Long-Term Real-World Effectiveness of Pharmacotherapies for Schizoaffective Disorder

Jonne Lintunen; Heidi Taipale; Antti Tanskanen; Ellenor Mittendorfer-Rutz; Jari Tiihonen; Markku Lähteenvuo


Schizophr Bull. 2021;47(4):1099-1107. 

In This Article


Study Population

A detailed description of the study population can be found in the Supplementary Appendix. Two cohorts of persons with SCHAFF identified from Finnish and Swedish nationwide registers were utilized for this study. In both countries, all residents have been assigned a unique personal identification number which was utilized in the linkage of country-specific registers.

The base of the Finnish cohort has been described previously.[16] It included all persons treated due to schizophrenia (the International Classification of Diseases [ICD-10] code F20) and SCHAFF (F25), with corresponding ICD-8 and-9 codes (295*), in inpatient care in Finland as recorded in the Hospital Discharge Register (HDR) maintained by the National Institute of Health and Welfare. The cohort included 7655 persons with SCHAFF.

The Swedish study base included all persons aged 16–64 with schizophrenia spectrum disorder diagnoses (ICD-10 F20, F25) and a registered treatment contact between July 1, 2005 until December 31, 2013 in Sweden.[17] The diagnoses were derived from inpatient and specialized outpatient care registers (the National Patient Register [NPR], maintained by the National Board of Health and Welfare) and disability pensions and sickness absences from the MiDAS register (maintained by the Swedish Social Insurance Agency). The cohort included 7525 persons with SCHAFF.

For both cohorts, the follow-up started on July 1, 2006. The follow-up time ended at death, change of diagnosis to schizophrenia (F20), or the end of the study (December 31, 2016 for the Swedish cohort and December 31, 2017 for the Finnish cohort), whichever occurred first. The different follow-up windows for the cohorts were due to reasons of registry data availability.


The main exposure measure was use of antipsychotics, which were defined as Anatomical Therapeutic Chemical (ATC) classification codes N05A (lithium N05AN01 excluded). In addition to drug substance level marked by ATC code, drug formulation information was utilized to further categorize antipsychotics into orals and LAIs. Most common antipsychotic monotherapies were assessed and all exposure periods including two or more antipsychotics used concurrently were defined as "polytherapy."

Adjunctive pharmacotherapy categories included mood stabilizers (carbamazepine N03AF01, valproic acid N03AG01, lamotrigine N03AX09, lithium N05AN01), antidepressants (N06A), and BZDRs (N05BA, N05CD, N05CF).

Drug use periods, ie, when drug use started and ended, were constructed with the PRE2DUP method which is based on a sliding average of defined daily doses (DDDs), purchase dates, amounts of drugs dispensed, and personal drug use patterns.[18] Each drug for each person was modeled separately, and oral antipsychotics and LAIs were also separated.


The main outcome measure was hospitalization due to psychosis (ICD-10 codes F20–F29), used as a marker for relapse. In addition, we analyzed a composite measure of treatment failure, including psychiatric hospitalization, any change in medication (switch, discontinuation, addition), and death. We also conducted a sensitivity analysis where the outcome was defined as any psychiatric hospitalization (ICD-10 codes F00–F99). In the analyses, an individual may have had recurring outcomes (except death).

Statistical Analyses

The analyses were conducted separately in the Finnish and Swedish cohorts using within-individual design.[19] In this design, each individual formed their own stratum. All time-invariant covariates (such as sex) were controlled for in the design and analyses were adjusted for time-varying covariates, ie, sequential order of treatments, use of other pharmacotherapies (mentioned in the exposure section), and time since cohort entry. Main analyses were performed using stratified Cox regression models yielding hazard ratios (HRs) and 95% confidence intervals (CI). Sensitivity analyses were conducted in between-individual design and analyzed with the traditional Cox model by adjusting for factors presented in supplementary table 1.

When comparing the effectiveness of specific antipsychotics, the reference was nonuse of antipsychotics. As antipsychotics are considered the first-line treatment, other medication categories were considered to be adjunctive therapies. These adjunctive pharmacotherapies were analyzed with antipsychotic use serving as reference (without adjunctive use). The results are reported for the combination of antipsychotics used concomitantly with certain adjunctive pharmacotherapies. This was done to observe whether combination use (which potentially has more adverse effects) has superior real-world effectiveness compared with antipsychotic only use. For antipsychotics, 16 most commonly used antipsychotics (in both countries) were reported (namely [oral if not otherwise specified] olanzapine, olanzapine LAI, clozapine, quetiapine, risperidone, risperidone LAI, aripiprazole, aripiprazole LAI, perphenazine, perphenazine LAI, paliperidone LAI, haloperidol, haloperidol LAI, zuclopenthixol, zuclopenthixol LAI, and levomepromazine) in addition to other second-generation (SG) antipsychotics, whereas other first-generation (FG) category was left out due to sparsity of users. In addition to monotherapies, polytherapy (more than one antipsychotic used concomitantly) was reported as an exposure for the main outcome. Categorization into FG-oral, FG-LAI, SG-oral, and SG-LAI is presented in supplementary table 2.

Five most commonly used oral antipsychotics (clozapine, olanzapine, quetiapine, risperidone, aripiprazole) and SG-LAIs as a category were analyzed in combinations with adjunctive pharmacotherapies. In these analyses, monotherapy of a specific antipsychotic was used as a reference to which that specific antipsychotic combined with an adjunctive pharmacotherapy class was compared with (eg, olanzapine monotherapy was used as a reference for olanzapine-mood stabilizer analyses). The results are presented as adjusted HRs with 95% CIs.

It was ensured that persons with uncertainty of diagnosis were removed. In the main analyses, individuals appeared in the analyses up until they received a diagnosis of schizophrenia, after which they did not contribute to data anymore. In the sensitivity analyses, anyone who received a schizophrenia diagnosis at any point during their follow-up was completely excluded from the analyses.

Nominal P values are displayed throughout the text, unless otherwise stated. P values were corrected for multiple comparisons using the Benjamini-Hochberg false discovery rate on a per graph basis and corrected P values <.05 were considered statistically significant.

Permissions were granted by pertinent institutional authorities at the Finnish National Institute for Health and Welfare (permission THL/847/5.05.00/2015), the Social Insurance Institution of Finland (65/522/2015), and Statistics Finland (TK53-1042-15). The Regional Ethics Board of Stockholm approved the Swedish part of this research project (decision 2007/762-31).