Time to discontinuation of antipsychotic drugs in a schizophrenia cohort: influence of current treatment strategies

Transcription

1 545614TPP / Therapeutic Advances in PsychopharmacologyRA Kroken, E Kjelby research-article2014 Therapeutic Advances in Psychopharmacology Original Research Time to discontinuation of antipsychotic drugs in a schizophrenia cohort: influence of current treatment strategies Rune A. Kroken, Eirik Kjelby, Tore Wentzel-Larsen, Liv S. Mellesdal, Hugo A. Jørgensen and Erik Johnsen Ther Adv Psychopharmacol 2014, Vol. 4(6) DOI: / The Author(s), Reprints and permissions: journalspermissions.nav Abstract Background: Rates of discontinuation of antipsychotic treatment for patients with schizophrenia are high and evidence is limited by selective inclusion and high attrition in randomized controlled trials. Aims: To study time to discontinuation of antipsychotic treatment for patients with schizophrenia. Method: All patients with schizophrenia (n = 396) discharged between 2005 and 2011 were followed until discontinuation (clinician or patient decided) of antipsychotic treatment or other endpoints. Univariate and multivariate survival analyses (with time on antipsychotic treatment as the dependent variable) using time-dependent variables were performed. Results: Clozapine displayed lower risk for all-cause (p < 0.001), clinician-decided (p = 0.012) and patient-decided (p = 0.039) discontinuation versus olanzapine oral treatment in the multivariate Cox regression. Second-generation long-acting injection antipsychotics (LAI) (p = 0.015) and first-generation long-acting injection antipsychotics (p = 0.013) showed significantly lower risks for patient-decided discontinuation than olanzapine oral. Conclusion: Higher effectiveness of clozapine and LAI treatment versus oral olanzapine were identified in a clinical cohort of patients with schizophrenia. Keywords: schizophrenia, antipsychotic, clozapine, treatment discontinuation, long-acting injection antipsychotics (LAI) Introduction Choosing time until drug discontinuation as the primary outcome measure in major antipsychotic effectiveness trials [Swartz et al. 2003; Fleischhacker et al. 2005] emphasizes the importance attributed to continuous antipsychotic drug treatment in schizophrenia. The alarmingly high discontinuation rates (e.g. 74% in 18 months, 42% in 12 months) [Lieberman et al. 2005; Kahn et al. 2008] consistently found across these studies highlight the clinicians need to be equipped with treatment strategies that optimize continuous antipsychotic drug treatment. Such strategies must also include the use of the most effective drugs and drug formulations. The evidence shows moderate differences in time until discontinuation among various oral antipsychotics in randomized pragmatic trials of effectiveness [Lieberman et al. 2005; Johnsen et al. 2010] whereas no difference in nonadherence is found between the oral and long-acting injection (LAI) formulations in a recent meta-analysis of randomized controlled trials (RCTs) [Leucht et al. 2011]. However, data from randomized studies may not suffice in informing the clinicians about optimal antipsychotic drug treatment in everyday clinical practice. The low and selective recruitment (10 30% of eligible patients) and high attrition rates (30 50%) [Leucht et al. 2008] invariably found in RCTs including also the pragmatic studies limit the generalizability of currently available evidence [Duggan et al. 2005; Asenjo Lobos et al. 2010]. Moreover, in RCTs of LAI antipsychotics the most severely ill patients do not consent to participate, which has the important consequence that the drugs are not tested in the most relevant Correspondence to: Rune A. Kroken, MD, PhD Division of Psychiatry, Haukeland University Hospital, Pb 23, 5812 Bergen, Norway rune.kroken@helsebergen.no Eirik Kjelby, MD Division of Psychiatry, Haukeland University Hospital, Bergen, Norway Tore Wentzel-Larsen, MSc Norwegian Centre for Violence and Traumatic Stress Studies, Oslo, Norway Centre for Child and Adolescent Mental Health, Eastern and Southern Norway, Oslo, Norway Centre for Clinical Research, Haukeland University Hospital, Bergen, Norway Liv S. Mellesdal, MSc Division of Psychiatry, Haukeland University Hospital, Bergen, Norway Hugo A. Jørgensen, MD, PhD Department of Clinical Medicine, Faculty of Medicine and Dentistry, University of Bergen, Norway Erik Johnsen, MD, PhD Division of Psychiatry, Haukeland University Hospital, Bergen, Norway Department of Clinical Medicine, Faculty of Medicine and Dentistry, University of Bergen, Norway 228

2 RA Kroken, E Kjelby et al. patient groups [Adams et al. 2001]. Bypassing the problems of selective recruitment and high attrition, well designed cohort studies may supplement the evidence base by delivering clinically relevant information from the total patient group for decision makers [Taylor et al. 2012]. Accordingly, to study the effectiveness of antipsychotic treatment in a representative clinical setting we performed an open cohort study of all patients discharged from hospital during a 5-year period and recorded antipsychotic treatment and concomitant psychosocial treatment (including involuntary treatment) in a time-dependent manner. Using this cohort, we have recently demonstrated the strong preventive effect of antipsychotic treatment on readmissions for patients with schizophrenia [Kroken et al. 2012]. Materials and methods Study design, setting and data sources All patients with a clinical diagnosis of schizophrenia according to the International Classification of Diseases, version 10 (ICD 10) [WHO, 1992] discharged subsequent to acute psychiatric hospital admission at the Psychiatric Clinic at Haukeland University Hospital, Norway were included in an open cohort study after discharge. The index admission was the first admission in the inclusion period and the study period was the time between discharge from index admission and the primary endpoint (i.e. the time when antipsychotic treatment used at inclusion was discontinued). Other endpoints were when patients moved out of the hospital recruitment area, died, were lost to follow up for other reasons, or still used discharge antipsychotic treatment at the end of the study. The Psychiatric Clinic receives all patients aged 18 years and over with schizophrenia in need of acute psychiatric hospital admission from a catchment area of 410,000 inhabitants. After discharge, the patients received outpatient treatment at the Psychiatric Clinic or they were referred to one of six community mental health centres (CMHCs) serving the different geographic areas of the hospital s catchment area and providing inpatient and outpatient treatment. Patient follow up could include several treatment periods at the hospital and the CMHCs. Data were obtained retrospectively from medical records or through contact with the treating specialist, general practitioner or the municipality psychiatric nurse. The medical records include notes from the treating clinician containing information from the consultations with patients, information gathered from relatives of the patients and from collaborators such as municipality nurses, and the results of laboratory tests, including serum levels of antipsychotics. Information regarding death dates and moving out of the recruitment area was verified using the Norwegian National Registry. Participants and study period All patients admitted to the Psychiatric Acute Unit between May 2005 and December 2010 with a discharge during the same period were eligible for the study. Patients who lived outside the catchment area were excluded. When there was doubt concerning classification of the information obtained, the case was discussed within the research team until a consensus was reached. Variables Main outcome: time until antipsychotic drug discontinuation. The main outcome was the time from discharge until antipsychotic drug discontinuation. The cause of discontinuation was categorized as follows: patient decided (specified according to cause: side effects, unknown), clinician decided (specified according to cause: side effects, lack of treatment effect, unknown) and unknown causes. The evaluation of adherence and discontinuation was based on accessible information from the medical records, including blood levels of antipsychotics. The evaluations were done by two of the authors individually (RAK and EK) and in cases of doubt discussed until consensus was reached. Discontinuations were classified as clinician decided when the clinician took the decision to stop prescribing the actual drug, even if this was preceded by the patient urging the doctor to do so. Classifications of discontinuations as patient decided were done when it was clear that the patient stopped taking the medication solely on their own initiative, without consulting the clinician responsible for the treatment. If information was evaluated as inadequate, the date of drug discontinuation was set to the last date with known drug intake. The discontinuation date for LAI antipsychotics was set to the date for the next scheduled injection. In line with other studies, temporary breaks in antipsychotic treatment for up to 2 weeks were not regarded as discontinuations [Mullins et al. 2008]. Drugs taken on demand were not recorded. The antipsychotic treatment was recorded with start 229

3 Therapeutic Advances in Psychopharmacology 4(6) and stop dates for each dosage period, allowing for the calculation of mean dosage over the entire treatment period. The antipsychotic doses (including the LAIs) were given in milligrams and as defined daily doses (DDDs) according to the World Health Organization Collaborating Centre for Drug Statistics Methodology [WHO Collaborating Centre for Drug Statistics Methodology, 2008]. The doses of the LAI antipsychotics were converted to daily doses by dividing the injected dose with the inter-injection time interval (days). The low-potency first-generation antipsychotics (FGAs) chlorpromazine, levomepromazine and chlorprothixene, in doses equal to or below 0.33 DDD (e.g. chlorpromazine 100 mg), were regarded as not prescribed with antipsychotic purposes and were excluded from statistical analysis. Antipsychotic treatment variables. Two variables for the analyses of time to discontinuation of antipsychotic treatment were generated: one for antipsychotic monotherapy with eight categories and a second for treatment with LAI antipsychotics with three categories. The eight categories of antipsychotic monotherapy (n = 289) were the following (with the numbers of patients using the different drugs): olanzapine oral (n = 87), risperidone oral (n = 32), quetiapine (n = 29), clozapine (n = 22), other second-generation antipsychotic (SGA) oral [n = 22: aripiprazole (n = 10) + ziprasidone (n = 7) + amisulpride (n = 5)], FGA oral [n = 19: zuclopenthixol (n = 7) + perphenazine (n = 9) + haloperidol (n = 2) + flupenthixol (n =1)], SGA LAI [n = 28: risperidone LAI (n = 26) + olanzapine LAI (n=2)], FGA LAI [n = 50: zuclopenthixol decanoate LAI (n = 24) + perphenazine decanoate LAI (n = 25) + pipotiazine palmitate LAI (n = 1)]. The three categories of LAI antipsychotic treatment (including patients receiving monotherapy and polytherapy, and accordingly, higher numbers than the LAI antipsychotics in the monotherapy variable) were risperidone LAI (n = 46), zuclopenthixol LAI (n = 35) and perphenazine LAI (n = 36). Olanzapine LAI (n = 5) was excluded as it was not available throughout the full study period and pipotiazine palmitate was excluded from the LAI antipsychotic categorization because it was only prescribed for one patient. Confounder variables. Confounding variables that were recorded due to their known influences on time to discontinuation of antipsychotic treatment [Mullins et al. 2008; Novick et al. 2010] were age, sex and number of previous admissions to psychiatric hospitals. In addition, the presence of alcohol or drug problems was recorded: if the patient received a comorbid clinical discharge ICD-10 diagnosis of F10.0 F19.9 at index admission; if the score of Health of the Nation Outcome Scales (HoNOS) [Wing et al. 1998] item 3 was 2 or more at index admission; or if the score on the Alcohol Use Scale (AUS) [Drake et al. 1990] or the Drug Use Scale (DUS) [Drake et al. 1996] was 3 or more. The HoNOS, AUS and DUS scales were administered by the resident physician at admission to index admission. Training in the use of the scales was performed but inter-rater reliability was not determined. Concomitant psychosocial treatment and involuntary treatment. After discharge from index admission, subsequent outpatient and inpatient treatment in the Psychiatric Clinic or in the CMHCs was recorded in a time-dependent manner. Single specialist consultations or evaluations without further treatment were not recorded as treatment periods. Treatment periods were categorized as follows: no treatment, voluntary outpatient treatment, involuntary outpatient treatment without, respectively with a compulsory antipsychotic treatment order, voluntary inpatient treatment and involuntary inpatient treatment without, respectively with a compulsory antipsychotic treatment order. Additional variables. At index admission various variables including Global Assessment of Functioning Split version [Karterud et al. 1998] and demographic variables were recorded. Statistical analyses Primary analyses were survival analyses by antipsychotic monotherapy and antipsychotic LAI treatment. For antipsychotic monotherapy, the time to discontinuation of antipsychotic treatment was explored using Kaplan Meier analysis and univariate and multivariate Cox regression analyses. Patients were censored from the survival analyses at the dates when any of the other endpoints occurred (death, moving out of the area, lost to follow up, and continued antipsychotic treatment at the end of the study period). The following variables were entered into the Cox models specified prior to analysis: age and sex; the clinical variables (comorbid drug/alcohol problems, the number of previous psychiatric hospital admissions, and psychiatric specialist treatment including time 230

4 RA Kroken, E Kjelby et al. periods with compulsory antipsychotic treatment orders) and antipsychotic monotherapy with olanzapine oral treatment as a reference category. Separate analyses were performed for all-cause, clinician-decided and patient-decided discontinuations. The proportional hazard assumption was checked as recommended by Thernau and Grambsch [Thernau and Grambsch, 2000] and was supplied with graphical checks using Schoenfeld residuals when indicated. Secondary analyses included survival analyses for time to discontinuation within the first year and discontinuations caused by side effects, but no significant adjusted hazard ratios (AHRs) were identified for the same predictors as in the primary analysis. In the multivariate Cox survival analyses with time to discontinuation of LAI antipsychotics as the dependent variable, the following confounding variables were entered: age, sex, comorbid drug/ alcohol problems and the number of previous psychiatric hospital admissions. Time to discontinuation of antipsychotic LAI treatment was also assessed using Kaplan Meier analysis. SPSS software, version (30 July 2009; IBM SPSS, Chicago, IL, USA) and R software (The R Foundation for Statistical Computing, Vienna, Austria) were used for the statistical analyses. Approvals This study was approved by the Regional Committee for Medical Research Ethics and the Norwegian Data Inspectorate. Results Descriptive data The sample consisted of 396 patients with a mean age of 42 years; 65.7% were men (Table 1). There was a median of four admissions per patient (including the index admission); the mean number of admissions was 6.8, and 22.7% of patients had a drug or alcohol problem. After discharge 19.4% were without specialist treatment, while 41.0% had at least one period with voluntary outpatient treatment and 33.9% had at least one period with involuntary outpatient treatment. At the time of discharge, 16 patients (4.0%) received no antipsychotic prescriptions, 289 patients received antipsychotic monotherapy (73.0%) and 31.1% of the total sample were treated with an antipsychotic LAI. Mean dosages [DDD (standard deviation, SD)] over the treatment period (Table 2) ranged from 0.77 (0.30) for risperidone oral to 2.00 (1.14) for ziprasidone; the mean dose was 1.30 DDD (excluding low-potency FGAs and the antipsychotic drug that was prescribed for one patient only) (Table 2). Clozapine was the antipsychotic with the longest duration of treatment in the study (861.9 days, SD = 616.8), followed by zuclopenthixol LAI, levomepromazine, risperidone LAI and perphenazine LAI. The mean follow-up period was days (SD = 591.6, range ) for patients on antipsychotic treatment at discharge. Primary analyses: prediction of discontinuation of antipsychotic monotherapeutic treatment and antipsychotic LAI treatment Monotherapy: univariate survival analyses. In the Cox analysis of the time to all-cause discontinuation (see Table 5, univariate survival analyses), clozapine [crude hazard ratio (CHR) = 0.25, p = 0.003] showed a significantly lower risk for discontinuation versus olanzapine oral. Kaplan Meier analysis of the time until discontinuation is displayed in Figure 1. Pairwise log-rank tests (with Holm s correction for multiple comparisons) revealed significantly longer times to discontinuation for clozapine versus olanzapine oral (p = 0.015), SGA oral other (p = 0.047), FGA oral (p = 0.009) and risperidone oral (p = 0.015). The Cox analysis of the time to clinician-decided discontinuation revealed a lower risk for discontinuation of clozapine compared with olanzapine oral (CHR = 0.20, p = 0.012). In addition, in the Cox analysis of the time to patient-decided discontinuation, clozapine significantly lowered this risk compared with olanzapine oral (CHR = 0.20, p = 0.025), with a trend for lower risk for FGA LAI treatment versus olanzapine oral (CHR = 0.49, p = 0.051). Monotherapy: multivariate survival analyses. In the multivariate Cox regression (Table 3) of the time to all-cause discontinuations of antipsychotic treatment, clozapine had a lower risk for discontinuation versus olanzapine oral (AHR = 0.17, p < 0.001), with a trend for lower risk for quetiapine versus olanzapine oral (AHR = 0.60, p = 0.093). In addition, in the multivariate Cox regression of the time to clinician-decided discontinuation (AHR = 0.20, p = 0.012) and the time to patient-decided discontinuation (AHR = 0.21, p = 0.039), clozapine exhibited a significantly lower risk for discontinuation versus olanzapine oral. In the analysis of the time to patient-decided discontinuation, SGA LAI (AHR = 0.26, 231

5 Therapeutic Advances in Psychopharmacology 4(6) Table 1. Baseline characteristics of the 396 patients with schizophrenia. Sociodemographic factors Age at index admission, mean years (SD) 42.4 (14.1) Male sex, n (%) 260 (65.7) Married/living with cohabitant, n (%) 36 (9.1) Education completed,* n (%) Primary school, 7 9 years 191 (48.2) Secondary school, 12 years 118 (29.8) College/university 46 (11.6) Non-Norwegian ethnicity, n (%) 39 (9.8) Caring for children under 18 years of age, $ n (%) 9 (2.3) Main income source, n (%) Employed, student or on unemployment benefits 19 (4.8) Disability pension, sick pay or social security 330 (83.3) Retired 17 (4.3) Other/missing 30 (7.6) Clinical factors GAF (split version) Symptom score, mean (SD) 31.2 (9.4) GAF (split version) Function score, mean (SD) 33.4 (10.5) Index admission at first admission to psychiatric hospital, n (%) 54 (13.6) First episode including first 3 years after onset of psychosis, n (%) 51 (12.9) Age at first psychiatric admission, mean years (SD) 31.1 (11.7) Number of admissions, median (range), mean 4 (1 73), 6.8 Length of stay (LOS) in days, mean (SD) 78.1 (195.1) Schizophrenia diagnosis, n (%) F (76.5) F (6.6) F (0.8) F (7.8) F F20 not specified 33 (8.4) Comorbid alcohol/drug problem at index admission, n (%) 90 (22.7) HoNOS problem-drinking/drug-taking score 2, n (%) 58 (14.7) AUS score 3, n (%) 34 (8.6) DUS score 3, n (%) 37 (9.3) Comorbid ICD-10 diagnosis F10.0 F (9.8) Specialist treatment after discharge from index admission, n (%) No specialist treatment 77 (19.4) Outpatient, voluntary 162 (41.0) Outpatient, involuntary 90 (22.7) Outpatient, involuntary, with antipsychotic treatment order 44 (11.1) Inpatient, voluntary 191 (48.2) Inpatient, involuntary 108 (27.3) Inpatient, involuntary, with antipsychotic treatment order 48 (12.1) Portions of patients with substantial (>5%) missing values: *10.4%, $ 7.1%, 20.5%, 11.4%, 11.4%. AUS, Alcohol Use Scale [Drake et al. 1990]; DUS, Drug Use Scale [Drake et al. 1996]; GAF, Global Assessment of Functioning Split Version [Karterud et al. 1998]; HoNOS, Health of the Nation Outcome Scale [Wing et al. 1998]; ICD-10, International Classifications of Mental and Behavioural Disorders [WHO, 1992]. p = 0.015) and FGA LAI (AHR = 0.35, p = 0.013) also showed significantly lower risks for discontinuation versus olanzapine oral. Antipsychotic LAI treatment: univariate and multivariate analyses. The univariate Cox analysis of the time to all-cause discontinuations of 232

6 RA Kroken, E Kjelby et al. Table 2. Antipsychotic dosages and treatment periods, arranged after mean duration of the antipsychotic treatment periods.* n (%) Mean daily dose Duration of treatment mg (SD) DDD (SD) Total days (SD) Discontinued n (pd cd m) days (SD) Censored n; days (SD) Clozapine 33 (8.7) (188.38) 1.54 (0.63) (616.8) 9 (3 6 0) (493.9) 24; (605.9) Zuclopenthixol LAI 35 (9.2) (6.63) 0.95 (0.44) (647.7) 18 (4 14 0) (457.9) 17; (533.5) Levomepromazine 21 (5.5) (51.58) 0.23 (0.17) (618.6) 11 (1 9 1) (332.3) 10; (482.0) Risperidone LAI 46 (12.1) 3.62 (1.29) 1.34 (0.48) (605.7) 28 (9 19 0) (379.3) 18; (654.7) Perphenazine LAI 36 (9.5) 8.74 (3.94) 1.25 (0.56) (640.0) 23( ) (436.0) 13; (664.6) Chlorprothixene 23 (6.1) (57.70) 0.35 (0.19) (543.3) 12 (1 11 0) (468.1) 11; (541.1) Amisulpride 14 (3.7) (365.2) 1.89 (0.91) (492.6) 8 (1 7 0) (223.2) 6; (473.3) Ziprasidone 15 (3.9) (90.95) 2.00 (1.14) (585.2) 9 (1 8 0) (294.3) 6; (684.6) Perphenazine oral 28 (7.4) (10.22) # 0.57 (0.34) (570.0) 23 (8 14 1) (364.4) 5; (745.7) Olanzapine oral 124 (32.6) (6.45)** 1.55 (0.65) (551.4) 74 ( ) (371.1) 50; (582.6) Quetiapine 41 (10.8) (291.3) $$ 1.47 (0.73) (366.7) 22 ( ) (432.6) 19; (600.2) Haloperidol 5 (1.3) (11.26) 1.30 (1.41) (371.3) 1 (0 1 0) ; (306.0) Chlorpromazine 9 (2.4) (56.21) 0.29 (0.19) (263.2) 8 (1 6 1) (280.9) 1; Aripiprazole 17 (4.5) (7.78) 1.18 (0.52) (504.2) 8 (5 3 0) (280.5) 9; (544.3) Risperidone oral 53 (13.9) 3.86 (1.48) 0.77 (0.30) (485.2) 42 ( ) (325.1) 11; (624.1) Zuclopenthixol oral 19 (5.0) (18.91) 0.70 (0.63) (454.0) 10 (3 6 1) (99.2) 9; (563.8) Olanzapine LAI $ 5 (1.3) (5.08) 1.78 (0.51) (219.5) 2 (0 2 0) (24.0) 3; (243.0) * Excluding antipsychotics with n < 5. $ Olanzapine LAI was not available for the full study period. Percentage of patients on medication (N = 380). Number of patients with available data for the calculation of mean antipsychotic dose: n = 32, n = 20, # n = 27, **n = 121, $$ n = 40, n = 16, n = 52. Mean. cd, clinician decided; DDD, defined daily dose; LAI, long-acting injection; m, missing; pd, patient decided; SD, standard deviation. antipsychotic LAI treatment (Table 4) revealed no significant hazard ratios, while in the multivariate Cox analysis, zuclopenthixol LAI had a significantly lower risk for discontinuation compared with risperidone LAI (AHR = 0.54, p = 0.046). In the Kaplan Meier analysis for time to all-cause discontinuations of antipsychotic LAI treatment, the log-rank test did not reveal significant differences. patients subjected to involuntary treatment. In addition, the mean follow-up time is in the longer range compared with corresponding studies [Haddad et al. 2009]. Another advantage of the study is the measurement of all inpatient and outpatient specialist treatment periods, including periods under involuntary treatment, which were entered in the analyses in a time-dependent manner. Discussion Main findings Our primary finding is the lower risks for allcause, clinician-decided and patient-decided discontinuations for clozapine versus olanzapine oral treatment. SGA LAI and FGA LAI showed significantly lower risks for patient-decided discontinuation than olanzapine oral. The major strengths of this study are the inclusive, consecutive clinical sample with a small dropout rate that also includes the most severely ill patients and The effectiveness of clozapine versus olanzapine oral Our finding of an all-cause discontinuation AHR of 0.17 for clozapine versus olanzapine oral is equivalent to an 83% reduced risk for discontinuation of clozapine versus olanzapine oral treatment. This result is unexpected and may be the result of at least three different factors. First, it is likely that our complete consecutive cohort study with a long follow-up period can detect drug differences that studies with selectively recruited patients and high attrition rates cannot. Findings 233

7 Therapeutic Advances in Psychopharmacology 4(6) (a) (b) Probability of still using index antipsychotic Probability of still using index antipsychotic olanzapine oral risperidone oral quetiapine clozapine SGA Other oral FGA oral SGA LAI FGA LAI p=0.005 (log-rank test) Years from index discharge Years from index discharge Figure 1. Kaplan Meier (KM) plots for the probability of continued use of antipsychotic monotherapy after discharge. (a) Dotted line represents 95% confidence interval. (b) Covariate: antipsychotic monotherapy. FGA, first-generation antipsychotic; LAI, long-acting injection; SGA, second-generation antipsychotic. from other cohort studies support this conclusion, for example, the results from a large cohort study from a restricted area in Scotland [Taylor et al. 2008] and the findings from a Finnish national register study [Tiihonen et al. 2006] both reveal the superior effectiveness of clozapine versus other SGAs. Second, clozapine is particularly efficacious in treatment-resistant schizophrenia [National Institute for Health and Care Excellence, 2009; Buchanan et al. 2010] and, although not systematically assessed, the high mean number of admissions in our cohort could indicate a high frequency of treatment resistance, generally estimated to be present in at least 30% of patients with schizophrenia [Kane et al. 1988; Suzuki et al. 2011]. Third, the result could, at least partly, be caused by confounding by indication as the selection process for clozapine treatment includes the consideration of the patient being able to cooperate with rigorous blood monitoring, and the possibility that these patients are inherently more adherent cannot be excluded. However, the direction of any bias is hard to predict as it is also conceivable that the clinical population for which clozapine currently is indicated (patients regarded as treatment resistant) as a group displays lower drug adherence compared with the total population of patients with schizophrenia. This could be caused by, for example, high and persistent symptom burden including negative symptoms and disappointing experiences with antipsychotic drugs for these patients. Moreover, nonadherence may be uncovered at an earlier stage in patients receiving clozapine treatment compared with patients using other antipsychotics because of the closer follow up in the clozapine group. Accordingly, although confounding by indication cannot be ruled out in this study, we find it unlikely that it explains the major part of the large effectiveness difference between clozapine and olanzapine oral. Olanzapine is widely regarded as one of the most efficacious antipsychotic drugs based on systematic reviews [Asenjo Lobos et al. 2010; Komossa et al. 2010], but high attrition rates limit the conclusions that can be drawn, for example, a mean attrition rate of 49.2% was revealed in a Cochrane systematic review of olanzapine versus other SGA drugs [Komossa et al. 2010]. The results from observational studies including this paper seems to question the superior effectiveness of olanzapine. Other SGA and quetiapine versus olanzapine oral The nonsignificant trends for a higher risk of allcause discontinuation in the pooled other SGA category (aripiprazole, ziprasidone and amisulpride, AHR = 2.30, p = 0.066) and a lower risk of all-cause discontinuation of quetiapine versus olanzapine oral (AHR = 0.60, p = 0.093) raise hypotheses that there exist additional gradients of effectiveness between oral antipsychotic drugs that this study could not confirm due to lack of power

8 RA Kroken, E Kjelby et al. Table 3. Multivariate Cox survival analyses. Predictors for days to discontinuation of antipsychotic monotherapy, adjusted hazard ratios, 95% confidence intervals and p values. Covariate Multivariate analyses, time to discontinuation, all causes Multivariate analyses, time to discontinuation, clinician decided Multivariate analyses, time to discontinuation, patient decided AHR 95% CI p value AHR 95% CI p value AHR 95% CI p value Clozapine: olanzapine oral < Quetiapine: olanzapine oral FGA LAI: olanzapine oral SGA LAI: olanzapine oral SGA oral, other: olanzapine oral FGA oral: olanzapine oral Risperidone oral: olanzapine oral Age (years) at index admission, age 30: age 20 Men: women Number of admissions, 4: < admissions Comorbid drug/alcohol problem: no Psychiatric specialist treatment Voluntary outpatient treatment: no Involuntary outpatient treatment: no Involuntary outpatient treatment with ATO: no Voluntary inpatient < treatment: no Involuntary inpatient <0.001 treatment: no Involuntary inpatient treatment with ATO: no Note: Bold values signifies p < AHR, adjusted hazard ratio; ATO, antipsychotic treatment order; CI, 95% confidence interval; FGA, first-generation antipsychotic; LAI, long-acting injection; SGA, second-generation antipsychotic;, unstable estimate. LAI antipsychotics versus olanzapine oral Our finding of a significantly lower risk for patientdecided discontinuation for SGA LAI and FGA LAI versus olanzapine oral treatment is striking and together with a large Finish cohort study [Tiihonen et al. 2011] supports the hypothesis that antipsychotic LAI treatment is useful in severely ill patients with schizophrenia and specifically in patients with adherence problems. The risk for patient-decided discontinuation of risperidone oral was nonsignificantly higher than olanzapine oral (p = 0.315). In addition, because the SGA LAI group primarily consisted of risperidone LAI, the lower risk of discontinuation of SGA LAI versus olanzapine oral indicates an effectiveness of mode of administration, rather than the drug itself. The AHR for all-cause discontinuations (versus olanzapine oral) of FGA LAI (0.66) and SGA LAI (0.76) did not reach significance; this could be the result of low power in the study. We believe these results call for renewed interest in antipsychotic LAI treatment and recognition that antipsychotic LAI drugs may be very effective tools in the treatment of schizophrenia. Risperidone LAI versus zuclopenthixol LAI Our finding of a significantly higher risk for allcause discontinuation of risperidone LAI 235

9 Therapeutic Advances in Psychopharmacology 4(6) Table 4. Univariate and multivariate Cox survival analyses. Predictors for days to termination of antipsychotic long-acting injection treatment, crude hazard ratios, adjusted hazard ratios, 95% confidence intervals and p values. Covariate Univariate analyses, time to termination, all causes Multivariate analyses, time to termination, all causes CHR 95% CI p value AHR 95% CI p value Zuclopenthixol LAI: risperidone LAI Perphenazine LAI: risperidone LAI Age (years) at index admission, age 30: age Men: women Number of admissions, 4: 3 admissions Comorbid drug/alcohol problem Note: Bold values signifies p < AHR, adjusted hazard ratio; CHR, crude hazard ratio; CI, 95% confidence interval; LAI, long-acting injection. Table 5. Univariate Cox survival analyses. Predictors for days to discontinuation of antipsychotic monotherapy, crude hazard ratios, 95% confidence intervals and p values. Covariate Univariate analyses, time to discontinuation, all causes Univariate analyses, time to discontinuation, clinician decided Univariate analyses, time to discontinuation, patient decided CHR 95% CI p value CHR 95% CI p value CHR 95% CI p value Clozapine: olanzapine oral FGA LAI: olanzapine oral SGA LAI: olanzapine oral Quetiapine: olanzapine oral SGA oral other: olanzapine oral FGA oral: olanzapine oral Risperidone oral: olanzapine oral Age (years) at index admission, age 30: age 20 Men: women Number of admissions, 4: admissions Comorbid drug/alcohol problem: no Psychiatric specialist treatment Voluntarily outpatient treatment: no Involuntary outpatient treatment: no Involuntary outpatient treatment with ATO: no Voluntarily inpatient treatment: < no Involuntary inpatient treatment: <0.001 no Involuntary inpatient treatment with ATO: no Note: Bold values signifies p < ATO, antipsychotic treatment order; CHR, crude hazard ratio; CI, 95% confidence interval; FGA, first-generation antipsychotic; LAI, long-acting injection; SGA, second-generation antipsychotic;, unstable estimate

10 RA Kroken, E Kjelby et al. compared with zuclopenthixol LAI was contrary to our hypothesis. This result indicates the high effectiveness of at least one of the medium-potent FGA LAIs compared with a widely used SGA LAI. The result is at odds with the only randomized head-to-head comparison that has been performed for these drugs [Rubio et al. 2006], in which risperidone LAI performed better on several measures, including the PANSS total score. A retrospective observational study [Shajahan et al. 2010], however, that compared several LAIs (risperidone, zuclopenthixol and flupenthixol) found longer time to discontinuation and lower hospitalization rates for patients treated with zuclopenthixol LAI. Also relevant is the finding in a large effectiveness study that another medium-potent oral FGA (perphenazine) was found to have a comparable effectiveness to several SGAs [Lieberman et al. 2005]. Risperidone LAI was used in a rather high dose in the present study (mean = 1.34 DDD) and risperidone is associated with more motor side effects than most other SGA drugs [Komossa et al. 2011]. The possibility that these factors contribute to the lower effectiveness of risperidone LAI versus zuclopenthixol LAI found in our study cannot be ruled out. The influence of age, sex and clinical variables on time to discontinuation of antipsychotic treatment The demonstrated effects of age, gender, number of previous admissions and comorbid drug or alcohol problems are expected and in line with conclusions from other studies [Mullins et al. 2008; Novick et al. 2010]. For the analyses of allcause, clinician-decided and patient-decided discontinuation, outpatient specialist treatment did not predict a longer time to discontinuation. As expected, inpatient treatment strongly influenced time to discontinuation. Limitations The limitations of our study include the use of clinical diagnosis as the inclusion criterion and less stringent measures of nonadherence. However, 32.4% of the patients on antipsychotic medication used antipsychotic LAIs, and for the 257 patients on oral treatment, 104 (40.5%) had at least one antipsychotic drug measurement above the minimum serum level. For the remaining 153 patients (40.2% of patients on medication), the validation of adherence rests on the information and evaluations from treating clinicians, general practitioners and municipality nurses and from clinical notes. The possibility that we have overrated drug adherence for oral drugs cannot be ruled out; thus the relative strengths of antipsychotic LAIs versus olanzapine oral may have been underestimated. Other systematic influences seem unlikely. Conclusion Our study of an inclusive sample of patients with schizophrenia from a specific catchment area reveals the high effectiveness of clozapine and LAI antipsychotics, and higher effectiveness of zuclopenthixol LAI versus risperidone LAI. The inclusion of compulsory antipsychotic treatment orders and the utilization of time-dependent variables in the analyses highlight the relevance of the results for real-world treatment settings for patients with schizophrenia. Acknowledgements We thank Ingvild Helle, RN, Jill Bjarke, RN, Petter Jakobsen, RN, Marianne Langeland, RN and Geirr Fitje, MBA at the Division of Psychiatry, Haukeland University Hospital, Bergen for their skilful and patient cooperation in this project. Funding The work was supported by research grants from the Western Norway Regional Health Authority and from the Division of Psychiatry, Haukeland University Hospital. Conflict of interest statement Kroken: has been reimbursed by Janssen Cilag for attending conferences. Kjelby: has been reimbursed by Bristol-Myers Squibb, Novartis, Lundbeck, Eli Lilly and AstraZeneca for attending conferences. Wentzel-Larsen and Mellesdal: no financial or other relationships relevant to this study. Jørgensen: has received honoraria from Eli Lilly for contribution to a brochure. Johnsen: has received honoraria for lectures given in meetings arranged by Bristol- Myers Squibb, Eli Lilly, and AstraZeneca companies, and for a contribution to an information brochure by Eli Lilly. Has been reimbursed by the Eli Lilly Company and the Janssen Cilag Company for attending conferences. References Adams, C., Fenton, M., Quraishi, S. and David, A. (2001) Systematic meta-review of depot antipsychotic drugs for people with schizophrenia. Br J Psychiatry 179:

11 Therapeutic Advances in Psychopharmacology 4(6) Asenjo Lobos, C., Komossa, K., Rummel-Kluge, C., Hunger, H., Schmid, F., Schwarz, S. et al. (2010) Clozapine versus other atypical antipsychotics for schizophrenia. Cochrane Database Syst Rev (11): CD Buchanan, R., Kreyenbuhl, J., Kelly, D., Noel, J., Boggs, D., Fischer, B. et al. (2010) The 2009 schizophrenia PORT psychopharmacological treatment recommendations and summary statements. Schizophr Bull 36: Drake, R., Osher, F., Noordsy, D., Hurlbut, S., Teague, G. and Beaudett, M. (1990) Diagnosis of alcohol use disorders in schizophrenia. Schizophr Bull 16: Drake, R., Rosenberg, S. and Mueser, K. (1996) Assessing substance use disorder in persons with severe mental illness. New Dir Ment Health Serv 70: Duggan, L., Fenton, M., Rathbone, J., Dardennes, R., El-Dosoky, A. and Indran, S. (2005) Olanzapine for schizophrenia. Cochrane Database Syst Rev (1): CD Fleischhacker, W., Keet, I. and Kahn, R. (2005) The European First Episode Schizophrenia Trial (EUFEST): rationale and design of the trial. Schizophr Res 78: Haddad, P., Taylor, M. and Niaz, O. (2009) Firstgeneration antipsychotic long-acting injections v. oral antipsychotics in schizophrenia: systematic review of randomised controlled trials and observational studies. Br J Psychiatry 195: S20 S28. Johnsen, E., Kroken, R., Wentzel-Larsen, T. and Jørgensen, H. (2010) Effectiveness of secondgeneration antipsychotics: a naturalistic, randomized comparison of olanzapine, quetiapine, risperidone, and ziprasidone. BMC Psychiatry 10: 26. Kahn, R., Fleischhacker, W., Boter, H., Davidson, M., Vergouwe, Y. and Keet, I. (2008) Effectiveness of antipsychotic drugs in first-episode schizophrenia and schizophreniform disorder: an open randomised clinical trial. Lancet 371: Kane, J., Honigfeld, G., Singer, J. and Meltzer, H. (1988) Clozapine for the treatment-resistant schizophrenic. A double-blind comparison with chlorpromazine. Arch Gen Psychiatry 45: Karterud, S., Pedersen, G., Loevdahl, H. and Friis, S. (1998) Global Assessment of Functioning-Split Version (S-GAF): Background and Scoring Guidelines. Oslo: Department of Psychiatry, Ullevål University Hospital. Komossa, K., Rummel-Kluge, C., Hunger, H., Schmid, F., Schwarz, S. and Duggan, L. (2010) Olanzapine versus other atypical antipsychotics for schizophrenia. Cochrane Database Syst Rev (3): CD Komossa, K., Rummel-Kluge, C., Schwarz, S., Schmid, F., Hunger, H. and Kissling, W. (2011) Risperidone versus other atypical antipsychotics for schizophrenia. Cochrane Database Syst Rev (1): CD Kroken, R., Mellesdal, L., Wentzel-Larsen, T., Jørgensen, H. and Johnsen, E. (2012) Timedependent effect analysis of antipsychotic treatment in a naturalistic cohort study of patients with schizophrenia. Eur Psychiatry 27: Leucht, S., Heres, S., Hamann, J. and Kane, J. (2008) Methodological issues in current antipsychotic drug trials. Schizophr Bull 34: Leucht, C., Heres, S., Kane, J., Kissling, W., Davis, J. and Leucht, S. (2011) Oral versus depot antipsychotic drugs for schizophrenia a critical systematic review and meta-analysis of randomised long-term trials. Schizophr Res 127: Lieberman, J., Stroup, T., McEvoy, J., Swartz, M., Rosenheck, R., Perkins, D. et al. (2005) Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med 12: Mullins, C., Obeidat, N., Cuffel, B., Naradzay, J. and Loebel, A. (2008) Risk of discontinuation of atypical antipsychotic agents in the treatment of schizophrenia. Schizophr Res 98: National Institute for Health and Care Excellence (2009) The NICE guideline on core interventions in the treatment and management of schizophrenia in adults in primary and secondary care updated edition. London: NICE. Available at: nice.org.uk/cg82/guidance/pdf/english (accessed 17 July 2014). Novick, D., Haro, J., Suarez, D., Perez, V., Dittmann, R.W. and Haddad, P. (2010) Predictors and clinical consequences of non-adherence with antipsychotic medication in the outpatient treatment of schizophrenia. Psychiatry Res 176: Rubio, G., Martínez, I., Ponce, G., Jiménez-Arriero, M., López-Muñoz, F. and Alamo, C. (2006) Long-acting injectable risperidone compared with zuclopenthixol in the treatment of schizophrenia with substance abuse comorbidity. Can J Psychiatry 51: Shajahan, P., Spence, E., Taylor, M., Daniel, D. and Pelosi, A. (2010) Comparison of the effectiveness of depot antipsychotics in routine clinical practice. The Psychiatrist 34: Suzuki, T., Remington, G., Mulsant, B., Rajji, T., Uchida, H., Graff-Guerrero, A. et al. (2011) Treatment resistant schizophrenia and response to antipsychotics: a review. Schizophr Res 133: Swartz, M., Perkins, D., Stroup, T., McEvoy, J., Nieri, J. and Haak, D. (2003) Assessing 238

12 RA Kroken, E Kjelby et al. clinical and functional outcomes in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia trial. Schizophr Bull 29: Taylor, M., Cavanagh, J., Hodgson, R. and Tiihonen, J. (2012) Examining the effectiveness of antipsychotic medication in first-episode psychosis. J Psychopharmacol 26: Taylor, M., Shajahan, P. and Lawrie, S. (2008) Comparing the use and discontinuation of antipsychotics in clinical practice: an observational study. J Clin Psychiatry 69: Thernau, T. and Grambsch, P. (2000) Modelling Survival Data: Extending the Cox Model. Berlin: Springer. Tiihonen, J., Haukka, J., Taylor, M., Haddad, P., Patel, M. and Korhonen, P. (2011) A nationwide cohort study of oral and depot antipsychotics after first hospitalization for schizophrenia. Am J Psychiatry 168: Tiihonen, J., Walhbeck, K., Lönnqvist, J., Klaukka, T., Ioannidis, J., Volavka, J. et al. (2006) Effectiveness of antipsychotic treatments in a nationwide cohort of patients in community care after first hospitalisation due to schizophrenia and schizoaffective disorder: observational follow-up study. Br Med J 333: 224. WHO (1992) International Statistical Classification of Diseases and Related Health Problems, 10th Revision. Geneva: World Health Organization. WHO Collaborating Centre for Drug Statistics Methodology (2008) ATC Classification Index With DDDs Oslo. Wing, J., Beevor, A., Curtis, R., Park, S., Hadden, S. and Burns, A. (1998) Health of the Nation Outcome Scales (HoNOS). Research and development. Br J Psychiatry 172: Visit SAGE journals online SAGE journals 239