Use of antipsychotic medications among HIV-infected individuals with schizophrenia

Schizophrenia Research 71 (2004) 435 – 444 www.elsevier.com/locate/schres

Use of antipsychotic medications among HIV-infected individuals with schizophrenia Ann Bagchi a,*, Usha Sambamoorthi a, Elizabeth McSpiritt a, Philip Yanos a,b, James Walkup a, Stephen Crystal a a

Institute for Health, Health Care Policy, and Aging Research, Rutgers, The State University of New Jersey, 30 College Avenue, New Brunswick, NJ 08901, USA b UMDNJ-New Jersey Medical School, 183 South Orange Avenue, P.O. Box 1709, Newark, NJ 07101-7897, USA Received 17 March 2003; received in revised form 2 February 2004; accepted 13 February 2004 Available online 27 April 2004

Abstract Persons with schizophrenia face elevated risk of infection with HIV. While HIV therapy is demanding, patients diagnosed with both conditions also require appropriate and consistent management of their psychiatric illness, for the same reasons that generally apply to persons with schizophrenia and because untreated psychiatric illness can interfere with full participation in HIV care. This study examines the correlates of use of and persistence on antipsychotic medications among HIV-infected individuals with schizophrenia, using merged New Jersey HIV/AIDS surveillance data and paid Medicaid claims. Persistence was defined as at least 2 months of medication use in a quarter. We identified 350 individuals who were dually diagnosed with HIV and schizophrenia. Overall, 81% of these beneficiaries had at least one claim for an antipsychotic medication at some point between 1992 and 1998. Multivariate techniques were used, including simple logistic regressions on use and robust longitudinal regressions that controlled for repeated observations on the same individual and treatment gaps. Among users of antipsychotic medications, persistence was very low at 37%. Racial/ethnic minorities were less likely to receive atypical antipsychotic medications. Use of atypical antipsychotics was associated with higher persistence. Our study confirmed past findings of racial disparities in the receipt of atypical antipsychotic medications. Findings suggest that use of atypical medications may benefit individuals dually diagnosed with HIV and serious mental illness. D 2004 Elsevier B.V. All rights reserved. Keywords: Medicaid; AIDS/HIV; Antipsychotic medications; Schizophrenia; Persistence

1. Introduction Previous research among New Jersey Medicaid recipients found that 5.7% of HIV-infected individuals

* Corresponding author. Tel.: +1-732-932-6949; fax: +1-732932-8592. E-mail address: [email protected] (A. Bagchi). 0920-9964/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2004.02.021

also suffered from schizophrenia (Walkup et al., 1999). Other studies confirm higher rates of serious mental illness among individuals with HIV infection than among the general population (Blank et al., 2002; Kilbourne et al., 2001; Cournos and McKinnon, 1997; Stoskopf et al., 2001), with a concomitant increase in the burden of care and treatment for HIV/AIDS. A number of studies document the significance of managing the symptoms of mental disorders as a means of

436

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

improving adherence to treatments for chronic conditions (Wang et al., 2002; Ruo et al., 2003; Sambamoorthi et al., 2000). As Kilbourne et al. (2001) noted, while HIV infection itself involves complex and intensive treatment, ‘‘patients may require equally intensive management of psychiatric comorbidity to fully participate in HIV care’’ (p. S27). Studies among individuals with schizophrenia indicate that although adherence to antipsychotic medications leads to significant improvements in quality of life, control of schizophrenia symptoms and the prevention of relapse (Terkelsen and Menikoff, 1995; Weiden and Olfson, 1995), persistence in the use of antipsychotics averages only about 50% (Young et al., 1999). The 1990s saw the development of atypical antipsychotic medications that showed promise in improving adherence to pharmacological treatments for schizophrenia (Olfson et al., 2000; Rosenheck et al., 2000), possibly due to lower incidence of extrapyramidal side effects than conventional antipsychotic medications (Arana, 2000; Conley, 2000; Conley and Meltzer, 2000; Garver, 2000; Miller, 2000). These advances could prove especially beneficial to HIVinfected individuals, since some exhibit particular vulnerability to the side effects inherent in typical antipsychotic medications (Horwath et al., 1996; Meyer et al., 1998). Our study provides the first estimates of antipsychotic use and persistence among a subset of persons with schizophrenia who are also infected with HIV.

2. Methods 2.1. Study population Because state Medicaid programs assume much of the cost for treating serious mental illness (McKusick et al., 1996) and because the Medicaid-eligible population includes many individuals with comorbid HIV/AIDS and schizophrenia (Blank et al., 2002; Walkup et al., 1999), Medicaid claims data offer important information on antipsychotic use in individuals living with HIV and schizophrenia. The study population was identified through a file match between New Jersey State’s HIV/AIDS Registry and the Medicaid eligibility file, details of which appear elsewhere (Sambamoorthi et al., 1999, 2001). We

restricted our analyses to adults with schizophrenia (n = 459) who participated in Medicaid between January 1992 and December 1998 and received a diagnosis of HIV or AIDS between January 1992 and March 1996. Because we did not have encounter data for managed care participants, we excluded the 109 individuals identified as enrollees in these programs; this gave us a final sample size of 350 beneficiaries. Statistical analyses revealed that individuals excluded for managed care enrollment were similar to the study population in terms of gender, race/ethnicity, region and primary diagnoses, although they were older and less likely to be on Medicare. 2.2. Measures We identified individuals with schizophrenia based on diagnostic codes conforming to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) recorded in health care encounters. Using an algorithm similar to one developed and validated by Lurie et al. (1992), we classified individuals with at least one inpatient claim or two outpatient claims with ICD-9-CM codes of 295 (schizophrenia) as persons with schizophrenia (Walkup et al., 1999). 2.3. Dependent variables We identified antipsychotics using the National Drug Codes (NDC) recorded in pharmacy claims. We distinguished between individuals who used typical antipsychotics only (including those who received medications by injection), atypicals only and both atypical and conventional antipsychotics. Following the method of an earlier study, in regression analyses we combined users of any atypical medication into one group (Nightengale et al., 1998). The list of NDCs used in this study and the classification of medications can be furnished upon request. 2.4. Independent variables 2.4.1. Demographic characteristics The Registry provided information on several demographic characteristics. Men served as the reference category for gender in multivariate analyses. Using white as the comparison group, individuals

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

were classified into one of three racial/ethnic groups: African American, Latino, and white. Because the effect of age is likely to be nonlinear, we included the following categories of age at AIDS/HIV diagnosis: 18 –29 years (the reference group), 30 –39 years and 40 years and older. We contrasted patterns of antipsychotic medication use between high-HIV-prevalence areas of the state (counties nearest to New York City and to Philadelphia, comprising Essex, Hudson, Passaic, Bergen, Union and Camden) and other areas of New Jersey (New Jersey DHSS, 1996). Based on data from the Registry, we utilized three categories for mode of HIV transmission: injection drug user, other patients and those with missing data. We originally included current substance abuse as an independent variable; however, it did not prove to be a significant predictor in any of the models tested, so we elected to exclude it from the analyses presented. 2.4.2. Additional variables Primary diagnosis (HIV vs. AIDS), year of diagnosis and vital status (deceased during or surviving to the end of the study period) served as controls for disease stage. Based on our previous research, we expected individuals dually enrolled in Medicare to have greater access to outpatient medical treatments (Sambamoorthi et al., 2001). Therefore, we included Medicare coverage, based on claim type, as a covariate in our study. Some of the patients in our sample participate in New Jersey’s AIDS Community Care Alternatives Program (ACCAP), an HIV-specific Medicaid home and community-based care waiver program which offers a variety of services, including case management and private duty nursing that could affect treatment patterns (Merzel et al., 1992; Sambamoorthi et al., 2001). We identified as participants in the waiver program individuals with at least one claim for waiver eligible services and included this as a covariate in all analyses. 2.5. Analytic procedures We tested for significant bivariate differences in use of antipsychotic medications using the chi-square test statistic. We performed logistic regressions to predict the factors associated with antipsychotic use. Among users of antipsychotic medications, we derived antipsychotic medication use and type of anti-

437

psychotic medication in each month counting forward from the date of first prescription. Monthly use was aggregated to quarters and beneficiaries were counted as consistent users if their record included at least 2 months of antipsychotic use per quarter. A prior study, using a quarters approach, defined as ‘‘regular’’ users those with at least one prescription claim per quarter (Svarstad et al., 2001). However, this approach designates a minimum threshold of adherence at 33% (i.e., four claims per year). Given the 50% persistence rate cited above, the present study utilized a more conservative approach, with 2 months of use representing a 67% adherence rate. The number of quarters can vary across individuals depending on the time of the first claim and end of follow-up period. To offset this variation we included a control variable for the number of quarters observed. Because repeated measurements from the same persons were used to measure persistence, we applied robust covariance methods (Diggle et al., 1994). The odds of persistent use of antipsychotic medications by person i in quarter t was modeled in a logistic regression model. This model is an extension of the general linear model with a complex error structure and was estimated with the generalized estimating equation (GEE) technique for binary outcome variables using STATA’s xtgee procedure.

3. Results 3.1. Characteristics of the sample The first two columns of Table 1 provide information on the characteristics of the study population. Males represented 54% of the 350 HIV-infected patients with schizophrenia. The majority of patients were African American (65%) with the remainder split equally between Latinos and whites (17% each). Fifty-six percent were aged 30 –39 years at the time of HIV diagnosis, another 16% were 18– 29 and the remaining 29% were 40 years old or older. Based on our indicators of disease status, the majority of patients were at an advanced stage of illness, with 57% receiving a Registry classification of AIDS. Injection drug use was the most common route of exposure, accounting for 63% of the sample. Seventyfive percent of the study cohort lived in the high HIV

438

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

Table 1 Use of antipsychotic medications among HIV-infected Medicaid recipients with schizophrenia N

All Gender Male Female Race/ethnicity White African American Latino Age at diagnosis 18 – 29 years 30 – 39 years 40 and older Mode of transmission Non-IDU Missing Injection drug users County of residence Elsewhere High prevalence Medicare participation Nonparticipant Participant Waiver status Non-ACCAP ACCAP Diagnosis as of 1996 HIV AIDS Vital status as of 1998 Nondecedents Decedents Year of HIV diagnosis Year 1991 – 1992 Year 1993 – 1994 Year 1995 – 1996

Sample % Any Logistic regression (%) use on use Odds ratio

95% CI

350 100

80.6

189 161

54.0 46.0

78.8 82.6

1.34

[0.76, 2.37]

62 226 61

17.7 64.6 17.4

83.9 80.5 77.0

0.97 0.72

[0.44, 2.13] [0.28, 1.85]

55 195 100

15.7 55.7 28.6

80.0 83.1 76.0

1.27 0.84

[0.57, 2.83] [0.36, 1.97]

80 50 220

22.9 14.3 62.9

76.3 84.0 81.4

1.73 1.44

[0.65, 4.59] [0.76, 2.74]

89 261

25.4 74.6

86.5 78.5

0.56

[0.28, 1.14]

225 125

64.3 35.7

81.8 78.4

0.73

[0.41, 1.30]

315 35

90.0 10.0

80.0 85.7

2.01

[0.71, 5.69]

150 200

42.9 57.1

82.7 79.0

0.96

[0.51, 1.80]

237 113

67.7 32.3

83.5 74.3w

0.52*

[0.28, 0.98]

89 202 59

25.4 57.7 16.9

82.0 79.2 83.1

0.80 0.82

[0.41, 1.57] [0.32, 2.09]

w denotes significant bivariate group differences at the 5% level, *p < 0.05.

prevalence urban areas of the state at the time of initial diagnosis. One-third of individuals died before the end of the follow-up period. Only 10% were enrolled

in the waiver program, and 36% were enrolled in both Medicaid and Medicare at some time during the study period. 3.2. Use of antipsychotic medications Bivariate group differences in the use of any antipsychotic medications and results from multivariate regressions also appear in Table 1. Of the 350 patients, 282 (81%) had at least one paid claim for an antipsychotic medication between January 1992 and December 1998. We found significant bivariate and multivariate differences in the use of antipsychotic medications by vital status. There were no other significant differences in use of antipsychotics across the study variables, suggesting that the majority of these dually diagnosed individuals had some access to needed medications. 3.3. Use of atypical antipsychotic medications Table 2 reports the findings on use of atypical antipsychotic medications. Among the 282 patients using antipsychotics, 34% had at least one claim for an atypical antipsychotic. In the bivariate comparisons, we found significant differences in the use of atypical medications by race and illness severity. A significantly lower proportion of racial minorities (34% for African Americans and 17% for Latinos) than whites (50%) had at least one prescription for an atypical. However, it should be noted from the outset that some of these sample cells are fairly small. For example, only 47 Latinos were included in the population of beneficiaries with any antipsychotic use. Therefore, the 17% reported above represents only eight individuals. Additional bivariate results indicate that individuals at a more advanced stage of HIV illness (those diagnosed with AIDS, those who died before the end of the study period and those with an earlier diagnosis of HIV) were less likely to have claims for atypicals than those with lower illness severity. Several of these findings held up in multivariate analyses. Latinos and deceased individuals had lower odds of a claim for an atypical medication, controlling for other factors, while those odds were 2.45 times greater among individuals most recently (1995 – 1996) diagnosed with HIV. The findings for African Amer-

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444 Table 2 Use of atypical antipsychotic medications among HIV-infected Medicaid recipients with schizophrenia Among antipsychotic users N All Gender Male Female Race/ethnicity White African American Latino Age at diagnosis 18 – 29 years 30 – 39 years 40 and older Mode of transmission Non-IDU Missing Injection drug users County of residence Elsewhere High prevalence Participation in medicare Nonparticipant Participant Waiver status Non-ACCAP ACCAP Diagnosis as of 1996 HIV AIDS Vital status as of 1998 Nondecedents Decedents Year of HIV diagnosis Year 1991 – 1992 Year 1993 – 1994 Year 1995 – 1996

%

Logistic regression on use of atypicals

% Any Odds atypicals ratio

Adjusted 95% CI

282 100.0 33.7 149 133

52.8 29.5 47.2 38.3

1.91*

52 182 47

18.4 50.0 64.5 33.5w 16.7 17.0w

0.53 [0.26, 1.09] 0.20** [0.07, 0.56]

44 162 76

15.6 29.5 57.4 36.4 27.0 30.3

1.52 1.33

[0.68, 3.38] [0.54, 3.25]

61 42 179

21.6 34.4 14.9 42.9 63.5 31.3

1.12 1.12

[0.45, 2.78] [0.56, 2.25]

77 205

27.3 37.7 72.7 32.2

0.87

[0.46, 1.62]

184 98

65.2 30.4 34.8 39.8

1.68

[0.93, 3.04]

252 30

89.4 34.9 10.6 23.3

1.09

[0.38, 3.15]

124 158

44.0 41.9 56.0 27.2w

0.63

[0.35, 1.13]

198 84

70.2 41.9 29.8 14.3w

0.24** [0.11, 0.52]

73 160 49

25.9 24.7 56.7 33.8w 17.4 46.9w

1.55 2.45*

[1.08, 3.39]

[0.78, 3.08] [1.04, 5.79]

w denotes significant bivariate group differences at the 5% level, *p < 0.05, **p < 0.01.

439

after consideration of other sociodemographic factors, women were more likely to recieve atypical antipsychotics than men. 3.4. Gaps in antipsychotic claims In analyzing treatment persistence, we found that 89% (250) of individuals with any use of antipsychotic medications exhibited gaps in their prescription claims. This was higher than the 63% reported in a recent study of first admission patients with a psychotic disorder (Mojtabai et al., 2002). The mean length of treatment gap also varied between the two studies, with our study population showing sensitivity to the presence of outliers. Table 3 presents analyses on the mean and median length of gaps across the set of independent variables. The results suggest that African Americans and surviving individuals experienced much longer treatment gaps than members of other ethnic groups or those who died during the study period. Both findings held up in logistic regression analyses, demonstrating a need to control for treatment gaps in further analyses. Preliminary analyses indicated significantly longer treatment gaps for ACCAP participants than nonparticipants. As these analyses utilize a continuous, highly skewed outcome variable and involve relatively small subsamples, they are vulnerable to the influence of small numbers of outlying cases. Outliers may explain this counterintuitive finding on the association of gap duration with ACCAP participation, as the ACCAP group included two individuals with extreme outlying values (one with a mean of 630 days and outlying value of 690 days and the other with a mean of 241 days and outlying value of 1107 days). Analyses excluding these two cases showed no significant differences in treatment gaps by waiver participation. The inclusion or exclusion of these two cases did not materially affect the other findings we report, and they are excluded in the results shown in the tables. 3.5. Persistence on antipsychotic medications

icans, those diagnosed with AIDS and individuals diagnosed with HIV in 1993 –1994 reduced to insignificance with the introduction of additional predictors. The multivariate analysis further reveals that,

Table 4 reports the bivariate and multivariate results for antipsychotic persistence. The bivariate results suggest significant differences in persistence

440

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

Table 3 Mean and median gaps in antipsychotic claimsy N

Total Gender Male Female Race/ethnicity White African American Latino Age at diagnosis 18 – 29 years 30 – 39 years 40 and older Mode of transmission Non-IDU Missing Injection drug users County of residence Elsewhere High prevalence Medicare Nonparticipant Participant Waiver status Non-ACCAP ACCAP Diagnosis as of 1996 HIV AIDS Vital status as of 1998 Nondecedents Decedents Year of HIV diagnosis Year 1991 – 1992 Year 1993 – 1994 Year 1995 – 1996 Type of drug Typicals only Atypicals onlya Atypical/typicala

Mean

Pr>AtA

Median

95% CI

Logistic regression OR

95% CI

280

75.6

40.7

148 132

75.0 76.2

0.92

40.3 40.7

[33.4, 47.1] [35.3, 49.1]

0.84

[0.43, 1.65]

52 180 47

50.9 84.8 64.3

0.04* 0.52

35.2 43.2 39.2

[27.2, 43.0] [36.8, 51.0] [29.1, 53.2]

4.29** 2.77

[1.37, 13.41] [0.73, 10.53]

44 162 74

99.2 72.7 67.6

0.13 0.11

54.9 41.6 29.3

[46.3, 79.1] [36.6, 48.4] [27.6, 38.3]

0.75 0.49

[0.32, 1.76] [0.18, 1.37]

60 42 178

75.4 84.0 73.5

0.67 0.90

37.4 46.7 40.2

[30.7, 47.6] [32.8, 64.3] [34.7, 46.9]

0.49 1.66

[0.14, 1.71] [0.73, 3.79]

77 203

72.4 76.7

0.76

33.6 44.0

[29.2, 40.8] [38.4, 51.4]

1.83

[0.83, 4.03]

183 97

77.4 72.2

0.68

38.2 46.4

[33.1, 44.1] [36.6, 54.6]

0.73

[0.36, 1.51]

252 28

76.3 69.2

0.73

40.7 38.1

[36.3, 46.7] [29.2, 57.2]

2.09

[0.68, 6.42]

123 157

82.5 69.7

0.30

44.7 43.4

[35.8, 55.5] [34.2, 44.7]

0.89

[0.45, 1.78]

196 84

86.8 47.6

0.00**

44.0 33.4

[38.7, 54.5] [29.6, 41.6]

0.27**

[0.10, 0.71]

73 158 49

85.9 74.4 63.6

0.43 0.25

44.0 38.5 35.7

[38.0, 67.6] [33.3, 48.6] [29.5, 46.3]

0.90 0.47

[0.43, 1.89] [0.16, 1.35]

187 9 84

58.5 84.6 71.5

0.38

34.4 46.5 38.7

[10.0, 45.3] [35.3, 53.4] [34.2, 45.8]

1.35

[0.67, 2.73]

y—Analyses exclude two outlying cases. *p < 0.05, **p < 0.01, a—categories collapsed to ‘‘Any Atypicals’’ in regression analyses.

across most of the independent variables. The profile of a persistent user, based on the bivariate findings, is an older (age 40 or more) white person with AIDS living outside the high prevalence areas. These individuals typically did not participate in Medicare and had at least some use of atypical medications. However, in multivariate analyses, once controls for gaps and the number of observed quarters are introduced, the only remaining differ-

ence was for type of antipsychotic medication. Given the small number of individuals taking only atypical medications, in the logistic model users of any atypical antipsychotic were combined into a single group. The results indicate that the odds of persistence were 4.25 [CI 3.35– 5.39] times greater among patients with one or more claims for atypical antipsychotics than among those taking only the older medications.

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444 Table 4 Persistence on antipsychotic medications among HIV-infected Medicaid recipients with schizophrenia Sample N Total Gender Male Female Race/ethnicity White African American Latino Age at diagnosis 18 – 29 years 30 – 39 years 40 and older Mode of transmission Non-IDU Missing Injection drug users County of residence Elsewhere High prevalence Medicare Nonparticipant Participant Waiver status Non-ACCAP ACCAP Diagnosis as of 1996 HIV AIDS Vital status as of 1998 Nondecedents Decedents Year of HIV diagnosis Year 1991 – 1992 Year 1993 – 1994 Year 1995 – 1996 Type of drug Typicals only Atypicals onlya Atypical/typicala Quarter After initiation w

%

280 100

Persistent Logistic regression quarters OR 95% CI (%) 37.3

148 132

52.9 38.3 47.1 36.0

1.12

[0.62, 2.00]

52 180 47

18.6 50.0w 64.3 36.0 16.8 27.3

0.90 0.58

[0.46, 1.77] [0.19, 1.74]

44 162 74

15.7 21.3w 57.9 38.7 26.4 44.3

1.92 2.48

[0.80, 4.57] [0.91, 6.79]

60 42 178

21.4 40.8w 15.0 41.4 63.6 34.8

1.11 1.02

[0.45, 2.75] [0.51, 2.03]

77 203

27.5 43.3w 72.5 35.0

1.00

[0.52, 1.94]

183 97

65.4 43.2w 34.6 33.8

1.58

[0.87, 2.87]

252 28

90.0 37.2 10.0 37.7

1.82

[0.53, 6.26]

123 157

43.9 35.6w 56.1 39.0

0.93

[0.51, 1.70]

196 84

70.0 19.7w 30.0 37.6

0.40

[0.15, 1.07]

73 158 49

26.1 38.4w 56.4 34.4 17.5 45.2

1.08 1.22

[0.54, 2.17] [0.52, 2.86]

187 9 84

66.8 31.2w 3.2 41.5 30.0 47.0

4.25** [3.35, 5.39] 0.96** [0.94, 0.97]

denotes significant bivariate group differences at the 5% level, **p < 0.01, a—categories collapsed to ‘‘Any Atypicals’’ in regression analyses.

441

4. Discussion Our study offers the first estimates of antipsychotic use and persistence among individuals dually diagnosed with HIV and schizophrenia. Despite the small sample sizes employed in this study, the analyses indicated lower use of atypical antipsychotics among HIV-infected Latinos with comorbid schizophrenia than among whites. Although a recent study of care in physicians’ offices and hospital outpatient departments suggested that the racial/ethnic gap in prescriptions for atypical antipsychotics narrowed in these settings between 1992 and 2000 (Daumit et al., 2003), the majority of studies document a persistent ethnic gap across time periods and treatment settings (Kreyenbuhl et al., 2003; Valenti et al., 2003; Copeland et al., 2003; Mark et al., 2002; Leslie and Rosenheck, 2001; Kuno and Rothbard, 2002). Since the analyses reported herein examined HIV-positive schizophrenics with comparable socioeconomic status and health coverage (including a prescription drug benefit), it emphasizes the need to examine factors other than economically based access barriers to explain the ethnic disparity observed in our findings. The rates of persistence found in our analyses agree with those reported from earlier estimates that used a similar, but less conservative, definition of medication adherence (Svarstad et al., 2001). Based on our findings, atypical medications are associated with advantages for adherence when compared with the older antipsychotics. This is the first such finding in a dually diagnosed population and could have significant implications for treatment of comorbid HIV and schizophrenia. Since improved adherence with the use of atypicals is hypothesized to be due to their lower side effect profiles and, in some cases, greater effectiveness, expanding their use might prove particularly beneficial for those requiring long-term treatment. However, use of atypicals probably interacts with a variety of other factors in shaping treatment persistence; more research is needed to clarify these relationships. Prior studies documented and analyzed gaps in antipsychotic prescription claims (Mojtabai et al., 2002, 2003). Several factors may account for the fact that the gaps identified in our study were significantly longer than in earlier studies (Dolder et al., 2002; Svarstad et al., 2001; Mojtabai et al., 2002). First, our

442

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

study used a longer follow-up period (up to 7 years) than that used in any prior publication on antipsychotic persistence. Many factors may intervene over such a long study period to influence persistence rates, regardless of the type of medication being used. We examined the possibility that patient hospitalization explained many of the gaps in prescription claims, but in very few cases did an inpatient stay co-occur with a break in medication treatment. Treatment recommendations from the PORT study suggest that patients should remain on antipsychotic medications for at least 1 year after symptom stabilization (Lehman et al., 1998a,b). In addition, the American Psychiatric Association (APA) has advised the continuation of medications for 5 years or longer for patients with multiple psychotic episodes (APA, 1997). However, as with many chronic conditions care must be customized for each individual. Breaks in refills may reflect legitimate patient-provider decisions to discontinue medications; however, our data offer no way to test this supposition. 4.1. Limitations Several features of our study may limit the generalizability of our findings. One limitation is the inability to determine whether or not individuals who filled antipsychotic prescriptions actually took the medication. Without that information it is possible that our models overestimate the use and persistence of antipsychotic medications. However, while electronic monitoring is reported to be more sensitive to non-adherence (Arnsten et al., 2001), pharmacy refill records are reported to be as satisfactory as the Medication Event Monitoring System for examining persistence of use (Frick et al., 1998). Additionally, by relying on administrative data on filled prescriptions, this study does not suffer from recall-bias associated with self-reported measurements of pharmacy utilization, which tend to overestimate use (Arnsten et al., 2001). Analyses revealed that a large number of patients in our sample (39%) switched from typical to atypical medications at some point during the study period. Since only 3% of our study population had claims exclusively for atypical antipsychotics, our analyses of atypical users effectively represent these ‘‘switchers,’’ and the higher adherence on atypicals cannot be

categorically attributed to atypical use. Personal communications with staff at the New Jersey’s Medicaid pharmacy program (Kocsardy, 2002) confirmed that state policy was unlikely to have influenced such switching. For example, New Jersey’s Medicaid program did not require a patient to first try conventional antipsychotics before being allowed a prescription for an atypical medication. Recent studies of antipsychotic prescribing patterns suggest that polypharmacy with antipsychotics is not uncommon (Tapp et al., 2003; Freudenreich and Goff, 2002). Future analyses of adherence should verify whether such prescribing occurs in the population being studied and test for the possible influence on treatment persistence. We did not have indicators of disease severity and year of diagnosis for schizophrenia although prior research has validated the claim-based algorithm used to identify patients with schizophrenia (Lurie et al., 1992). These are important factors in the consideration of medication usage as they have implications for antipsychotic prescribing patterns. The small sample size introduced limitations for statistical modeling, as only larger effects could be detected. This may be why, in a few cases, large differences in percentages did not translate into significant differences in the regression analyses. For example, despite a large difference in the percentage of nonACCAP versus ACCAP members using atypicals (35% and 23%, respectively), the effect size did not reach statistical significance. More importantly, the small sample size limited the number of comparisons and analyses that could be performed. Additionally, our study is based on a single payer (Medicaid) and may miss some episodes of care that might occur in settings not reimbursed by Medicaid (e.g. uncompensated care). However, it is believed that most of the care received by this population is reimbursed by Medicaid (McKusick et al., 1996). As they excluded managed care enrollees, the study results cannot be generalized to this population. 4.2. Conclusion HIV infection has come to be seen as a chronic condition. For individuals with schizophrenia and comorbid HIV, it will be increasingly important to understand how to improve their persistence on antipsychotic medications as survival times increase. This

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

research, using a long study period, is a step toward improving that understanding.

Acknowledgements This research was supported by grants R01 MH 58984 and P30 MH 43450 from NIMH, and P20HS11825 from AHRQ. The findings and opinions reported here are those of the authors and do not necessarily represent the views of any other individuals or organizations. The authors wish to acknowledge the research assistance provided by Michelle Kennedy, and thank the New Jersey Department of Human Services and Department of Health and Senior Services for their cooperation in making data available.

References American Psychiatric Association, 1997. Practice guidelines for the treatment of patients with schizophrenia. Am. J. Psychiatry 154 (Suppl. 4), 1 – 63. Arana, G.W., 2000. An overview of side effects caused by typical antipsychotics. J. Clin. Psychiatry 61 (Suppl. 8), 5 – 11. Arnsten, J.H., Demas, P.A., Farzadegan, H., Grant, R.W., Gourevitch, M.N., Chang, C.J., Buono, D., Eckholdt, H., Howard, A.A., Schoenbaum, E.E., 2001. Antiretroviral therapy adherence and viral suppression in HIV-infected drug users: comparison of self-report and electronic monitoring. Clin. Infect. Dis. 33, 1417 – 1423. Blank, M.B., Mandell, D., Aiken, L., Hadley, T., 2002. Co-occurrence of HIV and serious mental illness among Medicaid recipients. Psychiatr. Serv. 53 (7), 868 – 873. Conley, R.R., 2000. Risperidone side effects. J. Clin. Psychiatry 61 (Suppl. 8), 20 – 23. Conley, R.R., Meltzer, H.Y., 2000. Adverse events related to olanzapine. J. Clin. Psychiatry 61 (Suppl. 8), 26 – 29. Copeland, L.A., Zeber, J.E., Valenstein, M., Blow, F.C., 2003. Racial disparity in the use of atypical antipsychotic medications among veterans. Am. J. Psychiatry 160, 1817 – 1822. Cournos, F., McKinnon, K., 1997. HIV seroprevalence among people with severe mental illness in the United States: a critical review. Clin. Psychol. Rev. 17 (3), 259 – 269. Daumit, G., Crum, R.M., Guallar, E., Powe, N.R., Primm, A.B., Steinwachs, D.M., Ford, D.E., 2003. Outpatient prescriptions for atypical antipsychotics for African Americans, Hispanics and whites in the United States. Arch. Gen. Psychiatry 60, 121 – 128. Diggle, P., Liang, K.Y., Zeger, S., 1994. Analysis of Longitudinal Data. Oxford Univ. Press, Oxford. Dolder, C.R., Lacro, J.P., Dunn, L.B., Jeste, D.V., 2002. Antipsy-

443

chotic medication adherence: is there a difference between typical and atypical agents? Am. J. Psychiatry 159 (1), 103 – 108. Freudenreich, O., Goff, D.C., 2002. Antipsychotic combination therapy in schizophrenia: a review of efficacy and risks of current combinations. Acta Psychiatr. Scand. 106, 323 – 330. Frick, P.A., Gal, P., Lane, T.W., Sewell, P.C., 1998. Antiretroviral medication compliance in patients with AIDS. AIDS Patient Care STDS 12 (6), 463 – 470. Garver, D.L., 2000. Review of quetiapine side effects. J. Clin. Psychiatry 61 (Suppl. 8), 31 – 33. Horwath, E., Cournos, F., McKinnon, K., Guido, J.R., Herman, R., 1996. Illicit-drug injection among psychiatric patients without a primary substance use disorder. Psychiatr. Serv. 47 (2), 181 – 185. Kilbourne, A.M., Justice, A.C., Rabeneck, L., Rodriquez-Barradas, M., Weissman, S., 2001. General medical and psychiatric comorbidity among HIV-infected veterans in the post-HAART era. J. Clin. Epidemiol. 54, S22 – S28. Kocsardy, B., 2002. Personal Communication. June 5. Kreyenbuhl, J., Zito, J.M., Buchanan, R.W., Soeken, K.L., Lehman, A.F., 2003. Racial disparity in the pharmacological management of schizophrenia. Schizophr. Bull. 29, 183 – 193. Kuno, E., Rothbard, A.B., 2002. Racial disparities in antipsychotic prescription patterns for patients with schizophrenia. Am. J. Psychiatry 159 (4), 567 – 572. Lehman, A., Steinwachs, D., the Survey Co-investigators of the PORT Project, 1998a. At issue: translating research into practice: the Schizophrenia Patient Outcomes Research Team (PORT) treatment recommendations. Schizophr. Bull. 24, 1 – 10. Lehman, A., Steinwachs, D., the Survey Co-investigators of the PORT Project, 1998b. Patterns of usual care for schizophrenia: initial results from the Schizophrenia Patient Outcomes Research Team (PORT) client survey. Schizophr. Bull. 24, 11 – 20. Leslie, D.L., Rosenheck, R., 2001. The effect of institutional fiscal stress on the use of atypical antipsychotic medications in the treatment of schizophrenia. J. Nerv. Ment. Dis. 189, 377 – 383. Lurie, N., Popkin, M., Dysken, M., Moscovice, I., Finch, M., 1992. Accuracy of diagnoses of schizophrenia in Medicaid claims. Hosp. Community Psychiatry 43, 69 – 71. Mark, T.L., Dirani, R., Slade, E., Russo, P.A., 2002. Access to new medications to treat schizophrenia. J. Behav. Health Serv. Res. 29 (1), 15 – 29. McKusick, D., Mark, T., King, E., Harwood, R., Buck, J.A., Dilonardo, J., Genuardi, J.S., 1996. Spending for mental health and substance abuse treatment. Health Aff. 17 (3), 147 – 157. Merzel, C., Crystal, S., Sambamoorthi, U., Karus, D., Kurland, C., 1992. New Jersey’s Medicaid waiver for acquired immunodeficiency syndrome. Health Care Financ. Rev. 13, 27 – 44. Meyer, J.M., Marsh, J., Simpson, G., 1998. Differential sensitivities to risperidone and olanzapine in a human immunodeficiency virus patient. Biol. Psychiatry 44, 791 – 794. Miller, D.D., 2000. Review and management of clozapine side effects. J. Clin. Psychiatry 61 (Suppl. 8), 14 – 17. Mojtabai, R., Lavelle, J., Gibson, P.J., Sohler, N., Craig, T.J., Carlson, G.A., Bromet, E.J., 2002. Gaps in use of antipsychotics after discharge by first-admission patients with

444

A. Bagchi et al. / Schizophrenia Research 71 (2004) 435–444

schizophrenia, 1989 to 1996. Psychiatr. Serv. March 53 (3), 337 – 339. Mojtabai, R., Lavelle, J., Gibson, P.J., Bromet, E.J., 2003. Atypical antipsychotics in first admission schizophrenia: medication continuation and outcomes. Schizophr. Bull. 29 (3), 519 – 530. New Jersey State Department of Health and Senior Services, 1997. New Jersey HIV/AIDS Cases Reported as of December 31, 1996. Division of AIDS Prevention and Control, New Jersey Department of Health and Senior Services, Trenton. Nightengale, B.S., Crumly, J.M., Liao, J., Lawrence, B., Jacobs, E.W., 1998. Current topics in clinical psychopharmacology: economic outcomes of antipsychotic agents in a Medicaid population: traditional agents vs. risperidone. Psychopharmacol. Bull. 34 (3), 373 – 382. Olfson, M., Mechanic, D., Hansell, S., Boyer, C., Walkup, J., Weiden, P., 2000. Predicting medication noncompliance after hospital discharge among patients with schizophrenia. Psychiatr. Serv. 51, 216 – 222. Rosenheck, R., Chang, S., Choe, Y., Cramer, J., Xu, W., Thomas, J., Henderson, W., Charney, D., 2000. Medication continuation and compliance: a comparison of patients treated with clozapine and haloperidol. J. Clin. Psychiatry. 61, 5 – 11. Ruo, B., Rumsfeld, J.S., Hlatky, M.A., Liu, H., Browner, W.S., Whooley, M.A., 2003. Depressive symptoms and health-related quality of life: the Heart and Soul Study. JAMA 290 (2), 215 – 221. Sambamoorthi, U., Collins, S.R., Crystal, S., Walkup, J., 1999. Home-care use and expenditures among Medicaid beneficiaries with AIDS. Health Care Financ. Rev. 20 (4), 161 – 177. Sambamoorthi, U., Walkup, J., Olfson, M., Crystal, S., 2000. Antidepressant treatment and health services utilization among

HIV-infected Medicaid patients diagnosed with depression. J. Gen. Intern. Med. 15, 311 – 320. Sambamoorthi, U., Collins, S., Crystal, S., 2001. Dually eligible individuals with AIDS: characteristics and health services use. J. Health Soc. Policy 14 (1), 19 – 35. Stoskopf, C.H., Kim, Y.K., Glover, S.H., 2001. Dual diagnosis: HIV and mental illness, a population-based study. Community Ment. Health J. 37 (6), 469 – 479. Svarstad, B.L., Shireman, T.I., Sweeney, J.K., 2001. Using drug claims data to assess the relationship of medication adherence with hospitalization and costs. Psychiatr. Serv. 52, 805 – 811. Tapp, A., Ernst Wood, A., Secrest, L., Erdmann, J., Cubberley, L., Kilzieh, N., 2003. Combination antipsychotic therapy in clinical practice. Psychiatr. Serv. January 54 (1), 55 – 59. Terkelsen, K., Menikoff, A., 1995. Measuring the cost of schizophrenia: implications for the post-institutional era in the US. PharmacoEconomics 8, 199 – 222. Valenti, A.M., Narendran, R., Pristach, C.A., 2003. Who are patients on conventional antipsychotics? Schizophr. Bull. 29, 195 – 199. Wang, P.S., Bohn, R.L., Knight, E., Glynn, R.J., Mogun, H., Avorn, J., 2002. Noncompliance with antihypertensive medications: the impact of depressive symptoms and psychosocial factors. J. Gen. Intern. Med. 17, 504 – 511. Weiden, P., Olfson, M., 1995. Cost of relapse in schizophrenia. Schizophr. Bull. 21, 419 – 429. Young, J., Spitz, R., Hillbrand, M., Daneri, G., 1999. Medication adherence failure in schizophrenia: a forensic review of rates, reasons, treatments and prospects. J. Am. Acad. Psychiatry Law 27 (3), 426 – 444.