OBJECTIVES: Patient-level models are more and more used in health economics. Our objective was to review main characteristics of patient-level models used to evaluate cost-effectiveness of antipsychotics in schizophrenia.

METHODS: Cost-effectiveness studies for antipsychotics in schizophrenia using a patient-level model published after 2000 were identified systematically through Medline, Embase, congresses websites and grey literature. Main characteristics were extracted: population, treatment strategy(ies), economic outcomes and timeframes.

RESULTS: After screening 1889 records, 17 references presented results from a discrete event simulation (DES) model and 5 from a patient-level state-transition model were identified. Only 2 DES and 2 patient-level state-transition models were core models whereas others were model application studies. DES models were mostly adapted for general schizophrenia (n=7). Switch was allowed in all studies, with number of lines ranging from 2 to 10. Most studies were cost-utility analyses (n=11), or cost-effectiveness analyses considering cost per relapse/diabetes/cardiovascular events avoided, cost per cumulative PANSS score decreased, and cost per time in relapse-free gained. The timeframe was mostly 5 years (n=13); other timeframes were 2, 3 and 6 years. Patient-level state-transition models considered general patients (n=2) or patients with risk of non-adherence (n=3). Switch was considered in 4 studies, with 5 lines allowed. All studies were cost-utility analyses, with 1 year (n=4) and lifetime (n=1) timeframe considered.

CONCLUSIONS: Although patient-level models are widely used in economic evaluation of antipsychotics in schizophrenia, only 2 DES and 2 patient-level Markov core models were retrieved. Flexibility of these models in terms of structure, such as treatment switch and consideration of multiple clinical outcomes, may allow more and more adaptations in the future, to answer health technology assessment research questions.