By meta-analysing pooled studies and available individual participant data, we aim to provide new insight on olanzapine therapeutic drug monitoring in schizophrenia.

We conducted a computerized search of bibliographic databases (Pubmed, Cochrane library, Web of Science and PsycINFO) to identify studies that assessed the relationship between olanzapine plasma concentration and the change in patients’ clinical scores. We investigated this relationship with olanzapine plasma level 12h00 post-intake using a random-effects model.

7 studies were included in the pooled data analysis (781 patients). We found no difference in oral dose between responders and non-responders but a significantly higher concentration of 4.50 µg/L in responders (p < 0.01). Olanzapine concentration above the thresholds identified in each study was associated with response (odd ratio = 3.50, p = 0.0007). We identified that non-responder patients showed greater inter-individual variability than responders. In the individual data analysis (159 patients), we found no relationship between dose and clinical response but an association between plasma level and response in the shape of a parabolic curve. The Receiver Operating Characteristic curve found a threshold of 22.07 µg/L to identify responders (96% sensitivity, 86% specificity) and a threshold of 56.47 µg/L to identify a decreased probability of response.

In contrast to oral dose, our work confirmed that plasma olanzapine levels are associated with clinical response and should therefore be used to optimise treatment. We determined a treatment response threshold of 22.07 µg/L and suggest that a concentration above the therapeutic window may result in a decreased response.