Economic evaluations of strategies targeting pre‐diagnosis dementia populations: A systematic review

Abstract

This review aims to appraise and summarize the best available evidence from economic evaluations (EEs) of strategies targeting populations prior to a diagnosis of dementia. A comprehensive search was conducted in EMBASE, PubMed, EconLit, Web of Science, Cumulative Index to Nursing and Allied Health (CINAHL), and The NHS Economic Evaluation Database (NHS EED). Publications were dually screened and assessed against eligibility criteria, and reporting quality was appraised using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) framework. Seventeen full EEs met the inclusion criteria; all were conducted in high‐income countries and covered strategies in identification (29%), risk reduction (65%), and other strategies (6%). The number of publications from 2016 to 2025 tripled compared to the preceding periods. Most risk‐reduction strategies were cost‐effective and exhibited high reporting quality. The average reporting quality, based on the CHEERS framework, was 72%, with a noticeable improvement observed in articles published after 2020. Overall, the findings suggest that most risk‐reduction strategies may offer good value for money. However, evidence remains limited, specifically in low‐ and middle‐income countries; therefore, additional EEs are needed to guide future investment in strategies targeting pre‐diagnosis dementia populations.

Highlights

• Value for money of pre‐diagnosis dementia care remains under‐researched

• From 2016 to 2025, economic evaluations (Ees) tripled and reporting quality improved

• Risk‐reduction interventions were generally found to be cost‐effective

• Far fewer studies assessed cost‐effectiveness of prevention than post‐diagnosis interventions

• All studies originated in countries with established evidence‐based policy systems

1. INTRODUCTION

Dementia is a syndrome of progressive cognitive and functional decline that ultimately compromises independence in daily activities. It arises from multiple underlying neurodegenerative processes or diseases, such as Alzheimer's disease (AD), vascular dementia, dementia with Lewy bodies, frontotemporal dementias, and so forth. ¹ Historically, the treatment of dementia has focused on managing symptoms post‐diagnosis. However, decades of trials that failed to find a curative treatment have led to increasing recommendations on preventing, intervening, and managing dementia, with a focus on modifiable risk factors. ² , ³ , ⁴ , ⁵

Given rising life expectancy, the economic burden of dementia is escalating, with global costs projected to reach $2.8 trillion by 2030. ⁶ In the absence of curative treatment, growing evidence supports strategies targeting individuals at stages prior to having a diagnosis, indicating pre‐diagnosis as an emerging alternative investment decision to mitigate the growing burden of dementia.

The Lancet Commission on Dementia Prevention, Intervention and Care recommends that modifying 14 risk factors could potentially delay or prevent up to 45% of all cases. ⁵ However, this raises a critical question: Will effective strategies be economically viable and worth investing in? Although evidence regarding the effectiveness of strategies is crucial, it is often insufficient for informed decision‐making without evaluating their costs. For this reason, economic evaluation (EE), defined as the comparative analysis of alternative strategies regarding both costs and effectiveness, has become a vital tool for health care decision‐making. ⁷

EEs typically follow one of two designs: trial‐based or model‐based evaluations. Trial‐based evaluations are conducted alongside clinical trials, directly measuring and valuing resource use and effectiveness for participants within the trial period. In dementia research, conducting trials covering the entire disease trajectory, from pre‐diagnosis stages to advanced dementia, would be prohibitively expensive and logistically challenging. Therefore, model‐based evaluations are necessary to extrapolate outcomes beyond the limited duration and scope of trials. These models can incorporate data from multiple sources to generate long‐term estimates of costs and effectiveness. ⁸ , ⁹ Thus model‐based evaluations are equally important to determine the cost‐effectiveness of pre‐diagnosis investments.

Systematic reviews of EEs provide a structured means of summarizing existing evidence, assessing their quality and identifying research gaps. Previous reviews have examined diverse populations and interventions across the dementia spectrum, but have concentrated primarily on post‐diagnosis populations or specific interventions. For instance, Livingston et al. (2014) ¹⁰ and Eaglestone et al. (2023) ¹¹ reviewed non‐pharmacological interventions targeting individuals already diagnosed with mild cognitive impairment (MCI) or dementia. Braun et al. (2024) ¹² and Walsh et al. (2022) ¹³ synthesized evidence on population‐ and lifestyle‐based prevention strategies to reduce modifiable risk factors in the general population. Handels et al. (2019) ¹⁴ discussed methodological challenges in modeling primary prevention, whereas Darab et al. (2024) ¹⁵ and Kwon et al. (2025) ¹⁶ focused on modeling evaluations of diagnostic and treatment strategies among clinically diagnosed or biomarker‐confirmed populations.

Overall, these reviews reveal fragmented evidence across prevention, identification, and management strategies targeting mixed pre‐ and post‐diagnosis populations. To date, no comprehensive synthesis has focused exclusively on EEs of pre‐diagnosis dementia. Moreover, the reporting quality of EEs remains inadequate, potentially hindering their usefulness for health professionals and decision‐makers. ¹⁷ , ¹⁸ , ¹⁹ This systematic review, therefore, aims to identify, summarize, and critically appraise the current best available EE evidence on strategies targeting pre‐diagnosis dementia populations. Findings will help inform future investment decisions and promote greater standardization and transparency in emerging economic evidence, aligning with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS‐2022) checklist. ²⁰

To our knowledge, this is the first systematic review of EEs for pre‐diagnosis dementia, categorizing strategies into identification, risk reduction, and other strategies. In this review, pre‐diagnosis populations are defined to include the general population, at‐risk groups, and individuals with MCI without detectable neurodegenerative pathology. This broader framing reflects the real‐world context in which prevention and early‐intervention strategies are implemented, recognizing that many individuals may never receive a definitive diagnosis. Furthermore, to ensure that evidence from EEs reliably informs health professionals and decision‐makers, our newly developed framework, modified from CHEERS‐2022, is applied to assess the quality of the included studies objectively. ²¹

2. METHOD

2.1. Protocol and registration

This systematic review was registered with the International Prospective Register of Systematic Reviews (CRD42024521521), and the peer‐reviewed protocol was published elsewhere. ²² The results were reported following the Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) guidelines. ²³

2.2. Search strategy

Potentially relevant literature was identified through a systematic search across six electronic databases: EMBASE, PubMed, EconLit, Web of Science, Cumulative Index to Nursing and Allied Health (CINAHL), and the NHS Economic Evaluation Database (NHS EED). The search covered publications from January 1, 2000, to January 14, 2025. Detailed information on search terms was uploaded to a repository. ²⁴

2.3. Eligibility criteria

This review employed the Population, Intervention, Comparators, Outcomes, and Study design (PICOS) framework to define inclusion and exclusion criteria (see Table 1).

TABLE 1.

The Population, Intervention, Comparators, Outcomes, and Study design (PICO) criteria.

Criteria	Inclusion	Exclusion
Population	Individuals at any stage prior to having a diagnosis of dementia, including the general population, people at risk of dementia, and people with MCI without detectable dementia pathology (e.g. MCI due to AD). The pre‐diagnosis dementia population was commonly predefined based on authors explicitly indicating that their studies began with a “no dementia” cohort, a general MCI population, or at‐risk populations.	Studies that did not target pre‐diagnosis dementia populations, populations with unclear boundaries surrounding diagnosis (i.e. mixed pre‐ and post‐diagnosis) or individuals who might only exhibit MCI symptomatology, but where neurodegenerative pathology had been established (e.g., MCI due to AD).
Interventions	All strategies targeting our pre‐specified population at the pre‐diagnosis stage of dementia were included. Eligible interventions must have an intended effect before the diagnosis of dementia. The following types of strategies were considered: 1. Identification of individuals at risk of dementia; 2. Interventions targeting modifiable risk factors (e.g., physical activity, diet, cognitive training), and 3. Other strategies, such as healthcare service delivery models that operate before a dementia diagnosis is made. Eligibility required that the intervention produce effects during the pre‐diagnosis stage, rather than merely identifying undiagnosed individuals for subsequent intervention.	Articles were excluded if: 1. The strategies under investigation did not specifically target the defined pre‐diagnosis populations; 2. The intervention only identified undiagnosed individuals and then delivered risk reduction or management strategies after identification, that is, no intervention effect occurred during the pre‐diagnosis stage; 3. The study addressed both pre‐ and post‐diagnosis stages in a single study (e.g., combining early detection and treatment for diagnosed patients); such studies will be considered in a separate review.
Comparator	The comparison group contained any of the following: (1) usual care or standard of care or conventional interventions, (2) placebo, (3) doing nothing or without intervention, or (4) any alternatives aimed at a pre‐defined population.	Studies that did not include any comparator
Outcomes	Primary outcomes were cost‐effectiveness assessment, including costs, incremental costs, quality‐adjusted life‐year (QALY), incremental QALY, and incremental cost‐effectiveness ratio (ICER). Additional outcomes were included: life years gained, time of institutionalized delay, resource use, utility, and so forth.	Studies that did not report economic outcomes
Study design	Only full EEs were selected for this review. Full EEs were based on Drummond's classification scheme, including (1) cost‐effectiveness analysis (CEA), (2) cost‐utility analysis (CUA), (3) cost‐benefit analysis (CBA), and (4) cost‐consequence analysis (CCA).	Studies which were (1) Partial EEs such as cost‐of‐illness analysis, efficacy effectiveness evaluation, and cost‐outcome description; (2) reviews, editorials, letters, commentaries, conference abstracts; (3) proceedings, protocols, case studies, and full‐text unretrievable.
Settings	The review included studies conducted worldwide, without limiting to any clinical setting.	None

The target population included individuals at any stage prior to having a diagnosis of dementia (i.e., pre‐diagnosis dementia). This included the general population, people at risk of dementia, and people with MCI without detectable neurodegenerative pathology (e.g., MCI due to AD). This review included strategies that exclusively targeted the aforementioned populations. Articles were included if they were full EEs according to Drummond's classification schemes (Table S1), encompassing cost‐effectiveness analysis (CEA), cost‐utility analysis (CUA), cost‐benefit analysis (CBA), and cost‐consequence analysis (CCA). ⁷ No restrictions were applied to the study settings.

Articles were excluded if: (1) The population included both pre‐diagnosis and diagnosed individuals or MCI with detectable neurodegenerative pathology (e.g., MCI due to AD). (2) Intervention targeted both pre‐ and post‐diagnosis stages (e.g., combining identification and treatment) in a single study. (3) Partial EEs such as cost analysis, efficacy or effectiveness evaluation, and cost‐outcome description were excluded because these designs do not compare both costs and outcomes of competing alternatives and therefore cannot provide a full assessment of cost‐effectiveness or value for money. (4) Articles were excluded if they were reviews, editorials, letters, commentaries, viewpoints, perspectives, conference abstracts, protocols, case studies, or full‐text unretrievable.

2.4. Study selection

References were imported into Covidence for title and abstract screening and full‐text review. The selection of full‐text reviews was guided by inclusion and exclusion criteria. During the screening and review processes, at least two reviewers independently screened and evaluated each record, and any conflicts were resolved by a third reviewer.

2.5. Data extraction and synthesis

A standardized template was piloted, refined, and finalized for data extraction. Study details, including general information, methodology, and outcomes, were systematically extracted into the predefined template using Covidence. At least two reviewers independently conducted data extraction, while a third reviewer verified, compared, and resolved any discrepancies to ensure the accuracy and consistency of all extracted data.

Pre‐diagnosis strategies were categorized into three groups: identification, risk reduction, and other strategies. The latter category comprised articles that did not fit into either identification or risk reduction but nonetheless presented relevant considerations for policy. A narrative synthesis was performed to summarize the characteristics of the included articles. To facilitate comparison, all cost values were converted into international dollars (Int$) (2024) using the purchasing power parity from the International Monetary Fund dataset. ²⁵ If the included articles did not explicitly report the price year, we applied a year preceding the publication year to account for the typical time lag between study execution and publication.

Results were plotted on a cost‐effectiveness plane to enhance comparability across studies that reported incremental costs and quality‐adjusted life years (QALYs). Two willingness‐to‐pay (WTP) thresholds were applied to visualize the relative cost‐effectiveness of those strategies: $150,000/QALY, representing a less‐regulated market in the United States, ²⁶ and £12,936/QALY, reflecting a cost‐constrained system in the United Kingdom (UK). ²⁷ These thresholds were chosen to illustrate the spectrum between high and low benchmarks.

2.6. Quality appraisal

Two reviewers independently assessed the reporting quality of the included articles using our framework, and a third reviewer made comparisons and resolved any tiebreaking decisions. The process of piloting and developing a quality assessment framework, which was modified from the CHEERS‐2022 checklist, ²⁰ was described in the review protocol. ²² This framework included the title (2 sub‐items), abstract (4 sub‐items), introduction (3 sub‐items), methods (39 sub‐items), results (11 sub‐items), discussion (6 sub‐items), and other relevant information (2 sub‐items). ²¹

Articles were scored by summing the scores of all applicable sub‐items and dividing by the number of applicable sub‐items.

Twenty‐eight original items from the CHEERS‐2022 checklist were scored and visualized using traffic‐light colors based on the following rules:

• Green: If all applicable sub‐items within an original item have been fully fulfilled.

• Red: If all applicable sub‐items within an original item were completely lacking.

• Amber: If an original item was judged partially fulfilled.

• Black: Items marked as not applicable.

3. RESULT

We identified 9922 references, of which 17 articles were included in the final review (see Figure 1). Those articles were categorized into three groups, including identification (29%), risk reduction (65%), and other strategies (6%). The results are presented in the following sections: characteristics of the included articles (Section 3.1), intervention‐specific findings categorized by intervention type (Section 3.2), and quality assessment (Section 3.3). Details of the 17 included articles are provided in the supplemental material.

FIGURE 1.

Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) flowchart of the study review process.

3.1. Characteristics of included articles

The characteristics of the included articles are summarized in Table 2. All studies were conducted in high‐income countries, including the UK (n = 3), ²⁸ , ²⁹ , ³⁰ Canada (n = 3), ³¹ , ³² , ³³ Australia (n = 2), ³⁴ , ³⁵ the United States (n = 1), ³⁶ and South Korea. ³⁷ The remaining seven articles came from European Union (EU) countries, with three conducted in a single country, encompassing Germany, ³⁸ , ³⁹ and Spain, ⁴⁰ and three involving multiple EU countries. ⁴¹ , ⁴² , ⁴³ However, the results from the multi‐country studies were aggregated. One article did not explicitly state the country, but it was funded by EU organizations and reported currency in Euro (EUR). ⁴⁴ Those three multiple‐country studies ⁴¹ , ⁴² , ⁴³ and the study without the country‐specific statement ⁴⁴ were categorized as “Other EU countries” to illustrate geographic distribution and publication trends over time by strategy in Figure 2.

TABLE 2.

Included article characteristics (N = 17).

Characteristics	Included studies n (%)	Reference
Country
United Kingdom	3 (18%)	28 , 29 , 30
Canada	3 (18%)	31 , 32 , 33
Australia	2 (12%)	34 , 35
Germany	2 (12%)	38 , 39
Finland; Sweden *	2 (12%)	41 , 42
United States	1 (6%)	36
South Korea	1 (6%)	37
Spain	1 (6%)	40
France; Monaco *	1 (6%)	43
No statement *	1 (6%)	44
Sponsorship source
Public	13 (76%)	28 , 29 , 31 , 32 , 33 , 34 , 35 , 37 , 38 , 40 , 41 , 42 , 44
Mixed	3 (18%)	30 , 39 , 43
None	1 (6%)	36
Population
At‐risk groups	8 (47%)	29 , 30 , 33 , 34 , 38 , 40 , 41 , 43
MCI (without dementia pathology)	6 (35%)	31 , 32 , 36 , 37 , 39 , 44
General population; At‐risk groups	3 (18%)	28 , 35 , 42
Intervention category
Risk reduction	11 (65%)	28 , 29 , 30 , 31 , 34 , 37 , 38 , 39 , 41 , 42 , 43
Identification	5 (29%)	32 , 33 , 36 , 40 , 44
Other	1 (6%)	35
Comparator
Current practice	8 (47%)	28 , 32 , 33 , 35 , 40 , 41 , 42 , 44
Other	6 (35%)	29 , 31 , 34 , 36 , 38 , 39
Placebo	2 (12%)	37 , 43
Do nothing	1 (6%)	30
Outcome
QALY	13 (76%)	29 , 30 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 41 , 42 , 44
Other	4 (24%)	28 , 31 , 40 , 43
Study design
Model‐based study	11 (65%)	28 , 29 , 30 , 32 , 33 , 34 , 35 , 36 , 41 , 42 , 44
Trial‐based study	6 (35%)	31 , 37 , 38 , 39 , 40 , 43
Type of full economic evaluation
CUA	8 (47%)	29 , 32 , 33 , 34 , 36 , 37 , 38 , 42
CAE	4 (24%)	28 , 31 , 40 , 43
CAE; CUA	4 (24%)	30 , 35 , 39 , 41
CUA; CBA	1 (6%)	44
Perspective
Societal perspective	6 (35%)	29 , 32 , 34 , 38 , 39 , 42
Health care sector perspective; societal perspective	4 (24%)	28 , 36 , 37 , 41
Health care sector perspective	4 (24%)	31 , 33 , 35 , 44
No statement	1 (6%)	30
Payer/patient perspective	1 (6%)	43
Payer/patient perspective; health care sector perspective	1 (6%)	40
Time horizon
Lifetime	7 (41%)	28 , 29 , 30 , 34 , 35 , 36 , 41
10 years and over	3 (18%)	32 , 33 , 42
1–5 years	4 (24%)	38 , 40 , 43 , 44
<1 year	3 (18%)	31 , 37 , 39
Resource use measurement
Not described	8 (47%)	28 , 29 , 30 , 32 , 33 , 34 , 37 , 44
Questionnaire	3 (18%)	31 , 38 , 39
Other databases	3 (18%)	40 , 41 , 42
Literature	2 (12%)	35 , 36
Questionnaire; registry; case report form	1 (6%)	43
Utility measurement
Not applicable	6 (35%)	28 , 31 , 33 , 34 , 40 , 43
EQ‐5D‐3L	4 (24%)	30 , 35 , 39 , 41
EQ‐5D‐5L	1 (6%)	42
HUI‐2	2 (12%)	32 , 36
Not specified	2 (12%)	29 , 44
EQ‐5D‐3L; KQOL‐AD	1 (6%)	37
EQ‐5D‐5L; EQ VAS	1 (6%)	38

^*Report as “Other EU countries” in Figure 2.

FIGURE 2.

Geographic distribution of included articles (N = 17), by publication year, categories, and average Consolidated Health Economic Evaluation Reporting Standards (CHEERS) scores.

As shown in Figure 2, no studies were published before 2011. Between 2011 and 2015, four studies were included, two on identification ³² , ⁴⁰ and two on risk reduction. ³¹ , ⁴² From 2016 to 2025, the number of publications tripled, with 13 studies focusing on risk reduction (n = 9), ²⁸ , ²⁹ , ³⁰ , ³⁴ , ³⁷ , ³⁸ , ³⁹ , ⁴¹ , ⁴³ identification (n = 3), ³³ , ³⁶ , ⁴⁴ and other strategies (n = 1). ³⁵

Model‐based evaluations were the most common design (n = 11), with the Markov being the most frequently applied model (n = 7). ²⁸ , ²⁹ , ³² , ³⁵ , ³⁶ , ⁴¹ , ⁴² Model‐based studies typically assessed longer time horizons than trial‐based counterparts, with seven adopting a lifetime horizon, ²⁸ , ²⁹ , ³⁰ , ³⁴ , ³⁵ , ³⁶ , ⁴¹ whereas the longest time horizon reported for trial‐based evaluations was 3 years. ⁴³ Most model‐based evaluations used QALY as their primary outcome (n = 10), ²⁹ , ³⁰ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ⁴¹ , ⁴² , ⁴⁴ whereas only three trial‐based studies included QALY. ³⁷ , ³⁸ , ³⁹ The European Quality of Life 5 Dimensions 3 Level version (EQ‐5D‐3L) was the most used utility measurement (n = 4), ³⁰ , ³⁵ , ³⁹ , ⁴¹ followed by the Health Utilities Index Mark 2 (HUI‐2) (n = 2). ³² , ³⁶

Seven model‐based studies (64%) did not describe measurements of resource use to estimate costs. ²⁸ , ²⁹ , ³⁰ , ³² , ³³ , ³⁴ , ⁴⁴ Data from the existing literature ³⁵ , ³⁶ and other databases ⁴¹ , ⁴² were the main sources of resource use in the model‐based evaluations. In contrast, only one trial‐based article did not report the resource use measurements. ³⁷ Half of the trial‐based studies utilized questionnaires to measure the resources used. ³¹ , ³⁸ , ³⁹

3.2. Intervention‐specific results

Figure 3 illustrates the cost‐effectiveness outcomes of the strategies, showing both incremental costs and incremental QALYs, along with their CHEERS Framework scores and study designs.

FIGURE 3.

Cost‐effectiveness plane of eligible strategies.

In the identification category, incremental QALYs ranged from 0.00014 ³³ to 0.046, ⁴⁴ whereas incremental costs varied from −54 ³³ to 1198 ³² (Int$, 2024). Genetic test ONDRISeq yielded a small gain in QALYs alongside a slight cost reduction, making it a dominant strategy. ³³ The majority of risk‐reduction interventions were cost‐saving, with modest gains in QALYs. Incremental QALYs varied from 0.0033 ³⁷ to ≈0.0967, ²⁹ and incremental costs ranged from −3,974 ⁴² to 976 ²⁹ (Int$, 2024). The article on “other strategies” was cost‐saving, with zero incremental QALY and −60 (Int$, 2024) incremental cost, respectively. ³⁵

Detailed findings for each category are presented in the following sections, and a comprehensive summary of each included study is available in the supplemental material.

3.2.1. Identification

The evaluation of identification articles (n = 5) covered a diverse range of strategies, including screening services ⁴⁰ ; genetic screening for the apolipoprotein E (APOE) e4 allele ³² and for 21 AD‐related genes (ONDRISeq) ³³ ; and cerebrospinal fluid (CSF) biomarker testing. ³⁶ , ⁴⁴ Most articles (n = 4) considered the current practice as the comparator, which was defined variably as “standard of care,” “standard test workup,” or “the status quo.” ³² , ³³ , ⁴⁰ , ⁴⁴ The remaining article compared identification strategies against strategies without test information. ³⁶ Four of 10 articles did not report intervention costs, whereas the reported intervention costs ranged from 341 ³³ to 413 ³⁶ (Int$, 2024).

Iragorri et al. ³³ found that ONDRISeq dominated current practice in Ontario. Handels et al. ⁴⁴ and Djalalov et al. ³² concluded that CSF testing and APOE e4 were cost‐effective at WTP thresholds of €20,000 and CAD 39,000, respectively. Carnero‐Pardo et al. and Michaud et al. did not reach a definitive conclusion on cost‐effectiveness. ³⁶ , ⁴⁰

3.2.2. Risk reduction

The evaluation of risk‐reduction interventions (n = 11) could be subdivided into lifestyle‐related risk reduction (n = 5), prevention of health conditions (n = 2), and a combination of both (n = 4). Of the five articles that evaluated lifestyle strategies, three studies focused on physical activity ²⁸ , ³¹ , ³⁹ ; one on nutrition ³⁰ ; and one on a combined physical activity intervention, cognitive stimulation, and nutrition. ⁴³ Regarding the prevention of health conditions, Zhang et al. evaluated the multi‐domain cardiovascular disease prevention program, ⁴² and Lee et al. evaluated herbal medicine compared to a placebo. ³⁷ The remaining four articles evaluated the combination of health condition prevention and lifestyle‐related risk reduction. These articles evaluated a range of interventions, including antihypertensive therapy, diabetes prevention, hearing aids, and nicotine replacement therapy ²⁹ ; strategies to reduce noncommunicable diseases and promote adherence to healthy behaviors through training ³⁴ ; the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) prevention program, ⁴¹ which consists of cardiovascular risk‐factor control, physical activity, cognitive training and dietary counselling; and a comprehensive multicomponent intervention targeting cognitive decline. ³⁸ The latter included nutritional guidance, monitoring of vascular risk factors, promotion of physical activity, home‐based cognitive training, medication optimization, individually tailored social engagement activities, and access to additional support services. ³⁸

Seven articles reported intervention costs, which ranged from 0.06 ³⁰ to 818 ⁴¹ (Int$, 2024), whereas four other articles did not provide information on intervention costs. ²⁸ , ²⁹ , ³⁴ , ⁴³ Regarding cost‐effectiveness, four articles indicated that interventions were dominant, ³⁰ , ³¹ , ⁴¹ , ⁴² whereas three reported that interventions were cost‐effective. ³⁴ , ³⁸ , ³⁹ Mukadam et al. evaluated four interventions, finding that antihypertensive therapy, nicotine replacement therapy, and hearing aids were cost‐effective, whereas the diabetes prevention package had an incremental cost‐effectiveness ratio (ICER) that exceeded the WTP threshold of £20,000. ²⁹ Lee et al. concluded that herbal medicine was not cost‐effective. ³⁷ The remaining two articles did not explicitly state the cost‐effectiveness conclusion. ²⁸ , ⁴³

3.2.3. Other strategies

The only study classified under this category was Advanced Care Planning (ACP) conducted in Australia by Nguyen et al. ³⁵ This study examined the cost‐effectiveness of ACP compared to the “current situation”. The cost of implementing ACP was estimated to range from 601 to 735 (Int$, 2024) per individual and find a small saving in total cost, yielding a reduction of −60 (Int$, 2024), suggesting a net zero compared to not providing ACP as part of the current situation.

3.2.4. Quality assessment: The CHEERS framework

Results here are from the application of a novel CHEERS framework ⁴⁵ to objectively compare the reporting quality of all EEs that meet the inclusion criteria. Our 17 included articles had an average reporting quality score of 72% (± SD 12%), ranging from 48% to 89% These results are summarized in Figure 4, which provides the reader a useful source to understand what CHEERS items have been fully, partially, or not satisfied, and how, on aggregate, this results in the CHEERS score (where an item is not satisfied; specifics of item, or subitem, are available upon request).

FIGURE 4.

Reporting quality assessment using the framework.

Overall, the average CHEERS score of the included articles showed an upward trend, indicating improved reporting quality after 2020. The average CHEERS scores declined from 71% (2011–2015) to 64% (2016–2020), but increased to 78% in 2021–2025, suggesting greater adherence to reporting standards in recent years. The average quality scores for identification, risk‐reduction interventions, and other strategies were 69%, 73%, and 71%, respectively. Regarding study design, trial‐based evaluations had higher overall reporting quality than model‐based evaluations, with an average score of 81% versus 67%.

Item scores varied from 6% to 100%. Only Nguyen et al. reported on “characterizing distributional effects” (Item 19, 6%). ³⁵ VanBaal et al. and Brettschneider et al. included the “health economics analysis plan” section (Item 4, 12%). ²⁸ , ³⁸ Notably, no article mentioned the item “21 – approach to engagement with patients and others affected by the study” in their publication, making all results for the “effect of engagement with patients and others affected by the study” (Item 25) not applicable. It is noteworthy that these two items were added to the CHEERS checklist in 2022; thus, EEs published before 2022 did not report them. Two items received the highest score (100%), including item “27 – source of funding,” and item “28 – conflict of interest.”

4. DISCUSSION

4.1. Key findings

This review is the first attempt to comprehensively identify, summarize, and critically appraise all EEs of strategies targeting populations prior to a dementia diagnosis, covering identification, risk reduction, and other strategies. As of January 2025, 17 articles met our explicit inclusion criteria, with the majority focusing on risk‐reduction interventions.

The rates of production of EEs tripled since 2016, following groundwork that reframed dementia as partly preventable, from early “brain health” and risk‐factor framing and economic perspective pieces to quantified population‐attributable fractions for modifiable risks. ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ A 2015 inflexion then provided proof‐of‐concept and policy salience through the FINGER randomized trial, ⁵⁰ the Organization for Economic Co‐operation and Development (OECD)'s Addressing Dementia report, ⁵¹ and the launch of the Global Brain Health Institute, ⁵² which collectively amplified the “brain health” prevention narrative. Subsequently, the World Health Organization (WHO) Dementia risk reduction guidelines, ⁵³ World Dementia Council: the dementia landscape project, ⁵⁴ and the Lancet Commission report ⁵ further legitimized pre‐diagnosis strategies for funders and journals, accelerating demand for models that value upstream interventions.

Despite this increase, our findings indicate considerable scope for further growth, especially when contrasted with post‐diagnosis research. ⁵⁵ In that field, the volume of cost‐effectiveness evidence was approximately eight times greater; yet findings were mixed as to whether delaying intervention until diagnosis represents the most efficient use of resources. ⁵⁵ Among interventions targeting pre‐diagnosis dementia populations, risk reduction strategies showed the strongest evidence, with the majority being cost‐effective and associated with modest QALY gains.

The CHEERS framework provided objective evidence that the aggregate CHEERS scores appear to have improved notably during 2021–2025. Reporting standards were generally higher for modifiable risk reduction interventions—particularly in trial‐based EEs—than for other strategies. The three trial‐based evaluations stood out by demonstrating substantially higher reporting quality, compared to model‐based counterparts, thereby offering valuable real‐world evidence (one even yielded a net savings alongside strong QALY gains). The traffic‐light chart (Figure 4) highlights the items requiring attention to improve reporting quality in future EEs.

4.2. Geographic distribution of included articles

The geographic distribution of the included studies (Figure 2) highlights a dominance of high‐income countries, particularly those with strong policy frameworks for health research and evidence‐based decision‐making. The largest proportion originated from EU countries (41.2%), followed by Canada (17.6%), the UK (17.6%), and Australia (11.8%). This concentration likely stems from funding agencies in those countries that often mandate or encourage the inclusion of EEs in health research to optimize resource allocation and inform policy. For example, countries within the EU have a long‐standing emphasis on social welfare and public investment in health, which drives the demand for cost‐effectiveness evidence to guide policy. ⁵⁶ Moreover, agencies such as the Canadian Institutes of Health Research (CIHR), the National Institute for Health and Care Research (NIHR) in the UK, and the National Health and Medical Research Council (NHMRC) in Australia prioritize generating evidence on clinical and economic impacts to support evidence‐based decision‐making. ⁵⁷ , ⁵⁸ , ⁵⁹ In addition, global initiatives such as the Lancet Commission on Dementia have further supported the integration of EE into health strategy development for dementia. ⁵

Although a study from Asia indicates emerging interest in incorporating the value of EEs for pre‐diagnostic strategies, many countries, particularly those with limited resources, remain underrepresented. Given the growing burden of dementia worldwide, greater investment in EE research in low‐ and middle‐income countries is warranted. Encouragingly, many of the model‐based studies identified in this review could be adapted to local contexts, and fostering international collaboration may help to generate more inclusive and regionally relevant evidence.

4.3. Insights into risk reduction interventions

This review provides economic evidence to support the implementation of several risk‐reduction interventions outlined in the Lancet Commission report ⁵ and in line with WHO guidelines for reducing the risk of cognitive decline and dementia. ⁵³ Our review indicates emerging evidence that many strategies could represent a potential investment to mitigate the risk of dementia. Of particular note are interventions that demonstrated incremental QALY gains alongside reduced total cost compared with usual care. Examples include nicotine replacement therapy, ²⁹ hearing aids, ²⁹ a home‐based exercise program, ³⁹ a multi‐domain cardiovascular disease prevention program, ⁴² and the FINGER program, ⁴¹ which combines exercise, cognitive training, dietary counselling, and cardiovascular risk‐factor control. However, variation in reporting quality (see Figure 4) highlights the importance of critically assessing how EEs differ in methodology and quality, and the extent to which findings are sufficiently robust to inform policy. Some interventions may indeed merit further research before being recommended for large‐scale implementation.

4.4. Challenges in comparing cost‐effectiveness across categories

Comparing the cost‐effectiveness of various strategies, within and across categories, is inherently complex due to substantial heterogeneity among studies. Nevertheless, using a generic health outcome measure such as QALYs provides a standardized metric that allows for cross‐comparison. QALYs capture both mortality (life years gained) and morbidity (reflected in quality gains, as weighted by preferences or utility scores), making them essential for evaluating relative value across interventions. ⁷ Despite this, half of the included trial‐based evaluations in this review did not collect utility data for estimating QALYs. Future trial‐based evaluations should prioritize the collection of utility data to enable robust comparisons across strategies.

4.5. Cost‐effectiveness is not one‐size‐fits‐all: The role of country‐specific thresholds and decision‐making processes

The cost‐effectiveness of a strategy is determined not only by its incremental costs and effects, but also by the WTP threshold in a given jurisdiction. WTP thresholds indicate the monetary value society places on generating or preserving one unit of health, such as a QALY. ⁷

To facilitate comparison, the U.S. and the UK thresholds were employed to represent the upper and lower bounds, respectively (Figure 3). The results indicate that all interventions, except the diabetes prevention package, ²⁹ fell below the U.S. threshold and were therefore considered cost‐effective. When applying the more conservative UK threshold, four interventions, namely the diabetes prevention package, ²⁹ herbal medicine, ³⁷ the multicomponent intervention against cognitive decline, ³⁸ and apolipoprotein E (APOE) genetic testing, ³² exceeded the threshold and were not deemed cost‐effective.

Therefore, caution is warranted when applying cost‐effectiveness evidence across countries without incorporating an EE that aligns with the country‐specific WTP threshold. In countries without explicit WTP thresholds, decision‐making should consider cost‐effectiveness alongside other factors, such as budget impact, affordability, and equity. A transparent decision‐making process that involves various stakeholders is essential. ⁶⁰

4.6. Limitations

Despite the use of a comprehensive search strategy and a rigorous process, this review has several limitations. First, in line with A Measurement Tool to Assess Systematic Reviews, version 2 (AMSTAR 2), ⁶¹ we followed a predefined review process as set out in the scope of our review, which was published in a review protocol. ²² However, one limitation is the critique of all aspects of different forms of EE (i.e., trials v models), and specifically of the underlying assumptions of the included model‐based EEs and how varying assumptions influenced the cost‐effectiveness results. Second, the definition of pre‐diagnosis dementia populations may be ambiguous, as it is not a standardized clinical term; however, we clearly specified the criteria used to define these populations from an operational and health–economic viewpoint. Third, during the full‐text screening, reviewers excluded partial EE as an inappropriate study design in general without specifying the type of partial EE according to Drummond's classification. This made it impossible to determine how many partial EEs were assessed at the full‐text screening. Fourth, three included articles did not report the original cost year; to enable cost comparisons, we assumed the cost year to be 1 year prior to publication, which may not accurately reflect the actual values when adjusted to 2024. Finally, the reporting quality assessment was based on a newly developed framework that has not yet undergone peer review.

4.7. Future research

This review identifies several areas for improvement in future research. First, for EEs conducted across multiple countries, results should be reported separately for each participating country. Aggregated results in such cases are of limited value, as they do not account for differences in benchmarks and decision‐making processes across countries. Second, given that the majority of at‐risk populations and individuals with dementia live in developing countries, there is a substantial need for future EEs to identify cost‐effective strategies tailored to these settings. Third, the implementation and collection of utility data for estimating QALYs should be expanded in trial‐based evaluations. Fourth, most model‐based evaluations did not report resource use measurement in sufficient detail. Future model‐based studies should therefore enhance transparency by providing comprehensive reporting of resource‐use inputs. Finally, following the CHEERS 2022 checklist, authors should develop and share health economics analysis plans and actively involve patients and the public in study design. Such practices would improve the clarity, transparency, and policy relevance of reporting.

5. CONCLUSION

This review finds that most risk‐reduction interventions targeting pre‐diagnosis dementia populations were cost‐effective and reported high quality. These findings suggest that investing in such interventions may offer good value for the money. However, the evidence base remains limited relative to post‐diagnosis dementia research, constraining the strength of any definitive conclusions. To inform future investment decisions, there is a clear need for high‐quality, standardized EEs, particularly in low‐ and middle‐income settings, addressing the prevention of dementia before diagnosis.

CONFLICT OF INTEREST STATEMENT

All authors have no conflict of interest to declare.

Supporting information

Supporting Information

Supporting Information