a nationwide cohort study of oral and depot antipsychotics after first hospitalization for...

Article

Am J Psychiatry 168:6, June 2011 ajp.psychiatryonline.org 603

(Am J Psychiatry 2011; 168:603–609)

of first-episode patients experience only a single episode of psychosis (1).

Medication nonadherence rates are high in schizo-phrenia, as they are in many chronic medical conditions. Three systematic reviews on adherence to antipsychotic medication for psychosis noted that patients took an av-erage of 58% of the recommended amount of medication (7), that 41.2% of patients did not regularly take medica-tion as prescribed (8), that prescription completion rates ranged from 28% to 85% (9), and that irregular medication use rates ranged from 31% to 62% (9). Many of the stud-ies were based on interviews or questionnaires, so patients not willing to participate were not included. Thus, nonad-herence rates are probably higher in unselected patient populations. Current guidelines for schizophrenia include long-acting injections of antipsychotics as an option for maintenance treatment, particularly for overcoming es-tablished nonadherence (1, 2, 10). Two earlier systematic

A Nationwide Cohort Study of oral and Depot Antipsychotics After First Hospitalization

for Schizophrenia

Jari Tiihonen, M.D., Ph.D.

Jari Haukka, Ph.D.

Mark Taylor, F.R.C.Psych.

Peter M. Haddad, M.D., F.R.C.Psych.

Maxine X. Patel, M.D., M.R.C.Psych.

Pasi Korhonen, Ph.D.

Objective: Data on the effectiveness of antipsychotics in the early phase of schizo-phrenia are limited. The authors exam-ined the risk of rehospitalization and drug discontinuation in a nationwide cohort of 2,588 consecutive patients hospitalized for the first time with a diagnosis of schizophre-nia between 2000 and 2007 in Finland.

Method: The authors linked national databases of hospitalization, mortality, and antipsychotic prescriptions and com-puted hazard ratios, adjusting for the ef-fects of sociodemographic and clinical variables, the temporal sequence of the antipsychotics used, and the choice of the initial antipsychotic for each patient.

Results: Of 2,588 patients, 1,507 (58.2% ) collected a prescription for an antipsy-chotic during the first 30 days after hospi-tal discharge, and 1,182 (45.7% , 95% con-fidence interval [CI]=43.7–47.6) continued their initial treatment for 30 days or lon-ger. In a pairwise comparison between depot injections and their equivalent oral formulations, the risk of rehospitalization

for patients receiving depot medications was about one-third of that for patients re-ceiving oral medications (adjusted hazard ratio=0.36, 95% CI=0.17–0.75). Compared with oral risperidone, clozapine (adjusted hazard ratio=0.48, 95% CI=0.31–0.76) and olanzapine (adjusted hazard ratio=0.54, 95% CI=0.40–0.73) were each associated with a significantly lower rehospitaliza-tion risk. Use of any antipsychotic com-pared with no antipsychotic was associat-ed with lower mortality (adjusted hazard ratio=0.45, 95% CI=0.31–0.67).

Conclusions: In Finland, only a minority of patients adhere to their initial antipsy-chotic during the first 60 days after dis-charge from their first hospitalization for schizophrenia. Use of depot antipsychot-ics was associated with a significantly low-er risk of rehospitalization than use of oral formulations of the same compounds. Among oral antipsychotics, clozapine and olanzapine were associated with more fa-vorable outcomes. Use of any antipsychot-ic was associated with lower mortality.

After the acute phase of treatment for a first episode of psychotic illness, guidelines usually indicate that sub-sequent maintenance antipsychotic medication treat-ment should continue for at least 1 year, but consensus is lacking regarding the total duration of treatment if the patient remains asymptomatic (1, 2). A 5-year observa-tional study of patients with first-episode psychosis indi-cated that stopping antipsychotic medication increased the rate of relapse nearly fivefold compared with contin-ued treatment (3). Discontinuation rates for oral antipsy-chotics have been shown to be high: 74% at 18 months for chronic schizophrenia in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) (4) and 42% at 12 months for first-episode schizophrenia in the European First Episode Schizophrenia Trial (EUFEST) (5). However, the risk of relapse needs to be weighed against the like-lihood and severity of adverse effects caused by antipsy-chotic medication (6) and the fact that approximately 20%

This article is featured in this month’s AJP Audio and is an article that provides Clinical Guidance (p. 609).

oRAl AND DePot ANtIPSYCHotICS AFteR FIRSt HoSPItAlIzAtIoN FoR SCHIzoPHReNIA

604 ajp.psychiatryonline.org Am J Psychiatry 168:6, June 2011

MethodThis was a register-based case linkage study of all people in

Finland 16–65 years of age who had their first hospitalization in which schizophrenia was diagnosed (ICD-10 code F20) during the period of 2000–2007 and who had not collected (i.e., pur-chased) any antipsychotic prescription (Anatomic Therapeutic Chemical [ATC] code N05A) within 6 months before admission.

The study cohort was identified from the Finnish National Hospital Discharge Register, which is administrated by the Na-tional Institute for Health and Welfare. A total of 33,318 patients had at least one hospitalization due to schizophrenia-related ill-ness (ICD-10 codes F20–F25) during the study period. Of these, 7,434 experienced their first hospitalization during that period, and 2,588 had a strictly defined schizophrenia diagnosis (F20) during their first hospitalization. Detailed information about the prescription database and the procedure for calculating the dura-tion of antipsychotic use is presented in the data supplement that accompanies the online edition of this article.

Statistical Analysis

The outcome measures of interest were 1) the risk of all-cause discontinuation of the initial antipsychotic medication; 2) the risk of rehospitalization for schizophrenia; and 3) the risk of death. Only those patients who had received any antipsychotic medi-cation within the first 30 days after discharge were included in the analysis of all-cause discontinuation. The follow-up for each patient started at the end of the first hospitalization period. The end of follow-up for the whole study was December 31, 2007. For the analysis of all-cause discontinuation (N=1,507), patients’ follow-up ended at the time their initial treatment changed for any reason, at death, or at the end of the study period, whichever occurred first. For the analysis of rehospitalization (N=2,588), fol-low-up ended at the time of the first rehospitalization, at death, or at the end of the study period. For the mortality analysis, follow-up ended at death or at the end of the study period. Detailed in-formation about the statistical analysis is presented in the online data supplement.

Results

The mean age of the study population was 37.8 years (SD=13.7), and 62% were male. Of 2,588 patients with a first hospitalization, 1,507 (58.2%) used an antipsychot-ic medication during the first 30 days after discharge, and 1,182 (45.7% of the total, 95% confidence interval [CI]=43.7–47.6) continued the initial antipsychotic medi-cation (started within 30 days after hospital discharge) for 30 days or longer. Alternatively, 54.3% of our cohort either did not collect an antipsychotic prescription within 30 days of hospital discharge or used their initial antipsy-chotic medication for less than 30 days.

The association between clinical and sociodemograph-ic variables and the risk of all-cause antipsychotic discon-tinuation over a mean follow-up period of 2 years (5,221 person-years) is summarized in Table S1 in the online data supplement. During the follow-up period, 1,394 patients discontinued their initial antipsychotic medication (start-ed not only during the first 30 days but at any time during follow-up) for the following reasons: 50.6% (705 patients) because of a change in antipsychotic prescription, 34.3% (478 patients) because of discontinuation, 14.8% (207 pa-

reviews on depot injections (11, 12) suggested a nonadher-ence rate of only 24% (range=0%–54%), and more recently Shi et al. (13) calculated the mean medication possession ratio (cumulative number of days covered by depot divid-ed by 365 days) to be 91% for patients receiving depot first-generation antipsychotics.

Studies comparing clinical outcomes for depot and oral antipsychotics vary in study design and methodological rigor as well as in their primary findings. A meta-analysis (14) showed no difference in the risk of relapse for depot first-generation antipsychotics compared with oral medi-cations, although global improvement was more likely in the depot cohort. However, patients for whom depot medication is most indicated—those who are nonadherent with oral formulations—are likely to be underrepresented in randomized controlled trials. Observational studies are generally more supportive of depot medication; mirror-im-age (before-after) studies show a consistent benefit for de-pot medication over prior medication in terms of reducing total number of days in hospital, but a major problem in these studies is that, by definition, the initial (oral) medica-tion has failed (15, 16), and the temporal order of medica-tions (first oral, then depot injection) may have a substan-tial effect on the medications’ comparative effectiveness.

Depot antipsychotics are not widely prescribed after a first episode of psychosis (17), even though nonadherence rates in first-episode patients are high and are strongly linked to relapse (18). Relapse at this crucial period in ear-ly adulthood can have marked psychosocial consequenc-es in terms of education or work opportunities. The low use of depot antipsychotics at this point in the illness pro-cess may reflect a presumed dislike of injectable medica-tion among first-episode patients and negative attitudes associated with use of depot injections (19, 20). However, several studies have suggested that depot medications are effective and acceptable in patients with first-episode psy-chotic illness (17, 21). Unfortunately, there is a dearth of long-term data comparing the use of depot and oral anti-psychotics after first-episode psychosis (22).

Using case linkage of the comprehensive national data-bases available in Finland, we analyzed data from a large nationwide cohort of patients discharged after a first hospi-talization in which schizophrenia was diagnosed to assess risk of rehospitalization, all-cause discontinuation, and to-tal mortality for patients using any antipsychotic or no an-tipsychotic and for patients receiving depot formulations or the equivalent oral formulations. We also compared commonly used depot antipsychotics with the most com-monly used oral second-generation antipsychotics that are not available in depot formulation to further answer the question of which are the most effective antipsychotics in the maintenance phase after a first hospitalization for schizophrenia, as assessed by risk of rehospitalization. We hypothesized that there would be no differences between different antipsychotic drugs, irrespective of whether they were administered orally or as a depot injection.

tIIHoNeN, HAUKKA, tAYloR, et Al.


olanzapine (hazard ratio=0.54, 95% CI=0.40–0.73) were associated with a significantly lower risk of rehospitaliza-tion than oral risperidone. In the pairwise comparisons between depot agents and their equivalent oral formula-tions, no significant differences in rehospitalization risk were observed between pairs, but in the pooled analysis, depot agents were associated with a significantly lower risk of rehospitalization (adjusted hazard ratio=0.36, 95% CI=0.17–0.75, p=0.007; weighted mean=0.53, 95% CI=0.32–0.88) (Table 1).

There were 160 deaths during the follow-up period. Use of any antipsychotic (64 deaths/6,260 person-years) compared with no use of antipsychotic (96 deaths/4,509 person-years), when considered in terms of number of deaths per person-years, was associated with a lower risk of mortality (Cox model hazard ratio=0.49, 95% CI=0.34–0.72; marginal structural modeling hazard ratio=0.38, 95% CI=0.19–0.74). When the hospital episodes were excluded from the time at risk for death (16 deaths censored), the hazard ratio was reduced to 0.45 (95% CI=0.31–0.67). Be-cause of the low number of deaths, the confidence inter-vals were wide, and no statistically significant differences in mortality were observed between antipsychotic medi-cations.

Discussion

To our knowledge, this is the first study of the adherence and comparative effectiveness of specific antipsychotic treatments in a large unselected population of patients in a real-world setting after their first hospitalization with

tients) because of hospitalization, and 0.3% (four patients) because of death.

The all-cause drug discontinuation risks and the me-dian dosages for each medication are listed in Figure S1 in the online data supplement. Table 1 lists the pairwise comparisons for all-cause discontinuation between depot antipsychotics (mean age, 46.4 years; mean duration of first hospitalization, 95.9 days) and their equivalent oral formulations (mean age, 39.8 years; mean duration of first hospitalization, 63.5 days). Of a total depot monotherapy of 298 person-years, 73 person-years (24.5%) consisted of treatment that was started within 30 days after discharge from the first hospitalization. For haloperidol, perphen-azine, and risperidone, the depot formulation was associ-ated with a significantly lower risk of discontinuation than was the oral formulation. In the pooled analysis of depot and oral antipsychotics (haloperidol, perphenazine, ris-peridone, and zuclopenthixol), depot antipsychotics were associated with a 59% lower risk of discontinuation than oral antipsychotics (hazard ratio=0.41, 95% CI=0.27–0.61, p<0.0001).

During a mean follow-up period of 2.0 years (5,221 person-years), 1,496 patients (57.8%) were rehospitalized because of a relapse of schizophrenia symptoms. Com-pared with no use of antipsychotics, use of any antipsy-chotic was associated with a lower risk of rehospitalization (Cox model hazard ratio=0.38, 95% CI=0.34–0.43; marginal structural model hazard ratio=0.48, 95% CI=0.42–0.56). The risk of rehospitalization for individual antipsychot-ics is summarized in Table 2 and Figure 1. Only clozapine (fully adjusted hazard ratio=0.48, 95% CI=0.31–0.76) and

tABle 1. Pairwise Comparisons for Risk of All-Cause Discontinuation of the Initial Antipsychotic treatment and Risk of Rehospitalization After a First Hospitalization for Schizophreniaa

All-Cause Discontinuation Rehospitalization

ComparisonAdjusted Hazard

Ratiob 95% CI pAdjusted Hazard

Ratiob 95% CI p

Any depot injection compared with equivalent oral formulation 0.41 0.27–0.61 <0.0001 0.36 0.17–0.75 0.007

Haloperidol depot injection compared with oral haloperidol 0.27 0.08–0.88 0.03 0.12 0.01–1.13 0.06

Perphenazine depot injection compared with oral perphenazine 0.32 0.19–0.53 <0.0001 0.53 0.22–1.28 0.16

Risperidone depot injection compared with oral risperidone 0.44 0.31–0.62 <0.0001 0.57 0.30–1.08 0.09

Zuclopenthixol depot injection compared with oral zuclopenthixol 0.75 0.29–1.89 0.54 0.49 0.11–2.14 0.35

a All-cause discontinuation was analyzed among those patients who had started their initial antipsychotic treatment during the first 30 days after the index hospitalization (1,507 patients had started any antipsychotic, and 587 patients had started haloperidol, perphenazine, risperidone, or zuclopenthixol). In the analysis of risk of rehospitalization, those patients in the total cohort (N=2,588) who had used ha-loperidol, perphenazine, risperidone, or zuclopenthixol at any time during follow-up (967 person-years) were included in the analysis.

b The hazard ratios are adjusted by using the following as covariates: age at diagnosis, sex, duration of first hospital episode, and current and previous use of anxiolytics, hypnotics and sedatives, antidepressants, drugs used in addictive disorders, analgesics, antiparkinsonian drugs, blood glucose-lowering drugs, and lipid-modifying agents. For risk of rehospitalization, covariates also included previous use of anti-psychotics during the follow-up and the choice of initial antipsychotic (serving as a surrogate for the patient’s clinical status at baseline and thus reflecting the clinical correlates determining the selection of treatment). If pooled estimates were weighted using the person-years of each monotherapy category (for all-cause discontinuation: haloperidol 7/207, perphenazine 20/207, risperidone 167/207, and zuclopen-thixol 13/207 person-years; for risk of rehospitalization: haloperidol 31/967, perphenazine 172/967, risperidone 700/967, and zuclopen-thixol 64/967 person-years), the adjusted hazard ratios would have been 0.43 (95% CI=0.32–0.58, p<0.0001) for all-cause discontinuation and 0.53 (95% CI=0.32–0.88, p=0.0139) for rehospitalization.



blind randomized controlled trials, but is generally cho-sen by the patient and clinician. Not all discontinuations in our study can be assumed to indicate nonadherence; some patients would have stopped medication between prescriptions because of other events, such as medication switch, rehospitalization, and death. However, approxi-mately one-third of discontinuations did not precede such an event, and it is reasonable to assume that most of these cases do represent nonadherence; it is unlikely that clinicians would deliberately stop medication early when the diagnosis is schizophrenia.

The mean age of the patients was greater than 37 years. This relatively high age is probably due to delays in set-tling on a strictly defined diagnosis of schizophrenia, since 21% of the patients had a history of antipsychotic use in outpatient care (earlier than 6 months before the index hospitalization). A 1966 birth cohort study (N=11,017) in northern Finland (24) showed that until age 27, all 37 co-hort members who were listed in the national treatment register as being hospitalized with a schizophrenia diag-nosis were in fact found to have schizophrenia when re-searchers applied DSM-III-R criteria to the information in available hospital records (0% false positive)—but the study authors found an additional 34 patients hospitalized with other diagnoses of psychotic illness who also fulfilled DSM criteria for schizophrenia (48% false negatives). Thus, it is likely that many patients had symptoms and transient antipsychotic treatment before their first hospitalization with a schizophrenia diagnosis (although those patients who used antipsychotics in outpatient care during the 6 months preceding our study period were excluded).

a diagnosis of schizophrenia. The results showed that in Finland, a high-income country where the cost of antipsy-chotic medication is fully reimbursed, more than half of patients either did not collect an antipsychotic prescrip-tion within 30 days after discharge from their first hospi-talization or discontinued their initial antipsychotic med-ication within 30 days. This finding was not attributable to selection bias, since the study population included all patients in Finland with a first hospitalization for schizo-phrenia. No information was available on the medication used in the hospital, but it can be assumed that virtually all patients had received some kind of antipsychotic treat-ment (23). Because there is no compulsory outpatient treatment in Finland, our results on the comparative ef-fectiveness of specific antipsychotics can be generalized only to those patients who are willing to use antipsychotic treatment in outpatient care.

In two previous large randomized controlled trials, the all-cause discontinuation rates were 74% for all patients at 18 months (in the CATIE study [4]) and 42% at 12 months (in the EUFEST study [5]); at 2 months, the all-cause dis-continuation rates in those studies were substantially lower than 50%. This may be explained by selection and follow-up of the patients: in a real-world setting, patients may be less motivated or less cooperative, and routine clinical care and follow-up are less intensive than in ran-domized controlled trials. Also, the participants in CATIE were patients with a previously established pattern of an-tipsychotic treatment. On the other hand, one might have expected that continuation would be favored in clinical practice, since medication is not blinded, as it is in double-

tABle 2. Risk of Rehospitalization After a First Hospitalization for Schizophrenia, by Antipsychotic treatment Pattern (N=2,588 )

Treatment NaPerson-Years

Crude Relative

Risk 95% CI

Adjusted Hazard Ratiob 95% CI

Fully Adjusted

Hazard Ratioc 95% CI p

Haloperidol, depot 1 18 0.29 0.04–2.05 0.29 0.04–2.06 0.21 0.03–1.60 0.13Clozapine 58 412 0.72 0.53–0.99 0.51 0.37–0.72 0.48 0.31–0.76 0.001Olanzapine 134 904 0.76 0.59–0.98 0.71 0.55–0.91 0.54 0.40–0.73 <0.0001Other antipsychotics 10 82 0.63 0.33–1.20 0.69 0.36–1.33 0.56 0.29–1.10 0.09Risperidone, depot 15 119 0.65 0.38–1.11 0.94 0.55–1.63 0.57 0.30–1.08 0.09Perphenazine, depot 13 109 0.61 0.34–1.09 0.66 0.37–1.18 0.59 0.31–1.12 0.11Polypharmacy 80 321 1.28 0.96–1.71 1.25 0.92–1.69 0.92 0.66–1.28 0.62Zuclopenthixol, depot 6 52 0.59 0.26–1.35 0.66 0.29–1.52 0.95 0.37–2.44 0.92Risperidone, oral 113 581 1.00 1.00 1.00Perphenazine, oral 11 63 0.90 0.48–1.67 1.25 0.67–2.34 1.11 0.57–2.18 0.76Quetiapine 54 217 1.28 0.93–1.77 1.58 1.14–2.20 1.11 0.75–1.64 0.60No treatment 993 2,318 2.20 1.81–2.68 2.27 1.86–2.78 1.63 1.30–2.04 <0.0001Haloperidol, oral 5 13 1.98 0.81–4.84 1.77 0.71–4.36 1.79 0.63–5.09 0.28Zuclopenthixol, oral 3 12 1.29 0.41–4.05 1.88 0.60–5.96 1.93 0.57–6.58 0.29a N=number of rehospitalizations.b The hazard ratios are adjusted by using the following as covariates: age at diagnosis, sex, duration of first hospital episode, previous use of antipsychotics during follow-up, and current and previous use of anxiolytics, hypnotics and sedatives, antidepressants, drugs used in addic-tive disorders, analgesics, antiparkinsonian drugs, blood glucose-lowering drugs, and lipid-modifying agents.

c The fully adjusted hazard ratios (illustrated in Figure 1) were adjusted using the same covariates as the adjusted ratios but also included the choice of the initial antipsychotic for each patient (reflecting the clinical correlates determining the selection of treatment). The Bonferroni-corrected p value corresponding to a statistical significance level of 0.05 with 13 multiple comparisons is 0.05/13=0.0038.



discharge from hospital (23). However, no comprehensive data are available on how strictly these guidelines are fol-lowed throughout the country.

A recent meta-analysis (27) of 15 randomized controlled trials with a total of 2,522 patients with early psychosis did not observe significant differences in efficacy between atypical and conventional antipsychotics, but the study did not look at differences within these two groups of an-tipsychotics. In our study, of the oral medications studied, clozapine and olanzapine were associated with the low-est all-cause discontinuation and rehospitalization rates. These findings are in line with those from the CATIE (4, 28) and EUFEST (5) studies as well as our previous cohort study (29). It should be noted that about 6% of patients died during a mean follow-up of 4 years in this cohort of patients 16–65 years of age. Use of any antipsychotic was associated with significantly lower mortality compared with no antipsychotic treatment, which is also in keeping with previous cohort studies (29, 30), although the rela-tively low statistical power meant that no significant dif-ferences were observed between individual antipsychotic treatments.

In observational studies, adjusting for confounding fac-tors is important. Although we used a number of clinical and sociodemographic factors in the analyses, we had no direct measure for intrinsic treatment adherence or risk of relapse. Therefore, in the comparison of rehospitalization risk, we made a further adjustment by using the choice of the initial antipsychotic, reflecting the patient’s clinical

The use of defined daily dosage in our study may have caused some patients continuously using unusually low doses to have been misclassified as discontinuing the medication. On the other hand, our nonadherence results may be underestimates, as it was assumed that all patients who collect a prescription on time are fully adherent. It is probable, too, that many discontinuations that we classi-fied as being due to rehospitalization were attributable to nonadherence (that is, the prescription was filled, but the medication was not taken as prescribed). Our results in-dicate that nonadherence with antipsychotic medication is a major issue in first-episode schizophrenia and occurs very early on, perhaps contributing to a relatively poor long-term prognosis in schizophrenia.

Depot injections were associated with about a 50%–65% lower risk of rehospitalization than oral formulations of the same compounds even when the temporal sequence of medications used was taken into account. In a systematic metareview of randomized controlled trials (14), no dif-ference was observed in relapse rates between depot and oral antipsychotics. This is understandable, as patients who are adherent and cooperative enough to participate in a randomized controlled trial would not likely receive much additional benefit from a formulation designed to enhance adherence. Since nonadherent patients cannot be forced to participate in randomized controlled trials, observational studies are the only way to investigate this issue. A recent systematic review comparing depot and oral formulations (16), which included observational data, found variable results but noted several studies in which depot use was associated with less time in hospital and better global outcomes. In our nationwide cohort, about 8% of patients used depot injections as their first medica-tion in outpatient care, and about 10% of the treatment patient-years consisted of depot medication (see Table 2; also see Figure S1 in the online data supplement). This is a low rate of use compared with that seen in chronic schizo-phrenia. A recent review (25) concluded that depot agents were prescribed for one-quarter to one-third of patients in the U.K., depending on clinical setting. It can be assumed that depot agents are used especially among those patients with the lowest treatment adherence (8, 26). Also, in the Finnish Guidelines for Treatment of Schizophrenia (23), oral first- or second-generation antipsychotics are recom-mended for treatment of acute and chronic patients, and depot formulations are recommended for those patients with inadequate insight and treatment compliance. If the target population for depot medication were broadened, then patients with better insight and treatment adherence would be included, and a reduction in rehospitalization rate would be expected. However, this reduction can be achieved only among those patients who are willing to use antipsychotic treatment in outpatient care. The Finnish Guidelines for Treatment of Schizophrenia indicate that patients must be seen by the medical staff at least once a month in outpatient care during the first 6 months after

FIGURe 1. Risk of Rehospitalization After a First Hospitali-zation for Schizophrenia, by Antipsychotic treatment Pat-tern (N=2,588 )a

Haloperidol, depot

Clozapine

Olanzapine

Other antipsychotics

Risperidone, depot

Perphenazine, depot

Polypharmacy

Zuclopenthixol, depot

Risperidone, oral

Perphenazine, oral

Quetiapine

No treatment

Haloperidol, oral

Zuclopenthixol, oral

1 32 40

Hazard Ratio With 95% CI

a This is an illustration of the fully adjusted hazard ratios reported in Table 2.



treatment per se but that the treatments studied are not sufficiently efficacious or well tolerated, resulting in fre-quent switching of medication.

Received Aug. 30, 2010; revisions received Nov. 17 and Dec. 21, 2010; accepted Jan. 3, 2011 (doi: 10.1176/appi.ajp.2011.10081224). From the Department of Forensic Psychiatry, University of Eastern Finland, N iuvanniemi Hospital, Kuopio; the Department of Clinical Physiology, Kuopio University Hospital, Kuopio; the Department of M ental Health and Alcohol Research, National Institute for Health and Welfare, Helsinki; the Tampere School of Public Health, Univer-sity of Tampere, Finland; NHS Lothian, Edinburgh, U.K .; the Greater M anchester West M ental Health NHS Foundation Trust, Salford, U.K .; University of M anchester, M anchester, U.K .; the Department of Psy-chosis Studies, Institute of Psychiatry, K ing ’s College London; StatFinn Oy, Espoo, Finland; and EP ID Research Oy, Espoo. Address correspon-dence and reprint requests to Dr. T iihonen, Department of Forensic Psychiatry, University of Eastern Finland, N iuvanniemi Hospital, FI-70240 Kuopio, Finland; [email protected] (e-mail).

Dr. T iihonen has served as a consultant to Lundbeck, Organon, Janssen-Cilag , Eli Lilly, AstraZeneca, Hoffman-La Roche, and Bristol-M yers Squibb and has received fees for g iving expert opinions to Bristol-M yers Squibb and G laxoSmithKline and lecture fees from Janssen-Cilag , Bristol M yers-Squibb, Eli Lilly, Pfizer, Lundbeck, G laxo-SmithKline, and AstraZeneca. Dr. Haukka has been in research col-laboration w ith Janssen-Cilag and Eli Lilly and has been a member of the expert advisory group for Astellas. Dr. Taylor has received lecture and consultancy fees and conference expenses from AstraZeneca, Bristol-M yers Squibb, Eli Lilly, and Janssen-Cilag . Dr. Haddad has re-ceived lecture and consultancy fees and conference expenses from AstraZeneca, Bristol-M yers Squibb, Eli Lilly, and Janssen-Cilag . Dr. Pa-tel has received reimbursement for attendance at scientific confer-ences and has received fees for lecturing and consultancy and inves-tigator-initiated grants from Janssen-Cilag and Eli Lilly. Dr. Korhonen has received consultancy fees from Janssen-Cilag , Novartis Pharma, and Orion Pharma.

Supported by the Annual EVO Financing (special government subsidies from the M inistry of Health and Welfare, Finland) and by Janssen-Cilag . The funders w ere not involved in the conduct of the study or in the collection, management, analysis, or interpretation of the data. The research permission numbers to use the data w ere obtained from the Social Insurance Institute (Kela 14/522/2009), the National Institute for Health and Welfare (Dnro 206/5.05.00/2009), and Statistics Finland (TK -53-739-09).

References

1. National Institute for Health and Clinical Excellence (NICE): Schizophrenia: Core Interventions in the Treatment and Man-agement of Schizophrenia in Adults in Primary and Secondary Care (Update), Final Version (NICE Clinical Guideline 82). Lon-don, NICE, March 2009 (http://www.nice.org.uk/nicemedia/pdf/CG82NICEGuideline.pdf)

2. Buchanan RW, Kreyenbuhl J, Kelly DL, Noel JM, Boggs DL, Fisch-er BA, Himelhoch S, Fang B, Peterson E, Aquino PR, Keller W; Schizophrenia Patient Outcomes Research Team (PORT): The 2009 schizophrenia PORT psychopharmacological treatment recommendations and summary statements. Schizophr Bull 2010; 36:71–93

3. Robinson D, Woerner MG, Alvir JM, Bilder R, Goldman R, Geisler S, Koreen A, Sheitman B, Chakos M, Mayerhoff D, Lieberman JA: Predictors of relapse following response from a first epi-sode of schizophrenia or schizoaffective disorder. Arch Gen Psychiatry 1999; 56:241–247

4. Lieberman JA, Stroup TS, McEvoy JP, Swartz MS, Rosenheck RA, Perkins DO, Keefe RS, Davis SM, Davis CE, Lebowitz BD, Severe J, Hsiao JK; Clinical Antipsychotic Trials of Intervention Effective-ness (CATIE) Investigators: Effectiveness of antipsychotic drugs

status at baseline and thus the clinical correlates deter-mining the selection of the initial treatment, as a covariate. Such an adjustment resulted in major change in only one medication category: the rehospitalization risk for poly-pharmacy changed from a higher risk (1.25) to a lower risk (0.92) compared with oral risperidone. This suggests that patients receiving polypharmacy (several antipsychot-ics) as their first treatment in outpatient care are among the most severely ill patients (having a high intrinsic risk of relapse), and when this factor is adjusted, the hazard ratio for subsequent polypharmacy decreases. However, this did not change the statistical significance of the find-ing. When multiple comparisons are made, one must be cautious to draw conclusions based on the significance of the findings. When a Bonferroni correction was applied, clozapine and olanzapine differed significantly from oral risperidone in risk of rehospitalization.

Those patients who are more likely to be nonadher-ent with oral medication, and thus more likely to receive prescriptions for depot medication, are also likely to have poorer insight than other patients (8, 26). In the United States, patients treated with depot medications have also been observed to use more alcohol and illicit drugs, to be more likely to have been arrested, to have more severe psychopathology (such as psychotic symptoms and disor-ganized thinking), and to have more previous psychiatric hospitalizations than patients using oral antipsychotics (13). Therefore, it is unlikely that residual confounding would explain the better outcome among patients receiv-ing depot antipsychotics. No substantial differences were observed between the results obtained by Cox models and marginal structural modeling analyses using causal mod-els, which suggests that the results are not subject to major bias. In all previous studies of depot compared with oral formulations, it has not been possible to take into account the medication history and temporal sequence of the medication (first oral, then depot medication, compared with first depot, then oral medication). Our results indi-cate that depot medications are associated with substan-tially better outcomes even when this issue is accounted for. Although there is no compulsory outpatient care in Finland, the regular contact with the health care person-nel providing depot injections may have a beneficial ef-fect on patients’ outcome. The low number of patients receiving depot haloperidol resulted in wide confidence intervals, and therefore no firm conclusions can be drawn about its effectiveness.

We were not able to assess patients’ subjective quality of life during the treatments. Specific antipsychotics differ markedly in their side effect profiles (6), and all available antipsychotic medications are far from ideal with regard to efficacy and tolerability. In this study, about 50% of all-cause discontinuation was attributable to switch to anoth-er antipsychotic (or from monotherapy to polypharmacy or from polypharmacy to monotherapy). This implies that a majority of patients are not averse to pharmacological



18. Coldham EL, Addington J, Addington D: Medication adherence of individuals with a first episode of psychosis. Acta Psychiatr Scand 2002; 106:286–290

19. Patel MX, Haddad PM, Chaudhry IB, McLoughlin S, David AS: Psychiatrists’ use, knowledge, and attitudes to first- and sec-ond-generation antipsychotic long-acting injections: compari-sons over 5 years. J Psychopharmacol 2010; 24:1473–1482

20. Waddell L, Taylor M: Attitudes of patients and mental health staff to antipsychotic long-acting injections: systematic review. Br J Psychiatry Suppl 2009; 195:S43–S50

21. Weiden PJ, Schooler NR, Weedon JC, Elmouchtari A, Sunakawa A, Goldfinger SM: A randomized controlled trial of long-acting injectable risperidone vs continuation on oral atypical antipsy-chotics for first-episode schizophrenia patients: initial adher-ence outcome. J Clin Psychiatry 2009; 70:1397–1406

22. Patel MX, Taylor M, David A: Antipsychotic long-acting injec-tions: mind the gap. Br J Psychiatry Suppl 2009; 195:S1–S4

23. Finnish Medical Society Duodecim and the Finnish Psychiatric Association: Finnish Guidelines for Treatment of Schizophre-nia. http://www.terveysportti.fi/xmedia/hoi/hoi35050.pdf

24. Isohanni M, Mäkikyrö T, Moring J, Räsänen P, Hakko H, Par-tanen U, Koiranen M, Jones P: A comparison of clinical and research DSM-III-R diagnoses of schizophrenia in a Finnish na-tional birth cohort: clinical and research diagnoses of schizo-phrenia. Soc Psychiatry Psychiatr Epidemiol 1997; 32:303–308

25. Barnes TR, Shingleton-Smith A, Paton C: Antipsychotic long-acting injections: prescribing practice in the UK. Br J Psychiatry Suppl 2009; 52:S37–S42

26. Patel MX, de Zoysa N, Bernadt M, David AS: A cross-section-al study of patients’ perspectives on adherence to antipsy-chotic medication: depots versus oral. J Clin Psychiatry 2008; 69:1548–1556

27. Crossley NA, Contante M, McGuire P, Power P: Efficacy of atypi-cal v typical antipsychotics in the treatment of early psychosis: meta-analysis. Br J Psychiatry 2010; 196:434–439

28. McEvoy JP, Lieberman JA, Stroup TS, Davis SM, Meltzer HY, Rosenheck RA, Swartz MS, Perkins DO, Keefe RS, Davis CE, Se-vere J, Hsiao JK; CATIE Investigators: Effectiveness of clozapine versus olanzapine, quetiapine, and risperidone in patients with chronic schizophrenia who did not respond to prior atypical antipsychotic treatment. Am J Psychiatry 2006; 163:600–610

29. Tiihonen J, Wahlbeck K, Lönnqvist J, Klaukka T, Ioannidis JPA, Volavka J, Haukka J: Effectiveness of antipsychotic treatments in a nation-wide cohort of 2230 patients in community care af-ter the first hospitalization due to schizophrenia and schizoaf-fective disorder. Br Med J 2006; 333:224–227

30. Tiihonen J, Lönnqvist J, Wahlbeck K, Klaukka T, Niskanen L, Tanskanen A, Haukka J: 11-year follow-up study of mortality in patients with schizophrenia: a population-based cohort study (FIN11 study). Lancet 2009; 374:620–627

in patients with chronic schizophrenia. N Engl J Med 2005; 353:1209–1223

5. Kahn RS, Fleischhacker WW, Boter H, Davidson M, Vergouwe Y, Keet IP, Gheorghe MD, Rybakowski JK, Galderisi S, Libiger J, Hummer M, Dollfus S, López-Ibor JJ, Hranov LG, Gaebel W, Peuskens J, Lindefors N, Riecher-Rössler A, Grobbee DE; EUFEST study group: Effectiveness of antipsychotic drugs in first-epi-sode schizophrenia and schizophreniform disorder: an open randomised clinical trial. Lancet 2008; 371:1085–1097

6. Haddad PM, Sharma SG: Adverse effects of atypical antipsy-chotics: differential risk and clinical implications. CNS Drugs 2007; 21:911–936

7. Cramer JA, Rosenheck R: Compliance with medication regi-mens for mental and physical disorders. Psychiatr Serv 1998; 49:196–201

8. Lacro JP, Dunn LB, Dolder CR, Leckband SG, Jeste DV: Preva-lence of and risk factors for medication nonadherence in pa-tients with schizophrenia: a comprehensive review of recent literature. J Clin Psychiatry 2002; 63:892–907

9. Thieda P, Beard S, Richter A, Kane J: An economic review of compliance with medication therapy in the treatment of schizophrenia. Psychiatr Serv 2003; 54:508–516

10. Kane JM, Garcia-Ribera C: Clinical guideline recommendations for antipsychotic long-acting injections. Br J Psychiatry Suppl 2009; 195:S63–S67

11. Young JL, Zonana HV, Shepler L: Medication noncompliance in schizophrenia: codification and update. Bull Am Acad Psychia-try Law 1986; 14:105–122

12. Young JL, Spitz RT, Hillbrand M, Danieri G: Medication adher-ence failure in schizophrenia: a forensic review of rates, rea-sons, treatments, and prospects. J Am Acad Psychiatry Law 1999; 27:426–444

13. Shi L, Ascher-Svanum H, Zhu B, Faries D, Montgomery W, Marder SR: Characteristics and use patterns of patients taking first-generation depot antipsychotics or oral antipsychotics for schizophrenia. Psychiatr Serv 2007; 58:482–488

14. Adams CE, Fenton MK, Quraishi S, David AS: Systematic meta-review of depot antipsychotic drugs for people with schizo-phrenia. Br J Psychiatry 2001; 179:290–299

15. Davis JM, Matalon L, Watanabe MD, Blake L: Depot antipsy-chotic drugs: place in therapy. Drugs 1994; 47:741–773

16. Haddad PM, Taylor M, Niaz OS: First-generation antipsychotic long-acting injections v oral antipsychotics in schizophrenia: systematic review of randomized controlled trials and observa-tional studies. Br J Psychiatry Suppl 2009; 195:S20–S28

17. Emsley R, Oosthuizen P, Koen L, Niehaus DJH, Medori R, Rabi-nowitz R: Remission in patients with first-episode schizophre-nia receiving assured antipsychotic medication: a study with risperidone long-acting injection. Int Clin Psychopharmacol 2008; 23:325–331

Clinical Guidance: Effective Treatment of the First Episode of SchizophreniaFewer than 50% of patients in the Finnish health care system continue treatment for the first 2 months after an initial hospitalization for schizophrenia, according to Tiihonen et al. Choice of treatment affects relapse rate. Injectable long-acting antipsychotics have a 64% lower relapse rate than equivalent oral medication. Clozapine and olanzapine have significantly lower relapse rates, compared to risperidone. Antipsychotic use overall is associated with decreased mortality.

a nationwide cohort study of oral and depot antipsychotics after first hospitalization for...

Documents