Our findings

The results of our study indicated that extending the haematological monitoring interval from 4-weekly to 12-weekly did not increase the incidence of life-threatening agranulocytosis in people taking clozapine. Our findings provide additional evidence to support existing calls for extended monitoring to be adopted in routine practice, to help address the prevalent under-utilisation of clozapine.^{Reference Ingimarsson, MacCabe, Haraldsson, Jónsdóttir and Sigurdsson6,Reference Atkin, Kendall, Gould, Freeman, Liberman and O'Sullivan7,Reference Schulte10,Reference Whiskey, Barnard, Oloyede, Dzahini, Taylor and Shergill11}

Comparison with other studies

The safety of extended monitoring has been previously shown by several studies.^{Reference Ingimarsson, MacCabe, Haraldsson, Jónsdóttir and Sigurdsson6,Reference Schulte10,Reference Hata, Fujimoto, Kanai, Yoshiyama, Nakatani and Nakabayashi12,Reference Matsui, Ishibashi, Kawano, Oshibuchi, Ishigooka and Nishimura13} To our knowledge, this is the first and largest study to investigate the clinical impact of extended monitoring in the UK. During the pandemic, data from a group of 50 patients (27%) in the USA demonstrated no clinical decompensation during a 6-week observational period after receiving extended monitoring.^{Reference Nichols, Gannon, Conlogue, Sarpal, Montgomery and Sherwood8} Similarly, a retrospective study in Qatar described the absence of neutropenia and fewer admissions to psychiatric and medical services in those receiving extended monitoring.^{Reference Wadoo, Karim, ElTorki, Riaz, Latoo and Alabdulla14} Overall, these findings suggest that a significant proportion of patients can safely receive clozapine with extended monitoring. Although some observers have expressed concerns about reducing haematological monitoring for those receiving clozapine treatment,^{Reference Lahdelma and Appelberg15} predominantly because of the risk of late-onset neutropenia, these events remain rare and comparable to other medications that are not subject to the same monitoring scrutiny.^{Reference Andrès and Maloisel16} Indeed, initial guidelines for the psychotropic carbamazepine extensively advocated routine monitoring for blood dyscrasias, but this was eventually removed owing to the rarity of the adverse effect.^{Reference Daughton, Padala and Gabel17}

Haematological and clinical outcomes

In our study there was no increase in people admitted with life-threatening haematological adverse events related to clozapine use in the group who received extended monitoring. Although incidence rates were higher than reported in previous studies, all cases of mild neutropenia were likely related to infection with COVID-19.^{Reference Giles, Varghese, Shymko, Nguyen and Waters9} This is consistent with evidence that the risks of neutropenia and agranulocytosis are very low after 1 year of treatment.^{Reference Ingimarsson, MacCabe, Haraldsson, Jónsdóttir and Sigurdsson6,Reference Schulte10} In fact, more recent data, including meta-analyses, have challenged the notion that clozapine poses an increased risk of neutropenia compared with conventional antipsychotics.^{Reference Myles, Myles, Xia, Large, Kisely and Galletly18,Reference Myles, Myles, Xia, Large, Bird and Galletly19} Of note, in the intervention group, a higher proportion had identified benign ethnic neutropenia compared with the comparison group, which might have increased the likelihood of detecting mild to moderate neutropenia in the former cohort. Moreover, when compared with larger historical data our extended monitoring cohort had similar haematological outcomes. In a total of 21 473 patients, Kumar^{Reference Kumar20} reported a neutropenia incidence of 7.43 per 1000 patient-years. Although definitions of moderate to severe neutropenia were marginally different between studies, when calculated with Kumar as a control, the IRR is 0.59 (95% CI: 0.12–2.25, P = 0.36), compared with 0.48 (95% CI 0.02–28.15, P = 0.29) with our control group.

In our study, we found no difference in psychiatric hospital admission rates in those receiving extended monitoring and the comparison group. In addition, clozapine doses and concomitant medications remained unchanged during the observation period, suggesting relative stability of mental state despite reduced face-to-face service contact. Notably, previous studies have suggested that some of the superior clinical benefits with clozapine, such as reduced mortality and self-harm compared with other antipsychotics, are independent of the increased service utilisation and clinician contact.^{Reference Wimberley, MacCabe, Laursen, Sørensen, Astrup and Horsdal21,Reference Hayes, Downs, Chang, Jackson, Shetty and Broadbent22} Moreover, there was no substantial difference in clozapine non-adherence between the two groups despite fewer contacts. Nevertheless, it is also plausible that the absence of group differences in our study was due to the relatively short follow-up.

COVID-19 infection

Undoubtedly, it is challenging to disentangle the effects of the clozapine policy changes from those of the COVID-19 pandemic, including social distancing and the resultant reduction in face-to-face contacts with mental health services. Nevertheless, we attempted to do this in our study through a mirror-image design, with a control group receiving care as usual. Contrary to expectations, we found that 5% of participants receiving extended monitoring contracted COVID-19, compared with 6% of those receiving treatment as usual, suggesting that the reduced contact was not effective in achieving the intended prevention or reduction of COVID-19 infection. Nevertheless, these data provide justification for the policy change to improve service utilisation, as previous studies have shown stringent clozapine monitoring to be a significant barrier to clozapine utilisation.^{Reference Parkes, Mantell, Oloyede and Blackman3} Moreover, it could be argued that this intervention may contribute to improved patient satisfaction.^{Reference Parkes, Mantell, Oloyede and Blackman3} Although COVID-19 has undoubtedly had a negative impact on clinical care in many areas, this policy change appears to provide a real opportunity to implement changes to improve clozapine utilisation rates in the UK and possibly beyond.

Clinical implications

There are wide international disparities between guidelines on the monitoring of clozapine treatment and many of the differences in recommendations appear arbitrary in nature. From a scientific standpoint, there has been much debate long before the pandemic regarding the necessity of haematological monitoring of clozapine in perpetuity, particularly regarding its indefinite nature and thresholds for discontinuation.^{Reference Oloyede, Blackman, Whiskey, Bachmann, Dzahini and Shergill23} Cases of clozapine-related agranulocytosis usually occur in the first 6 months of treatment,^{Reference Atkin, Kendall, Gould, Freeman, Liberman and O'Sullivan7} which suggests that this adverse drug reaction can be identified in the early stages by standard monitoring. In addition to this, there is accumulating evidence questioning the need for haematological monitoring beyond the first 6–12 months of clozapine treatment.^{Reference Schulte10,Reference Myles, Myles, Xia, Large, Kisely and Galletly18,Reference Myles, Myles, Xia, Large, Bird and Galletly19} In fact, authors have noted that the risk of neutropenia is comparable to that with other psychotropics not subject to the same blood testing requirements and that the stringency of current practice is a result of isolated cases early in clozapine's development. The impact of these early cases has distorted the perception of clozapine's relative risks, a perception that continues to dominate its widely reported clinical benefits for seriously and persistently ill patients, especially considering the limited alternative treatment strategies to address inadequate antipsychotic response. Coupled with this is the recently gained awareness that twice-monthly or monthly monitoring will have a low likelihood of capturing the characteristic rapid drop in neutrophil granulocytes during clozapine-induced agranulocytosis. Instead, a more pragmatic strategy of intensive haematological monitoring in the first year of clozapine initiation followed by selective haematological monitoring in case of febrile illnesses or pharyngitis needs to be explored. In their commendable review, over 10 years ago the Netherlands Clozapine Collaboration Group recommended lowering the frequency of blood tests to four times a year (off-label) for mentally competent and adequately informed patients who request it.^{Reference Oloyede, Blackman, Whiskey, Bachmann, Dzahini and Shergill23} As previously demonstrated, stringent monitoring of people receiving clozapine treatment imposes an unnecessary burden on them.^{Reference Parkes, Mantell, Oloyede and Blackman3} Such monitoring also has important economic implications,^{Reference Girardin, Poncet, Blondon, Rollason, Vernaz and Chalandon24} contributing to inequalities in access to healthcare while providing negligible benefit. It is our opinion that the relaxation of clozapine blood monitoring requirements prompted by the pandemic provides an opportunity for revising the protocol to be followed once the public health emergency has resolved.

Strengths, limitations and future studies

This study has several strengths. Our study's within-participant design helped to minimise the impact of individual-level confounders on the number of in-patient admissions by effectively comparing patients against themselves. Furthermore, inclusion of a control group separates the impact of the monitoring from that of the pandemic itself. However, a key limitation of the study is that patients who received extended monitoring were selected based on an absence of prior haematological events and demonstrated adherence to clozapine. This limits the generalisability of our results because participants were selected who were deemed at low risk of complications with extended monitoring. Nevertheless, as mentioned above, the utility of monthly monitoring has been questioned from a clinical and economic perspective. A further limitation is that our study includes a relatively short follow-up period. Future data may be required to determine clinical impact on a long-term basis. However, it has been demonstrated in one of our care settings that most patients will continue clozapine treatment long term and of those who do discontinue treatment, most do so within the first 4 years.^{Reference Gee, Shergill and Taylor25} As most of our patients were established on clozapine long before extended monitoring was introduced, it is plausible that discontinuation rates will remain low if extended monitoring is implemented in similar populations. Although our findings are encouraging, it should be noted that our sample is small considering that true clozapine-induced severe neutropenia is a rare adverse drug reaction. Our sample size of 459 participants in the extended monitoring group provides a power of 96% to detect one case of neutropenia and 27% to detect agranulocytosis. Despite insufficient power to detect agranulocytosis, we are still able to estimate the upper limit of the 95% CI for agranulocytosis risk. Applying the rule of three to estimate the upper limit of the 95% CI of agranulocytosis risk under the extended monitoring scheme, we can be 95% confident that the maximum risk does not exceed 0.6%, i.e. the 95% CI is 0–0.6%.^{Reference Hanley and Lippman-Hand26} Moreover, from a clinical perspective, it is plausible that a larger sample would have resulted in more non-clozapine induced, transient cases of neutropenia in the standard monitoring group compared with the extended monitoring group, simply due to the increased frequency of monitoring. It is unknown whether a reduction in the amount of physical contact patients had with clinicians was compensated by increased contact by telephone or other means, so the amount of monitoring patients receive might have been compromised. Finally, our study is underpowered to identify potential differences in haematological events. Moreover, the different monitoring between intervention and control groups makes comparison of haematological outcomes very difficult to interpret. Despite these limitations, this is the largest study to investigate the effectiveness of extended monitoring in patients receiving clozapine treatment. Future larger studies are required to confirm our preliminary findings.