Assessing the Preparedness of the Australian Health Care System Infrastructure for an Alzheimer’S Disease-Modifying Therapy

Research Report AUSTRALIA

SANGITA M. BAXI, FEDERICO GIROSI, JODI L. LIU Assessing the Preparedness of the Australian Health Care System Infrastructure for an Alzheimer’s Disease-Modifying Therapy

SUMMARY AND KEY FINDINGS

■■ Disease-modifying therapies to prevent or delay the progression of Alzheimer’s disease (AD) are under clinical investigation. AD is the leading cause of dementia, and over 800,000 Australians with mild cognitive impairment in 2018 could potentially benefit from a therapy that reduces the risk of progression to dementia.

■■ The delivery of a therapy to people with early-stage disease presents potential infrastructure challenges in diagnosing and treating a large population. The objective of this analysis is to illustrate the magnitude of capacity challenges for the Australian health care system if a hypothetical therapy becomes available in 2023.

■■ We use a simulation model to compare the expected number of patients with the capacity of the Australian health care system and assess three possible constraints: visits to dementia specialists, biomarker testing and infusion delivery. We make many assumptions, including those related to specifications of a hypothetical therapy, patient uptake rates and projected capacity growth trends.

■■ If capacity follows historical growth trends, our analysis suggests that Australians could wait over ten months in the initial year. We find that the most pressing constraint is the availability of dementia specialists to evaluate and diagnose patients, followed by average waiting times of a few months for biomarker testing to confirm diagnosis and infusion delivery of the treatment. The first year without any wait times would be 2034. By eliminating wait lists that delay people from accessing the therapy, we estimate that 54,000 Australians would not progress from mild cognitive impairment due to AD to Alzheimer’s dementia between 2023 and 2033.

■■ Although our analysis includes simplifying assumptions and is not predictive of the future, our intention is to facilitate discussions among stakeholders. Multiple stakeholders would need to coordinate reimbursement, workforce and capacity planning, and outreach efforts to ensure that people with early-stage AD would have timely access if a therapy becomes available.

Introduction At present, no disease-modifying therapy (DMT) is available for AD; however, 83 DMT agents have An estimated 376,000 Australians live with demen- been evaluated in more than 200 clinical tr ials from tia as of 2018 (Australian Institute of Health and 2002 to 2012 (Cummings, Morstorf and Zhong, Welfare [AIHW], 2018). Since 2013, dementia has 2014).1 In 2018, there were several DMTs in develop- been the second leading overall cause of death in ment (Cummings et al., 2018), most of which target Australia and accounted for 13,729 deaths in 2017 beta-amyloid, tau or both proteins (Cummings, (Australian Bureau of Statistics [ABS], 2018c). Morstorf and Zhong, 2014; Cummings, Morstorf and Rates of dementia differ among subpopulations; for Lee, 2016; Cummings et al., 2017; Cummings et al., example, dementia prevalence among Aboriginal 2018). Selected candidate therapies currently in phase and Torres Strait Islanders is estimated to be two 2 and phase 3 clinical trials are presented in Ta ble 1. to five times that of nonindigenous Australians (AIHW, 2018). Alzheimer’s dementia is the most common type of dementia. Alzheimer’s Disease International (2009) estimates 50 to 75 percent of dementia is Although a therapy is due to Alzheimer’s disease (AD). AD is a progres- sive neurodegenerative disorder characterised by not yet available, early changes in the brain including the accumulation of beta-amyloid plaques and tau tangles. These facilitation of policy changes accumulate over many years, progress- discussions and actions ing through stages of cognitive impairment and eventually leading to irreversible neurological could help minimise degeneration and Alzheimer’s dementia (Perl, 2010). Although AD is the most common cause of demen- future infrastructure tia, people often exhibit mixed dementia in which more than one type of dementia occurs, such as issues in which an Alzheimer’s dementia in combination with vascu- lar dementia or Lewy body dementia (Dementia approved therapy Australia, n.d.b). Risk factors for AD include age, genetics and lifestyle; while there are nonpharma- cannot be accessed cological approaches to reducing risk and managing symptoms, there is currently no way to prevent AD due to capacity (Dementia Australia, n.d.a). constraints.

2 TABLE 1 Selected Alzheimer’s Disease-Modifying Therapy Candidates in Phase 2 and Phase 3 Clinical Trials, as of April 2019

Clinical Expected Primary National Clinical Candidate Sponsor Trial Phase Completion Date Trial Identifier

Anti-beta-amyloid antibodies

Gantenerumab Hoffmann-La Phase 3 May 2022 NCT03443973, Roche NCT03444870

BAN2401 Eisai/Biogen Phase 3 March 2024 NCT03887455

LY3002813 a Eli Lilly Phase 2 October 2020 NCT03367403

Anti-tau antibodies

ABBV-8E12 AbbVie Phase 2 April 2021 NCT02880956

RO7105705 Genentech Phase 2 September 2020 NCT03289143

BACE inhibitors

Elenbecestat (E2609) Eisai/Biogen Phase 3 June 2021 NCT03036280

CNP520 Novartis/Amgen/ Phase 2/3 July 2024 NCT03131453 Banner Alzheimer’s Institute

Vaccines

CAD106 (anti-beta amyloid) Novartis Phase 2/3 August 2024 NCT02565511

AADvac1 (anti-tau) Axon Phase 2 June 2019 NCT02579252 Neuroscience SOURCE: RAND analysis of company websites and ClinicalTrials.gov website as of April 4, 2019. NOTES: Anti-beta-amyloid and anti-tau antibodies are monoclonal antibodies that are typically administered by intravenous or subcutaneous infusions. Beta-secretase (BACE) inhibitors are oral drugs that block BACE, a beta-site amyloid precursor protein-cleaving enzyme thought to prevent the accumulation of beta-amyloid. Alzheimer’s vaccines are injections of antigens or preformed antibodies with the aim of triggering antibody responses. a LY3002813 is being evaluated alone and in combination with LY3202626, a BACE inhibitor.

Despite incomplete understanding of the dis- an approved therapy cannot be accessed due to ease mechanisms and drug targets, there has been capacity constraints. If a DMT becomes available, some optimism for fu ture therapies in recent years, one of the infrastructure challenges that health particularly for therapies to delay or prevent disease systems would face is how to diagnose and treat a progression in patients in early stages of AD such as large number of people with early-stage AD. Based mild cognitive impairment (MCI), rather than thera- on MCI prevalence published in a meta-analysis and pies that aim to cure or reverse Alzheimer’s dementia Australian population projections by age, we esti- after manifest dementia occurs (Aisen, 2017; Lancet mate that approximately 802,225 Australians aged Neurology, 2015). However, the future availability of 50 and older had MCI in 2018 (ABS, 2018e; Petersen an Alzheimer’s DMT remains uncertain, and phase et al., 2018). 3 trials for two leading amyloid candidates were dis- The objective of this analysis is to assess continued in early 2019 (Biogen, 2019; F. Hoffman-La the magnitude of the potential health system Roche Ltd, 2019). challenges in the diagnosis and treatment of early- Although a therapy is not yet available, early stage AD in Australia, given the hypothetical facilitation of policy discussions and actions could availability of a DMT (including approval and help minimise future infrastructure issues in which public reimbursement). The modelling results

3 FIGURE 1 Conceptual Framework of the Disease States and Clinical Pathway

Untreated normal cognition or MCI Progression of MCI due to AD without treatment Screening phase Diagnostic phase Treated MCI First Second due to AD Age 50+ dementia dementia specialist visit specialist visit Treatment phase Reduced progression Treatment of MCI due Seek cognitive (reduces to AD after assessment? Biomarker risk of Treatment Yes treatment testing indicated for MCI progression Manifest due to AD? Yes indicated? from MCI Alzheimer’s due to AD to dementia Cognitive Yes Alzheimer’s assessment dementia) for MCI Biomarker test

MCI detected?

Yes Positive for Yes biomarkers?

Yes Seek specialist?

NOTE: Individuals may exit the model if the next step is not indicated at each decision point (diamonds) and due to all-cause mortality (not shown in the figure).

reflect a thought experiment of future challenges with a simplified clinical pathway in which people rather than predictive modelling. This study move through screening, diagnosis and treatment follows our previous analyses in the United States, within the health system if a DMT for AD was avail- six European countries and Canada (Liu et al., 2017; able. People in the normal cognition and untreated Hlávka, Mattke and Liu, 2018; Liu et al., 2019). MCI states go through the screening and diagnosis clinical phases. Within the MCI state, people may have undiagnosed/untreated MCI, MCI due to cau ses Conceptual Framework and other than AD, untreated MCI due to AD or treated Simulation Model MCI due to AD depending on what step they are in Conceptual Framework within the clinical pathway. People are considered to have treated MCI due to AD after undergoing the Figure 1 shows the conceptual framework that is the treatment clinical phase. Untreated and treated basis for our simulation model. There are two major MCI due to AD progress to Alzheimer’s dementia at components to the framework: the disease trajec- different rates. tory (depicted in the grey boxes) and the clinical Within the normal cognition and MCI states, pathway (depicted in the white boxes). The frame- the Screening Phase includes cognitive assessment of work has a trajectory through disease states—normal older adults that could occur in primary care set- cognition, MCI and Alzheimer’s dementia—overlaid tings via an assessment of functional deficits and a

4 short instrument such as the Folstein Mini-Mental State Examination (MMSE) (Folstein, Folstein and McHugh, 1975) or the General Practitioner We model three Assessment of Cognition (2016; see also Brodaty et al., 2002).2 These assessments are not specific for MCI or capacity constraints: dementia and are a general screen for mental function that a provider can administer in approximately less availability of dementia than 15 minutes. Those who exhibit cognitive impairment might be evaluated for reversible and treatable specialists, biomarker causes (i.e., hematologic, metabolic, mood or pharma- testing and treatment cological causes). If no other treatable or reversible features are delivery. detected and the impairment is consistent with a potential diagnosis of AD, an individual would likely be referred to a dementia specialist for further evaluation (Diagnostic Phase). The further evaluation may exits the model each year based on age-adjusted all- include additional cognitive and functional assess- cause mortality rates. ments.3 After further evaluation, a dementia special- Second, a systems dynamic model simulates ist may refer people for testing of amyloid and/or health care system capacity constraints within tau biomarkers to determine if the MCI is likely due the MCI state. Within the MCI state, pe ople are to AD. A dementia specialist would then determine diagnosed (for MCI due to AD) and treated, and whether treatment is indicated for individuals who the outflows from these steps are constrained by have a positive biomarker test.4 infrastructure capacity. We model three capac- If indicated, patients could be treated with a ity constraints: availability of dementia specialists, therapy that would reduce the risk of progression biomarker testing and treatment delivery. The from MCI to Alzheimer’s dementia (Treatment model includes two dementia specialist visits in the Phase). While pe ople have untreated MCI due to diagnostic phase. The allocation of capacity between AD, the disease continues to progress. Relative to these two visits is optimised such that specialists treated MCI patients, untreated MCI patients have a do not take on a new patient for an initial visit if higher risk of progressing from MCI to a later stage they do not have the capacity to provide confirma- of the disease with manifest dementia, at which tory visits for existing patients in the same year. As point we assume a DMT would no longer with the annual transitions between disease state in be effective. the stylised model, the clinical pathway for screening, diagnosis and treatment is also modelled as an Simulation Model annual step (i.e., individuals may progress through these phases in one year if capacity allows). We Our simulation model consists of two parts that have applied this model to assess the preparedness reflect the disease trajectory and clinical pathway of the health care systems in the United States (Liu in the framework. First, a Markov model simu- et al., 2017), six European countries (Hlávka, Mattke lates annual transitions between disease states. and Liu, 2018) and Canada (Liu et al., 2019). See Individuals move through the disease states— the model overview in the Appendix for additional from no cognitive impairment (i.e., no MCI and details. no Alzheimer’s dementia) to MCI to Alzheimer’s For this analysis, we incorporate Australian data dementia—according to transition probabilities on the population, disease prevalence, mortality derived from the liter ature (see Table A.1 in the and historical workforce and infrastructure into the Appendix). In each state, a share of individuals also model. See Ta ble A-1 for model parameter values and their respective sources.

5 Model Assumptions input collected in the original development of In this analysis of a hypothetical future therapy, the model in Liu et al., 2017). we incorporate many simplifying assumptions. • The number of individuals with MCI is These assumptions include the year of approval, derived from MCI prevalence estimates from approved indication, age eligibility to receive treat- a meta-analysis of 20 studies (Petersen et al., 7 ment, treatment modality, treatment dosage and 2018). Of those who screen positive for MCI, duration, treatment effectiveness and the treatment we assume 50 percent would follow up with safety profile. Our assumptions about patient uptake, a dementia specialist referral to seek further contraindications, treatment delivery and effective- evaluation (based on expert input collected in ness are generally consistent with those used in our Liu et al., 2017). prior United States, European and Canadian stud- • We assume that dementia specialists would ies that were informed by expert input (Liu et al., conduct further evaluation of people with 2017; Hlávka, Mattke and Liu, 2018; Liu et al., 2019). MCI that could be due to AD. In Australia, We apply the same assumptions for this analysis of the specialists who are likely to be involved in the Australian health care system unless otherwise diagnosing MCI due to AD are geriatricians, noted. To inform our assumptions for the Australian neurologists and old-age or neuropsychia- context, we reviewed the literature and consulted trists. We assume approximately 15 pe rcent of two experts familiar with clinical practice and care all psychiatrists practice old-age psychiatry or delivery in Australia. neuropsychiatry, based on a national sur- The key assumptions in our analysis are as vey of the psychiatry workforce in Australia follows: (Royal Australian and New Zealand College of Psychiatrists [RANZCP], 2014). If the • A DMT for patients with MCI due to evaluation confirms MCI and does not find AD becomes available in 2023.5 We assume an alternative explanation for MCI (e.g., prior the therapy would be delivered through a stroke) or a reason to not pursue treatment course of intravenous infusions, based on the (e.g., presence of another life-limiting disease), route of administration for anti-beta-amyloid specialists would refer individuals for bio- monoclonal antibody therapies that are the marker testing. Of th ose with confirmed MCI furthest along in clinical trials. possibly due to AD, we assume that 90 percent • We assume that individuals age 50 and older of patients would be referred and consent to are eligible for annual cognitive screening. biomarker testing, based on expert input in We model the population 50 years and older the original development of the model (Liu because later-stage clinical trials include ages et al., 2017). as young as 50 (e.g., U.S. National Library of • We assume that biomarker testing may be per- Medicine, 2019a; 2019b).6 Screening starts in formed with a positron emission tomography the year prior (2022) as patients and provid- (PET) scan for amyloid deposits in the brain ers anticipate approval of the therapy in 2023. or with lumbar puncture to retrieve cerebro- General practitioners could conduct cognitive spinal fluid (CSF) for assessment of amyloid assessments, wh ether through annual screen- and/or tau proteins.8 The relative use of th ese ings or assessments when early symptoms biomarker tests in the future with a DMT is of cognitive impairment are suspected. We uncertain and would depend on factors such assume the capacity to conduct preliminary as reimbursement levels, costs and patient cognitive and functional assessments would preferences. Based on our assessment of the be unconstrained. We assume that 80 percent liter ature and input from Australian experts, of individuals age 50 and older would be we selected a base-case assumption that screened each year starting in the year prior 80 percent of tests would be performed using to the availability of a DMT (based on expert PET and 20 percent would be performed

6 using CSF.9 Last, we assume that 45 percent of uptake, which would be dependent on fac tors such people with MCI have clinically relevant bio- as the therapeutic profile, effectiveness, outreach and marker levels, based on published biomarker awareness and public reimbursement of a DMT in studies (Doraiswamy et al., 2014; Ong et al., Australia. As the safety profile and possible adverse 2015). reactions are unknown with a hypothetical therapy, • If an individual’s biomarker level is clini- the model does not account for specific treatment cally relevant, she or he returns to a dementia monitoring. specialist who determines wh ether treatment Although our model focuses on three key is indicated. The individual is referred for capacity constraints, there would likely be other treatment if there are no contraindications capacity challenges. For example, with screening, and the individual consents. Of people with the capacity of general practitioners to conduct cog- MCI who test above a certain biomarker level, nitive assessments may be limited given competing we assume that 80 percent would be eligible needs with ageing populations. On the other hand, for and seek treatment (based on expert input) behavioural issues such as awareness of screening (Liu et al., 2017). availability or stigma associated with a positive • We assume that the therapy would be deliv- screen for predementia may reduce patient uptake ered by intravenous infusion every four weeks for screening. for one year, following protocols for a typi- Our model evaluates the capacity in Australia cal immunotherapy. We further assume that as a whole. We do not examine capacity issues at treatment reduces the relative risk of progres- the state and territory level or for different popula- sion from MCI due to AD to Alzheimer’s tions. There would be additional capacity challenges dementia by 50 pe rcent relative to untreated and access issues in rural and remote areas and in MCI due to AD. closing the health status gap between the general population and Aboriginal and Torres Strait Islander people (Smith et al., 2008; Eades et al., 2010; AIHW, Limitations 2018; Australian Government Department of the As with our previous studies, this analysis has Prime Minister and Cabinet, 2019). Moreover, while limitations. First, our model is based on a styl- we model national capacity, the actual expansion of ised disease trajectory and clinical pathway that diagnostic and treatment delivery capacity would includes the many simplifying assumptions out- depend on national, state and territory decisions on lined above. The disease trajectory consists of three priorities, reimbursement, capital investment and states (normal cognition, MCI and Alzheimer’s clinical guidelines. dementia); the model does not allow for rever- With these limitations in mind, our analysis is sion of MCI back to normal cognition. This meant to convey the magnitude of potential capac- reversion has been observed at fairly high rates ity challenges in hypothetical scenarios to inform but occurs at lower rates in clinic-based studies policy and planning decisions, not to predict future than community-based studies (Malek-Ahmadi, outcomes. 2016). Although reversions occur, individuals who revert are at higher risk for retransitioning to MCI Patient Demand and (Koepsell and Monsell, 2012). In addition, the definition of MCI has varied considerably over time Capacity Estimates and across studies. Patient Demand In the clinical pathway, we make assumptions about the availability and specifications of the ther- The expected patient demand in the screening and apy, which are dependent on clinical trial results. We diagnostic phases of the clinical pathway in 2022 also make assumptions around provider and patient (the year prior to treatment availability) is shown in Figure 2. In 2022, we estimate that th ere would be

7 FIGURE 2 workforce data from the Medical Board of Australia Expected Patient Demand in Screening (2018). For the projected workforce, we use projections and Diagnostic Phases in 2022 (millions) from the Health Workforce 2025 Medical Specialties Volume 3 report (AFHW, 2012) for neurologists and Age 50+ 8.91 geriatricians through 2025 and projections from the Australia’s Future Health Workforce—Psychiatry Cognitive 7.13 screening report (Department of Health, 2016) for psychiatrists through 2030. Beyond the available published projec- Screening 0.91 Screening Phase positive for MCI tions (after 2025 for neurologists and geriatricians and after 2030 for psychiatrists), we apply the population Evaluation 0.46 by specialist growth rates for the population aged 50 and older, in effect assuming that the number of specialists per Biomarker 0.41 capita remains steady in future years (Beckett and testing Morrison, 2007; Tsai, Eliasziw and Chen, 2012) to Biomarker 0.18 meet the population’s demand (Hara et al., 2018); positive this assumption appears reasonable as the histori- Diagnostic Phase Treatment 0.15 cal and published workforce projection trends are indicated similar to population growth trends (Medical Board of Australia, 2018; AFHW, 2012; 2016; ABS, 2018e). 909,000 Australians who screen positive for MCI, Table 2 shows the projected number of specialists, 409,000 patients who would seek and be eligible for which we estimate will increase from 5,695 in 2022 to biomarker testing and 147,000 patients who would be 7,147 in 2035, or by about 25 percent. determined eligible for treatment. Although further evaluation would typically be conducted by these three types of specialists, not all Specialist Workforce geriatricians, neurologists and psychiatrists would Geriatricians, neurologists and neuropsychiatrists conduct cognitive evaluations of people with MCI. are the specialists who typically diagnose cognitive Geriatricians typically see older patients, who have impairment and dementia in Australia (Dementia higher prevalence of early-stage AD and also later- Australia, 2006; 2012; 2018). We start with historical stage AD and possibly manifest dementia. Neurologists

TABLE 2 Projected Workforce of Specialists That May Diagnose Early-Stage Alzheimer’s Disease

Year Neurologists Geriatricians Psychiatrists

2022 694 882 4,119

2025 734 955 4,236

2030 787 1,024 4,809

2035 850 1,106 5,191 SOURCES: Historical data are from the Medical Board of Australia (2018). Published workforce projections for geriatricians and neurologists (AFHW, 2012) and psychiatrists (AFHW, 2016) were applied through 2025 and 2030, respectively. After 2025 (geriatricians and neurologists) and 2030 (psychiatrists), projections are based on the population growth rate from the ABS (2018e) Series B projections. NOTE: The full psychiatrist workforce is displayed in this table. For this analysis, we assume 15 percent of all psychiatrists in Australia would be involved in diagnosing Alzheimer’s pathology in people with MCI.

8 are more likely to see younger patients; however, not all neurologists would evaluate pe ople with cognitive disorders. Psychiatrists provide a broad range of services There is not a standard related to mental health conditions as well as psycho- therapy and neuropsychiatrists are a relatively small way of diagnosing MCI subspecialty in Australia. Based on a survey of psychiatrists reporting their clinical practice area (RANZCP, due to AD in clinical 2014), we assume that 15 percent of psychiatrists in Australia would provide visits to evaluate MCI patients practice; rather, a and diagnose Alzheimer’s pathology. For this analy diagnosis is made sis, we make assumptions about the specialists’ excess capacity and willingness to provide more visits for the only after careful MCI population to model specialist workforce capacity. Overall, we assume that these specialists could and thorough clinical increase their capacity for visits by 5 percent in order to meet the demand of new MCI patients seeking further consultation. evaluation. This simplifying assumption could be conceptualised as all specialists in this group providing Although current clinical trial protocols typically 5 percent more visits, half of the specialists conducting require an amyloid PET scan to identify amyloid 10 percent more visits, or other proportions of involved deposits in trial participants, reimbursement and specialists. access to technology will dictate the use of diagnostic modalities in routine clinical practice. Increasing Diagnostic Technology PET scanner capacity would require substantial capital investment in equipment, staff and building There is not a standard way of diagnosing MCI upgrades. There has been a marked increase in the due to AD in clinical practice; rather, a diagnosis number of PET scanners over the last several years, is made only after careful and thorough clinical and there are currently 82 PET scanners in Australia, consultation. Tests can include additional cogni- including a new scanner in the Northern Territory10 tive tests or biomarker tests, including tests to rule (AFHW, 2018; Fyles, 2018; Organisation for out other medical conditions that might present Economic Co-operation and Development [OECD], similarly to dementia. MCI due to AD is gener- 2018c). Access to the radioactive tracer is another ally confirmed with biomarkers (beta-amyloid constraint; cyclotrons produce the radioactive tracers and/or tau) (Portet et al., 2006; Palmqvist et al., used in PET imaging and are generally located near 2015). In research settings, providers typically use PET facilities due to the short half-life of the tracers. two different methods to confirm the presence of Aside from a new PET scanner and cyclotron in the the Alzheimer’s pathology: neuroimaging using Northern Territory, most scanners and cyclotrons are PET with fluorodeoxyglucose (FDG) tracers; and based in the eastern and southern coasts of Australia, CSF measurement of beta-amyloid and tau levels leaving individuals in the northern and western (Dementia Australia, 2012). Neither modality is coasts without ready access to this technology. approved in Australia for clinical diagnosis, and Measurement of liquid biomarkers in CSF is the Medical Services Advisory Committee (MSAC, less costly than PET scans, and lumbar punctures 2015b; 2016) recently did not endorse a request for to retrieve CSF could be performed in many facili- Medicare Benefits Schedule reimbursement for the ties. However, only one laboratory, the National use of FDG PET imaging for diagnosis of AD. Association of Testing Authorities / International It is unclear wh ether PET imaging, CSF assays Laboratory Accreditation Cooperation, is cur- or both types of biomarker testing would be used in rently accredited to conduct CSF diagnostic test- Australia if an Alzheimer’s DMT becomes available. ing in Australia (National Dementia Diagnostics

9 Laboratory, 2019). Although standardisation of some may need retraining and support to become thresholds and laboratory assays has been a chal- more comfortable with the procedure. Currently, lenge in the use of CSF biomarkers, there has been lumbar punctures are typically performed in hospital significant progress in validating CSF biomarkers settings by physicians. However, lumbar punctures as an alternative to PET (Bjerke and Engelborghs, could be done with only a procedure room and a 2018; Hansson et al., 2018). While using CSF could postprocedure recovery room if needed. result in lower costs relative to PET, quality con- In contrast, we assume that access to PET scans trol of laboratory measurements would need to be would be limited by the capacity of scanners and established. cyclotrons required for amyloid PET scanning. Given that current PET and laboratory capacity Cyclotrons produce the radioactive tracers that bind to for biomarker testing to diagnose MCI due to AD is beta-amyloid and are typically located within approxi- limited to research and specialised clinical settings mately 320 kilometres of PET sites or within four in the absence of a DMT, investment in one or both hours transit time (Giamis, 2012) because the half-life methods would be required in order to expand diag- of radiopharmaceutical containing the fluorine-18 (18F) nostic capacity for early-stage Alzheimer’s patients. isotope is relatively short.12 There are two registered Given the uncertainty over reimbursement levels and entries for FDG tracers in Australia: 2-deoxy-2-(18F) which method would be preferred in Australia in the fluoro-D-glucose and FDGen (fludeoxyglucose [18F] future, we analyse scenarios with the following two injection), though only the former lists neurological sets of assumptions: disorders as an indication (MSAC, 2015a). Th ere are • Base-case biomarker testing assumption: at least seven sites, including two commercial entities 80 percent of biomarker testing performed that produce radiolabelled-FDG (MSAC, 2015a). using PET and 20 percent using CSF. This Figure 3 shows the locations of current PET assumes that PET would be the primary scanners in Australia. Cyclotron coverage area is method used.11 We assume that capacity greatest across the eastern and southern coastal areas growth in PET scanners and cyclotrons, where most of the population resides, and tracers including capital and personnel training, are generally flown from cyclotrons to sites with would be similar to recent growth rates in PET scanners but no cyclotron. However, th ere are Australia (see Figure A-2 in the Appendix). still relatively few PET scanners to accommodate the • Alternative biomarker testing assumption: expected number of people with MCI who could seek 50 percent of biomarker testing performed PET imaging, and only one scanner in the Northern using PET and 50 pe rcent using CSF. This Territory. assumes that there would be investment in CSF testing, which could include establishing Infusion Delivery protocols for lumbar punctures and laboratory measurements, in addition to continued We assume that a DMT would be delivered intrave- growth in PET capacity similar to recent nously through a course of infusions, typically taking growth rates. Although CSF would be less place in hospital outpatient clinics. Many infusion costly than PET, we assume that th ere would therapies currently in development would be deliv- still be investment in PET, particularly in ered every few weeks for a period of 12 to 24 months. urban areas and for people who are unable to While other modalities and treatment durations may undergo a lumbar puncture due to anatomical eventually be adopted in clinical practice, we model reasons or contraindications, or who have an a hypothetical therapy that would be administered aversion to the procedure. approximately every four weeks over the course of one year for a total of 14 infusions per patient. In an In this analysis, we do not model capacity alternative scenario, we also assess when infusion constraints for CSF testing. In theory, physicians are delivery is not a barrier, modelling expanded capacity capable of performing lumbar punctures, although

10 FIGURE 3 PET Scanners in Australia

PET Units 1 Population per 2 Sq Km 3 0–0 0–2 2–9 9–302 302–7,792

SOURCES: ABS, 2018a; 2019a; Department of Health, 2018; OECD, 2018c. NOTE: This map shows 82 PET scanners at 60 sites. There are also 19 cyclotrons, including 5 units in Sydney (a cyclotron in Darwin is scheduled to be installed in 2019 [Fyles, 2018]) (IAEA, 2019). Population density is shown by local government areas.

or a therapy that can be delivered by another route of infusion clinics could be dedicated to administering administration (i.e., subcutaneous or oral). the Alzheimer’s DMT. In this analysis, we use an index approach consistent with our prior European and Canadian analyses to estimate infusion capacity.13 Our esti- Simulation Results Under mates of the capacity to deliver infusions are based Selected Capacity Scenarios on the relative capacity of the Australian health care Base- Case Scenario system constructed from four indicators: hospital beds, active nurses, MRI scanners and PET scan- The base-case scenario reflects historical capac- ners (see Table A-2 in the Appendix). We use OECD ity trends projected forward and our assumptions (2018c) data to develop the Australian capacity index related to specialist availability, biomarker testing relative to the United States. Then, we use the index performed and infusion capacity. to scale the historical and projected per capita Figure 4 shows the wait lists under these assump- infusion capacity in the United States to estimate tions. Initially, the main constraint is the availability infusion capacity in Australia (see Figure A-3 in of specialists to evaluate patients who screen positive the Appendix). We assume the same relative rate of for MCI. We estimate that the specialists’ capacity for growth in infusion capacity as the United States. As visits would be 314,000 visits in 2022, or 65 percent in our prior analyses, we assume that excess capac- of the projected 486,000 pe ople with MCI who may ity for new patients in existing infusion clinics is seek evaluation by a specialist and who would need 10 percent, while 80 percent of the capacity in new two visits in our model. The backlog of patients due to

11 FIGURE 4 FIGURE 5 Projected Wait Lists for Alzheimer’s Projected Wait Times for Alzheimer’s Disease Diagnosis, Testing and Disease Diagnosis, Testing and Treatment Treatment

Dementia Biomarker Infusion Dementia Biomarker Infusion specialist visits testing treatment specialist visits testing treatment

0.25 12

10 0.20

8 0.15 6 0.10 4

0.05 or treatment, months 2 Number of patients waiting Average time delay in diagnosis for diagnosis or treatment, millions 0 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035

NOTE: Our base-case scenario assumes 80 percent of biomarker testing NOTE: Our base-case scenario assumes 80 percent of biomarker conducted via PET and 20 percent via CSF. testing conducted via PET and 20 percent via CSF.

limited specialist capacity would continue until 2025, barriers to diagnosis and treatment of people with at which point the backlog of patients would have been MCI due to AD. The key assumptions for each of the seen and only incident cases would require evaluation three alternate scenarios are shown in Table 3. going forward. As the backlog of patients moves past In alternative scenario 1, we assume that use of the specialist wait lists, the waits for biomarker testing CSF biomarker testing would be expanded such that and then infusion treatments increase. The wait lists 50 percent of biomarker testing is conducted using for biomarker testing are relatively short and are elimi- PET and 50 percent using CSF. Capacity for CSF test- nated by 2026. The wait lists for infusion treatment ing could rapidly be expanded if planning, policies, extend longer and would continue un til 2034. regulations and reimbursements encourage the estab- Figure 5 illustrates average waiting times while lishment of protocols training for lumbar punctures patients are on the wait lists for diagnosis and treat- (e.g., in outpatient settings) as needed and development. With PET as the predominant biomarker test ing standard laboratory procedures for the assays. In in the base case, we estimate that biomarker testing addition, the development of new technologies such would initially result in average wait times of about as blood-based assays (Doecke et al., 2012; Burnham two to three months in each year. The average wait- et al., 2016; Gupta et al., 2017; Hampel et al., 2018) ing times for infusions are also relatively short, with have made progress in recent years and would help a maximum of two months, but the waits would be facilitate expansion of biomarker testing capacity. If sustained until 2034. CSF can be used for 50 percent of the biomarker testing, we estimate that the waits for biomarker testing Alternative Scenarios would be eliminated (Figure 6). Thus, the extent to which PET and CSF biomarker testing are used in We now assess alternative scenarios that reflect Australia would affect patients’ access to the therapy. concerted efforts to expand capacity and eliminate However, even in this scenario with investment in

12 TABLE 3 Capacity Assumptions Across Scenarios

Scenario Specialists Biomarker Testing Infusions

Base case Neurologists, geriatricians, 80% PET with historical Level estimated using a general and 15% of psychiatrists with capacity projected forward, health care capacity index, with 5% excess capacity for visits 20% CSF with no capacity current capacity projected constraint forward

Alternative 1: Same as base case 50% PET with historical Same as base case Investment in CSF capacity projected forward, 50% CSF with no capacity constraint

Alternative 2: Same as base case 50% PET with historical No capacity constraint Investment in CSF and infusion capacity projected forward, delivery (or nonintravenous 50% CSF with no capacity administration) constraint

Alternative 3: No capacity constraint No capacity constraint No capacity constraint No capacity constraints

FIGURE 6 Projected Wait Lists and Times for Alzheimer’s Disease Diagnosis, Testing and Treatment Under Alternative Scenario 1

Dementia specialist visits Biomarker testing Infusion treatment

0.25 12

10 0.20

8 0.15 6 0.10 4

0.05 Average time delay in 2 Number of patients waiting for diagnosis or treatment, millions diagnosis or treatment, months 0 0 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035

CSF, delays in access would still extend through 2033 weeks for a total of 14 doses, 2.1 million infusions due to waits for specialists and infusions. would need to be administered to treat the approxi- In alternative scenario 2, the elimination of the mately 147,000 Australians with MCI due to AD infusion delivery constraint could reflect adequate estimated to be eligible for treatment as of 2023. capacity growth of infusion services or a nonin- Eliminating the infusion constraint may be accom- travenous therapy modality (i.e., subcutaneous or plished if building additional infusion centre capac- oral administration). Assuming infusions would ity becomes a priority and/or if home infusions can be delivered approximately ev ery 4 weeks over 52 be utilised widely.

13 TABLE 4 Summary of Projected Wait Times for Alzheimer’s Disease Diagnosis, Testing and Treatment, by Scenario

Average Wait Time in Months

Scenario 2022 2025 2030 2035

Base case 10 4 1 0

Alternative 1: 9 3 1 0 Investment in CSF

Alternative 2: 9 0 0 0 Investment in CSF and infusion delivery (or nonintravenous administration)

Alternative 3: 0 0 0 0 No capacity constraints

Table 4 shows a comparison of the projected wait Alzheimer’s Dementia Cases Avoided times in each scenario. In alternative scenario 3, we in the Base and Alternative Scenarios present the case in which all three capacity constraints Figure 7 shows the cumulative incident Alzheimer’s and the associated wait times are eliminated. Key dementia cases between 2023 and 2033 in the base- assumptions leading to this scenario include adequate case and alternative scenarios. In the base-case training and task shifting such that health care pro- scenario with historical capacity trends projected for- viders could evaluate all patients seeking diagnosis, ward, we estimate that 344,000 people with MCI due as well as adequate expansion of PET and/or CSF to AD would not develop to Alzheimer’s dementia due capacity and infusion services. Taking into account to treatment between 2023 and 2033. These results are the extensive time and resources required to train a contingent on the assumed treatment effectiveness significant number of specialists as well as challenges of a 50 percent relative risk reduction in the progres- in shifting tasks and preparing a broad workforce to sion of MCI due to AD to Alzheimer’s dementia evaluate MCI patients with diagnostic criteria cur- after treatment. Progression to Alzheimer’s dementia rently used only in research and specialised clini- occurs at a higher rate for pe ople who are on wait lists cal settings, this outcome is unlikely. However, this during this period. scenario demonstrates an upper bound of Alzheimer’s The alternative scenarios depict expanded capac- dementia cases that could be avoided if all capacity ity that reduce wait lists. Relative to the base-case constraints we re overcome. Whether these constraints scenario, alternative scenario 1 (investment in CSF) could be fully addressed is subject to fu ture policies, would lead to 5,000 additional avoided dementia cases technological advances and drug development. and alternative scenario 2 (investment in CSF and In addition to the alternative scenarios, we infusion services) would lead to an additional 19,000 conducted sensitivity analysis around our capacity avoided cases. In alternative scenario 3, in which all projections. In the Appendix, we present low and capacity constraints are alleviated, we estimate an high projected capacity scenarios for specialist visits, additional 54,000 cases of dementia would be avoided PET scanners and infusions (Figures A.1–A.6). Under relative to the base-case scenario. In total, we estimate the high projected capacity scenario, wait lists would that 398,000 cases of Alzheimer’s dementia cases be eliminated by 2028. However, under the low pro- could be avoided if a treatment was available and all jected capacity scenario, the wait times would extend constraints were addressed between 2023 and 2033. beyond 2050 due to the limited dementia specialist See Figure A.9 for the cumulative Alzheimer’s visits to accommodate patient demand. dementia cases avoided under the sensitivity

14 FIGURE 7 Projected Cumulative Number of New Alzheimer’s Dementia Cases Avoided Although the fu ture Between 2023 and 2033 Under Different Scenarios availability of an 450,000 Alzheimer’s DMT is +54,000 400,000 +19,000 344,000 +5,000 350,000 uncertain, particularly 300,000 in light of recent 250,000 discontinued trials, 200,000

150,000 there are currently still 100,000 several therapies und er 50,000

0 investigation in clinical Base-case Alternative Alternative Alternative scenario scenario 1 scenario 2 scenario 3 trials. NOTE: The cumulative new Alzheimer’s dementia cases avoided in the base-case scenario is relative to no treatment being available. For the alternative scenarios, the labels indicate the additional number of cumulative new Alzheimer’s dementia cases avoided relative to the base scenario (dashed line). DMT becomes available in 2023. People with MCI due to AD who are not diagnosed and treated in a timely fashion would be at higher risk of progress- scenarios with either low or high projected capacity ing to Alzheimer’s dementia when their dementia compared with the base-case scenario. could potentially be avoided. If a treatment becomes available in 2023 and assuming the capacity of the Discussion health system follows historical trends, Australians could wait over ten months to undergo diagnosis and A DMT would be a breakthrough innovation biomarker testing in the initial year, and wait lists to reduce the progression of early-stage AD to could extend through 2033. If a hypothetical therapy Alzheimer’s dementia. Alzheimer’s dementia exacts reduces relative risk of progression by 50 percent and a substantial burden on patient and caregiver health, all of the capacity constraints are overcome, 398,000 as well as significant costs to health systems and indi- Alzheimer’s dementia cases could be avoided between viduals as the estimated total direct costs of dementia 2023 and 2033. were $8.8 billion AUD in 2016 (Brown, Hansnata, As in other countries, there are several differ and La, 2017). Although the future availability of an ent policy and planning solutions that may help Alzheimer’s DMT is uncertain, particularly in light address these capacity constraints (Liu et al., 2017; of recent discontinued trials, there are currently still Hlávka, Mattke and Liu, 2018; Liu et al., 2019). several therapies under investigation in clinical tr ials. The initial bottleneck in the clinical pathway If a therapy becomes available, the large number of would be specialist visits to evaluate and diagnose people who could benefit from diagnosis and treat- early stage Alzheimer’s dementia. In Australia, access ment would present health system challenges that to specialists and some technologies would likely be would require timely attention to address. more problematic in rural and remote areas. One The objective of this analysis is to examine consideration is that a greater share of the popula- the magnitude of potential capacity challenges if a tion aged 50 or older lives outside of major cities

15 compared to th ose under age 50 (ABS, 2018d). An The availability of inexpensive biomarker testing option to increase access to cognitive screening could reduce the need for PET and CSF biomarkers services could be to expand upon the delivery of to confirm diagnosis of MCI due to AD as well as care provided by services such as the Royal Flying the number of cases needing referral to specialists. Doctor Service (2019) or the Heart of Australia Our alternative scenarios illustrate the possibility of (2019). In addition, telemedicine consultations could reducing waiting times, which could be due to capac- be used to bridge the access gap for those living in ity expansion due to infrastructure building or to remote communities. As well as geographic chal- advances in technology. Alternative scenario 1 results lenges, there could be outreach or awareness issues in no constraints on biomarker testing that could be in reaching indigenous populations such as the due to investment in CSF biomarkers or the emer- Aboriginal and Torres Strait Islanders, who are gence of other diagnostic technology. Alternative disproportionally represented in rural, remote and scenario 3 reflects a case in which there are no capac- very remote areas (ABS, 2018f; AIHW, 2018). The ity constraints. Blood biomarker tests that could be Kimberley Indigenous Cognitive Assessment (KICA- used for routine screening and diagnosis could help Cog), KICA-Screen and modified KICA (mKICA) achieve alternative scenario 3 by alleviating both the (LoGiudice et al., 2006; Dementia Australia, 2019) burden on specialists (e.g., by reducing false positives are screening tools validated for use in older indig- through increased or routine screening or bio- enous populations in Australia; use of such tools, in marker tests ordered by nonspecialists) and on PET conjunction with culturally relevant care and other equipment. methods, could better address dementia needs in this Based on our analyses, biomarker testing is less population. of a constraint in Australia than in other countries; Better screening and diagnostic tests could this may be due to the age of the population and a also assist in identification of early-stage AD. In relatively well-established PET scanner infrastruc- the Australian Imaging, Biomarkers and Lifestyle ture. For example, in 2017, 33 percent of Australians (2006) study of ageing, researchers are evaluating were aged 50 or older, compared with 37 pe rcent of individuals aged 60 or older who have MCI or AD, Canadians (ABS, 2019a; Statistics Canada, 2018). In as well as healthy controls, over ten years, study- 2015, Australia had 2.64 PET scanners per million ing biomarkers, cognitive parameters and lifestyle persons, while the United States had 5.12, France factors as both indicators and predictors of future had 1.95 and Canada had 1.31 scanners per million outcomes to better understand diagnosis and pro- persons (OECD, 2018c). While there are still some gression of the disease (see also Ellis et al., 2009). In expected waits, act ual capacity and use will depend addition to PET and CSF biomarkers, researchers on factors such as Medicare and other payer reim- are working to develop blood and retinal tests that bursement policies. Furthermore, access to PET would make screening and diagnosis more readily scanners varies across Australia, with most of the accessible and less expensive than current screen- scanners available in major cities along the southern ing and diagnostic tests. For example, researchers in and eastern coasts. Australia and Japan are developing a new blood test Last, infusion capacity will also depend on for AD (Nakamura et al., 2018; National Dementia how quickly capacity could be expanded as well as Diagnostics Laboratory, 2019). The availability of Medicare reimbursement levels. Such expansion accurate blood or retinal tests in the future has the could follow the trends observed in infusion expan- potential to overcome some of the capacity con- sion for oncology and multiple sclerosis therapies, straints around biomarker testing, including cost, including expanded centres or home-based treat- infrastructure, equipment and training. Such a test ments (chemo@home, n.d.; South Australia Health, could also allow physicians to screen populations 2014). more broadly or for routine screening, rather than waiting for the presentation of cognitive impairment symptoms.

16 Conclusions would likely be ordered and interpreted by a specialist. Second, there is precedent based on symptom-management treatments A DMT for AD would be a breakthrough to reduce for Alzheimer’s dementia that require a diagnosis confirmed by a specialist in order to receive Medicare reimbursement (Dementia the burden on pe ople with Alzheimer’s, their care- Australia, 2019). givers and health care systems. Although th ere is 5 Our U.S. and European analyses, published in 2017 and 2018, uncertainty in the future of therapies that are in assumed that a therapy would become available in 2020. Our development, it may be useful for policymakers, pay- Canadian analysis, published in 2019, assumed a therapy would ers and providers to consider potential challenges in become available in 2021. In March 2019, the leading phase 3 trial for an Alzheimer’s DMT scheduled for trial completion delivering such a therapy to the large population of in 2020 was discontinued (Biogen, 2019). As of April 2019, the people with early-stage AD. leading phase 3 trial has a primary completion date in 2022 (U.S. We have estimated the magnitude of potential National Library of Medicine, 2019a). Thus, for this current analysis, we selected 2023 as the first year that a DMT would be capacity challenges under a number of assumptions if available. a hypothetical Alzheimer’s DMT became available in 6 Our U.S. and European analyses assumed that the age eligibil- 2023. The goal of our analysis is not to provide pre- ity would be 55 and older. For the Canadian analysis and this cise projections, but to facilitate discussions among analysis, we assume eligibility would be ages 50 and older as the stakeholders that could help address infrastructure later stage clinical trials tend to target ea rlier age groups and some trials targeting later age groups have been terminated (e.g., challenges. As we have found in analyses of other U.S. National Library of Medicine, 2019c). The inclusion of ages countries, delays in access to a therapy in Australia 50–54 in our analysis has relatively little impact on the results could result in people progressing from early stages because MCI prevalence is lower among younger ages. of cognitive impairment to full-blown dementia due 7 There is considerable variation in how MCI has been defined to capacity limitations. Our analysis suggests that not over time and across studies. The meta-analysis conducted by Petersen et al. (2018) includes some studies using narrower defi- expanding capacity beyond historical trends could nitions to th ose using broader definitions as well as studies from result in approximately 54,000 Australians progress- different countries. Australian studies have reported similar MCI ing from MCI due to AD to Alzheimer’s dementia prevalence (Anstey et al., 2013; Brodaty et al., 2017); however, there is variation. By relying on the prevalence estimates by age while waiting for diagnosis and treatment between from the meta-analysis, the implicit assumption is that variation 2023 and 2033. As the development of Alzheimer’s in the underlying prevalence of MCI is greater across studies therapies continues, our hope is that discussions on (and definitions) than across countries. ways to expand capacity and increase awareness also 8 This assumption differs from our prior analysis in the United States, where a PET scan is the only currently FDA-approved continue. modality for clinical use. In our analyses of European countries, we assumed that 90 percent of biomarker testing would be performed by CSF biomarker testing and only 10 percent would Notes be PET imaging for patients with contraindications to lumbar puncture. The 80 percent PET and 20 percent CSF screening 1 Clinical trials of AD treatments include DMTs and symptom- assumption is consistent with our analysis in Canada. reducing agents. In 2018, 63 percent of clinical trials were 9 In alternative scenarios, we assume that 50 percent PET and investigating DMTs (either small molecule or immunotherapy) 50 pe rcent CSF would be used. For more details about this alter- (Cummings et al., 2018). native assumption, see the next section. 2 Other cognitive assessments may be used (Cognitive Decline 10 Currently, these scanners are primarily used to diagnose and Partnership Centre, 2015), such as the Standardised MMSE evaluate oncology indications. (Molloy, 1991) and the Montreal Cognitive Assessment (MoCA) (Ciesielska et al., 2016). 11 Approximately 71 percent of the Australian population resides in major cities (ABS, 2016; 2018b), defined as populations of 3 Cognitive and functional assessments may involve the Alzheim- 100,000 or more, where PET scanners are more likely to be avail- er’s Disease Assessment Scale Cognitive subscale (ADAS-Cog) able. (Dementia Australia, 2006; Skinner et al., 2012; Cognitive Decline Partnership Centre, 2015), the Neuropsychiatry Unit 12 Fluorodeoxyglucose (18F) has a half-life of 110 minutes (Interna- Cognitive Screening Tool (NUCOG) (Walterfang, Siu and tional Atomic Energy Agency [IAEA], 2012). Velakoulis, 2006; Walterfang and Velakoulis, 2013) or the Clini- 13 cians Interview Based Impression of Severity (CIBIS) (Knopman To assess the validity of our index approach, we examined data et al., 1994). from the government’s Pharmaceutical Benefits Scheme and found similar trends. 4 We assume that a diagnosis of MCI due to AD would need to be confirmed by a specialist for two reasons. First, biomarker tests

17 References Australian Institute of Health and Welfare, “Australia’s Health 2018,” 2018. As of April 10, 2019: ABS—See Australian Bureau of Statistics. https://www.aihw.gov.au/reports/australias-health/australias- health-2018/contents/table-of-contents AFHW—See Australia’s Future Health Workforce. Australia’s Future Health Workforce, Health Workforce 2025: AIHW—See Australian Institute of Health and Welfare. Medical Specialties Volume 3, November 2012. As of April 10, Aisen, P. S., “Continuing Progress in Alzheimer’s Disease Trials: 2019: Cause for Optimism,” Journal of Prevention of Alzheimer’s http://pandora.nla.gov.au/pan/133228/20150419-0017/www.hwa. Disease, Vol. 4, No. 4, 2017, pp. 211–212. gov.au/sites/uploads/HW2025_V3_FinalReport20121109.pdf Alzheimer’s Disease International, World Alzheimer Report, ———, “Australia’s Future Health Workforce (AFHW) Reports,” 2009. As of June 12, 2019: 2016. As of April 10, 2019: https://www.alz.co.uk/research/files/WorldAlzheimerReport.pdf http://www.health.gov.au/internet/main/publishing.nsf/Content/ australias-future-health-workforce-reports Anstey, K. J., N. Cherbuin, R. Eramudugolla, K. Sargent-Cox, S. Easteal, R. Kumar, and P. Sachdev, “Characterizing Mild Beckett, Megan K., and Peter A. Morrison, Forecasting the Supply Cognitive Disorders in the Young-Old over 8 Years: Prevalence, of and Demand for Physicians in the Inland Southern California Estimated Incidence, Stability of Diagnosis, and Impact on Area, Santa Monica, Calif.: RAND Corporation, TR524, 2007. As IADLs,” Alzheimer’s & Dementia, Vol. 9, No. 6, November 2013, of June 12, 2019: pp. 640–648. As of June 12, 2019: https://www.rand.org/pubs/technical_reports/TR524.html https://www.ncbi.nlm.nih.gov/pubmed/23474041 Biogen, “Biogen and Eisai to Discontinue Phase 3 Engage and Australian Bureau of Statistics, “1270.0.55.003—Australian Emerge Trials of Aducanumab in Alzheimer’s Disease,” 2019. As Statistical Geography Standard (ASGS): Volume 3—Non ABS of April 10, 2019: Structures, July 2016,” 2018a. As of April 10, 2019: http://investors.biogen.com/news-releases/news-release-details/ http://www.abs.gov.au/AUSSTATS/[email protected]/ biogen-and-eisai-discontinue-phase-3-engage-and-emerge-trials DetailsPage/1270.0.55.003July%202016?OpenDocument Bjerke, M., and S. Engelborghs, “Cerebrospinal Fluid Biomarkers ———, “3235.0—Regional Population by Age and Sex, Australia, for Early and Differential Alzheimer’s Disease Diagnosis,” 2017,” 2018b. As of April 22, 2019: Journal of Alzheimer’s Disease, Vol. 62, No. 3, 2018, pp. 1199– http://www.abs.gov.au/ausstats/[email protected]/mf/3235.0 1209. As of June 12, 2019: http://www.ncbi.nlm.nih.gov/pubmed/29562530 ———, “Australia’s Leading Causes of Death, 2017,” 2018c. As of April 10, 2019: Brodaty, H., L. Aerts, J. D. Crawford, M. Heffernan, N. A. Kochan, https://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20 S. Reppermund, K. Kang, K. Maston, B. Draper, J. N. Trollor, Subject/3303.0~2017~Main%20Features~Australia’s%20 and P. S. Sachdev, “Operationalizing the Diagnostic Criteria for leading%20causes%20of%20death,%202017~2 Mild Cognitive Impairment: The Salience of Objective Measure in Predicting Incident Dementia,” American Journal of Geriatric ———, “Census of Population and Housing: Reflecting Psychiatry, Vol. 25, No. 5, May, 2017, pp. 485–497. Australia—Stories from the Census, 2016,” 2018d. As of April 10, 2019: Brodaty, H., D. Pond, N. M. Kemp, G. Luscombe, L. Harding, K. https://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20 Berman, and F. A. Huppert, “The GPCOG: A New Screening Test for Subject/2071.0~2016~Main%20Features~Small%20Towns~113 Dementia Designed for General Practice,” Journal of the American Geriatrics Society, Vol. 50, No. 3, March, 2002, pp. 530–534. ———, “Population Projections, Australia, 2017 (base)—2066: ‘Table B9. Population Projections, by Age and Sex, Australia— Brown, Laurie, Erick Hansnata, and Hai Anh La, Economic Series B,’ ” 2018e. As of January 15, 2019: Cost of Dementia in Australia 2016–2056, Canberra, Australia: https://www.abs.gov.au/AUSSTATS/[email protected]/ NATSEM at the Institute for Governance and Policy Analysis, DetailsPage/3222.02017%20(base)%20-%202066?OpenDocument University of Canberra, February, 2017. ———, “Summary Commentary—Estimates of Aboriginal and Burnham, S. C., C. C. Rowe, D. Baker, A. I. Bush, J. D. Doecke, Torres Strait Islander Australians,” 2018f. As of April 22, 2019: N. G. Faux, S. M. Laws, R. N. Martins, P. Maruff, S. L. Macaulay, https://www.abs.gov.au/ausstats/[email protected]/mf/3238.0.55.001 S. Rainey-Smith, G. Savage, D. Ames, C. L. Masters, W. Wilson, and V. L. Villemagne, “Predicting Alzheimer Disease from ———, “Census 2016, T01 Age by Sex (LGA),” 2019a. As of a Blood-Based Biomarker Profile: A 54-Month Follow-Up,” April 22, 2019: Neurology, Vol. 87, No. 11, September 13, 2016, pp. 1093–1101. http://stat.data.abs.gov.au/Index.aspx Centers for Disease Control and Prevention, “Ambulatory ———, “Deaths, Year of Registration, Summary Data, Sex, Health Care Data: Questionnaires, Datasets, and Related States, Territories and Australia,” 2019b. As of April 10, 2019: Documentation,” August 31, 2017. As of October 4, 2017: http://stat.data.abs.gov.au/Index.aspx?DatasetCode=DEATHS_ https://www.cdc.gov/nchs/ahcd/ahcd_questionnaires.htm SUMMARY chemo@home, homepage, n.d. As of April 10, 2019: Australian Government Department of the Prime Minister and https://chemoathome.com.au Cabinet, Closing the Gap Report, 2019. As of June 12, 2019: https://ctgreport.pmc.gov.au/sites/default/files/ctg-report-2019. Ciesielska, N., R. Sokolowski, E. Mazur, M. Podhorecka, A. pdf?a=1 Polak-Szabela, and K. Kedziora-Kornatowska, “Is the Montreal Cognitive Assessment (MoCA) Test Better Suited Than the Mini- Australian Imaging, Biomarker and Lifestyle Study of Ageing, Mental State Examination (MMSE) in Mild Cognitive Impairment homepage, 2006. As of April 10, 2019: https://aibl.csiro.au (MCI) Detection Among People Aged over 60? Meta-Analysis,” Psychiatria Polska, Vol. 50, No. 5, October 31, 2016, pp. 1039–1052.

18 Cognitive Decline Partnership Centre, Clinical Practice Department of Health, Australia’s Future Health Workforce— Guidelines for Dementia in Australia—Public Consultation Draft, Psychiatry, 2016. As of April 10, 2019: 2015. As of June 12, 2019: http://www.health.gov.au/internet/main/publishing.nsf/Conte https://sydney.edu.au/medicine/cdpc/documents/about/ nt/597F2D320AF16FDBCA257F7C0080667F/$File/AFHW%20 Dementia-Guideline-Draft-for-Public-Consultation.pdf Psychiatry%20Report.pdf Cummings, J., G. Lee, T. Mortsdorf, A. Ritter, and K. Zhong, ———, “Positron Emission Tomography (PET),” 2018. As of “Alzheimer’s Disease Drug Development Pipeline: 2017,” April 10, 2019: Alzheimer’s & Dementia (N.Y.), Vol. 3, No. 3, September 2017, http://www.health.gov.au/internet/main/publishing.nsf/Content/ pp. 367–384. As of June 12, 2019: pet-nuclear-medicine-imaging http://www.ncbi.nlm.nih.gov/pubmed/29067343 Doecke, J. D., S. M. Laws, N. G. Faux, W. Wilson, S. C. Burnham, Cummings, J., G. Lee, A. Ritter, and K. Zhong, “Alzheimer’s C. P. Lam, A. Mondal, J. Bedo, A. I. Bush, B. Brown, K. De Ruyck, Disease Drug Development Pipeline: 2018,” Alzheimer’s & K. A. Ellis, C. Fowler, V. B. Gupta, R. Head, S. L. Macaulay, Dementia (N.Y.), Vol. 4, 2018, pp. 195–214. As of June 12, 2019: K. Pertile, C. C. Rowe, A. Rembach, M. Rodrigues, R. Rumble, http://www.ncbi.nlm.nih.gov/pubmed/29955663 C. Szoeke, K. Taddei, T. Taddei, B. Trounson, D. Ames, C. L. Masters, R. N. Martins, Alzheimer’s Disease Neuroimaging Cummings, J., T. Morstorf, and G. Lee, “Alzheimer’s Drug- Initiative, Biomarker Australian Imaging, and Group Lifestyle Development Pipeline: 2016,” Alzheimer’s & Dementia (N.Y.), Research, “Blood-Based Protein Biomarkers for Diagnosis of Vol. 2, No. 4, November 2016, pp. 222–232. As of June 12, 2019: Alzheimer Disease,” Archives of Neurology, Vol. 69, No. 10, http://www.ncbi.nlm.nih.gov/pubmed/29067309 October 2012, pp. 1318–1325. As of June 12, 2019: Cummings, J. L., T. Morstorf, and K. Zhong, “Alzheimer’s http://www.ncbi.nlm.nih.gov/pubmed/22801742 Disease Drug-Development Pipeline: Few Candidates, Frequent Doraiswamy, P. M., R. A. Sperling, K. Johnson, E. M. Reiman, Failures,” Alzheimer’s Research & Therapy, Vol. 6, No. 4, 2014, T. Z. Wong, M. N. Sabbagh, C. H. Sadowsky, A. S. Fleisher, A. p. 37. As of June 12, 2019: Carpenter, A. D. Joshi, M. Lu, M. Grundman, M. A. Mintun, http://www.ncbi.nlm.nih.gov/pubmed/25024750 D. M. Skovronsky, M. J. Pontecorvo, and AV45-A11 Study Group, Dall, T. M., M. V. Storm, R. Chakrabarti, O. Drogan, C. M. “Florbetapir F 18 Amyloid PET and 36-Month Cognitive Decline: Keran, P. D. Donofrio, V. W. Henderson, H. J. Kaminski, J. C. A Prospective Multicenter Study,” Molecular Psychiatry, Vol. 19, Stevens, and T. R. Vidic, “Supply and Demand Analysis of the No. 9, September 2014, pp. 1044–1051. As of June 12, 2019: Current and Future US Neurology Workforce,” Neurology, http://www.ncbi.nlm.nih.gov/pubmed/24614494 Vol. 81, No. 5, July 30, 2013, pp. 470–478. As of February 1, 2019: Eades, S. J., B. Taylor, S. Bailey, A. B. Williamson, J. C. Craig, https://www.ncbi.nlm.nih.gov/pubmed/23596071 S. Redman, and Search Investigators, “The Health of Urban Dementia Australia, “Research—Risk Reduction and Aboriginal People: Insufficient Data to Close the Gap,” Medical Prevention,” n.d.a. As of June 12, 2019: Journal of Australia, Vol. 193, No. 9, November 1, 2010, https://www.dementia.org.au/about-dementia/dementia- pp. 521–524. research/reduction-and-prevention Ellis, K. A., A. I. Bush, D. Darby, D. De Fazio, J. Foster, P. ———, “Types of Dementia,” n.d.b. As of June 12, 2019: Hudson, N. T. Lautenschlager, N. Lenzo, R. N. Martins, P. https://www.dementia.org.au/information/about-dementia/ Maruff, C. Masters, A. Milner, K. Pike, C. Rowe, G. Savage, C. types-of-dementia Szoeke, K. Taddei, V. Villemagne, M. Woodward, D. Ames, and AIBL Research Group, “The Australian Imaging, Biomarkers ———, House of Representatives Standing Committee on Health and Lifestyle (AIBL) Study of Aging: Methodology and Baseline and Ageing—Inquiry into Dementia: Early Diagnosis and Characteristics of 1112 Individuals Recruited for a Longitudinal Intervention, 2012. As of June 12, 2019: Study of Alzheimer’s Disease,” International Psychogeriatrics, https://www.aph.gov.au/parliamentary_business/committees/ Vol. 21, No. 4, August 2009, pp. 672–687. house_of_representatives_committees?url=haa/dementia/subs/ sub044%20-%20alzheimers%20australia%20act%20-%202%20 F. Hoffman-La Roche Ltd, “Media Release: Roche to Discontinue may%202012.pdf Phase III CREAD 1 and 2 Clinical Studies of Crenezumab in Early Alzheimer’s Disease (AD)—Other Company Programmes ———, “PBS Subsidies for Cholinesterase Inhibitors,” 2016. As of in AD Continue,” 2019. As of April 10, 2019: June 12, 2019: https://www.roche.com/media/releases/med-cor-2019-01-30.htm https://www.dementia.org.au/files/helpsheets/Helpsheet- DementiaQandA02-PBSSubsidiesForCholinesteraseInhibit Folstein, M. F., S. E. Folstein, and P. R. McHugh, “ ‘Mini-Mental ors_english.pdf State’: A Practical Method for Grading the Cognitive State of Patients for the Clinician,” Journal of Psychiatric Research, ———, “Early Diagnosis of Dementia—The Importance of Early Vol. 12, No. 3, November 1975, pp. 189–198. Diagnosis of Dementia and Information about the Diagnosis Process,” 2018. As of June 12, 2019: Fyles, Natasha, Minister of Health, “Media Statement—Boosting https://www.dementia.org.au/information/diagnosing-dementia/ Health Services: Territory’s First Pet Scanner Fully Operational,” early-diagnosis-of-dementia Northern Territory Government Newsroom, December 12, 2018. As of June 12, 2019: —— — , “Cognitive Screening and Assessment,” 2019. As of http://newsroom.nt.gov.au/mediaRelease/28509 June 12, 2019: https://www.dementia.org.au/information/for-health- General Practitioner Assessment of Cognition, homepage, 2016. professionals/clinical-resources/cognitive-screening-and- As of April 10, 2019: assessment http://gpcog.com.au/ Giamis, Anthony M., “Self-Referencing Radiation Detector for Use with a Radiopharmaceutical Quality Control Testing System,” Google Patents, 2012. As of June 12, 2019: https://patents.google.com/patent/US8937287

19 Gupta, V. B., E. Hone, S. Pedrini, J. Doecke, S. O’Bryant, I. James, Lancet Neurology, “Alzheimer’s Disease: A Time for Cautious A. I. Bush, C. C. Rowe, V. L. Villemagne, D. Ames, C. L. Masters, Optimism,” Lancet Neurology, Vol. 14, No. 8, August 2015, p. 779. R. N. Martins, and AIBL Research Group, “Altered Levels of As of June 12, 2019: Blood Proteins in Alzheimer’s Disease Longitudinal Study: http://www.ncbi.nlm.nih.gov/pubmed/26194922 Results from Australian Imaging Biomarkers Lifestyle Study of Ageing Cohort,” Alzheimer’s & Dementia (Amst), Vol. 8, 2017, Liu, Jodi L., Jakub P. Hlávka, Daniel T. Coulter, Sangita M. pp. 60–72. As of June 12, 2019: Baxi, Soeren Mattke, and Courtney A. Gidengil, Assessing the http://www.ncbi.nlm.nih.gov/pubmed/28508031 Preparedness of the Canadian Health Care System Infrastructure for an Alzheimer’s Treatment, Santa Monica, Calif.: RAND Hampel, H., S. E. O’Bryant, J. L. Molinuevo, H. Zetterberg, C. L. Corporation, RR-2744-BIOG, 2019. As of July 18, 2019: Masters, S. Lista, S. J. Kiddle, R. Batrla, and K. Blennow, “Blood- https://www.rand.org/pubs/research_reports/RR2744.html Based Biomarkers for Alzheimer Disease: Mapping the Road to the Clinic,” Nature Reviews Neurology, Vol. 14, No. 11, November Liu, Jodi L., Jakub P. Hlávka, Richard Hillestad, and Soeren 2018, pp. 639–652. As of June 12, 2019: Mattke, Assessing the Preparedness of the U.S. Health Care System http://www.ncbi.nlm.nih.gov/pubmed/30297701 Infrastructure for an Alzheimer’s Treatment, Santa Monica, Calif.: RAND Corporation, RR-2272-BIOG, 2017. As of June 12, 2019: Hansson, O., J. Seibyl, E. Stomrud, H. Zetterberg, J. Q. https://www.rand.org/pubs/research_reports/RR2272.html Trojanowski, T. Bittner, V. Lifke, V. Corradini, U. Eichenlaub, R. Batrla, K. Buck, K. Zink, C. Rabe, K. Blennow, L. M. Shaw, LoGiudice, D., K. Smith, J. Thomas, N. T. Lautenschlager, O. P. Swedish BioFINDER Study Group, and Alzheimer’s Disease Almeida, D. Atkinson, and L. Flicker, “Kimberley Indigenous Neuroimaging Initiative, “CSF Biomarkers of Alzheimer’s Cognitive Assessment Tool (KICA): Development of a Disease Concord with Amyloid-Beta PET and Predict Clinical Cognitive Assessment Tool for Older Indigenous Australians,” Progression: A Study of Fully Automated Immunoassays in International Psychogeriatrics, Vol. 18, No. 2, June 2006, BioFINDER and ADNI Cohorts,” Alzheimer’s & Dementia, Vol. 14, pp. 269–280. No. 11, November 2018, pp. 1470–1481. As of June 12, 2019: Mabbot, Vanna, and Paul Storey, “Australian Statistics on http://www.ncbi.nlm.nih.gov/pubmed/29499171 Medicines 2015,” 2016. As of April 10, 2019: Hara, K., S. Kunisawa, N. Sasaki, and Y. Imanaka, “Fu ture http://www.pbs.gov.au/info/statistics/asm/asm-2015 Projection of the Physician Workforce and Its Geographical Malek-Ahmadi, M., “Reversion from Mild Cognitive Impairment Equity in Japan: A Cohort-Component Model,” BMJ Open, to Normal Cognition: A Meta-Analysis,” Alzheimer Disease & Vol. 8, No. 9, September 17, 2018, e023696. As of June 12, 2019: Associated Disorders, Vol. 30, No. 4, October–December 2016, http://www.ncbi.nlm.nih.gov/pubmed/30224401 pp. 324–330. Heart of Australia, homepage, 2019. As of April 29, 2019: Medical Board of Australia, “Statistics—Registration Data,” https://www.heartofaustralia.com 2018. As of April 10, 2019: Hlávka, Jakub P., Soeren Mattke, and Jodi L. Liu, Assessing the https://www.medicalboard.gov.au/news/statistics.aspx Preparedness of the Health Care System Infrastructure in Six Euro Medical Service Advisory Committee, F-18 Fluorodeoxyglucose pean Countries for an Alzheimer’s Treatment, Santa Monica, Calif.: (FDG) Positron Emission Tomography (PET) for the Diagnosis of RAND Corporation, RR-2503-BIOG, 2018. As of June 12, 2019: Alzheimer’s Disease—MSAC Application No. 1195 Assessment https://www.rand.org/pubs/research_reports/RR2503.html Report, February 2015a. As of June 12, 2019: Hodkinson, H. M., “Evaluation of a Mental Test Score for http://www.msac.gov.au/internet/msac/publishing.nsf/Content/7 Assessment of Mental Impairment in the Elderly,” Age Ageing, 660CFC0BBA595EACA25801000123BAA/$File/1195FinalAssess Vol. 1, No. 4, November 1972, pp. 233–238. mentReport-accessible.pdf House of Representatives Standing Committee on Health and ———, Public Summary Document—Application No. 1195—F-18 Ageing, Inquiry into Dementia: Early Diagnosis and Intervention, Fluorodeoxyglucose Positron Emission Tomography (FDG PET) for 2012, item 44. As of June 12, 2019: the Diagnosis of Alzheimer’s Disease, 2015b. As of June 12, 2019: https://www.aph.gov.au/parliamentary_business/committees/ http://www.msac.gov.au/internet/msac/publishing.nsf/Content/7 house_of_representatives_committees?url=haa/dementia/subs.htm 660CFC0BBA595EACA25801000123BAA/$File/1195Final-PSD- Accessible.pdf IAEA—See International Atomic Energy Agency. ———, “1195—F-18 Flurodeoxyglucose Positron Emission International Atomic Energy Agency, “Cyclotron Master List,” Tomography (FDG PET) for the Diagnosis of Alzheimer’s 2019. As of April 10, 2019: Disease—Application Detail,” 2016. As of April 10, 2019: https://nucleus.iaea.org/sites/accelerators/Lists/ http://www.msac.gov.au/internet/msac/publishing.nsf/ Cyclotron%20Master%20List/public_cyclotron_db_ Content/1195-public view.aspx#InplviewHash8ad68ad5-364e-45e3-9804- b6386fb07a48=SortField%3DCountry-SortDir%3Dasc- Mitchell, A. J., and M. Shiri-Feshki, “Rate of Progression of Mild FilterField1%3Dcountry-FilterValue1%3Daustralia Cognitive Impairment to Dementia—Meta-Analysis of 41 Robust Inception Cohort Studies,” Acta Psychiatrica Scandinavica, Knopman, D. S., M. J. Knapp, S. I. Gracon, and C. S. Davis, “The Vol. 119, No. 4, April 2009, pp. 252–265. As of June 12, 2019: Clinician Interview-Based Impression (CIBI): A Clinician’s https://www.ncbi.nlm.nih.gov/pubmed/19236314 Global Change Rating Scale in Alzheimer’s Disease,” Neurology, Vol. 44, No. 12, December 1994, pp. 2315–2321. Molloy, D. W., E. Alemayehu, and R. Roberts, “Reliability of a Standardized Mini-Mental State Examination Compared with the Koepsell, T. D., and S. E. Monsell, “Reversion from Mild Traditional Mini-Mental State Examination,” American Journal Cognitive Impairment to Normal or Near-Normal Cognition: of Psychiatry,Vol. 148, No. 1, January 1991, pp. 102–105. Risk Factors and Prognosis,” Neurology, Vol. 79, No. 15, October 2012, pp. 1591–1598. As of June 12, 2019: MSAC—See Medical Service Advisory Committee. https://www.ncbi.nlm.nih.gov/pubmed/23019264

20 Nakamura, Akinori, Naoki Kaneko, Victor L. Villemagne, Rossetti, H. C., C. Munro Cullum, L. S. Hynan, and L. H. Lacritz, Takashi Kato, James Doecke, Vincent Doré, Chris Fowler, “The CERAD Neuropsychologic Battery Total Score and the Qiao-Xin Li, Ralph Martins, Christopher Rowe, Taisuke Progression of Alzheimer Disease,” Alzheimer Disease and Tomita, Katsumi Matsuzaki, Kenji Ishii, Kazunari Ishii, Yutaka Associated Disorders, Vol. 24, No. 2, April–June, 2010, pp. 138– Arahata, Shinichi Iwamoto, Kengo Ito, Koichi Tanaka, Colin L. 142. As of February 1, 2019: Masters, and Katsuhiko Yanagisawa, “High Performance https://www.ncbi.nlm.nih.gov/pubmed/20505431 Plasma Amyloid-B Biomarkers for Alzheimer’s Disease,” Nature, Vol. 554, February 8, 2018, pp. 249–254. Royal Australian and New Zealand College of Psychiatrists, Workforce Survey Report: The Australian Psychiatric Workforce, National Dementia Diagnostics Laboratory, “National Dementia June 2014. As of June 12, 2019: Diagnostics Laboratory (NDDL),” 2019. As of June 12, 2019: https://www.ranzcp.org/files/resources/reports/ranzcp- https://www.florey.edu.au/science-research/scientific-services- workforce-survey-report-australian-membersh.aspx facilities/national-dementia-diagnostics-laboratory Royal Flying Doctor Service, homepage, 2019. As of April 10, 2019: OECD—See Organisation for Economic Co-Operation and https://www.flyingdoctor.org.au/ Development. Skinner, J., J. O. Carvalho, G. G. Potter, A. Thames, E. Ong, K. T., V. L. Villemagne, A. Bahar-Fuchs, F. Lamb, N. Zelinski, P. K. Crane, L. E. Gibbons, and Alzheimer’s Disease Langdon, A. M. Catafau, A. W. Stephens, J. Seibyl, L. M. Neuroimaging Initiative, “The Alzheimer’s Disease Assessment Dinkelborg, C. B. Reininger, B. Putz, B. Rohde, C. L. Masters, Scale-Cognitive-Plus (ADAS-Cog-Plus): An Expansion of the and C. C. Rowe, “A Beta Imaging with 18F-Florbetaben in ADAS-Cog to Improve Responsiveness in MCI,” Brain Imaging Prodromal Alzheimer’s Disease: A Prospective Outcome Study,” and Behavior, Vol. 6, No. 4, December 2012, pp. 489–501. As of Journal of Neurology Neurosurgery and Psychiatry, Vol. 86, No. 4, June 12, 2019: 2015, pp. 431–436. http://www.ncbi.nlm.nih.gov/pubmed/22614326 Organisation for Economic Co-Operation and Development, Smith, K., L. Flicker, N. T. Lautenschlager, O. P. Almeida, D. “Health Care Resources,” database, 2018a. As of December 7, 2018: Atkinson, A. Dwyer, and D. LoGiudice, “High Prevalence of https://stats.oecd.org/Index.aspx?DataSetCode=HEALTH_ Dementia and Cognitive Impairment in Indigenous Australians,” REAC Neurology, Vol. 71, No. 19, November 4, 2008, pp. 1470–1473. ———, “Health Care Resources: Hospital Beds,” database, 2018b. South Australia Health, “Regional Cancer Services for As of December 7, 2018: Kangaroo Island Consumers,” 2014. As of April 10, 2019: https://stats.oecd.org/index.aspx?queryid=30183# https://www.sahealth.sa.gov.au/wps/wcm/connect/779a358 04227fafe9169fd892ec2192f/14046.5+CHEMOKangaroo+isl ———, “OECD Health Statistics 2018,” 2018c. As of April 10, and.pdf?MOD=AJPERES&CACHEID=ROOTWORKSPACE- 2019: 779a35804227fafe9169fd892ec2192f-lOSfXTK http://www.oecd.org/health/health-data.htm Spackman, D. E., S. Kadiyala, P. J. Neumann, D. L. Veenstra, and Palmqvist, S., H. Zetterberg, N. Mattsson, P. Johansson, S. D. Sullivan, “Measuring Alzheimer Disease Progression with Alzheimer’s Disease Neuroimaging Initiative, L. Minthon, Transition Probabilities: Estimates from NACC-UDS,” Current K. Blennow, M. Olsson, O. Hansson, and Swedish BioFINDER Alzheimer Research, Vol. 9, No. 9, November 2012, pp. 1050– Study Group, “Detailed Comparison of Amyloid PET and CSF 1058. As of February 1, 2019: Biomarkers for Identifying Early Alzheimer Disease,” Neurology, https://www.ncbi.nlm.nih.gov/pubmed/22175655 Vol. 85, No. 14, October 6, 2015, pp. 1240–1249. As of June 12, 2019: http://www.ncbi.nlm.nih.gov/pubmed/26354982 Statistics Canada, “Table 17-10-0057-01 Projected Population, by Projection Scenario, Age and Sex, as of July 1 (x 1,000),” 2018. As Perl, D. P. “Neuropathology of Alzheimer’s Disease,” Mount Sinai of April 22, 2019: Journal of Medicine, Vol. 85, No. 14, January–February 2010, https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1710005701 pp. 32–42. As of June 12, 2019: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2918894/ Tombaugh, T. N., I. McDowell, B. Kristjansson, and A. M. Hubley, “Mini-Mental State Examination (MMSE) and the Petersen, R. C., O. Lopez, M. J. Armstrong, T. S. D. Getchius, M. Modified MMSE (3MS): A Psychometric Comparison and Ganguli, D. Gloss, G. S. Gronseth, D. Marson, T. Pringsheim, Normative Data,” Psychological Assessment, Vol. 8, No. 1, 1996, G. S. Day, M. Sager, J. Stevens, and A. Rae-Grant, “Practice pp. 48–59. Guideline Update Summary: Mild Cognitive Impairment: Report of the Guideline Development, Dissemination, and Tsai, Tsuen-Chiuan, Misha Eliasziw, and Der-Fang Chen, Implementation Subcommittee of the American Academy of “Predicting the Demand of Physician Workforce: An Neurology,” Neurology, Vol. 90, No. 3, January 16, 2018, pp. 126– International Model Based on “Crowd Behaviors,” BMC Health 135. As of June 12, 2019: Services Research, Vol. 12, No. 1, March 26, 2012, p. 79. As of http://www.ncbi.nlm.nih.gov/pubmed/29282327 June 12, 2019: https://bmchealthservres.biomedcentral.com/track/ Portet, F., P. J. Ousset, P. J. Visser, G. B. Frisoni, F. Nobili, P. pdf/10.1186/1472-6963-12-79 Scheltens, B. Vellas, J. Touchon, and M. C. I. Working Group of the European Consortium on Alzheimer’s Disease, “Mild Cognitive U.S. National Library of Medicine, “Safety and Efficacy Study of Impairment (MCI) in Medical Practice: A Critical Review of Gantenerumab in Participants with Early Alzheimer’s Disease the Concept and New Diagnostic Procedure. Report of the MCI (AD),” Clinical Trials.gov, 2019a. As of April 4, 2019: Working Group of the European Consortium on Alzheimer’s https://clinicaltrials.gov/ct2/show/NCT03443973 Disease,” Journal of Neurology, Neurosurgery, and Psychiatry, Vol. 77, No. 6, June 2006, pp. 714–718. As of June 12, 2019: ———, “A Study to Confirm Safety and Efficacy of BAN2401 in http://www.ncbi.nlm.nih.gov/pubmed/16549412 Participants with Early Alzheimer’s Disease (Clarity AD),” Clinical Trials.gov, 2019b. As of April 4, 2019: RANZCP—See Royal Australian and New Zealand College of https://clinicaltrials.gov/ct2/show/NCT03887455 Psychiatrists.

21 ———, “A Study of Lanabecestat (LY3314814) in Early Walterfang, Mark, and Dennis Velakoulis, “Neuropsychiatry Unit Alzheimer’s Disease Dementia,” Clinical Trials.gov, 2019c. As of Cognitive Screening Tool (NUCOG),” 2013. As of June 12, 2019: April 4, 2019: http://www.nucog.com.au/index.html https://clinicaltrials.gov/ct2/show/NCT02972658 Yesavage, J. A., R. O’Hara, H. Kraemer, A. Noda, J. L. Taylor, S. Vassilaki, M., R. H. Cha, J. A. Aakre, T. M. Therneau, Y. E. Geda, Ferris, M. C. Gely-Nargeot, A. Rosen, L. Friedman, J. Sheikh, M. M. Mielke, D. S. Knopman, R. C. Petersen, and R. O. Roberts, and C. Derouesne, “Modeling the Prevalence and Incidence of “Mortality in Mild Cognitive Impairment Varies by Subtype, Sex, Alzheimer’s Disease and Mild Cognitive Impairment,” Journal and Lifestyle Factors: The Mayo Clinic Study of Aging,” Journal of Psychiatric Research, Vol. 36, No. 5, September–October 2002, of Alzheimer’s Disease, Vol. 45, No. 4, 2015, pp. 1237–1245. As of pp. 281–286. February 1, 2019: https://www.ncbi.nlm.nih.gov/pubmed/25697699 Walterfang, M., R. Siu, and D. Velakoulis, “The NUCOG: Validity and Reliability of a Brief Cognitive Screening Tool in Neuropsychiatric Patients,” Australian & New Zealand Journal of Psychiatry, Vol. 40, Nos. 11–12, November–December 2006, pp. 995–1002.

22 About the Authors Sangita M. Baxi is a doctoral fellow at the Pardee RAND Graduate School and assistant policy researcher at the RAND Corporation. Her research interests include noncommunicable diseases in global settings and technology and innovations in health care. Federico Girosi is a senior policy researcher at the RAND Corporation and the Chief Scientist of the Digital Health Cooperative Research Centre (Sydney, Australia). His research interests include health care modelling for the purpose of policy analysis and the application of data science to solve prob lems of immediate interest to public and private health care stakeholders. Jodi L. Liu is an associate policy researcher at the RAND Corporation. Her research focuses on health care financing and payment policies and frequently uses simulation modelling to analyse effects on health care services, insurance coverage and expenditures.

23 AUSTRALIA

The RAND Corporation is a research organization that develops solutions to public policy challenges to help make communities throughout the world safer and more secure, healthier and more prosperous.

RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors. is a registered trademark.

Limited Print and Electronic About This Report Distribution Rights

This report illustrates the magnitude of health care system infrastructure chal- This document and trademark(s) lenges in the diagnosis and treatment of early-stage Alzheimer’s disease with a contained herein are protected hypothetical disease-modifying therapy in Australia. by law. This representation of RAND intellectual property is This research was sponsored by Biogen and carried out within the Access and provided for noncommercial use Delivery Program in RAND Health Care. only. Unauthorized posting of this RAND Health Care, a division of the RAND Corporation, promotes healthier publication online is prohibited. Permission is given to duplicate this societies by improving health care systems in the United States and other document for personal use only, as countries. We do this by providing health care decisionmakers, practitioners long as it is unaltered and complete. and consumers with actionable, rigorous, objective evidence to support their Permission is required from RAND most complex decisions. For more information, see www.rand.org/health-care. to reproduce, or reuse in another We thank Colin Masters and Perminder Sachdev for sharing insights into clini- form, any of our research documents cal practices and policies in Australia. We also thank Kaarin Anstey and Peter for commercial use. For information on reprint and linking permissions, Dortmans for their valuable feedback in the quality assurance review of this please visit www.rand.org/pubs/ report. Last, we are grateful to Patricia Smith for administrative assistance. permissions.

About RAND Australia For more information on this publication, visit RAND Australia is RAND’s Canberra-based subsidiary that analyses defence, www.rand.org/t/RR2891 security, economic and social issues for Australian clients. With a commit- © Copyright 2019 Biogen Inc. ment to core values of quality and objectivity, RAND Australia combines local research talent with world-class experts from across RAND’s global presence to www.rand.org/ solve complex Australian public policy problems. australia

RR-2891-BIOG

Assessing the Preparedness of the Australian Health Care System Infrastructure for an Alzheimer’S Disease-­Modifying Therapy

Assessing the Preparedness of the Australian Health Care System Infrastructure for an Alzheimer’S Disease-Modifying Therapy