Authors' reply

27 June 2022

We thank Mittal et al for their interest in our article. Their comments can be summarised into four issues. First, they correctly state that the current study presents limitations that are typical of such studies (unmeasured covariates, unstandardised diagnoses, the lack of an a priori hypothesis). However, this is an unlikely explanation for the major finding of the current study, i.e. that the five groups defined according to the medication possession ratio (MPR) before clozapine initiation achieved relatively similar and high MPRs afterward, as the five groups should be affected similarly by these issues.

Second, although MPR has its limitations, it is the recognised method of choice for studies in which other more direct adherence measures (e.g. blood dosages, pill counts) are impossible. In the present study, the validity of the MPRs is reflected in the observation that the relatively high MPR following clozapine initiation was associated with a low mean number of days in hospital (see Fig. 2 in the original article).¹ This is consistent with previous observations that MPR is strongly correlated with clinical outcomes such as hypertension² and lower rates of psychiatric hospital admission.^{2, 3} Finally, this limitation of MPR is unlikely to account for the main conclusion of this paper, as all five groups should be affected similarly.

Third, Mittal et al state that the overrepresentation of men may reflect some unmeasured effects that have influenced the decision to initiate clozapine and adherence to treatment before and after clozapine initiation. As they state, this predominance of males is typical of studies on treatment-resistant schizophrenia, suggesting that the current sample is representative of this population. In addition, as the gender proportion was similar among the five groups, this factor is unlikely to account for the relatively similar levels of adherence after clozapine initiation. Finally, some of these factors (e.g. social support and comorbidities) are unlikely to change as a consequence of clozapine initiation; they are thus unlikely explanations for the increase in level of adherence after clozapine initiation. Nevertheless, we agree that these issues might have influenced the decision to initiate clozapine, which makes it difficult to determine to what extent these results would apply in other contexts.

Fourth, Mittal et al correctly point out that the high proportion of patients with relatively high MPRs before the initiation of clozapine reflects the typical inclination of clinicians to preferentially use clozapine in patients judged more likely to adhere to treatment. However, this fact does not explain the most provocative finding of the current study, i.e. that high adherence rates can also be achieved in patients in whom low levels of adherence would have been expected. Moreover, Mittal et al mention that an assertive approach may be responsible for explaining better adherence to treatment. Although this is a possibility, it is reassuring to know that clozapine initiated in the real world setting was maintained, even after 3 year follow-up (Supplementary Fig. 3 and Supplementary Table 2).¹ Clozapine is often managed in Quebec through intensive community outreach teams that routinely monitor treatment adherence. However, these factors have not been measured directly. Therefore, it is reasonable to assume that patients will maintain their treatment once it has been initiated with management that meets current Quebec and Canadian practices. This could be a reflection of the quality of Quebec's public services in long-term follow-up in the community, at least among those willing to initiate best practices, such as the use of clozapine in eligible patients.

In conclusion, although the issues raised by Mittal et al do not invalidate the main findings of the current study, they usefully emphasise the importance of repeating such studies in settings in which clozapine would be introduced more systematically in patients with low levels of adherence and to better assess the impact of clinical interventions aimed at supporting adherence to clozapine.