medicaid program varied substantially across psychiatrists. Our findings illustrate the importance of understanding physicians’ prescribing behavior and indicate that even among specialties regularly prescribing a therapeutic category, some physicians rely heavily on a small number of agents. Health systems may need to offer educational interventions to clinicians in order to improve their ability to tailor treatment decisions to the needs of individual patients.

Appendix

Table A1.

Flow Chart for the Study Population

Table A2.

Weighting Structure Constructed in the Multiple Membership Modeling

The affiliation type identifies the relationship a psychiatrist has with a health care organization (e.g., attending, affiliated, or admitting in a hospital; staff, consulting, or treating in a nursing home), which was used to construct the weighting structure to represent the extent of membership for a psychiatrist to each of his or her affiliated organizations. Medical group affiliations were given the same weight as the affiliation type “attending” in inpatient facilities. We categorized the degree of membership into 2 groups based on this information: (i) “strong relationship” if a psychiatrist was affiliated with a medical group, had an attending relationship with a hospital, practiced at an outpatient location of a hospital (affiliated provider), or was contractually on staff at a nursing home (staff); (ii) “weak relationship” if a psychiatrist admitted patients to a hospital but was not designated as an attending or affiliated provider, consulted or treated patients at a nursing home without being on staff.

In the main analysis, we assigned the weighting ratio of “strong relationship” to “weak relationship” to be 5:1 (total weights summed to 1 for each psychiatrist), assuming affiliated organizations in the “strong relationship” group would be more influential to the psychiatrist’s prescribing behavior than organizations in the “weak relationship” group. To check the robustness of the results, in sensitivity analyses we explored other weighting schemes: 1:1 (i.e., equal weights), 10:1, and 2:1.

Table A3.

Sensitivity Analysis Results for Number of Ingredients

Variables	Coefficients (standard errors)
	Equal weights 1:1	Weighting ratio 10:1	Weighting ratio 2:1
Patient caseload characteristics
Female patients (%)	−0.01 (0.01)	−0.01 (0.01)	−0.01 (0.01)
SSI-eligible patients (%)	0.02 (0.01)**	0.02 (0.01)**	0.02 (0.01)**
Non-Hispanic whites (%)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Patients <18 years old (%)	−0.03 (0.00)***	−0.03 (0.00)***	−0.03 (0.00)***
Patients ≥50 years old (%)	0.00 (0.01)	0.00 (0.01)	0.00 (0.01)
Patients with serious mental illnesses (%)	0.02 (0.01)***	0.02 (0.01)***	0.02 (0.01)***
Patients with 2+ non-mental comorbidities (%)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)
Patients enrolled in fee-for-services (%)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Physician characteristics
Physician sex (ref = male)
Female	−0.29 (0.18)*	−0.29 (0.18)*	−0.29 (0.18)*
Physician age (ref = <40)
40-49	0.19 (0.25)	0.17 (0.25)	0.18 (0.25)
50-59	−0.03 (0.25)	−0.06 (0.25)	−0.04 (0.25)
≥60	0.27 (0.27)	0.24 (0.27)	0.26 (0.27)
Type of medical school attended (ref = ranked ≥21)
US top 20	−0.27 (0.30)	−0.26 (0.3)	−0.26 (0.30)
Foreign schools	0.12 (0.17)	0.11 (0.17)	0.11 (0.17)
Total number of antipsychotic prescriptions	0.002 (0.000)***	0.002 (0.000)***	0.002 (0.000)***
Number of other antipsychotic prescribers in the same organizations (ref = 0)
1-9	0.17 (0.33)	0.18 (0.33)	0.18 (0.33)
≥10	0.30 (0.34)	0.35 (0.34)	0.32 (0.34)
Practice location (ref = otherwise)
Urban only	−0.24 (0.27)	−0.24 (0.27)	−0.24 (0.27)
Organizational setting
Organization specialty (ref = otherwise)
Any affiliation with a behavioral health organization	0.79 (0.45)*	0.80 (0.45)*	0.79 (0.45)*
Organizational affiliation type (ref = outpatient only)
Inpatient only	0.35 (0.48)	0.36 (0.49)	0.36 (0.49)
Both inpatient and outpatient	0.12 (0.28)	0.05 (0.28)	0.09 (0.28)
Organization size	0.00 (0.00)*	0.00 (0.00)	0.00 (0.00)*
Intercept	5.13 (0.95)***	5.07 (0.96)***	5.11 (0.96)***

Notes:

^*Significant at 10 %;

^**significant at 5 %;

^***significant at 1 %.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved. AMA Masterfile, 2011.

Table A4.

Sensitivity Analysis Results for Share of Most Preferred Ingredient

Variables	Coefficients (standard errors)
	Equal weights 1:1	Weighting ratio 10:1	Weighting ratio 2:1
Patient caseload characteristics
Female patients (%)	0.02 (0.03)	0.02 (0.00)	0.02 (0.03)
SSI-eligible patients (%)	−0.16 (0.03)***	−0.16 (0.00)***	−0.16 (0.03)***
Non-Hispanic whites (%)	−0.03 (0.02)*	−0.03 (0.00)**	−0.03 (0.02)*
Patients <18 yrs (%)	0.07 (0.02)***	0.07 (0.00)***	0.07 (0.02)***
Patients ≥50 yrs (%)	−0.08 (0.04)**	−0.08 (0.00)**	−0.08 (0.04)**
Patients with serious mental illnesses (%)	−0.11 (0.03)***	−0.11 (0.00)***	−0.11 (0.03)***
Patients with 2+ non-mental comorbidities (%)	0.06 (0.04)	0.06 (0.00)	0.06 (0.04)
Patients enrolled in fee-for-services (%)	0.02 (0.01)	0.02 (0.00)	0.02 (0.01)
Physician characteristics
Physician sex (ref = male)
Female	1.22 (0.80)	1.23 (1.00)	1.22 (0.80)
Physician age (ref = <40)
40-49	1.06 (1.14)	1.05 (1.00)	1.05 (1.14)
50-59	1.26 (1.13)	1.24 (1.00)	1.24 (1.13)
≥60	1.92 (1.22)	1.92 (1.00)	1.92 (1.22)
Type of medical school attended (ref = ranked ?21)
US top 20	1.89 (1.33)	1.9 (1.00)	1.90 (1.33)
Foreign schools	1.40 (0.76)*	1.41 (1.00)*	1.40 (0.76)*
Total number of antipsychotic prescriptions	−0.003 (0.000)***	−0.003 (0.000)***	−0.003 (0.000)***
Number of other antipsychotic prescribers in the same organizations (ref = 0)
1-9	0.07 (1.46)	0.06 (1.00)	0.06 (1.46)
≥10	1.27 (1.47)	1.26 (1.00)	1.28 (1.47)
Practice location (ref = otherwise)
Urban only	0.78 (1.20)	0.79 (1.00)	0.78 (1.20)
Organizational setting
Organization specialty (ref = otherwise)
Any affiliation with a behavioral health organization	1.46 (1.95)	1.42 (2.00)	1.44 (1.95)
Organizational affiliation type (ref = outpatient only)
Inpatient only	2.77 (2.08)	2.74 (2.00)	2.76 (2.08)
Both inpatient and outpatient	1.48 (1.23)	1.50 (1.00)	1.49 (1.22)
Organization size	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)
Intercept	50.75 (4.23)***	50.85 (4.00)***	50.79 (4.23)***

Notes:

^*Significant at 10 %;

^**significant at 5 %;

^***significant at 1 %.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved. AMA Masterfile, 2011.

Table A5.

Sensitivity Analysis Results for HHI of Ingredients

Variables	Coefficients (standard errors)
	Equal weights 1:1	Weighting ratio 10:1	Weighting ratio 2:1
Patient caseload characteristics
Female patients (%)	3.80 (2.24)*	3.80 (2.24)*	3.80 (2.24)*
SSI-eligible patients (%)	−13.14 (2.32)***	−13.14 (2.32)***	−13.14 (2.32)***
Non-Hispanic whites (%)	−2.03 (1.27)	−2.12 (1.26)*	−2.07 (1.27)*
Patients <18 yrs (%)	8.04 (1.61)***	8.00 (1.61)***	8.03 (1.61)***
Patients ≥50 yrs (%)	−5.63 (3.03)*	−5.59 (3.03)*	−5.60 (3.03)*
Patients with serious mental illnesses (%)	−9.00 (2.26)***	−9.00 (2.25)***	−9.00 (2.26)***
Patients with 2+ non-mental comorbidities (%)	3.77 (3.07)	3.69 (3.08)	3.74 (3.08)
Patients enrolled in fee-for-services (%)	1.73 (1.09)	1.80 (1.09)*	1.77 (1.09)
Physician characteristics
Physician sex (ref = male)
Female	96.15 (58.85)*	97.88 (58.88)*	96.77 (58.87)*
Physician age (ref = <40)
40-49	66.89 (83.84)	61.14 (83.81)	64.86 (83.87)
50-59	81.43 (82.90)	75.34 (82.84)	78.99 (82.91)
≥60	107.42 (89.67)	102.16 (89.71)	105.51 (89.73)
Type of medical school attended (ref = ranked ≥21)
US top 20	127.76 (97.91)	132.42 (97.91)	129.44 (97.92)
Foreign schools	62.04 (55.90)	62.31 (56.00)	62.18 (55.94)
Total number of antipsychotic prescriptions	−0.22 (0.035)***	−0.22 (0.035)***	−0.22 (0.035)***
Number of other antipsychotic prescribers in the same organizations (ref = 0)
1-9	−65.34 (107.82)	−66.78 (107.66)	−65.97 (107.79)
≥10	−36.41 (110.21)	−32.50 (109.72)	−34.82 (110.08)
Practice location (ref = otherwise)
Urban only	39.62 (88.73)	41.70 (88.82)	40.30 (88.77)
Organizational setting
Organization specialty (ref = otherwise)
Any affiliation with a behavioral health organization	52.18 (145.26)	45.43 (145.65)	49.71 (145.51)
Organizational affiliation type (ref = outpatient only)
Inpatient only	135.49 (155.87)	136.15 (155.77)	135.49 (155.94)
Both inpatient and outpatient	95.71 (91.53)	102.52 (90.58)	98.63 (91.14)
Organization size	0.07 (0.07)	0.04 (0.07)	0.06 (0.07)
Intercept	3737.34 (312.77)***	3754.72 (312.65)***	3744.86 (312.81)***

Notes:

^*Significant at 10 %;

^*significant at 5 %;

^***significant at 1 %.

Sources: IMS Health, HealthCare Organizational Services, 2011, IMS Health Incorporated. All rights reserved. AMA Masterfile, 2011.