- Email: [email protected]
Effect of initial ziprasidone dose on treatment persistence in schizophrenia
Schizophrenia Research 83 (2006) 277 – 284 www.elsevier.com/locate/schres
Effect of initial ziprasidone dose on treatment persistence in schizophrenia C. Daniel Mullins a,*, Fadia T. Shaya a, Julie Magno Zito a, Nour Obeidat a, John Naradzay a, David J. Harrison b a
University of Maryland School of Pharmacy, 515 W. Lombard Street, Room 262, Baltimore, MD 21201, United States b Pfizer Inc., U.S. Outcomes Research Department, 235 E 42nd Street, New York, NY 10017, United States Received 20 October 2005; received in revised form 13 January 2006; accepted 18 January 2006 Available online 20 March 2006
Abstract Objective: To determine the relationship between ziprasidone initial dose and treatment persistence among patients diagnosed with schizophrenia. Method: Adult Medicaid recipients (N = 1096) diagnosed with schizophrenia who had ziprasidone prescription claims between July 1, 2001 and September 30, 2003, were categorized by initial dose: low (20–60 mg per day, n = 464), medium (61–119 mg per day, n = 320) and high dose (120–160 mg per day, n = 312). Treatment persistence up to 365 days was measured using refill patterns, allowing 15-day gaps between expected refill dates. Multivariate survival analysis explored the simultaneous impact of age, gender, race, previous hospitalization, and concomitant medication usage, in addition to initial dose of ziprasidone. Sensitivity analysis tested the robustness of results with different definitions for persistence and allowable gaps between refills. Results: Discontinuation rates across the observation period (maximum, 12 months per individual) were lower for patients initiated with high-dose than low-dose ziprasidone ( P = 0.001). Other factors significantly associated with greater discontinuation of medication were monotherapy (versus combination therapy) and hospitalization within the 6 months prior to the index date of therapy. Black race was associated with greater discontinuation, although this was not consistent across sensitivity analyses. Conclusions: Patients with schizophrenia started on high doses of ziprasidone have lower discontinuation rates in a retrospective Medicaid database than patients started on low doses. These results were robust across various sensitivity analyses. D 2006 Elsevier B.V. All rights reserved. Keywords: Persistence; Schizophrenia; Ziprasidone; Medicaid; Atypical antipsychotic
1. Introduction
* Corresponding author. Tel.: +1 410 706 0879; fax: +1 410 706 5394. E-mail address: [email protected] (C.D. Mullins). 0920-9964/$ - see front matter D 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2006.01.013
Successful management of schizophrenia is dependent on patient adherence to recommendations for consistent and long-term antipsychotic therapy (Keith and Kane, 2003). Adherence to therapy can substan-
278
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
tially reduce the chance of relapse in schizophrenia patients (National Institute of Mental Health, 2002; Keith and Kane, 2003). Furthermore, relapse entails a substantial clinical and economic burden (Weiden and Olfson, 1995; Weiden et al., 2004; Gilmer et al., 2004; Knapp et al., 2004). Lack of adherence to therapy is a key concern in the management of schizophrenia (National Institute of Mental Health, 2003). In the recent Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) study, funded by the National Institute of Mental Health (NIMH), 74% of patients discontinued therapy within 18 months (Phase I of the study), citing reasons such as lack of efficacy or intolerable side effects (Lieberman et al., 2005). Numerous patient-related, medication-related, and environmental and social risk factors have been proposed to affect adherence (Keith and Kane, 2003; Perkins, 2002; Lacro et al., 2002). Of the medicationrelated risk factors, drug dose has been suggested to play a part in nonadherence, but clear evidence has not yet been published regarding whether higher or lower doses of antipsychotic agents improve adherence in actual clinical practice. Clinician understanding of initial dose titration and target doses has evolved largely through clinical experience for most atypical antipsychotic agents. Ziprasidone is one of such agents, for which there is growing empirical evidence that appropriate dosing in therapy is an important determinant of clinical success (Davis and Chen, 2004). Ziprasidone was approved for use in the US in 2001 and its prescribing information recommends an initial dose for schizophrenia of 20 mg bid, with titration as needed up to 80 mg bid (US Package Insert). Subsequent clinical trial as well as postmarketing experience have indicated that optimal response for ziprasidone is generally obtained at higher doses: Short-term clinical trials have shown daily doses of 120 to 160 mg to be more effective than lower doses in treating acute schizophrenia episodes and in lowering medication discontinuation rates (Kane, 2003). Similarly, daily doses of 80 to 120 mg were found to be efficacious in maintenance therapy (Davis and Chen, 2004; Arato et al., 2002), although a dose–response effect was not consistently discernable (Arato et al., 2002). Symptomatic but stable patients switched from other antipsychotics to ziprasidone responded to doses of 80 to 100 mg, but patients who completed the study took slightly higher doses
(median daily dose of 120 mg) than those who discontinued (Weiden et al., 2003). In the CATIE schizophrenia trial, a higher proportion of ziprasidonetreated patients compared to other agents received maximal doses of the drug (160 mg per day), suggesting an improved effectiveness of ziprasidone at higher doses (Lieberman et al., 2005). Finally, a retrospective database study found that commercially insured patients with an initial prescription for ziprasidone of 120 to 160 mg per day had a significantly lower risk of subsequent treatment discontinuation than did patients started at 40 mg to 79 mg per day (Joyce et al., 2006-this issue). Such findings suggest a need for further evaluation of the relationship between ziprasidone dosing and medication adherence. The objective of this study was to determine whether the previously observed relationship between higher initial ziprasidone dose and lower risk of nonadherence could be extended to a non-commercial, Medicaid population. Specifically, we wanted to test the hypothesis that Medicaid patients initiated on a high dose of ziprasidone would subsequently exhibit lower rates of treatment discontinuation than patients initiated on a low dose of ziprasidone.
2. Methods 2.1. Data set The data set contained all pharmacy and medical service encounter and fee-for-service (FFS) claims from the Maryland Medicaid FFS and HealthChoice programs for recipients who received at least one pharmacy claim for selected antipsychotic drugs between January 1, 2001 and December 31, 2003. Antipsychotic medicines included atypical and conventional antipsychotics, lithium, and anticonvulsants that are frequently used as mood stabilizers. In addition, each recipient’s beginning and end dates of Medicaid eligibility across the 3-year study period were recorded to determine periods of continuous enrollment surrounding the index date, defined as the date of the recipient’s first ziprasidone prescription fill date. Demographic and Medicaid coverage information (i.e., Medicaid coverage group, race, and gender) was extracted from the index (first) prescription claim.
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
2.2. Study population Maryland Medicaid recipients between the ages of 18 and 64 years were included in the analysis if their index (first) prescription for ziprasidone occurred between July 1, 2001 and September 30, 2003. The recipient identifiers from the pharmacy data were then merged with the eligibility data to limit the study population to those ziprasidone users who had a schizophrenia disorder diagnosis (295.xx) any time during the 3-year study period. Furthermore, recipients were required to have continuous Medicaid eligibility for a minimum of 6 months prior to their index ziprasidone prescription. Recipients receiving injectable ziprasidone were excluded since their medical indication might be different from that of oral users. The injectable form is used for the treatment of acute agitation and not in the long-term management of schizophrenia. Age was calculated using date of birth as of the ziprasidone index date. Race was also identified in order to explore its effect on persistence. The racial distribution was consistent with the Maryland Medicaid demographics profile, indicating that most patients were either black or white. The numbers for other races were too small to provide sufficient sample sizes for analysis and, therefore, recipients whose race was other than black or white were excluded. 2.3. Data analysis 2.3.1. Operational definitions of study variables 2.3.1.1. Dose group definitions. Ziprasidone initial dose was categorized as bhighQ, bmediumQ or blowQ. The recommended dose range for ziprasidone is 40 to
279
160 mg per day, administered bid (US Package Insert). The antipsychotic efficacy threshold, 60% D2 occupancy, occurs at approximately 120 mg per day with ziprasidone. Therefore, the lower bound of the high dose range was defined as 120 mg per day, with an upper bound of 160 mg. The medium dose group, 61 to 119 mg per day, and the low dose group, 20 to 60 mg per day, were used to encompass the other commonly prescribed daily doses of ziprasidone, 40 mg and 80 mg per day. Initial doses were calculated from the index prescription by multiplying the quantity dispensed by the tablet’s strength and dividing by the days’ supply that was indicated on the claim. 2.3.1.2. Discontinuation. Time to discontinuation of drug therapy, the major study outcome, was calculated using prescription fills for oral ziprasidone. Starting from the index date of the first prescription, the time to discontinuation was calculated as the days between the index date and the date of the first gap of more than 15 days between prescriptions. Prescribing gaps were calculated by first adding days’ supply to the dispensing date of each prescription to identify the date the prescription should have been exhausted. The first gap in prescribing of 15 days or more was identified as a discontinuation event. Other events could have occurred during patient follow-up, namely disenrollment from Medicaid, psychiatric-related hospitalization, nonpsychiatric-related hospitalization, or persistence until the end of the follow-up period. Patients who fell into these other event categories were considered censored. The first event occurring after the ziprasidone index date determined the recipient’s status (Table 1). In order for a hospitalization to be classified as a psychiatric-related hospitalization, a psychiatric reve-
Table 1 Categories used to identify patient status at the end of follow-up Discontinuation Disenrollment Psychiatric-related hospitalization Nonpsychiatric hospitalization Persistence
Recipient stopped ziprasidone or had a gap in fill dates of 15 days or greater. (A gap in fill dates of 30 days or more was explored in the sensitivity analysis.) Recipient disenrolled from Medicaid before discontinuing ziprasidone. Lack of Medicaid eligibility for more than 1 day was categorized as loss of continuous eligibility. First inpatient hospitalization occurring after the ziprasidone index date lasting more than one day with a revenue code of 114, 124, 134, 144, 154 or 204 recorded within 7 days of the admission date. First nonpsychiatric inpatient-related hospitalization occurring after the ziprasidone index date lasting 7 days or longer. Recipient stayed on ziprasidone until the end of the study period.
280
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
nue code had to appear on a claim within 7 days of the admission date; otherwise, the hospitalization was identified as a nonpsychiatric hospitalization. 2.3.2. Operational definitions of other covariates 2.3.2.1. Prior health care utilization. A set of variables was created to examine health care utilization prior to the index date, eventually creating a numeric variable of count of psychiatric hospitalizations during differing lengths of pre-index periods. We counted the number of unique inpatient hospitalization episodes and outpatient1 and psychiatric outpatient2 visits that were associated with a mental health diagnosis (ICD-9-CM = 290.xx to 319.xx) during a 180-day pre-index period, and the total number of inpatient hospital days 30 days prior to the index date. The 180-day period was used for the analysis. 2.3.2.2. Comorbidities. The number of psychiatric comorbidities was calculated for each patient. Comorbidities were determined from the three diagnosis fields on all physician, inpatient, and outpatient hospital claims, provided that the beginning date of service for the claim occurred any time during the recipient’s continuous Medicaid eligibility. Fifteen unique classifications were used for each mental health diagnosis code, excluding schizophrenia: substance abuse, psychotic disorder, bipolar disorder, depression, adjustment disorder, anxiety disorder, personality disorder, tics, conduct disorder, oppositional disorder, attention deficit hyperactivity disorder, mental retardation, learning disorder, developmental disorders, and other psychiatric disorders. For the diagnosis to be considered present, the recipient had to have at least one claim with a diagnosis from that category. The number of mental health comorbidities was summed over all 15 categories. 2.3.2.3. Concomitant drug use. To account for concomitant use of psychotropic drugs, we compared individuals who used other psychotropic medications when initiated on ziprasidone with those who received 1
CPT = 99 201–5; 99 211–15; 99 241–45; 99 261–63; 99 271–75; 99 281–84. 2 CPT = 90 801–02; 90 804–15; 90 845–57; 90 862; 90 875.
ziprasidone monotherapy. To create a variable that measured concomitant medication use we performed the following: psychotropic drug claims were grouped into eleven drug classes, and psychotropic drug use episodes were defined. Psychotropic drug use was considered concurrent with ziprasidone use under two conditions: (1) the ziprasidone index date occurred between the begin and end dates for two prescription fills for the other psychotropic drugs, and (2) the prescription for the other psychotropic medication began within 30 days of the ziprasidone index date and at least 30 days prior to the ziprasidone end date. Use of at least one concurrent psychotropic medication was sufficient to classify the patient as a concurrent psychotropic medication user. 2.3.3. Statistical analysis Descriptive statistics were calculated to compare the distribution of age, gender, race, number of mental health hospitalizations within 180 days prior to the index date, presence of concurrent medications, and number of mental health diagnosis categories during the study period across the three ziprasidone dosage groups. Chi-square tests were used to test the differences across the dosage groups for categorical variables. Patients were followed for up to 365 days after the initiation of ziprasidone. Survival analyses using Kaplan–Meier survival curves were conducted. The survival time for each recipient was calculated as the number of days from the ziprasidone initiation date to the date of discontinuation. If recipients disenrolled, had psychiatric- or nonpsychiatric-related hospitalization, or reached the end of the study period before they discontinued ziprasidone treatment, recipients were considered as bcensoredQ in the survival analysis. Cox proportional hazard models were used to assess the relative effects of initial drug doses and confounding factors on the hazard ratios of discontinuation. Covariates in the models included age, gender, black race, initial ziprasidone daily dose, and number of mental health hospitalizations within 180 days prior to the index date. The reference group for daily dose was the low dose of initial ziprasidone (i.e., 20 mg to 60 mg per day). To demonstrate the relative effect on persistence of initiating treatment with high-dose versus low-dose ziprasidone, the proportional difference between the
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Kaplan–Meier survival curves of the two strata was calculated. The percentage increase in persistence was calculated by the following equation, where bRPDQ indicates Relative Percentage Difference in survival, bSurvRateQ indicates the Kaplan Meier survival rate (which reflects the chance of persisting longer than time t), and bHQ and bLQ indicate high and low initial dose, respectively. An RPD greater than zero reflects a greater likelihood of continuing ziprasidone on high doses relative to low doses. RPD ¼
SurvRateH;t SurvRateL;t SurvRateL;t
281
allowing a less stringent 30-day gap between ziprasidone prescriptions before a patient would become classified as bdiscontinuedQ. We also conducted a logistic regression analysis to estimate the odds of persisting on ziprasidone or else discontinuing, using the original 15-day gap before discontinuation (vis-a`-vis the hazard of discontinuing ziprasidone or else becoming censored, in the primary survival analysis). Finally, we repeated he logistic regression analysis, allowing for a 30-day gap between prescriptions.
3. Results
2.3.4. Sensitivity analysis Our primary analysis involved the estimation of the relationship between drug dose and subsequent medication discontinuation using Medicaid pharmacy claims data. The results were based on a precise definition of bdiscontinuationQ, i.e. stopping ziprasidone or having a gap in prescription fill dates of 15 days or greater. Recognizing that the results might be influenced by the definition of discontinuation and the statistical analysis of the data, we conducted a number of sensitivity analyses to test the robustness of the results. First, we conducted a survival analysis
3.1. Primary analysis The final study population consisted of 1096 adults with schizophrenia on low (N = 464), medium (N = 320), or high (N = 312) initial doses of ziprasidone. The mean age of the population was 39 years (median = 39.0, S.D. = 10.7). Patients were similar across dose strata, although patients in the high-dose group were significantly more likely to be white and have paranoid schizophrenia (Table 2). Patients in the low-dose group had significantly fewer hospitalizations than those in the high-dose group in the pre-index period (67.9% vs. 53.2% with no prior hospitalizations,
Table 2 Baseline demographic characteristics for Medicaid schizophrenia patients on low, medium or high doses of ziprasidone (N = 1096) Demographic characteristics
Category
Low, 20–60 mg (n = 464)
Medium, 61–119 mg (n = 320)
High, 120–160 mg (n = 312)
p-value (Chi-square test)
Age group
18–39 40–54 55–64 Female Male White Black No Yes 0 admissions 1 admission 2 admissions 3+ admissions None At least 1 0 1 2 3 4+
219 210 35 255 209 194 270 222 242 315 86 29 34 237 227 131 99 89 59 86
157 138 25 166 154 151 169 145 175 207 71 29 13 144 176 70 68 52 61 69
174 116 22 148 164 168 144 117 195 166 93 30 23 158 154 67 67 62 44 72
0.2057
Gender Race Paranoid schizophrenia Psychiatric hospitalizations 180 days prior to index date
Concurrent medications Comorbidities
(47.2) (45.3) (7.5) (55.0) (45.0) (41.8) (58.2) (47.8) (52.2) (67.9) (18.5) (6.3) (7.3) (51.1) (48.9) (28.2) (21.3) (19.2) (12.7) (18.5)
(49.1) (43.1) (7.8) (51.9) (48.1) (47.2) (52.8) (45.3) (54.7) (64.7) (22.2) (9.1) (4.1) (45.0) (55.0) (21.9) (21.3) (16.3) (19.1) (21.6)
(55.8) (37.2) (7.1) (47.4) (52.6) (53.8) (46.2) (37.5) (62.5) (53.2) (29.8) (9.6) (7.4) (50.6) (49.4) (21.5) (21.5) (19.9) (14.1) (23.1)
0.1208 0.0043* 0.0155* 0.0005*
0.2047 0.1002
282
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Table 3 Results for the primary analysis using a Cox model: hazard ratios for discontinuation of ziprasidone therapy Variable
Hazard 95% CI ratio
p value
Age Black Male High initial dose Psychiatric hospitalization 180 days prior to index date Concurrent medications z1
1.001 1.200 0.983 0.805
0.993–1.008 1.011–1.425 0.830–1.164 0.675–0.959
1.365 0.225
1.137–1.640 0.0009* 0.186–0.272 b0.0001*
0.8422 0.0373* 0.8399 0.0153*
Hazard ratios greater than 1.00 imply that the variable is associated with greater hazard of discontinuation of therapy. CI = confidence interval. * p b 0.05 based on the Wald’s Chi-square statistic.
p = 0.0002). The low- and high-dose strata did not differ significantly by age, use of concomitant psychotropic medications, or nonschizophrenia diagnoses. Although persistence among the medium-dose group was intermediate between that for the high- and low-dose groups, there were no statistically significant differences between the high-dose and medium-dose groups or between the low- and medium-dose groups, with respect to persistence. Therefore, for the multivariate analyses, only the high- and low-dose
groups were compared. Sensitivity analyses confirmed that, in all cases, persistence in the medium-dose group fell between that in the high- and low-dose groups. The Cox proportional hazard model (Table 3) demonstrated that high initial dose is associated with a 20% lower risk of discontinuation (HR = 0.805, p = .0153). Concurrent medication use is also associated with a lower rate of discontinuation (HR = 0.225, p b .0001). Black race (HR = 1.200, p = .0373) and a psychiatric hospitalization in the 6 months prior to the index date (HR = 1.365, p = .0009) were associated with a higher rate of discontinuation. The relative impact on persistence of initiating ziprasidone with a high-dose regimen as compared with a low-dose regimen is reflected in Fig. 1. Patients in the high-dose group were 10% more likely to be persistent at the 30-day mark. The difference in relative persistence increased over time, to a maximum of almost 50% by day 250, yielding a mean difference of 22.5%. 3.2. Sensitivity analysis Results of the Cox proportional hazard analysis allowing for a 30-day gap between prescriptions produced results similar to those of the primary analysis. Specifically, higher ziprasidone dose and absence of a hospitalization in the 180 pre-index days remained predictors of greater persistence.
Fig. 1. The relative percentage difference in persistence for patients initiating therapy on higher doses of ziprasidone.
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Results of the two logistic regression analyses concurred with those of the primary analysis. In fact, the effects of a higher initial dose of ziprasidone were more pronounced. Allowing for a 15-day gap, being on high-dose ziprasidone more than doubled the odds of persisting on therapy (OR = 2.27, p = 0.0046). Being on concurrent medication also greatly and significantly increased the odds of persistence, while having a previous hospitalization significantly decreased those odds. Finally, allowing for a 30-day gap between prescriptions also yielded similar results, with the use of high-dose ziprasidone being significantly associated with greater odds of persisting (OR = 2.306, p = 0.001).
4. Discussion The purpose of the study was to evaluate whether adherence to therapy differed according to the initial dose of ziprasidone. Our results indicate that in a naturalistic Medicaid setting, initiating ziprasidone therapy at 120 to 160 mg per day was statistically significantly associated with a lower discontinuation rate than was initiation of ziprasidone therapy at 20 to 60 mg per day. Taken together with findings from a similar naturalistic study in a privately insured population (Joyce et al., 2006-this issue), these results support the clinical benefits of starting patients at a higher dose of ziprasidone. Other factors appeared to also significantly influence medication discontinuation rates. Previous hospitalizations and black race were associated with higher discontinuation rates, while being on concurrent psychotropic medication was associated with a lower rate of discontinuation. Sensitivity analyses using alternative definitions of persistence on therapy and altering the outcome from time to discontinuation to odds of discontinuation yielded consistent results: patients initiated on a high dose of ziprasidone were significantly less likely to discontinue their medication than were patients initiated on a low dose. Our study has some limitations. A considerable number of confounding factors could potentially be acting to affect adherence (Keith and Kane, 2003; Perkins, 2002; Lacro et al., 2002) some of which are not captured in a Medicaid claims database and, therefore, are not adjusted for in this study. Some of the inclusion criteria for the study could also have introduced a form of selection bias, especially considering that the high-
283
dose patients were statistically significantly different at baseline from the low-dose patients in distribution of gender, race, hospitalization in the past 180 days, and type of schizophrenia. With the exception of the type of schizophrenia, all of these variables were controlled for in the analyses, but the argument that these baseline differences arose from unobserved biases can still be made. Furthermore, limitations arise as a result of using refill patterns from Medicaid pharmacy claims to determine discontinuation of therapy. Refill patterns do not directly measure adherence or even persistence with certainty. There are alternative operational definitions that could have been used for persistence, although no single method has proved to be a standard method of choice (Sikka et al., 2005; Steiner and Prochazka, 1997). Despite these limitations, the direction and statistical significance of the association between ziprasidone dose and medication continuation as a measure of adherence has important clinical implications. Our results are consistent with what is known of the dose– response profile of ziprasidone (Davis and Chen, 2004). In controlled clinical trials, higher doses of ziprasidone produced improvements in schizophrenia symptoms and were associated with lower discontinuation rates (Kane, 2003; Weiden et al., 2003). There is thus a body of scientific evidence indicating that higher doses of ziprasidone can produce a stronger effect on schizophrenia control as well as lower rates of drug discontinuation. Our results support this evidence.
5. Conclusion Our results indicate that initiation of ziprasidone therapy with higher doses (120 to 160 mg per day) may lower the rate of subsequent treatment discontinuation in schizophrenia patients, controlling for multiple demographic and clinical factors. This analysis of schizophrenia patients in a Medicaid setting lends support to previous clinical studies that have shown a positive association between higher ziprasidone doses, lower discontinuation rates, and improved patient outcomes. Our results are similar to those observed in a commercially insured population of patients with schizophrenia, which also supported initiation of ziprasidone treatment at 120 to 160 mg per day (Joyce et al., 2006-this issue).
284
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Acknowledgments This research was supported by Pfizer Inc, New York NY, USA. The data in this study were provided through cooperation with the Maryland Department of Health and Mental Hygiene. Portions of this work were presented at the International Society for Pharmacoeconomics and Outcomes 10th Annual International Meeting, May 15–18, 2005, and at the American Psychiatric Association’s 158th Annual Meeting, May 21–26, 2005. The authors gratefully acknowledge the editorial assistance of Lisa Blatt and the technical assistance of James Gardner and Diane McNally from the Pharmaceutical Research Computing unit at the University of Maryland School of Pharmacy.
References Arato, M., O’Connor, R., Meltzer, H.Y., 2002. A 1-year, doubleblind, placebo-controlled trial of ziprasidone 40, 80 and 160 mg/ day in chronic schizophrenia: the Ziprasidone Extended Use in Schizophrenia (ZEUS) study. Int. Clin. Psychopharmacol. 17, 207 – 215. Davis, J.M., Chen, N., 2004. Dose response and dose equivalence of antipsychotics. J. Clin. Psychopharmacol. 24, 192 – 208. Gilmer, T.P., Dolder, C.R., Lacro, J.P., Folsom, D.P., Lindamer, L., Garcia, P., Jeste, D.V., 2004. Adherence to treatment with antipsychotic medication and health care costs among Medicaid beneficiaries with schizophrenia. Am. J. Psychiatry 161, 692 – 699. Joyce, A.T., Harrison, D.J., Loebel, A.D., Carter, C.T., Ollendorf, D.A., 2006-this issue. Effect of initial ziprasidone dose on length of therapy in schizophrenia. Schizophr. Res. 83, 285 – 292. doi:10.1016/j.schres.2006.01.009. Kane, J.M., 2003. Oral ziprasidone in the treatment of schizophrenia: a review of short-term trials. J. Clin. Psychiatry 64 (suppl 19), 19 – 25. Keith, S.J., Kane, J.M., 2003. Partial compliance and patient consequences in schizophrenia: our patients can do better. J. Clin. Psychiatry 64, 1308 – 1315.
Knapp, M., King, D., Pugner, K., Lapuerta, P., 2004. Nonadherence to antipsychotic medication regimens: associations with resource use and costs. Br. J. Psychiatry 184, 509 – 516. Lacro, J.P., Dunn, L.B., Dolder, C.R., Leckband, S.G., Jeste, D.V., 2002. Prevalence of and risk factors for medication nonadherence in patients with schizophrenia: a comprehensive review of recent literature. J. Clin. Psychiatry 63, 892 – 909. Lieberman, J.A., Stroup, T.S., McEvoy, J.P., Swartz, M.S., Rosenheck, R.A., Perkins, D.O., Keefe, R.S., Davis, S.M., Davis, C.E., Lebowitz, B.D., Severe, J., Hsiao, J.K., 2005. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N. Engl. J. Med. 353, 1209 – 1223. National Institute of Mental Health. Schizophrenia. Bethesda (MD): National Institute of Mental Health, National Institutes of Health, US Department of Health and Human Services; 1999 [reprinted 2002; cited 2005 September 2]. (NIH Publication Number: NIH 02-3517). 32 pages. Available from: http:// www.nimh.nih.gov/publicat/schizoph.cfm. National Institute of Mental Health: Research on adherence interventions for mental disorders. Bethesda (MD): National Institute of Mental Health, National Institutes of Health, US Department of Health and Human Services; 2003 Publication PA-03-111. Perkins, D.O., 2002. Predictors of noncompliance in patients with schizophrenia. J. Clin. Psychiatry 63, 1121 – 1128. Sikka, R., Xia, F., Aubert, R.E., 2005. Estimating medication persistency using administrative claims data. Am. J. Manag. Care 11, 449 – 457. Steiner, J.F., Prochazka, A.V., 1997. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J. Clin. Epidemiol. 50, 105 – 116. US Package Insert for Geodon (ziprasidone, HCl) and Geodon (ziprasidone mesylate) for Injection, Revised May, 2005. Weiden, P.J., Olfson, M., 1995. Cost of relapse in schizophrenia. Schizophr. Bull. 21, 419 – 429. Weiden, P.J., Simpson, G.M., Potkin, S.G., O’Sullivan, R.L., 2003. Effectiveness of switching to ziprasidone for stable but symptomatic outpatients with schizophrenia. J. Clin. Psychiatry 64, 580 – 588. Weiden, P.J., Kozma, C., Grogg, A., Locklear, J., 2004. Partial compliance and risk of rehospitalization among California Medicaid patients with schizophrenia. Psychiatr. Serv. 55, 886 – 891.
Effect of initial ziprasidone dose on treatment persistence in schizophrenia C. Daniel Mullins a,*, Fadia T. Shaya a, Julie Magno Zito a, Nour Obeidat a, John Naradzay a, David J. Harrison b a
University of Maryland School of Pharmacy, 515 W. Lombard Street, Room 262, Baltimore, MD 21201, United States b Pfizer Inc., U.S. Outcomes Research Department, 235 E 42nd Street, New York, NY 10017, United States Received 20 October 2005; received in revised form 13 January 2006; accepted 18 January 2006 Available online 20 March 2006
Abstract Objective: To determine the relationship between ziprasidone initial dose and treatment persistence among patients diagnosed with schizophrenia. Method: Adult Medicaid recipients (N = 1096) diagnosed with schizophrenia who had ziprasidone prescription claims between July 1, 2001 and September 30, 2003, were categorized by initial dose: low (20–60 mg per day, n = 464), medium (61–119 mg per day, n = 320) and high dose (120–160 mg per day, n = 312). Treatment persistence up to 365 days was measured using refill patterns, allowing 15-day gaps between expected refill dates. Multivariate survival analysis explored the simultaneous impact of age, gender, race, previous hospitalization, and concomitant medication usage, in addition to initial dose of ziprasidone. Sensitivity analysis tested the robustness of results with different definitions for persistence and allowable gaps between refills. Results: Discontinuation rates across the observation period (maximum, 12 months per individual) were lower for patients initiated with high-dose than low-dose ziprasidone ( P = 0.001). Other factors significantly associated with greater discontinuation of medication were monotherapy (versus combination therapy) and hospitalization within the 6 months prior to the index date of therapy. Black race was associated with greater discontinuation, although this was not consistent across sensitivity analyses. Conclusions: Patients with schizophrenia started on high doses of ziprasidone have lower discontinuation rates in a retrospective Medicaid database than patients started on low doses. These results were robust across various sensitivity analyses. D 2006 Elsevier B.V. All rights reserved. Keywords: Persistence; Schizophrenia; Ziprasidone; Medicaid; Atypical antipsychotic
1. Introduction
* Corresponding author. Tel.: +1 410 706 0879; fax: +1 410 706 5394. E-mail address: [email protected] (C.D. Mullins). 0920-9964/$ - see front matter D 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.schres.2006.01.013
Successful management of schizophrenia is dependent on patient adherence to recommendations for consistent and long-term antipsychotic therapy (Keith and Kane, 2003). Adherence to therapy can substan-
278
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
tially reduce the chance of relapse in schizophrenia patients (National Institute of Mental Health, 2002; Keith and Kane, 2003). Furthermore, relapse entails a substantial clinical and economic burden (Weiden and Olfson, 1995; Weiden et al., 2004; Gilmer et al., 2004; Knapp et al., 2004). Lack of adherence to therapy is a key concern in the management of schizophrenia (National Institute of Mental Health, 2003). In the recent Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) study, funded by the National Institute of Mental Health (NIMH), 74% of patients discontinued therapy within 18 months (Phase I of the study), citing reasons such as lack of efficacy or intolerable side effects (Lieberman et al., 2005). Numerous patient-related, medication-related, and environmental and social risk factors have been proposed to affect adherence (Keith and Kane, 2003; Perkins, 2002; Lacro et al., 2002). Of the medicationrelated risk factors, drug dose has been suggested to play a part in nonadherence, but clear evidence has not yet been published regarding whether higher or lower doses of antipsychotic agents improve adherence in actual clinical practice. Clinician understanding of initial dose titration and target doses has evolved largely through clinical experience for most atypical antipsychotic agents. Ziprasidone is one of such agents, for which there is growing empirical evidence that appropriate dosing in therapy is an important determinant of clinical success (Davis and Chen, 2004). Ziprasidone was approved for use in the US in 2001 and its prescribing information recommends an initial dose for schizophrenia of 20 mg bid, with titration as needed up to 80 mg bid (US Package Insert). Subsequent clinical trial as well as postmarketing experience have indicated that optimal response for ziprasidone is generally obtained at higher doses: Short-term clinical trials have shown daily doses of 120 to 160 mg to be more effective than lower doses in treating acute schizophrenia episodes and in lowering medication discontinuation rates (Kane, 2003). Similarly, daily doses of 80 to 120 mg were found to be efficacious in maintenance therapy (Davis and Chen, 2004; Arato et al., 2002), although a dose–response effect was not consistently discernable (Arato et al., 2002). Symptomatic but stable patients switched from other antipsychotics to ziprasidone responded to doses of 80 to 100 mg, but patients who completed the study took slightly higher doses
(median daily dose of 120 mg) than those who discontinued (Weiden et al., 2003). In the CATIE schizophrenia trial, a higher proportion of ziprasidonetreated patients compared to other agents received maximal doses of the drug (160 mg per day), suggesting an improved effectiveness of ziprasidone at higher doses (Lieberman et al., 2005). Finally, a retrospective database study found that commercially insured patients with an initial prescription for ziprasidone of 120 to 160 mg per day had a significantly lower risk of subsequent treatment discontinuation than did patients started at 40 mg to 79 mg per day (Joyce et al., 2006-this issue). Such findings suggest a need for further evaluation of the relationship between ziprasidone dosing and medication adherence. The objective of this study was to determine whether the previously observed relationship between higher initial ziprasidone dose and lower risk of nonadherence could be extended to a non-commercial, Medicaid population. Specifically, we wanted to test the hypothesis that Medicaid patients initiated on a high dose of ziprasidone would subsequently exhibit lower rates of treatment discontinuation than patients initiated on a low dose of ziprasidone.
2. Methods 2.1. Data set The data set contained all pharmacy and medical service encounter and fee-for-service (FFS) claims from the Maryland Medicaid FFS and HealthChoice programs for recipients who received at least one pharmacy claim for selected antipsychotic drugs between January 1, 2001 and December 31, 2003. Antipsychotic medicines included atypical and conventional antipsychotics, lithium, and anticonvulsants that are frequently used as mood stabilizers. In addition, each recipient’s beginning and end dates of Medicaid eligibility across the 3-year study period were recorded to determine periods of continuous enrollment surrounding the index date, defined as the date of the recipient’s first ziprasidone prescription fill date. Demographic and Medicaid coverage information (i.e., Medicaid coverage group, race, and gender) was extracted from the index (first) prescription claim.
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
2.2. Study population Maryland Medicaid recipients between the ages of 18 and 64 years were included in the analysis if their index (first) prescription for ziprasidone occurred between July 1, 2001 and September 30, 2003. The recipient identifiers from the pharmacy data were then merged with the eligibility data to limit the study population to those ziprasidone users who had a schizophrenia disorder diagnosis (295.xx) any time during the 3-year study period. Furthermore, recipients were required to have continuous Medicaid eligibility for a minimum of 6 months prior to their index ziprasidone prescription. Recipients receiving injectable ziprasidone were excluded since their medical indication might be different from that of oral users. The injectable form is used for the treatment of acute agitation and not in the long-term management of schizophrenia. Age was calculated using date of birth as of the ziprasidone index date. Race was also identified in order to explore its effect on persistence. The racial distribution was consistent with the Maryland Medicaid demographics profile, indicating that most patients were either black or white. The numbers for other races were too small to provide sufficient sample sizes for analysis and, therefore, recipients whose race was other than black or white were excluded. 2.3. Data analysis 2.3.1. Operational definitions of study variables 2.3.1.1. Dose group definitions. Ziprasidone initial dose was categorized as bhighQ, bmediumQ or blowQ. The recommended dose range for ziprasidone is 40 to
279
160 mg per day, administered bid (US Package Insert). The antipsychotic efficacy threshold, 60% D2 occupancy, occurs at approximately 120 mg per day with ziprasidone. Therefore, the lower bound of the high dose range was defined as 120 mg per day, with an upper bound of 160 mg. The medium dose group, 61 to 119 mg per day, and the low dose group, 20 to 60 mg per day, were used to encompass the other commonly prescribed daily doses of ziprasidone, 40 mg and 80 mg per day. Initial doses were calculated from the index prescription by multiplying the quantity dispensed by the tablet’s strength and dividing by the days’ supply that was indicated on the claim. 2.3.1.2. Discontinuation. Time to discontinuation of drug therapy, the major study outcome, was calculated using prescription fills for oral ziprasidone. Starting from the index date of the first prescription, the time to discontinuation was calculated as the days between the index date and the date of the first gap of more than 15 days between prescriptions. Prescribing gaps were calculated by first adding days’ supply to the dispensing date of each prescription to identify the date the prescription should have been exhausted. The first gap in prescribing of 15 days or more was identified as a discontinuation event. Other events could have occurred during patient follow-up, namely disenrollment from Medicaid, psychiatric-related hospitalization, nonpsychiatric-related hospitalization, or persistence until the end of the follow-up period. Patients who fell into these other event categories were considered censored. The first event occurring after the ziprasidone index date determined the recipient’s status (Table 1). In order for a hospitalization to be classified as a psychiatric-related hospitalization, a psychiatric reve-
Table 1 Categories used to identify patient status at the end of follow-up Discontinuation Disenrollment Psychiatric-related hospitalization Nonpsychiatric hospitalization Persistence
Recipient stopped ziprasidone or had a gap in fill dates of 15 days or greater. (A gap in fill dates of 30 days or more was explored in the sensitivity analysis.) Recipient disenrolled from Medicaid before discontinuing ziprasidone. Lack of Medicaid eligibility for more than 1 day was categorized as loss of continuous eligibility. First inpatient hospitalization occurring after the ziprasidone index date lasting more than one day with a revenue code of 114, 124, 134, 144, 154 or 204 recorded within 7 days of the admission date. First nonpsychiatric inpatient-related hospitalization occurring after the ziprasidone index date lasting 7 days or longer. Recipient stayed on ziprasidone until the end of the study period.
280
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
nue code had to appear on a claim within 7 days of the admission date; otherwise, the hospitalization was identified as a nonpsychiatric hospitalization. 2.3.2. Operational definitions of other covariates 2.3.2.1. Prior health care utilization. A set of variables was created to examine health care utilization prior to the index date, eventually creating a numeric variable of count of psychiatric hospitalizations during differing lengths of pre-index periods. We counted the number of unique inpatient hospitalization episodes and outpatient1 and psychiatric outpatient2 visits that were associated with a mental health diagnosis (ICD-9-CM = 290.xx to 319.xx) during a 180-day pre-index period, and the total number of inpatient hospital days 30 days prior to the index date. The 180-day period was used for the analysis. 2.3.2.2. Comorbidities. The number of psychiatric comorbidities was calculated for each patient. Comorbidities were determined from the three diagnosis fields on all physician, inpatient, and outpatient hospital claims, provided that the beginning date of service for the claim occurred any time during the recipient’s continuous Medicaid eligibility. Fifteen unique classifications were used for each mental health diagnosis code, excluding schizophrenia: substance abuse, psychotic disorder, bipolar disorder, depression, adjustment disorder, anxiety disorder, personality disorder, tics, conduct disorder, oppositional disorder, attention deficit hyperactivity disorder, mental retardation, learning disorder, developmental disorders, and other psychiatric disorders. For the diagnosis to be considered present, the recipient had to have at least one claim with a diagnosis from that category. The number of mental health comorbidities was summed over all 15 categories. 2.3.2.3. Concomitant drug use. To account for concomitant use of psychotropic drugs, we compared individuals who used other psychotropic medications when initiated on ziprasidone with those who received 1
CPT = 99 201–5; 99 211–15; 99 241–45; 99 261–63; 99 271–75; 99 281–84. 2 CPT = 90 801–02; 90 804–15; 90 845–57; 90 862; 90 875.
ziprasidone monotherapy. To create a variable that measured concomitant medication use we performed the following: psychotropic drug claims were grouped into eleven drug classes, and psychotropic drug use episodes were defined. Psychotropic drug use was considered concurrent with ziprasidone use under two conditions: (1) the ziprasidone index date occurred between the begin and end dates for two prescription fills for the other psychotropic drugs, and (2) the prescription for the other psychotropic medication began within 30 days of the ziprasidone index date and at least 30 days prior to the ziprasidone end date. Use of at least one concurrent psychotropic medication was sufficient to classify the patient as a concurrent psychotropic medication user. 2.3.3. Statistical analysis Descriptive statistics were calculated to compare the distribution of age, gender, race, number of mental health hospitalizations within 180 days prior to the index date, presence of concurrent medications, and number of mental health diagnosis categories during the study period across the three ziprasidone dosage groups. Chi-square tests were used to test the differences across the dosage groups for categorical variables. Patients were followed for up to 365 days after the initiation of ziprasidone. Survival analyses using Kaplan–Meier survival curves were conducted. The survival time for each recipient was calculated as the number of days from the ziprasidone initiation date to the date of discontinuation. If recipients disenrolled, had psychiatric- or nonpsychiatric-related hospitalization, or reached the end of the study period before they discontinued ziprasidone treatment, recipients were considered as bcensoredQ in the survival analysis. Cox proportional hazard models were used to assess the relative effects of initial drug doses and confounding factors on the hazard ratios of discontinuation. Covariates in the models included age, gender, black race, initial ziprasidone daily dose, and number of mental health hospitalizations within 180 days prior to the index date. The reference group for daily dose was the low dose of initial ziprasidone (i.e., 20 mg to 60 mg per day). To demonstrate the relative effect on persistence of initiating treatment with high-dose versus low-dose ziprasidone, the proportional difference between the
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Kaplan–Meier survival curves of the two strata was calculated. The percentage increase in persistence was calculated by the following equation, where bRPDQ indicates Relative Percentage Difference in survival, bSurvRateQ indicates the Kaplan Meier survival rate (which reflects the chance of persisting longer than time t), and bHQ and bLQ indicate high and low initial dose, respectively. An RPD greater than zero reflects a greater likelihood of continuing ziprasidone on high doses relative to low doses. RPD ¼
SurvRateH;t SurvRateL;t SurvRateL;t
281
allowing a less stringent 30-day gap between ziprasidone prescriptions before a patient would become classified as bdiscontinuedQ. We also conducted a logistic regression analysis to estimate the odds of persisting on ziprasidone or else discontinuing, using the original 15-day gap before discontinuation (vis-a`-vis the hazard of discontinuing ziprasidone or else becoming censored, in the primary survival analysis). Finally, we repeated he logistic regression analysis, allowing for a 30-day gap between prescriptions.
3. Results
2.3.4. Sensitivity analysis Our primary analysis involved the estimation of the relationship between drug dose and subsequent medication discontinuation using Medicaid pharmacy claims data. The results were based on a precise definition of bdiscontinuationQ, i.e. stopping ziprasidone or having a gap in prescription fill dates of 15 days or greater. Recognizing that the results might be influenced by the definition of discontinuation and the statistical analysis of the data, we conducted a number of sensitivity analyses to test the robustness of the results. First, we conducted a survival analysis
3.1. Primary analysis The final study population consisted of 1096 adults with schizophrenia on low (N = 464), medium (N = 320), or high (N = 312) initial doses of ziprasidone. The mean age of the population was 39 years (median = 39.0, S.D. = 10.7). Patients were similar across dose strata, although patients in the high-dose group were significantly more likely to be white and have paranoid schizophrenia (Table 2). Patients in the low-dose group had significantly fewer hospitalizations than those in the high-dose group in the pre-index period (67.9% vs. 53.2% with no prior hospitalizations,
Table 2 Baseline demographic characteristics for Medicaid schizophrenia patients on low, medium or high doses of ziprasidone (N = 1096) Demographic characteristics
Category
Low, 20–60 mg (n = 464)
Medium, 61–119 mg (n = 320)
High, 120–160 mg (n = 312)
p-value (Chi-square test)
Age group
18–39 40–54 55–64 Female Male White Black No Yes 0 admissions 1 admission 2 admissions 3+ admissions None At least 1 0 1 2 3 4+
219 210 35 255 209 194 270 222 242 315 86 29 34 237 227 131 99 89 59 86
157 138 25 166 154 151 169 145 175 207 71 29 13 144 176 70 68 52 61 69
174 116 22 148 164 168 144 117 195 166 93 30 23 158 154 67 67 62 44 72
0.2057
Gender Race Paranoid schizophrenia Psychiatric hospitalizations 180 days prior to index date
Concurrent medications Comorbidities
(47.2) (45.3) (7.5) (55.0) (45.0) (41.8) (58.2) (47.8) (52.2) (67.9) (18.5) (6.3) (7.3) (51.1) (48.9) (28.2) (21.3) (19.2) (12.7) (18.5)
(49.1) (43.1) (7.8) (51.9) (48.1) (47.2) (52.8) (45.3) (54.7) (64.7) (22.2) (9.1) (4.1) (45.0) (55.0) (21.9) (21.3) (16.3) (19.1) (21.6)
(55.8) (37.2) (7.1) (47.4) (52.6) (53.8) (46.2) (37.5) (62.5) (53.2) (29.8) (9.6) (7.4) (50.6) (49.4) (21.5) (21.5) (19.9) (14.1) (23.1)
0.1208 0.0043* 0.0155* 0.0005*
0.2047 0.1002
282
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Table 3 Results for the primary analysis using a Cox model: hazard ratios for discontinuation of ziprasidone therapy Variable
Hazard 95% CI ratio
p value
Age Black Male High initial dose Psychiatric hospitalization 180 days prior to index date Concurrent medications z1
1.001 1.200 0.983 0.805
0.993–1.008 1.011–1.425 0.830–1.164 0.675–0.959
1.365 0.225
1.137–1.640 0.0009* 0.186–0.272 b0.0001*
0.8422 0.0373* 0.8399 0.0153*
Hazard ratios greater than 1.00 imply that the variable is associated with greater hazard of discontinuation of therapy. CI = confidence interval. * p b 0.05 based on the Wald’s Chi-square statistic.
p = 0.0002). The low- and high-dose strata did not differ significantly by age, use of concomitant psychotropic medications, or nonschizophrenia diagnoses. Although persistence among the medium-dose group was intermediate between that for the high- and low-dose groups, there were no statistically significant differences between the high-dose and medium-dose groups or between the low- and medium-dose groups, with respect to persistence. Therefore, for the multivariate analyses, only the high- and low-dose
groups were compared. Sensitivity analyses confirmed that, in all cases, persistence in the medium-dose group fell between that in the high- and low-dose groups. The Cox proportional hazard model (Table 3) demonstrated that high initial dose is associated with a 20% lower risk of discontinuation (HR = 0.805, p = .0153). Concurrent medication use is also associated with a lower rate of discontinuation (HR = 0.225, p b .0001). Black race (HR = 1.200, p = .0373) and a psychiatric hospitalization in the 6 months prior to the index date (HR = 1.365, p = .0009) were associated with a higher rate of discontinuation. The relative impact on persistence of initiating ziprasidone with a high-dose regimen as compared with a low-dose regimen is reflected in Fig. 1. Patients in the high-dose group were 10% more likely to be persistent at the 30-day mark. The difference in relative persistence increased over time, to a maximum of almost 50% by day 250, yielding a mean difference of 22.5%. 3.2. Sensitivity analysis Results of the Cox proportional hazard analysis allowing for a 30-day gap between prescriptions produced results similar to those of the primary analysis. Specifically, higher ziprasidone dose and absence of a hospitalization in the 180 pre-index days remained predictors of greater persistence.
Fig. 1. The relative percentage difference in persistence for patients initiating therapy on higher doses of ziprasidone.
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Results of the two logistic regression analyses concurred with those of the primary analysis. In fact, the effects of a higher initial dose of ziprasidone were more pronounced. Allowing for a 15-day gap, being on high-dose ziprasidone more than doubled the odds of persisting on therapy (OR = 2.27, p = 0.0046). Being on concurrent medication also greatly and significantly increased the odds of persistence, while having a previous hospitalization significantly decreased those odds. Finally, allowing for a 30-day gap between prescriptions also yielded similar results, with the use of high-dose ziprasidone being significantly associated with greater odds of persisting (OR = 2.306, p = 0.001).
4. Discussion The purpose of the study was to evaluate whether adherence to therapy differed according to the initial dose of ziprasidone. Our results indicate that in a naturalistic Medicaid setting, initiating ziprasidone therapy at 120 to 160 mg per day was statistically significantly associated with a lower discontinuation rate than was initiation of ziprasidone therapy at 20 to 60 mg per day. Taken together with findings from a similar naturalistic study in a privately insured population (Joyce et al., 2006-this issue), these results support the clinical benefits of starting patients at a higher dose of ziprasidone. Other factors appeared to also significantly influence medication discontinuation rates. Previous hospitalizations and black race were associated with higher discontinuation rates, while being on concurrent psychotropic medication was associated with a lower rate of discontinuation. Sensitivity analyses using alternative definitions of persistence on therapy and altering the outcome from time to discontinuation to odds of discontinuation yielded consistent results: patients initiated on a high dose of ziprasidone were significantly less likely to discontinue their medication than were patients initiated on a low dose. Our study has some limitations. A considerable number of confounding factors could potentially be acting to affect adherence (Keith and Kane, 2003; Perkins, 2002; Lacro et al., 2002) some of which are not captured in a Medicaid claims database and, therefore, are not adjusted for in this study. Some of the inclusion criteria for the study could also have introduced a form of selection bias, especially considering that the high-
283
dose patients were statistically significantly different at baseline from the low-dose patients in distribution of gender, race, hospitalization in the past 180 days, and type of schizophrenia. With the exception of the type of schizophrenia, all of these variables were controlled for in the analyses, but the argument that these baseline differences arose from unobserved biases can still be made. Furthermore, limitations arise as a result of using refill patterns from Medicaid pharmacy claims to determine discontinuation of therapy. Refill patterns do not directly measure adherence or even persistence with certainty. There are alternative operational definitions that could have been used for persistence, although no single method has proved to be a standard method of choice (Sikka et al., 2005; Steiner and Prochazka, 1997). Despite these limitations, the direction and statistical significance of the association between ziprasidone dose and medication continuation as a measure of adherence has important clinical implications. Our results are consistent with what is known of the dose– response profile of ziprasidone (Davis and Chen, 2004). In controlled clinical trials, higher doses of ziprasidone produced improvements in schizophrenia symptoms and were associated with lower discontinuation rates (Kane, 2003; Weiden et al., 2003). There is thus a body of scientific evidence indicating that higher doses of ziprasidone can produce a stronger effect on schizophrenia control as well as lower rates of drug discontinuation. Our results support this evidence.
5. Conclusion Our results indicate that initiation of ziprasidone therapy with higher doses (120 to 160 mg per day) may lower the rate of subsequent treatment discontinuation in schizophrenia patients, controlling for multiple demographic and clinical factors. This analysis of schizophrenia patients in a Medicaid setting lends support to previous clinical studies that have shown a positive association between higher ziprasidone doses, lower discontinuation rates, and improved patient outcomes. Our results are similar to those observed in a commercially insured population of patients with schizophrenia, which also supported initiation of ziprasidone treatment at 120 to 160 mg per day (Joyce et al., 2006-this issue).
284
C.D. Mullins et al. / Schizophrenia Research 83 (2006) 277–284
Acknowledgments This research was supported by Pfizer Inc, New York NY, USA. The data in this study were provided through cooperation with the Maryland Department of Health and Mental Hygiene. Portions of this work were presented at the International Society for Pharmacoeconomics and Outcomes 10th Annual International Meeting, May 15–18, 2005, and at the American Psychiatric Association’s 158th Annual Meeting, May 21–26, 2005. The authors gratefully acknowledge the editorial assistance of Lisa Blatt and the technical assistance of James Gardner and Diane McNally from the Pharmaceutical Research Computing unit at the University of Maryland School of Pharmacy.
References Arato, M., O’Connor, R., Meltzer, H.Y., 2002. A 1-year, doubleblind, placebo-controlled trial of ziprasidone 40, 80 and 160 mg/ day in chronic schizophrenia: the Ziprasidone Extended Use in Schizophrenia (ZEUS) study. Int. Clin. Psychopharmacol. 17, 207 – 215. Davis, J.M., Chen, N., 2004. Dose response and dose equivalence of antipsychotics. J. Clin. Psychopharmacol. 24, 192 – 208. Gilmer, T.P., Dolder, C.R., Lacro, J.P., Folsom, D.P., Lindamer, L., Garcia, P., Jeste, D.V., 2004. Adherence to treatment with antipsychotic medication and health care costs among Medicaid beneficiaries with schizophrenia. Am. J. Psychiatry 161, 692 – 699. Joyce, A.T., Harrison, D.J., Loebel, A.D., Carter, C.T., Ollendorf, D.A., 2006-this issue. Effect of initial ziprasidone dose on length of therapy in schizophrenia. Schizophr. Res. 83, 285 – 292. doi:10.1016/j.schres.2006.01.009. Kane, J.M., 2003. Oral ziprasidone in the treatment of schizophrenia: a review of short-term trials. J. Clin. Psychiatry 64 (suppl 19), 19 – 25. Keith, S.J., Kane, J.M., 2003. Partial compliance and patient consequences in schizophrenia: our patients can do better. J. Clin. Psychiatry 64, 1308 – 1315.
Knapp, M., King, D., Pugner, K., Lapuerta, P., 2004. Nonadherence to antipsychotic medication regimens: associations with resource use and costs. Br. J. Psychiatry 184, 509 – 516. Lacro, J.P., Dunn, L.B., Dolder, C.R., Leckband, S.G., Jeste, D.V., 2002. Prevalence of and risk factors for medication nonadherence in patients with schizophrenia: a comprehensive review of recent literature. J. Clin. Psychiatry 63, 892 – 909. Lieberman, J.A., Stroup, T.S., McEvoy, J.P., Swartz, M.S., Rosenheck, R.A., Perkins, D.O., Keefe, R.S., Davis, S.M., Davis, C.E., Lebowitz, B.D., Severe, J., Hsiao, J.K., 2005. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N. Engl. J. Med. 353, 1209 – 1223. National Institute of Mental Health. Schizophrenia. Bethesda (MD): National Institute of Mental Health, National Institutes of Health, US Department of Health and Human Services; 1999 [reprinted 2002; cited 2005 September 2]. (NIH Publication Number: NIH 02-3517). 32 pages. Available from: http:// www.nimh.nih.gov/publicat/schizoph.cfm. National Institute of Mental Health: Research on adherence interventions for mental disorders. Bethesda (MD): National Institute of Mental Health, National Institutes of Health, US Department of Health and Human Services; 2003 Publication PA-03-111. Perkins, D.O., 2002. Predictors of noncompliance in patients with schizophrenia. J. Clin. Psychiatry 63, 1121 – 1128. Sikka, R., Xia, F., Aubert, R.E., 2005. Estimating medication persistency using administrative claims data. Am. J. Manag. Care 11, 449 – 457. Steiner, J.F., Prochazka, A.V., 1997. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J. Clin. Epidemiol. 50, 105 – 116. US Package Insert for Geodon (ziprasidone, HCl) and Geodon (ziprasidone mesylate) for Injection, Revised May, 2005. Weiden, P.J., Olfson, M., 1995. Cost of relapse in schizophrenia. Schizophr. Bull. 21, 419 – 429. Weiden, P.J., Simpson, G.M., Potkin, S.G., O’Sullivan, R.L., 2003. Effectiveness of switching to ziprasidone for stable but symptomatic outpatients with schizophrenia. J. Clin. Psychiatry 64, 580 – 588. Weiden, P.J., Kozma, C., Grogg, A., Locklear, J., 2004. Partial compliance and risk of rehospitalization among California Medicaid patients with schizophrenia. Psychiatr. Serv. 55, 886 – 891.