National surveillance using a clinical quality indicator for prolonged antipsychotic use among older Australians with dementia who access aged care services

Dementia guidelines recommend antipsychotics are only used for behavioral and psychological symptoms when non‐drug interventions fail, and to regularly review use. Population‐level clinical quality indicators (CQIs) for dementia care in permanent residential aged care (PRAC) typically monitor prevalence of antipsychotic use but not prolonged use. This study aimed to develop a CQI for antipsychotic use >90 days and examine trends, associated factors, and variation in CQI incidence; and examine duration of the first episode of use among individuals with dementia accessing home care packages (HCPs) or PRAC.


| INTRODUCTION
Most people with dementia experience behavioral and psychological symptoms at some point, 1 at which time non-pharmacological strategies should be the mainstay for managing these symptoms.3][4] Understanding when and how medicines should be used for managing the behavioral and psychological symptoms of dementia has been identified as a key priority to support optimal medicines use in people with dementia. 5tipsychotics are high-risk medicines that contribute to sedative load, anticholinergic burden, and cognitive impairment in older people. 6Initiation of an antipsychotic within 3 months of entering permanent residential aged care (PRAC) is associated with a 20% higher risk of mortality among residents with dementia (adjusted hazard ratio 1.20, 95% CI 1.18-1.22). 7][9][10] Yet, many point prevalence studies throughout the world have reported widespread use of antipsychotic medicines in older people with dementia.2][13][14] In PRAC facilities, where appropriate management of behavioral and psychological symptoms of dementia is affected by factors such as resident acuity, workforce shortages and an unmet need for dementia care training, antipsychotics are frequently utilized.A meta-analysis of 57 studies reported a pooled prevalence of antipsychotic use of 26.1% (95% CI 25.1-27.2) among all residents of PRAC facilities. 15A recent study reported an increase in the prevalence of antipsychotic use by older Australians with dementia from 8% in the 9-12 months before PRAC entry to 32.9% within 3 months of entering PRAC. 7[18][19] Clinical trials examining the efficacy of antipsychotics for managing the behavioral and psychological symptoms of dementia are generally limited to a maximum of 12 weeks. 20In 2015, changes to the Australian product information for risperidone advised that treatment should not exceed 12 weeks when it is used for managing psychosis or persistent agitation or aggression among people with moderate to severe Alzheimer's disease. 21Prolonged antipsychotic use could place individuals with dementia at increased risk of medicines-related harm.However, international population-based clinical quality indicator (CQI) programs for routinely monitoring dementia care quality in PRAC facilities often monitor antipsychotic prevalence but not prolonged use. 22,23Three in every 10 older Australians with dementia accessing an aged care service in 2015-16 received an antipsychotic, 24 but duration of use was not examined.
Understanding duration of antipsychotic use in this population is necessary to ascertain if national efforts are improving care practices and inform next steps.Building on existing work to develop and test CQIs for inclusion in the newly established Australian Dementia Network (ADNet) Registry (a nationwide clinical quality registry for people newly diagnosed with mild cognitive impairment or dementia), 24,25 this study aimed to (i) develop a CQI for antipsychotic use of >90 days among people with dementia and examine annual trends, associated factors, and geographical and PRAC facility variation in CQI incidence and (ii) examine duration of use among individuals with dementia accessing aged care services who were new antipsychotic users.

| Study design and data source
Two retrospective analyses (repeated cross-sectional and cohort study) were conducted using data from the Registry of Senior Australians (ROSA) National Historical cohort.Described previously, ROSA is a national integrated aged care, healthcare, and social welfare platform which captures all individuals assessed for and/or who accessed government-funded aged care services from 2002 onwards. 26,27These services include HCPs and PRAC: HCPs are government-funded bundled aged care services for communitydwelling older people to provide assistance at home, while PRAC (similar to long-term care facilities or care homes) is funded for individuals who cannot live at home independently. 28Specific datasets analyzed in this study captured deidentified individual-level information from aged care eligibility and entry into PRAC assessments, aged care service records (i.e., dates and types of services received), prescription claims for medicines dispensed via the Pharmaceutical Benefits Scheme (PBS, coded by PBS item codes 29 and the World Health Organization Anatomical Therapeutic Chemical (ATC) classification system 30 ), and the Australian Institute of Health and Welfare (AIHW) National Death Index.

| Study cohort
The study cohorts comprised individuals aged 65-105 years with dementia, who did not identify as Aboriginal or Torres Strait Islander, and who accessed a government-subsidized HCP and/or PRAC between 1 July 2011 and 30 June 2016 (n = 270,253).Dementia was ascertained from aged care assessments or dispensing of a medicine used in the treatment of the cognitive symptoms (i.e., acetylcholinesterase inhibitor, memantine) or behavioral and psychological symptoms of dementia (i.e., risperidone, when subsidized for this indication) from July 2008 onwards, which enabled a minimum 3year lead in time to accurately capture the entire cohort of people with dementia. 31,32Individuals with schizophrenia or Huntington's disease diagnoses recorded in their aged care assessments were excluded (n = 3109) as antipsychotic treatment is an accepted component of management of these conditions and these individuals are often excluded from other CQIs assessing antipsychotic prevalence of use in aged care populations. 33alyses were undertaken separately for individuals who accessed an HCP and those who accessed PRAC.These two cohorts were defined annually and were not mutually exclusive.Individuals were followed until they ceased care due to death or for another reason (e.g., transferring from HCP to PRAC).Nationally, in 2021-22 the main reasons for exiting HCP episodes were transfer to PRAC (54.4%) or death (36.1%), while individuals commonly exited a PRAC episode at death (83.1%) or on transfer to another PRAC facility (8.9%) to commence a new PRAC episode. 34

| CQI development and testing
The CQI of interest was the proportion of people with dementia dispensed an antipsychotic for >90 days, calculated using the definition in Table S1.This indicator was prioritized by the ADNeT Registry due to changes to the product information for risperidone in 2015. 21Currently, risperidone is the only second-generation antipsychotic that is government-subsidized for managing psychotic symptoms and aggression among individuals with Alzheimer's disease and it is the most commonly dispensed antipsychotic to residents of PRAC facilities. 27Antipsychotic use was determined from prescription claims using ATC subgroup N05A, excluding N05AB04 (prochlorperazine) and N05AN01 (lithium).Because Australian pharmaceutical claims data do not include dosing instructions or duration of use, claims data in 2016 were used to estimate the duration of use for each antipsychotic preparation dispensed, using the waiting time distribution. 35The prescription duration represents the number of days in which 75% of individuals receive a subsequent dispensing of the same medicine (see Table S2).Individuals were considered to be taking an antipsychotic from the date of first dispensing to the date of last dispensing, plus the length of the prescription duration.Individuals who switched antipsychotics or were dispensed more than one type of antipsychotic were considered to be persistent provided use was continuous.Discontinuation of antipsychotic treatment was defined as a gap of more than two prescription durations between the end of a prescription duration and the start of any new antipsychotic prescriptions.

| Examination of antipsychotic use among new antipsychotic users
To further examine duration of antipsychotic use, the cohort was restricted to new HCP recipients and individuals residing in PRAC who first initiated an antipsychotic between July 2011-June 2016 and within the first six months of their PRAC stay and had at least two prescription claims for an antipsychotic up to 30 June 2017 (Figure S1).The PRAC cohort was restricted to individuals who initiated an antipsychotic within the first 6 months of PRAC because (i) antipsychotic prevalence of use peaks on PRAC entry and plateaus thereafter, 27 (ii) the median length of stay in PRAC is relatively short (i.e., 2.2 years for individuals with dementia), 34 and (iii) there may be differences in dementia severity and care needs for individuals entering PRAC for the first time compared to those who have lived in PRAC for multiple years.Antipsychotic use and discontinuation were defined as described above.

| Statistical analysis
Analyses were undertaken separately for the HCP and PRAC cohorts due to anticipated differences in the cumulative incidence of antipsychotic use.Descriptive statistics were used to characterize each cohort.
For the CQI analysis, we estimated the crude CQI cumulative incidence proportion and CQI incidence rate annually between 2011-12 and 2015-16 (i.e., to enable a lead in time of at least 3 years to capture the cohorts) using person-days as the denominator in each financial year.Poisson regression (or negative binomial regression if overdispersion was detected) was used to estimate changes in annual antipsychotic CQI incidence, with adjustment for age, sex, and comorbidity score.
After assessing associations between the covariates of interest and CQI incidence in 2015-16 bivariately, covariates with a p-value of ≤0.1 were assessed in backward stepwise multivariable Poisson (HCP cohort) or negative binomial regression models (for PRAC cohort due to overdispersion).The Akaike Information Criterion was used to assess model fit and correction for multiple hypothesis testing was undertaken using the stepdown Holm-Bonferroni method.
Funnel plots were constructed using estimates derived from logistic regression models that adjusted for age, sex, and comorbidity score to examine variation in CQI performance by geographical area (using the Statistical Areas Level 3 (SA3) categorization 39 ) and PRAC facility in 2015-16.We used the c-statistic, scaled Brier score, and the Hosmer-Lemeshow test to examine model fit and the 95% and 99.8% CIs around the cohort mean were estimated using the Wilson method. 40Only geographical areas or PRAC facilities with ≥20 individuals were shown in funnel plots to limit potential reidentification.
In a separate analysis, time to antipsychotic discontinuation among new antipsychotic users was estimated using the cumulative incidence function, with adjustment for the competing risk of death using the Fine-Gray method. 41Log rank tests were used to compare duration of first episode of use by year of initiation.Median time to antipsychotic recommencement was determined using Kaplan Meier curves.Follow-up for individuals was censored on 30/6/2017, PRAC entry (for HCP cohort) or transfer to another PRAC facility (for PRAC cohort), whichever occurred first.SAS version 9.4 (SAS Institute Inc.) was used for analyses.

Ethical approval
Ethical approval was provided by the University of South Australia   Little geographical variation in the adjusted proportion of individuals receiving an antipsychotic for >90 days was observed in either cohort, with 15.2% of SA3 areas in the HCP cohort and 2.8% in the PRAC cohort siting outside the 95% CI for this CQI (Table S5, Figure S2).Some facility-level variation was observed in the PRAC cohort, in which the adjusted proportion of residents treated for >90 days ranged from 0% to 55%, and 0.5% of PRAC facilities were above the upper 95% CI and 6.8% were below the lower CI around the mean (Figure 1).

| Examination of antipsychotic use among new antipsychotic users
Of the 15,767 HCP recipients treated with an antipsychotic over the five-year study period, 2367 (15.0%) initiated therapy while accessing the HCP and received at least two supplies during this time (Figure S1).Of the 71,311 incident residents of PRAC facilities receiving an antipsychotic, 15,597 (21.9%) initiated therapy within six months of PRAC entry and received at least two supplies.Characteristics of these individuals are shown in Table S6.The median time to antipsychotic initiation was 235 days (IQR 92-481) after HCP commencement and 31 days (IQR 7-80) after PRAC entry.In both cohorts, the most frequently initiated antipsychotics were risperidone (83.1% HCP, 81.2% PRAC), followed by haloperidol which was initiated by 14.7% of HCP recipients and 16.9% of residents of PRAC facilities (with 90.5% and 80.2% receiving oral or depot formulations, respectively).Antipsychotic therapy was often initially prescribed by a general medical practitioner (GP) (84.7% and 96.3% of newly dispensed antipsychotics in HCP and PRAC cohorts, respectively).Figure 2 presents the cumulative incidence of antipsychotic discontinuation and the competing risk of death in the HCP cohort.

| DISCUSSION
In this national study of 270,253 older Australians with dementia who accessed aged care services, small decreases (i.e., 3% annually) in the incidence of prolonged antipsychotic use for >90 days were observed between 2011-12 and 2015-16.Prior research in the same cohort reported that 31.3% of aged care recipients were dispensed an antipsychotic at least once in 2015-16, 24 while the present study found that 21.8% of PRAC recipients (who comprise most of the cohort) received an antipsychotic for >90 days that year.
These results, together with our examination of antipsychotic persistence among individuals in whom therapy was newly commenced, suggests that most older people with dementia taking an antipsychotic are treated for >90 days.This practice did not align with local recommendations during the study period, 42 nor current guidelines 20 that recommended initial treatment not exceed 12 weeks when antipsychotics are used for managing the behavioral and psychological symptoms of dementia.We also found that prior antipsychotic use was strongly associated with higher incidence of prolonged use in both cohorts, and little geographical and PRAC facility variation was observed.Importantly, this study demonstrates the feasibility of using population-based data for monitoring prolonged antipsychotic use in older people.
Decreased use of antipsychotics has been observed in studies conducted in PRAC facilities in the United States and Netherlands, 43,44 but not the United Kingdom. 45The 3% annual decrease in prolonged use observed in our study likely reflects changes to the product information for risperidone 21 and large-scale policy and quality improvement initiatives 46  diagnosis coding 52 which may also need to be monitored if the CQI tested in our study is implemented.
For individuals newly commencing an antipsychotic, we found that risperidone was the most used antipsychotic and treatment was most often prescribed by a GP.This is consistent with a previous study which examined antipsychotic initiation in the 100 days after PRAC entry using data from ROSA. 7 Risperidone is the only atypical that is PBS-subsided for managing behavioral and psychological symptoms among people with Alzheimer's disease 29 and local guidelines recommend risperidone be used first-line where pharmacotherapy is necessary. 2,20The median duration of antipsychotic use among new users was 334 days in the HCP cohort and 555 days among those receiving PRAC, with antipsychotics restarted in 41% and 34% of HCP and PRAC recipients who discontinued treatment, respectively.
Similarly, a nationwide register-based study of community-dwelling older people with Alzheimer's disease in Finland reported a median duration of 219 days (IQR 85-583), with 44% of individuals recommencing thereafter. 53Two other studies examining the duration of antipsychotic use in Australian PRAC facilities reported mean durations of use of 212 and 803 days, respectively, although these analyses were based on medicines administration records. 54,55These studies included prevalent users and used different methods to assess antipsychotic persistence, but together with our results reinforce that prolonged use of antipsychotics in people with dementia is common practice, especially in PRAC facilities.
Our findings provide additional evidence suggesting the need for system-level changes and person-centered services to limit antipsychotic initiation and support more timely withdrawal among older people with dementia.Additional supports for individuals with dementia during care transitions may also be required.Facilitators to deprescribing antipsychotics in people with dementia accessing PRAC include systems for reviewing appropriateness of use, policies and leadership support for deprescribing, and interprofessional collaboration. 56Comprehensive medicines reviews are one potential strategy to limit antipsychotic initiation and trigger deprescribing.
However, few individuals accessing HCP or PRAC receive comprehensive medicines reviews, 57 and the modest impact this stand-alone intervention has on overall antipsychotic use suggests additional supportive interventions may be required. 58There is some evidence that multifaceted interventions informed by behavior change theories can support successful withdrawal of antipsychotics in PRAC settings 46,59 and resources to guide review, tapering and withdrawal are available. 60,61To expand on the current government-subsidized comprehensive medicines review process, a new model of onsite pharmacist care will soon be implemented in Australian PRAC facilities.Further research is required to determine the impact of this program on prolonged antipsychotic use.Although a strength of this study is the use of aged care assessment data to ascertain dementia status, under-ascertainment is still possible as dementia may be underdiagnosed and underreported.Furthermore, the prevalence of use of medicines for managing the cognitive symptoms of dementia in Australia is low compared to countries such as Sweden, France, Germany and the UK. 62We excluded individuals with schizophrenia or Huntington's disease diagnoses from the CQI denominator in accordance with other CQIs assessing antipsychotic prevalence of use, 33 but did not exclude other mental health conditions such as bipolar disorder, which is not recorded in aged care assessments.We utilized PBS claims data to determine antipsychotic exposure and hence supplies from a doctor's bag, non-PBS funded prescriptions and those for hospital inpatients are not captured.The inability to capture non-PBS funded prescriptions may have led to an underestimation of the use of antipsychotics such as quetiapine and olanzapine, which are not government subsidized for managing the behavioral and psychological symptoms of dementia in Australia. 63,64ditional limitations included the inability to ascertain appropriateness of antipsychotic use, use for managing symptoms at end-of-life, or reasons for cessation.Consistent with other analyses of prescription claims data that do not capture dosing instructions, 35,58  initiating an antipsychotic continue treatment for more than 90 days.

| Strengths and limitations
Hence, antipsychotic use can be further minimized among older people with dementia who access aged care services.The CQI developed and tested in this study should be considered for use by care providers, government, and registries to monitor the quality and safety of antipsychotic use and implement strategies to improve dementia care quality.
(reference: 200489) and AIHW (reference: EO2022/4/1376) Human Research Ethics Committees.People identifying as Aboriginal or Torres Strait Islander were unable to be included in this study in accordance with existing ethical approvals.

F I G U R E 1
Abbreviation: CI, confidence interval.a Adjusted for age, sex, and Rx-risk comorbidity score.b p < 0.001.

F I G U R E 3
during this period.Our findings are in keeping with another population-based study which found the monthly rate of risperidone use among Australian veterans living in PRAC facilities decreased during 2012-2015, with further decreases observed immediately following changes to prescribing criteria in August 2015.47It is important to note that data for the present study were only available to mid-2016 at the time of cohort construction and changes in the prevalence of prolonged antipsychotic use and overall duration of use may have occurred since this period.Additional research to examine the impact of more recent initiatives to curb inappropriate antipsychotic use among people with dementia (e.g., national clinical guidelines,2,20 changes to PBS-subsidy criteria for risperidone,29 implementation of a national mandatory quality indicator program in PRAC facilities,48 and other aged care reforms in response to the Royal Commission into Aged Care Quality and Safety 49 ) as well as possible impacts due to the COVID-19 pandemic 50 are needed.Undertaking similar analysis using the prospective ADNeT Registry cohort which commenced recruitment in 2020 may be able to assess the impact of these recent initiatives on antipsychotic use.Recent international studies have also reported potential unintended consequences coinciding with initiatives to improve antipsychotic use in PRAC facilities such as substitution with other sedative medicines51 and changes in F I G U R E 2 Cumulative incidence of antipsychotic discontinuation and the competing risk of death among individuals with dementia who received a HCP and were new users of an antipsychotic (n = 2367 individuals).Shaded area represents the 95% CI. *n = 1495 (63%) of the HCP cohort were censored due to permanent residential aged care entry.HCP, home care package.Cumulative incidence of antipsychotic discontinuation and the competing risk of death among individuals with dementia who accessed permanent residential aged care (PRAC) and were new users of an antipsychotic (n = 15,597 individuals).Shaded area represents the 95% CI.

of 11 -
prescription durations were derived from claims data (based on the number of days in which three-quarters of individuals receive a prescription refill) to estimate antipsychotic duration of use; hence overestimation or underestimation is possible.Our analysis presumes that all medicines dispensed were consumed which means exposure misclassification is also possible for (i) individuals who ceased treatment and did not consume the full quantity dispensed or (ii) those who restarted therapy soon after treatment cessation.While the CQI developed in this study measures a specific aspect of care quality, there is a need to consider a broader range of factors, such as patient preferences and values, resources, care coordination, and cultural factors, when determining the quality of dementia care.5 | CONCLUSIONS Although there was a small decrease in the incidence of prolonged antipsychotic use between 2011-12 and 2015-16 among aged care recipients with dementia, study findings indicate most individuals 8 SLUGGETT ET AL.

Table 1 )
. The median length of follow up was 675 days (IQR 305-1165) for the HCP cohort and 512 days (IQR 198-1016) for the PRAC cohort.Characteristics of both cohorts in the first and final study years are shown in TableS3.
T A B L E 1 Clinical quality indicator incidence (antipsychotic use >90 days) in 2011-12 and 2015-16.
31ere dementia clinical quality registries or health datasets with linked prescription claims data are available.This cohort is known to represent nearly 40% of Australians with dementia.31Ourexamination of duration of use is particularly important given the majority of existing CQIs examine the prevalence of antipsychotic use, which might capture individuals with a once-off dispensing (e.g., dispensed at hospital discharge but not continued at home, or dispensed for end-of-life care) and not prolonged use.
Study strengths include the development of a new CQI for prolonged antipsychotic use among a nationwide cohort of individuals with dementia who access aged care services.This CQI could be applied in other countries