{"paper_id":"4f19a4d7-da18-4ff1-83bc-c5e38bd33345","body_text":"1 \nPrescribing of antipsychotics for people diagnosed \nwith severe mental illness in UK primary care:  \nA 20-year investigation of who receives treatment, \nwith which agents, and at what doses. \nAuthors \nAlvin Richards-Belle1, Naomi Launders1, Sarah Hardoon1, Kenneth K.C. Man2,3,4,5,  \nElvira Bramon1,6, David P.J. Osborn1,6, Joseph F. Hayes1,6 \n \nAffiliations \n1 Division of Psychiatry, University College London, London, United Kingdom \n2 Research Department of Practice and Policy, School of Pharmacy, University College London,  \nLondon, United Kingdom \n3 Centre for Medicines Optimisation Research and Education, University College London \nHospitals NHS Foundation Trust, London, United Kingdom \n4 Department of Pharmacology and Pharmacy, Li Ka Shing Faculty of Medicine, University of \nHong Kong, Hong Kong \n5 Laboratory of Data Discovery for Health (D24H), Hong Kong Science Park, Hong Kong \n6 Camden and Islington NHS Foundation Trust, London, United Kingdom \n \nCorresponding author: \nAlvin Richards-Belle, Division of Psychiatry, University College London, London, United Kingdom \nEmail: alvin.richards-belle.21@ucl.ac.uk\n \n \nWord count: 3,960 \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n \n 2 \nABSTRACT \nBackground \nContemporary data relating to antipsychotic prescribing in UK primary care for patients \ndiagnosed with severe mental illness (SMI) are lacking. \n \nAims \nTo describe contemporary patterns of antipsychotic prescribing in UK primary care for patients \ndiagnosed with SMI. \n \nMethods \nCohort study of patients with an SMI diagnosis (i.e., schizophrenia, bipolar disorder, other non-\norganic psychoses) first recorded in primary care between 2000-2017 derived from Clinical \nPractice Research Datalink. Patients were considered exposed to antipsychotics if prescribed at \nleast one antipsychotic in primary care between 2000-2019. We compared characteristics of \npatients prescribed and not prescribed antipsychotics; calculated annual prevalence rates for \nantipsychotic prescribing; and computed average daily antipsychotic doses stratified by patient \ncharacteristics. \n \nResults \nOf 309,378 patients first diagnosed with an SMI in primary care between 2000-2017, 212,618 \n(68.7%) were prescribed an antipsychotic between 2000-2019. Antipsychotic prescribing \nprevalence was 426 (95% CI, 420-433) per 1,000 patients in the year 2000, reaching a peak of \n550 (547-553) in 2016, decreasing to 470 (468-473) in 2019. The proportion prescribed \nantipsychotics was higher amongst patients diagnosed with schizophrenia (81.0%) than with \nbipolar disorder (64.6%) and other non-organic psychoses (65.7%). Olanzapine, quetiapine, \nrisperidone, and aripiprazole accounted for 78.8% of all prescriptions. Higher mean olanzapine \nequivalent total daily doses were prescribed to patients with the following characteristics: \nschizophrenia diagnosis, ethnic minority status, male sex, younger age, and greater deprivation. \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 3 \nConclusions \nAntipsychotic prescribing is dominated by olanzapine, quetiapine, risperidone, and aripiprazole. \nTwo thirds of patients with diagnosed SMI were prescribed antipsychotics in primary care, but \nthis proportion varied according to SMI diagnosis. There were disparities in both receipt and \ndose of antipsychotics across subgroups - further efforts are needed to understand why certain \ngroups are prescribed higher doses and whether they require dose optimisation to minimise side \neffects. \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 4 \nINTRODUCTION \nAntipsychotic medications are primarily indicated for the management of psychotic symptoms \nassociated with severe mental illnesses (SMI), such as schizophrenia and bipolar disorder, and \nwere prescribed to 810,000 patients in England alone in 2022/23 (an increase of 22% from \n2015/16).\n1 Amongst patients diagnosed with schizophrenia-spectrum disorders, antipsychotic \nuse (versus non-use) is associated with a significantly lower long-term mortality rate. 2 Despite \nthis overall benefit, antipsychotic agents vary in their propensity for adverse reactions - with \ncardiometabolic effects, such as weight gain, dyslipidaemia, and hyperglycaemia, major \nconcerns of “second-generation” antipsychotics, such as olanzapine, quetiapine, and \nrisperidone.\n3  \n \nIn the United Kingdom (UK), primary care services are responsible for the long-term prescribing \nof antipsychotics to patients diagnosed with SMI. Data relating to antipsychotic prescribing in \nprimary care are therefore essential for monitoring trends and identifying priorities for quality \nimprovement and research. However, contemporary data on the SMI population are limited. Most \nrecent reports have focussed on other diagnoses, such as dementia\n4 or personality disorders,5 or \non all-cause prescribing in children and young people 6 and adults.7 \n \nEarlier studies have documented antipsychotic prescribing practice in primary care for patients \ndiagnosed with SMI.8,9 Prah et al. investigated trends in schizophrenia, 1998-2007,8 and Hayes \net al. investigated bipolar disorder, 1995-2009.9 Both studies (1) illustrated the shift from \nprescribing first- to second-generation antipsychotics, (2) highlighted olanzapine, risperidone, \nand quetiapine as the most frequently prescribed antipsychotics, and (3) documented increases \nin the proportion of time spent receiving antipsychotic treatment, particularly for women (those \naged \n≥ 45 diagnosed with schizophrenia8 and those 18-30 diagnosed with bipolar disorder9). A \nmore recent study reported further increases in antipsychotic prescribing to patients diagnosed \nwith bipolar disorder (from 37% of patients in 2001 to 45% by 2018), with quetiapine, olanzapine, \nand aripiprazole now the most frequently prescribed.\n10 Whether prescribing for schizophrenia \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 5 \nand other psychoses has followed these trends is unknown, particularly following the licensing of \naripiprazole in 2004.11 \n \nSeveral studies report disparities in aspects of antipsychotic prescribing in UK primary care. A \n2006 study, in one London borough, compared the primary care management of Black versus \nWhite patients diagnosed with psychosis and reported that Black patients had greater odds of \nbeing prescribed long-acting injectable antipsychotics.\n12 A study (2005-2015) of diverse \npsychiatric diagnoses identified that men were, on average, prescribed higher antipsychotic \ndoses than women - but results were not stratified by SMI diagnosis. 13 Further contemporary \nexploration of these and other potential disparities, including stratification by age and deprivation, \nare needed in order to inform efforts to achieve equity of care. \n \nAim and objectives \nIn order to inform future quality improvement and research into the safer prescribing of \nantipsychotics, the overall aim of this study was to describe contemporary (2000-2019) patterns \nof antipsychotic prescribing for people diagnosed with SMI in UK primary care. Specific \nobjectives were: \n1. to compare the characteristics of patients diagnosed with SMI prescribed and not \nprescribed antipsychotics in primary care; \n2. to describe the most frequently prescribed antipsychotics in primary care and how this \nmay have changed over time; and \n3. to describe the average prescribed daily antipsychotic dose over the first year of \nprescribing and explore whether doses vary according to diagnosis, ethnicity, age, sex, \nand deprivation. \n \nMETHODS \nStudy design and data source \nWe conducted a longitudinal cohort study, using data from Clinical Practice Research Datalink \n(CPRD), to investigate antipsychotic prescribing from 1 January 2000 to 31 December 2019 in a \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 6 \ncohort of people first diagnosed with SMI in primary care between 1 January 2000 and 31 \nDecember 2017. The study design is summarised in Supplementary Figure 1. \n \nCPRD encompasses two databases (Aurum\n14 and GOLD15) which, collectively, contain the de-\nidentified primary care records of over 62 million (current and historic) patients from participating \nNational Health Service primary care practices. Over 98% of the UK population are registered in \nprimary care and CPRD has been shown to be broadly representative, with coverage of almost a \nquarter of the current population.\n16,17 CPRD contains coded information on consultations, \nprescriptions, observations, and referrals. We used data from the May 2022 and April 2023 \nbuilds of Aurum and GOLD, respectively. \n \nEthics and consent \nAll procedures involving patients were approved by the East Midlands - Derby Research Ethics \nCommittee (reference: 21/EM/0265). This study was reviewed by the Independent Scientific \nAdvisory Committee of CPRD (protocol no. 21_000729). All data sent by GP practices to CPRD \nare anonymised and therefore individual patient consent was not required (patients are able to \nopt-out from their data being shared). \n \nParticipants \nThe cohort comprised patients actively registered in primary care between 2000-2019 identified \nas first receiving an SMI diagnosis in their primary care record between 2000-2017. SMI \ndiagnosis was defined as a recorded Read or EMIS® code indicating diagnosis of schizophrenia, \nbipolar disorder, or other non-organic psychoses (e.g., psychotic episodes, schizoaffective \ndisorders, delusional disorder, non-organic psychosis not otherwise specified) (see the online \nrepository for the code list, which was verified by a clinician (JFH)). SMI diagnoses are typically \nmade by psychiatrists in secondary care and subsequently communicated to primary care. The \nvalidity of SMI diagnoses recorded in primary care has been established.\n18 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 7 \nOutcomes: Antipsychotics \nPatients were considered exposed to antipsychotics if prescribed at least one antipsychotic in \nprimary care during the study period (2000-2019). Antipsychotics could be initiated by general \npractitioners or specialists (e.g. psychiatrists), but must have been issued through primary care \n(standard practice for longer-term community prescriptions in the UK).\n19 Unless otherwise \nspecified, antipsychotic prescription could pre-date the recording of SMI diagnosis in primary \ncare (provided it was within the study period), given that antipsychotics may be initiated before a \nspecific SMI diagnosis is formulated and/or communicated to primary care. Whilst antipsychotic \nprescription could pre-date SMI diagnosis, the requirement for first-recorded SMI diagnosis \nbetween 2000-2017 allowed for each patient to accrue at least up to two years follow-up post-\ndiagnosis (assuming they remained alive and registered in primary care). \n \nPrescriptions of antipsychotics (objective 1 and 2) \nAntipsychotic medications (current and withdrawn) were identified through review of national and \ninternational reference sources.20,21 Search strategies, based on antipsychotic generic and \ncommon brand names (Supplementary Table 1), were developed to identify relevant product \ncodes in CPRD code dictionaries. For patients in the cohort, product codes were then used to \nextract data from prescription records, including product name, ingredient, prescription issue \ndate, strength, formulation, route of administration, quantity, duration, and de-identified free-text \ncontaining dosing instructions. We included both oral and injectable antipsychotics, but did not \ninclude prochlorperazine as an antipsychotic given it is primarily used as an antiemetic. \n \nAntipsychotic dose (objective 3) \nWe derived the total daily prescribed oral antipsychotic dose for each of (up to) the first 12 \nprescription dates for patients newly initiating antipsychotics in the study period (i.e., where \npatients had no identified prescriptions for antipsychotics in primary care prior to the study \nperiod). We considered doses of all tablet (e.g., extended release, sublingual) and liquid, but not \ninjectable, antipsychotic formulations. Free-text dosage instructions (e.g., “take five tablets per \nday”) were converted to numerical quantities using a text-mining algorithm implemented in the R \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 8 \npackage doseminer.22 To enable comparison across agents, calculated doses were then \nconverted to olanzapine equivalents according to the Defined Daily Dose (DDD) method 23 using \nchlorpromazineR24 (cariprazine and droperidol were not reported in the DDD method, 23 \nequivalence formulae for these antipsychotics came from references 25 and 26, respectively). In \nthe case of multiple prescriptions issued on the same date, we considered up to three unique \nprescriptions of each antipsychotic prescribed on a given date (>3 unique prescriptions of one \nmedication was considered potentially erroneous). \n \nStratifying variables \nWe extracted additional variables from CPRD, to characterise the cohort and for stratified \nanalyses. These included: year of birth, sex, ethnicity, geographic region, relative deprivation, \ndate of first SMI diagnosis, SMI diagnosis, and prescriptions of antidepressants and mood \nstabilisers. Where a patient had multiple ethnicity categories recorded, the most frequently \nrecorded was used, or the most recent, if frequencies were equal. For patients registered in \nEngland, if ethnicity was not coded in CPRD, ethnicity data were sourced from linked Hospital \nEpisode Statistics (HES) data,\n27 where available. Geographic region refers to the location of the \nprimary care practice at which the patient was registered at - and included Northern Ireland, \nScotland, Wales, and nine regions across England (defined according to Office for National \nStatistics categories). Linked small area-level data were used for patients registered in England \nto provide information on relative deprivation (quintile of the 2019 English Index of Multiple \nDeprivation), derived according to patients’ residential postcode (or the primary care practice \npostcode as a proxy, if not available). Where a patient had more than one SMI diagnosis \nrecorded over time, the most recent diagnostic category was used as we considered this more \nlikely to be accurate given a more complete clinical history, retaining the first diagnosis date.\n28 \nWe used binary indicators for prescriptions of antidepressants and mood stabilisers (defined \naccording to British National Formulary [BNF] chapters 4.3 and 4.2.3,\n20 respectively) during the \nstudy period. Follow-up time was calculated as the amount of time in years that patients were \nregistered in primary care during the study period (accounting for end of registration, death, or \nadministrative censoring). \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 9 \nStatistical analysis \nAll analyses were conducted in R (version 4.3.1), with code available in the online repository \n(https://github.com/Alvin-RB/antipsychotics_descriptive_study_cprd ). Descriptive statistics were \nused to characterise the cohort, stratified by antipsychotic exposure status (objective one). To \ndescribe antipsychotic prescribing trends (objective two), we first calculated the number of \npatients prescribed each antipsychotic at least once - overall and separately for long-acting \ninjectables, and reported data for antipsychotics prescribed to \n≥ 50 patients. We then calculated \nperiod prevalence rates for the prescribing of antipsychotics, overall and for each antipsychotic, \nstandardised to 1,000 patients, for each year 2000-2019. Within each year, the numerator was \nthe number of patients that received at least one prescription for the antipsychotic over a \ndenominator of the number of patients alive, diagnosed with SMI, and remaining registered in \nprimary care. Allowing for recording delays, diagnosis could be recorded up to two calendar \nyears after prescription to be considered “diagnosed” in the given year. Period prevalence rates \nfor the top 15 most frequently prescribed antipsychotic medications were represented using line \ngraphs, overall and stratified by SMI diagnosis. Line graphs were also used to visualise the mean \ntotal daily prescribed oral antipsychotic dose (with 95% confidence intervals) for up to the first 12 \nprescription dates, stratified by diagnosis, ethnicity, age, sex, and deprivation (objective three). \nAmong those prescribed an antipsychotic more than once, we focused on the first 12 prescription \ndates amongst patients identified as new users of antipsychotics in the study period to ensure \ncomparable prescribing periods across patients. Assuming an average prescription duration of \n28-30 days, we anticipated that this would approximate patients’ first year of prescribing. If the \nnumber of daily doses prescribed was missing for a given prescription, it was imputed \n(Supplementary Table 2). For missing daily dose values, the previous dose was carried forward \nfor the missing observation, only if the dose at the subsequent time-point was the same. \n \nRESULTS \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 10\nObjective one: Characteristics of patients prescribed and not prescribed \nantipsychotics in primary care \nFrom a total of 514,526 patients ever receiving a SMI diagnosis in the CPRD database during \nthe study period, 309,378 were identified as having an SMI diagnosis first recorded in their \nprimary care record between 2000-2017. From these, 212,618 (68.7%) were prescribed an \nantipsychotic in primary care at least once between 2000-2019, whilst 96,760 (31.3%) were not \n(Table 1). \n \n[Table 1] \n \nPatients prescribed and not prescribed antipsychotics were broadly similar demographically \n(Supplementary Figure 2), but some regional differences were observed - with greater \nproportions prescribed antipsychotics in the North West of England, Northern Ireland, and Wales. \nThe proportion prescribed antipsychotics was higher amongst patients diagnosed with \nschizophrenia (81.0%) than with bipolar disorder (64.6%) and other non-organic psychoses \n(65.7%).  Amongst those not prescribed antipsychotics, over a fifth (22.7%) were prescribed \nmood stabilisers and over half (53.3%) antidepressants in the study period, but these proportions \nwere higher amongst those prescribed antipsychotics (31.6% and 69.4%, respectively). The \nmedian time registered in primary care during the study period was shorter amongst those not \nreceiving antipsychotics (3.4 vs. 5.6 years). Comparisons are stratified by SMI diagnosis in \nSupplementary Tables 3-5. \n \nAmong those prescribed antipsychotics, almost all (98.2%) received at least one oral \nprescription. The median time from SMI diagnosis to first oral antipsychotic prescription was 28  \n(IQR, -78 to 651) days and from first to most recent or last antipsychotic prescription was 3.5 \n(IQR, 0.8 to 8.5) years. Over a third (34.4%) were prescribed an antipsychotic a median (IQR) of \n16 (3 to 53) months prior to having an SMI diagnosis recorded in their primary care record. Of \nthose prescribed an antipsychotic overall, 8.5% were prescribed a long-acting injectable, but this \nproportion ranged from 4.5% to 16.4% amongst those diagnosed with bipolar disorder and \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 11\nschizophrenia, respectively (Supplementary Tables 3-5). Stratified by ethnicity, the proportion \nprescribed a long-acting injectable was highest amongst Black patients (9.2%), very similar \namongst Asian and Mixed patients (6.8% and 6.9%, respectively) and lowest amongst White \npatients and those of other ethnicities (5.5% and 4.5%, respectively). \n \nObjective two: Antipsychotic prescribing trends \nAfter excluding 1,171 prescriptions (across 764 patients) considered potentially erroneous \nduplicates, the 212,618 patients diagnosed with SMI and prescribed an antipsychotic had a total \nof 11,745,996 prescriptions, covering 33 different medications, between 2000-2019. Olanzapine \nwas prescribed at least once to 91,961 (43.3%) patients and was the most frequently prescribed, \nfollowed by quetiapine (n=70,250, 33.0%), risperidone (n=63,893, 30.1%), and aripiprazole \n(n=44,344, 20.9%) (Supplementary Figure 3). These four antipsychotics accounted for 78.8% of \nall prescriptions. The most frequently prescribed first-generation antipsychotics were \nchlorpromazine (n=17,195, 8.1%) and haloperidol (n=17,119, 8.1%). Clozapine was infrequently \nprescribed in primary care (n=5,346, 2.5%). Trends were similar when considering first- and \nsecond-line medications (Supplementary Table 6). The most frequently prescribed long-acting \ninjectables were flupentixol and zuclopenthixol (Supplementary Figure 4). \n \nThe overall prevalence of antipsychotic prescribing was 426 (95% CI, 420 to 433) per 1,000 \npatients in the year 2000, reaching a peak of 550 (95% CI, 547 to 553) in 2016, then decreasing \nto 470 (95% CI, 468 to 473) in 2019 (Supplementary Figure 5). Annual prevalence rates for \nindividual antipsychotics varied over time (Supplementary Figure 6) and according to SMI \ndiagnosis. Amongst patients with a diagnosis of schizophrenia, olanzapine was most frequently \nprescribed, and, for most of the time-period, this was followed by risperidone (Figure 1). \nHowever, in 2015, aripiprazole overtook risperidone. Amongst those with a diagnosis of bipolar \ndisorder, olanzapine had been the most frequently prescribed up until to 2009, after which it was \novertaken by quetiapine (Figure 2). Amongst patients diagnosed with other non-organic \npsychoses, prescribing prevalences for quetiapine, aripiprazole, and risperidone were all \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 12\nrelatively similar since 2016, but olanzapine was the most frequently prescribed throughout \n(Supplementary Figure 7). \n \n[Figure 1] \n \n[Figure 2] \n \nObjective three: Variation in average prescribed daily antipsychotic dose over \npatients’ first year of prescribing \nOf the 212,618 patients prescribed antipsychotics between 2000-2019, 194,979 were identified \nas newly prescribed an oral antipsychotic in primary care during the study period. After \nexclusions (15,703 for receiving just one prescription and 42 due to having no known doses \nacross their first 12 prescription dates), a total of 179,234 patients, with 1,780,077 prescription \ndates, were included. \n \nMean total daily prescribed oral antipsychotic doses varied across subgroups, but all tended to \nincrease slightly over the first 12 prescription dates. Stratified by SMI diagnosis, patients \ndiagnosed with schizophrenia were prescribed the highest doses (mean [SD] daily dose at 12\nth \nprescription date: 10.7 [7.4] mg olanzapine equivalent dose), whilst those diagnosed with bipolar \ndisorder were prescribed the lowest doses (7.2 [6.0] mg) (Figure 3). When stratified by ethnicity, \nBlack patients were prescribed the highest doses (9.7 [6.9] mg olanzapine equivalent dose), \nfollowed by Mixed (9.5 [6.7] mg), Other (9.0 [6.6] mg), then Asian (8.8 [6.7] mg), whilst White \npatients were prescribed the lowest doses (8.1 [6.7] mg) [doses were lower in those with missing \nethnicity data (7.7 [6.5] mg), but very similar to those of White patients] (Figure 3). Mean daily \ndoses were higher in males compared to females (Supplementary Figure 8), in younger \ncompared to older (65+) patients (Supplementary Figure 9), and in patients in more versus less \ndeprived areas (Supplementary Figure 10). \n \n[Figure 3] \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 13\nDISCUSSION \nUsing a large, longitudinal sample of 309,378 patients diagnosed with SMI between 2000-2017, \nwe provide contemporary data (2000-2019) on antipsychotic prescribing practice in UK primary \ncare. We identify several important findings relevant to informing future quality improvement and \nresearch into safer prescribing, including (1) prescribing is dominated by olanzapine, quetiapine, \nrisperidone, and aripiprazole - accounting for 79% of all prescriptions; (2) disparities in \nprescribed antipsychotic and dose exist - namely higher doses prescribed to patients with \ncharacteristics such as ethnic minority status and greater deprivation and (3) almost a third of \npatients with a contemporaneous SMI diagnosis are not prescribed antipsychotics in primary \ncare. \n \nOverall, olanzapine was the most frequently prescribed antipsychotic throughout the study \nperiod. Stratified by diagnosis, this remained true for schizophrenia and other non-organic \npsychoses, but not for bipolar disorder, where, since 2010, quetiapine was most frequently \nprescribed. Adverse cardiometabolic effects are a major concern of second-generation \nantipsychotics and when antipsychotics are ranked according to their impact on cardiometabolic \nparameters, olanzapine is consistently identified as one of the worst-ranking, particularly for \nchanges in body weight, body mass index, and low-density lipoprotein cholesterol.\n3 The \ncontinued popularity of olanzapine may be due to a perceived greater efficacy compared to other \nantipsychotics,\n29 despite most antipsychotics being considered broadly comparable in efficacy. 30 \nAlternatively, for patients well-established on olanzapine, it may be due to the perceived relapse \nrisk presented by switching to a different antipsychotic with less cardiometabolic bur den. \n \nThe 2004 licensing of aripiprazole led to a major change in prescribing, whereby prescriptions of \naripiprazole have increased year-on-year - now making aripiprazole one of the most frequently \nprescribed antipsychotics. This is an important development as current evidence suggests that \naripiprazole is associated with less adverse cardiometabolic effects, especially when compared \nto olanzapine and quetiapine.\n3,31 However, some reviews have reported aripiprazole to be less \nefficacious than some antipsychotics, such as olanzapine and risperidone, 29,32 although others \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 14\nreport no differences.30 Aripiprazole is also suggested to exacerbate psychotic symptoms \namongst patients with significant prior antipsychotic exposure. 33 Aripiprazole is still one of the \nmost recently licensed antipsychotics and current popularity might reflect effectiveness of \npharmaceutical marketing or a novelty effect in the face of limited innovations in the development \nof new antipsychotics. These issues highlight the difficulty, but necessity, of evaluating the \nrisk/benefit ratio of individual antipsychotics. \n \nIf it were possible to optimise current prescribing, then efforts focusing on olanzapine, quetiapine, \nrisperidone, and aripiprazole could have a large impact on the SMI population given their very \nwidespread use (79% of all antipsychotic prescriptions). Studies of the comparative safety and \neffectiveness of aripiprazole are particularly warranted given aripiprazole’s potential to reduce \ncardiometabolic risk alongside concerns of possibly lesser effectiveness. Conversely, some \nantipsychotics are rarely prescribed and so there is limited opportunity to learn about their \nrelative risks and benefits in pharmacoepidemiologic studies using routine clinical data. \n \nTo inform future quality improvement and research, we sought to describe current practice and \nidentify subgroups that may potentially be at more risk of dose-dependent adverse reactions. We \nfound that, on average, higher doses were prescribed to patients with the following \ncharacteristics: diagnosis of schizophrenia, ethnic minority status, male sex, younger age, and \ngreater deprivation. In addition, we replicated higher use of long-acting injectables amongst \nBlack patients.\n12,34 To our knowledge, this is the first study to report disparities according to \nethnicity and deprivation in prescribed antipsychotic dose in UK primary care. For ethnicity - \npatients from all ethnic minorities were prescribed higher doses than White patients. Confidence \nintervals for ethnic minority groups were inevitably wider than, but never overlapped with, the \nWhite group - owing to smaller sample sizes reflecting minority status. We did not aim to \nestimate whether certain characteristics are causally  related to being prescribed higher doses or \nto identify potential mediating factors, and therefore our analyses were unadjusted, as \nrecommended for descriptive studies.\n35 Clearly, multiple factors may influence decisions to \nprescribe at a certain dose, and further research is needed to disentangle the effects of these \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 15\nfactors in order to explain, and potentially inform efforts to reduce, these disparities. Causal \ninference approaches accounting for a wide range of potential confounders (e.g., markers of \nseverity, access to care care), alongside qualitative approaches examining clinical decision-\nmaking, would be informative. \n \nFinally, there was a trend of declining antipsychotic prescribing rates over the later study years \nand, overall, almost one third of patients with a contemporaneous SMI diagnosis were not \nprescribed antipsychotics in primary care. This is potentially concerning given reports of worse \noutcomes, including higher mortality, amongst patients diagnosed with schizophrenia-spectrum \ndisorders that are not prescribed antipsychotics.\n2 Noting that less than 2% of these patients were \nidentified as prescribed an antipsychotic in primary care prior to the study period, It is difficult to \nascertain if the remainder were truly unexposed based on primary care records alone. Although \nthe shorter follow-up time reduced the opportunity to identify antipsychotic prescriptions, this \ngroup still had a median follow-up 3.4 years - seemingly sufficient to detect regular prescribing. \nNevertheless, a small proportion will likely have been prescribed antipsychotics exclusively in \nsecondary care (e.g., as inpatients) - an issue particularly relevant for clozapine and long-acting \ninjectables. Alternative explanations might include: patients declining antipsychotics and/or \ninstead receiving psychological interventions or non-antipsychotic pharmacotherapies \n(particularly for those diagnosed with bipolar disorder); patients with brief or less severe \npsychotic episodes; or perhaps some were not in contact with services following diagnosis \n(although many were prescribed other psychiatric medications). \n \nStrengths and limitations \nA major strength of this study is the large longitudinal cohort of patients diagnosed with SMI, \nderived from CPRD - which is broadly representative of the UK population.\n14,15 CPRD includes all \nprescriptions issued in primary care and is therefore accurate in terms of planned treatment, and \nprescriptions issued repeatedly suggest adherence to that medication. We covered a 20-year \nperiod, enabling the identification of contemporary trends in prescribing for SMI, whereas other \nrecent studies focused on other diagnoses or on all-cause prescribing. When analysing \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 16\nantipsychotic dose, it was important to consider dose over multiple time-points in order to capture \npotential changes, rather than just the starting dose, which may not have accurately reflected \nongoing management. \n \nThis study also has limitations. First, we included only prescriptions issued from primary care \n(and so were not able to comment in detail on clozapine prescribing) and did not have data on \ndispensing or individual patient adherence (although repeat prescriptions issued with a regular \ncadence suggest adherence). Studies combining prescribing and dispensing data across primary \nand secondary care are needed to characterise the complete national picture on antipsychotic \nprescribing; such studies might soon be feasible with the continued development of national data \nresources.\n36 Second, we studied broad ethnic groups, consistent with UK-census high-level \nethnicity categories, and focused on between-group, rather than within-group, heterogeneity. \nStudies of more specific ethnic groups are needed, but w ill be challenging due to smaller sample \nsizes and greater misclassification risk. Moreover, although ethnicity should be self-reported in \nprimary care, we cannot verify this assumption. Third, the study period went up to 2019 and \ntherefore did not cover the COVID-19 pandemic period. Initial evidence from an England-wide \nanalysis suggests that antipsychotic prescribing remained relatively stable in the SMI population \nduring the pandemic period,\n37 but studies with a greater SMI focus are warranted. \n \nConclusion \nAntipsychotic prescribing is dominated by olanzapine, quetiapine, risperidone, and aripiprazole. \nTwo thirds of patients with diagnosed SMI were prescribed antipsychotics in primary care, but \nthis proportion varied according to SMI diagnosis. There were disparities in both receipt and \ndose of antipsychotics across subgroups - further efforts are needed to understand why certain \ngroups are prescribed higher doses and whether they require dose optimisation to minimise side \neffects. \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 17\nAcknowledgements \nThis study is based in part on data from the Clinical Practice Research Datalink obtained under \nlicence from the UK Medicines and Healthcare products Regulatory Agency. The data is \nprovided by patients and collected by the NHS as part of their care and support. The \ninterpretation and conclusions contained in this study are those of the authors alone. \n \nData availability \nData underlying this study were accessed via Clinical Practice Research Datalink (CPRD) under \napproved protocol no. 21_000729. Authors are not able to share the data directly, however data \ncan be accessed directly from Clinical Practice Research Datalink (CPRD) following approval \nand licensing (see https://cprd.com/\n for further details). \n \nAnalytic code availability \nThe analytic code supporting the findings are available in the online repository \n(https://github.com/Alvin-RB/antipsychotics_descriptive_study_cprd ). \n \nAuthor contributions \nARB, EB, DPJO, and JFH forumated the research questions and designed the study. ARB \nanalysed the data. ARB wrote the first draft of the manuscript and NL, SH, KKCM, EB, DPJO \nand JFH critically reviewed the manuscript for important intellectual content. All authors approved \nthe final version to be published and agree to be accountable for all aspects of the work in \nensuring that questions related to the accuracy or integrity of any part of the work are \nappropriately investigated and resolved. \n \nFunding \nARB is funded by the Wellcome Trust through a PhD Fellowship in Mental Health Science. This \nresearch was funded in whole or in part by the Wellcome Trust. For the purpose of Open Access, \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 18\nthe author has applied a CC BY public copyright licence to any Author Accepted Manuscript \n(AAM) version arising from this submission. \n \nKKCM reports grants from the CW Maplethorpe Fellowship, the European Union Horizon 2020, \nthe UK National Institute of Health Research, the Hong Kong Research Grant Council, the Hong \nKong Innovation and Technology Commission, and reports personal fees from IQVIA, unrelated \nto the current work. \n \nEB acknowledges the support of: Medical Research Council (G1100583, MR/W020238/1), \nNational Institute of Health Research (NIHR200756), Mental Health Research UK - John Grace \nQC Scholarship 2018, Economic Social Research Council’s Co-funded doctoral award, The \nBritish Medical Association’s Margaret Temple Fellowship, Medical Research Council New \nInvestigator and Centenary Awards (G0901310, G1100583), NIHR BRC at UCLH (Biomedical \nResearch Centre at University College London Hospitals NHS Foundation Trust and University \nCollege London). \n \nDPJO is supported by the University College London Hospitals NIHR Biomedical Research \nCentre and the NIHR North Thames Applied Research Collaboration. This funder had no role in \nstudy design, data collection, data analysis, data interpretation, or writing of the report. The views \nexpressed in this article are those of the authors and not necessarily those of the NHS, the \nNIHR, or the Department of Health and Social Care. \n \nJFH is supported by UKRI grant MR/V023373/1, the University College London Hospitals NIHR \nBiomedical Research Centre and the NIHR ARC North Thames. \n \nDeclaration of interest \nJFH has received consultancy fees from Wellcome Trust and funding grants from juli Health. All \nother authors declare no potential competing interests. \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 19\nREFERENCES \n1  NHS Business Services Authority. Medicines Used in Mental Health. England 2022/23. 2023. \n(https://nhsbsa-opendata.s3.eu-west-\n2.amazonaws.com/mumh/mumh_annual_2223_v001.html#23_Antipsychotics). \n2  Vermeulen J, van Rooijen G, Doedens P, Numminen E, van Tricht M, de Haan L. \nAntipsychotic medication and long-term mortality risk in patients with schizophrenia; a \nsystematic review and meta-analysis. Psychol Med 2017; 47: 2217–28. \n3  Pillinger T, McCutcheon RA, Vano L, Mizuno Y, Arumuham A, Hindley G, et al. Comparative \neffects of 18 antipsychotics on metabolic function in patients with schizophrenia, predictors \nof metabolic dysregulation, and association with psychopathology: a systematic review and \nnetwork meta-analysis. Lancet Psychiatry 2020; 7: 64–77. \n4  Stocks SJ, Kontopantelis E, Webb RT, Avery AJ, Burns A, Ashcroft DM. Antipsychotic \nPrescribing to Patients Diagnosed with Dementia Without a Diagnosis of Psychosis in the \nContext of National Guidance and Drug Safety Warnings: Longitudinal Study in UK General \nPractice. Drug Saf 2017; 40: 679–92. \n5  Hardoon S, Hayes J, Viding E, McCrory E, Walters K, Osborn D. Prescribing of antipsychotics \namong people with recorded personality disorder in primary care: a retrospective nationwide \ncohort study using The Health Improvement Network primary care database. BMJ Open \n2022; 12: e053943. \n6  Radoj č ić  MR, Pierce M, Hope H, Senior M, Taxiarchi VP, Trefan L, et al. Trends in \nantipsychotic prescribing to children and adolescents in England: cohort study using 2000–\n19 primary care data. The Lancet Psychiatry 2023; 10: 119–28. \n7  Marston L, Nazareth I, Petersen I, Walters K, Osborn DPJ. Prescribing of antipsychotics in UK \nprimary care: a cohort study. BMJ Open 2014; 4: e006135. \n8  Prah P, Petersen I, Nazareth I, Walters K, Osborn D. National changes in oral antipsychotic \ntreatment for people with schizophrenia in primary care between 1998 and 2007 in the \nUnited Kingdom. Pharmacoepidemiol Drug Saf 2012; 21: 161–9. \n9  Hayes J, Prah P, Nazareth I, King M, Walters K, Petersen I, et al. Prescribing Trends in \nBipolar Disorder: Cohort Study in the United Kingdom THIN Primary Care Database 1995–\n2009. PLoS ONE 2011; 6: e28725. \n10  Ng VWS, Man KKC, Gao L, Chan EW, Lee EHM, Hayes JF, et al. Bipolar disorder \nprevalence and psychotropic medication utilisation in Hong Kong and the United Kingdom. \nPharmacoepidemiol Drug Saf 2021; 30: 1588–600. \n11  EMA. Abilify. European Medicines Agency. 2 022. \n(https://www.ema.europa.eu/en/medicines/human/EPAR/abilify). \n12  Pinto R, Ashworth M, Seed P, Rowlands G, Schofield P, Jones R. Differences in the primary \ncare management of patients with psychosis from two ethnic groups: a population-based \ncross-sectional study. Fam Pract 2010; 27: 439–46. \n13  Bazo-Alvarez JC, Morris TP, Carpenter JR, Hayes JF, Petersen I. Effects of long-term \nantipsychotics treatment on body weight: A population-based cohort study. J \nPsychopharmacol 2020; 34: 79–85. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 20\n14  Wolf A, Dedman D, Campbell J, Booth H, Lunn D, Chapman J, et al. Data resource profile: \nClinical Practice Research Datalink (CPRD) Aurum. International Journal of Epidemiology \n2019; 48: 1740–1740g. \n15  Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data \nResource Profile: Clinical Practice Research Datalink (CPRD). International Journal of \nEpidemiology 2015; 44: 827–36. \n16  Clinical Practice Research Datalink. CPRD Aurum May 2022. 2022. (https://cprd.com/cprd-\naurum-may-2022-dataset). \n17  Clinical Practice Research Datalink. CPRD GOLD July 2023. 2023. (https://cprd.com/cprd-\ngold-july-2023-dataset). \n18  Nazareth I, King M, Haines A, Rangel L, Myers S. Accuracy of diagnosis of psychosis on \ngeneral practice computer system. BMJ 1993; 307: 32–4. \n19  NHS England. Responsibility for prescribing between Primary & Secondary/Tertiary Care. , \n2018 (https://www.england.nhs.uk/wp-content/uploads/2018/03/responsib ility-prescribing-\nbetween-primary-secondary-care-v2.pdf). \n20  The National Institute for Health and Care Excellence. BNF: British National Formulary. \n2022. (https://bnf.nice.org.uk/drug/). \n21  WHO Collaborating Centre for Drug Statistics Methodology. WHOCC - ATC/DDD Index. \n(https://www.whocc.no/atc_ddd_index/). \n22  Jani M, Yimer BB, Selby D, Lunt M, Nenadic G, Dixon WG. “Take up to eight tablets per \nday”: Incorporating free-text medication instructions into a transparent and reproducible \nprocess for preparing drug exposure data for pharmacoepidemiology. \nPharmacoepidemiology and Drug Safety 2023; 32: 651–60. \n23  Leucht S, Samara M, Heres S, Davis JM. Dose Equivalents for Antipsychotic Drugs: The \nDDD Method. Schizophr Bull 2016; 42: S90–4. \n24  Brown E, Shah P, Kim J. chlorpromazineR: Convert Antipsychotic Doses to Chlorpromazine \nEquivalents. 2024. (https://docs.ropensci.org/chlorpromazineR/). \n25  Leucht S, Crippa A, Siafis S, Patel MX, Orsini N, Davis JM. Dose-Response Meta-Analysis \nof Antipsychotic Drugs for Acute Schizophrenia. Am J Psychiatry  2020; 177: 342–53. \n26  Gardner DM, Murphy AL, O’Donnell H, Centorrino F, Baldessarini RJ. International \nconsensus study of antipsychotic dosing. Am J Psychiatry 2010; 167: 686–93. \n27  Shiekh SI, Harley M, Ghosh RE, Ashworth M, Myles P, Booth HP, et al. Completeness, \nagreement, and representativeness of ethnicity recording in the United Kingdom’s Clinical \nPractice Research Datalink (CPRD) and linked Hospital Episode Statistics (HES). Population \nHealth Metrics 2023; 21: 3. \n28  Launders N, Kirsh L, Osborn DPJ, Hayes JF. The temporal relationship between severe \nmental illness diagnosis and chronic physical comorbidity: a UK primary care cohort study of \ndisease burden over 10 years. Lancet Psychiatry 2022; 9: 725–35. \n29  Leucht S, Schneider-Thoma J, Burschinski A, Peter N, Wang D, Dong S, et al. Long-term \nefficacy of antipsychotic drugs in initially acutely ill adults with schizophrenia: systematic \nreview and network meta-analysis. World Psychiatry 2023; 22: 315–24. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 21\n30  Schneider-Thoma J, Chalkou K, Dörries C, Bighelli I, Ceraso A, Huhn M, et al. Comparative \nefficacy and tolerability of 32 oral and long-acting injectable antipsychotics for the \nmaintenance treatment of adults with schizophrenia: a systematic review and network meta-\nanalysis. The Lancet 2022; 399: 824–36. \n31  Ribeiro ELA, de Mendonça Lima T, Vieira MEB, Storpirtis S, Aguiar PM. Efficacy and safety \nof aripiprazole for the treatment of schizophrenia: an overview of systematic reviews. Eur J \nClin Pharmacol 2018; 74: 1215–33. \n32  Huhn M, Nikolakopoulou A, Schneider-Thoma J, Krause M, Samara M, Peter N, et al. \nComparative efficacy and tolerability of 32 oral antipsychotics for the acute treatment of \nadults with multi-episode schizophrenia: a systematic review and network meta-analysis. \nThe Lancet 2019; 394: 939–51. \n33  Takeuchi H, Remington G. A systematic review of reported cases involving psychotic \nsymptoms worsened by aripiprazole in schizophrenia or schizoaffective disorder. \nPsychopharmacology 2013; 228: 175–85. \n34  Das-Munshi J, Bhugra D, Crawford MJ. Ethnic minority inequalities in access to treatments \nfor schizophrenia and schizoaffective disorders: findings from a nationally representative \ncross-sectional study. BMC Medicine 2018; 16: 55. \n35  Lesko CR, Fox MP, Edwards JK. A Framework for Descriptive Epidemiology. American \nJournal of Epidemiology 2022; 191: 2063–70. \n36  Department of Health & Social Care. Data saves lives: reshaping health and social care with \ndata. (https://www.gov.uk/government/publications/data-saves-lives-reshaping- health-and-\nsocial-care-with-data/data-saves-lives-reshaping-health-and-social-care-with-data). \n37  Macdonald O, Green A, Walker A, Curtis H, Croker R, Brown A, et al. Impact of the COVID-\n19 pandemic on antipsychotic prescribing in individuals with autism, dementia, learning \ndisability, serious mental illness or living in a care home: a federated analysis of 59 million \npatients’ primary care records in situ using OpenSAFELY. BMJ Ment Health 2023; 26: \ne300775. \n38  Selby D, Yimer BB, Jani M, Nenadic G, Lunt M, Dixon W. drugprepr: prepare electronic \nprescription record data to estimate drug exposure. 2021. \n(https://research.manchester.ac.uk/en/publications/drugprepr-prepare-electronic-\nprescription-record-data-to-estimate). \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 22\nTable 1. Characteristics of patients prescribed and not prescribed antipsychotics in \nprimary care between 2000-2019. \n Not prescribed \nantipsychotic,  \nN = 96,760 \nPrescribed \nantipsychotic,  \nN = 212,618 \nDEMOGRAPHICS   \nSex, n (%)   \n  Female 45,521 (47.0%) 102,449 (48.2%) \n  Male 51,232 (53.0%) 110,158 (51.8%) \n  Unknown 7 11 \nEthnicity, n (%)   \n  Asian 4,443 (5.2%) 12,695 (6.7%) \n  Black 6,199 (7.2%) 14,920 (7.9%) \n  Mixed 2,048 (2.4%) 4,629 (2.4%) \n  Other 1,536 (1.8%) 3,282 (1.7%) \n  White 71,337 (83.4%) 153,493 (81.2%) \n  Unknown 11,197 23,599 \nGeographic region, n (%)   \n  East Midlands 2,053 (2.1%) 4,051 (1.9%) \n  East of England 3,901 (4.0%) 8,374 (3.9%) \n  London 21,401 (22.1%) 43,706 (20.6%) \n  North East 2,620 (2.7%) 5,384 (2.5%) \n  North West 13,844 (14.3%) 35,906 (16.9%) \n  Northern Ireland 828 (0.9%) 3,685 (1.7%) \n  Scotland 6,378 (6.6%) 15,842 (7.5%) \n  South East 15,888 (16.4%) 33,762 (15.9%) \n  South West 10,738 (11.1%) 19,829 (9.3%) \n  Wales 5,137 (5.3%) 11,782 (5.5%) \n  West Midlands 11,124 (11.5%) 24,781 (11.7%) \n  Yorkshire & The Humber 2,848 (2.9%) 5,516 (2.6%) \nIMD quintile, n (%)1   \n  1 (Least deprived) 10,698 (13.1%) 20,434 (11.6%) \n  2 12,871 (15.7%) 25,232 (14.3%) \n  3 15,629 (19.1%) 32,090 (18.2%) \n  4 20,147 (24.6%) 43,938 (25.0%) \n  5 (Most deprived) 22,500 (27.5%) 54,373 (30.9%) \n  Unknown 14,915 36,551 \nTime actively registered in study period (years), median \n(IQR) \n3.4 (1.2, 9.0) 5.6 (2.0, 12.9) \nMENTAL HEALTH   \nSMI diagnosis, n (%)   \n  Bipolar disorder 38,413 (39.7%) 70,137 (33.0%) \n  Other non-organic psychoses 45,207 (46.7%) 86,611 (40.7%) \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 23\n Not prescribed \nantipsychotic,  \nN = 96,760 \nPrescribed \nantipsychotic,  \nN = 212,618 \n  Schizophrenia 13,140 (13.6%) 55,870 (26.3%) \nAge at first SMI diagnosis, median (IQR) 36 (25, 52) 37 (27, 53) \nAge at first SMI diagnosis (category), n (%)   \n  <30 36,018 (37.2%) 68,413 (32.2%) \n  30-39 19,506 (20.2%) 46,368 (21.8%) \n  40-64 26,881 (27.8%) 63,867 (30.0%) \n  65+ 14,355 (14.8%) 33,970 (16.0%) \nYear of SMI diagnosis, median (IQR) 2008 (2004, 2012) 2008 (2004, 2012) \nPrescribed a mood stabiliser, n (%)2 21,954 (22.7%) 67,241 (31.6%) \nPrescribed an antidepressant, n (%)2 51,558 (53.3%) 147,574 (69.4%) \nNo mood stabiliser or antidepressant, n (%)2   \n  At least one antidepressant or mood stabiliser 57,306 (59.2%) 162,522 (76.4%) \n  No antidepressant/mood stabiliser 39,454 (40.8%) 50,096 (23.6%) \nTime from SMI diagnosis to end of follow-up (years), \nmedian (IQR) \n5.7 (2.2, 10.6) 6.8 (3.2, 11.8) \nANTIPSYCHOTICS   \nAntipsychotic initiation time-period, n (%)3   \n  <2000 - 13,895 (6.5%) \n  2000-2009 - 94,067 (44.2%) \n  2010-2019 - 104,656 (49.2%) \nPrescribed antipsychotic prior to SMI diagnosis date, n \n(%) \n- 73,038 (34.4%) \nEver prescribed oral antipsychotic, n (%) - 208,693 (98.2%) \nTime from SMI diagnosis to first oral antipsychotic \n(days), median (IQR) \n- 28 (-78, 651) \nAge at first oral antipsychotic, median (IQR) - 38 (28, 54) \nAge at first oral antipsychotic category, n (%)   \n  <30 - 59,971 (28.2%) \n  30-49 - 49,187 (23.1%) \n  40-64 - 66,226 (31.1%) \n  65+ - 33,309 (15.7%) \nEver prescribed LAI antipsychotic, n (%) - 17,976 (8.5%) \nTime from SMI diagnosis to first LAI antipsychotic \n(years), median (IQR) \n- 3.1 (0.4, 7.6) \nAge at first LAI, median (IQR) - 44 (32, 61) \nTime from first to last antipsychotic (years), median \n(IQR) \n- 3.5 (0.8, 8.5) \nIMD, index of multiple deprivation, severe mental illness; LAI, long-acting injectable. \n1 Amongst patients registered at primary care practices in England only. \n2 During the study period, 2000-2019. \n3 Amongst the 96,760 patients not prescribed an antipsychotic during the study period, 1,450 (1.50%) were prescribed an \nantipsychotic prior to the year 2000. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 24\n \nFigure 1. Annual prevalence rates for the prescribing of antipsychotics to patients diagnosed with schizophrenia. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 25\n \nFigure 2. Annual prevalence rates for the prescribing of antipsychotics to patients diagnosed with bipolar disorder. \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint \n\n \n 26\n \n \nFigure 3. Mean total daily prescribed oral antipsychotic dose over the first 12 prescriptions – stratified by severe mental illness diagnosis \nand ethnicity. \nGraphs show the mean olanzapine equivalent dose (mg) at each time-point, with 95% confidence intervals. The table beneath the g raph shows the corresponding mean (SD) olanzapine equivalent \ndoses at prescription date 1, 6 and 12, with the number of observations at each time-point. The overall median time between pre scription dates was 28 days.  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted March 27, 2024. ; https://doi.org/10.1101/2024.03.26.24304727doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}