Cost-effectiveness of physical activity interventions for prevention and management of cognitive decline and dementia—a systematic review

Background Although increasing physical activity (PA) has been suggested to prevent and manage cognitive decline and dementia, its economic impact on healthcare systems and society is largely unknown. This study aimed to summarize evidence on the cost-effectiveness of PA interventions to prevent and manage cognitive decline and dementia. Methods Electronic databases, including PubMed/MEDLINE, Embase, and ScienceDirect, were searched from January 2000 to July 2023. The search strategy was driven by a combination of subject-heading terms related to physical activity, cognitive function, dementia, and cost-effectiveness. Selected studies were included in narrative synthesis, and extracted data were presented in narrative and tabular forms. The risk of bias in each study was assessed using the Consolidated Health Economic Evaluation Reporting Standards and Consensus on Health Economic Criteria list. Results Five of the 11 identified studies focused on individuals with existing dementia. Six of the 11 identified studies focused on individuals with no existing dementia, including 3 on those with mild cognitive impairment (MCI), and 3 on those with no existing MCI or dementia. PA interventions focused on individuals with no existing dementia were found to be cost-effective compared to the control group. Findings were mixed for PA interventions implemented in individuals with existing dementia. Conclusions PA interventions implemented before or during the early stage of cognitive impairment may be cost-effective in reducing the burden of dementia. More research is needed to investigate the cost-effectiveness of PA interventions in managing dementia. Most existing studies used short-term outcomes in evaluating the cost-effectiveness of PA interventions in the prevention and management of dementia; future research should consider adding long-term outcomes to strengthen the study design. Supplementary Information The online version contains supplementary material available at 10.1186/s13195-023-01286-7.


Introduction
As the global population ages, the prevalence of dementia continues to rise, which becomes one of the greatest clinical, public health, and social challenges.Worldwide, it is estimated that around 55 million people were living with dementia in 2019, and the number may increase to 139 million in 2050 [1].Dementia severely erodes functioning and quality of life for people affected and creates devastating burdens and stress for their families and healthcare systems.The economic consequences of dementia are enormous-the global societal cost of dementia in 2019 was estimated to be $1.3 trillion [1].This brings about a critical question: given the limited healthcare and public health resources, how can we better allocate resources to curb the burden of dementia effectively and efficiently?[2].
Physical activity (PA) is defined as any bodily movement produced by skeletal muscles that results in energy expenditure [3].Several systematic reviews and meta-analyses suggested that PA may reduce or delay the development of several modifiable risk factors for cognitive declines, such as obesity, diabetes, and hypertension [4][5][6].Other literature reviews and studies found that PA interventions could be effective in improving cognition among individuals with mild cognitive impairment (MCI) [7][8][9] and individuals with dementia [10][11][12].Although increasing PA has been proposed to facilitate healthy aging and suggested as a protective factor for cognitive decline and dementia, the economic implications of using PA interventions for reserving cognitive function or reducing the burden of dementia remain unclear.To inform decision-makers on resource allocation, it is necessary to consider whether the effectiveness (or benefit) of PA intervention outweighs its cost, given the preference of the population [13].In this study, we conduct a systematic review of cost-effectiveness analysis (CEA) of PA interventions for the prevention and management of cognitive decline and dementia.Our objective is to synthesize current evidence on the cost-effectiveness of PA interventions related to cognitive function, MCI, and dementia.We aim to inform future intervention design and policymaking for reducing the burden of dementia.Our systematic review will also shed light on future research directions in the economic evaluation of PA interventions for reducing cognitive decline and preventing and managing dementia.

Methods
We followed the systematic review's Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Checklist (Supplemental material) [14].The review protocol was registered in the PROSPERO system (CRD42022365200).

Data sources and search strategy
A literature search was conducted using PubMed/MED-LINE, Embase, and ScienceDirect, with publication dates ranging from January 2000 to July 2023.As guidelines for diagnosing and treating MCI and dementia have been evolving, we only included studies over the past two decades and excluded studies before 2000 so that findings from the included studies are relatively comparable [15,16].The search strategy was driven by a combination of subject-heading terms related to PA, cognitive function, MCI, dementia, and cost-effectiveness.The following search terms were used in PubMed: ((physical exercise) OR (physical activity) OR (leisure time) OR (sport) OR (muscle stretching exercise) OR (fitness) OR (physical activities) OR (exercise training) OR (physical training)) AND ((quality-adjusted) OR (cost-utility) OR (cost-effectiveness) OR (health economics) OR (economic evaluation)) AND ((mild cognitive impairment) OR (dementia) OR (Alzheimer's disease)).This approach was adapted accordingly to search on Embase and ScienceDirect and the completed search strategy was listed in Supplementary Table S1.We also searched the reference lists of the selected articles and other review articles to identify studies missing from the database search.The literature pool was exported to EndNote X9.

Study selection
The inclusion criteria of our population of interest included adults aged over 18 years old.We excluded studies that focus on children and animal models.Interventions of interest included a wide range of PA interventions aimed at preventing dementia or managing existing MCI or dementia.PA interventions included both exercise (i.e., a subset of PA planned and structured to improve or maintain physical fitness) and daily activities (e.g., occupational, sports, household, or other activities that result in energy expenditure) [3].For comparison, we included adults who received standard-of-care (i.e., standard information or treatments provided by neurologists or clinicians) or different frequency, duration, and intensity of PA interventions.All economic outcomes related to cost-effectiveness were included in the review.Cost-effectiveness was assessed using the incremental cost-effectiveness ratio (ICER), which measures the additional cost required to gain an additional unit of effectiveness or benefit.The ICER was calculated by comparing the costs and outcomes of the intervention with those of the comparator or standard of care [17].An intervention was considered cost-effective if the calculated ICER was lower than the willingness to pay threshold (i.e., the maximum amount society is willing to pay for an additional unit of health benefit) [17].Willingness to pay thresholds are not uniformly defined but are often related to the economic wealth of a society or country [17].We included studies that reported cost-effectiveness given a variety of health outcomes, such as quality-adjusted life years and other relevant outcomes reported in the included studies.Review articles, editorials, letters, research notes, conference abstracts, and protocol-only articles were excluded.Articles not written in English were excluded.Two co-authors (W.L. and K.R.K.) conducted a comprehensive search to identify relevant studies and removed duplicates using the automatic function in EndNote and manual hand search.During the initial screening phase, W.L. and K.R.K. independently assessed titles and abstracts to determine whether studies met the eligibility criteria.During the following screening phase, W.L. and K.R.K. independently examined the full text of the remaining studies to determine eligibility.Disagreements were resolved by consensus or in consultation with a third reviewer (Y.L.).All decisions were recorded through EndNote and Excel spreadsheets.

Data extraction
Two co-authors (W.L. and K.R.K.) independently extracted data from the selected articles using a standardized data extraction form and recorded results in Microsoft Excel.Discrepancies were resolved through discussion and consensus between the two co-authors or in consultation with a third reviewer (Y.L.).Extracted data included the following: (1) study identification (first author, year, title, country); (2) type of the CEA study, including trial-based CEA (performed alongside clinical trial), model-based CEA (developed using best available evidence from the literature), and "Hybrid" CEA ("Intrial" results extrapolated using modeling techniques) [18]; (3) other study design including time horizon (i.e., the time over which the costs and effects were measured), intervention, sample size, and inclusion criteria; (4) health economic properties (perspective, discount rate, sensitivity analysis, sources of cost data, measures of health outcome); ( 5) main health economic outcomes (e.g., ICER); and (6) conclusions.All the co-authors checked the extracted data and confirmed the decision.

Risk of bias assessment
As recommended by the Cochrane collaboration, two co-authors (W.L. and K.R.K.) independently assessed the risk of bias of the included studies using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Checklist and Consensus on Health Economic Criteria (CHEC)-list [19].The CHEER Checklist is a checklist with 24 items designed to focus on the reporting quality of economic evaluations [20].A study was considered to have good reporting quality if it reported 20-24 items, moderate quality if it reported 14-19 items, and low quality if it reported less than 14 items.The CHEClist is a checklist with 19 items designed to focus on the methodological quality of economic evaluations [21].A study was considered to have good methodological quality if it reported 15-19 items, moderate methodological quality if it reported 11-14 items, and low methodological quality if it reported less than 11 items.Disagreements were resolved by consensus or consultation with a third reviewer (Y.L.).

Equity, diversity, and inclusion statement
Our research team consisted of seven women and three men from a variety of disciplines, including two junior researchers.Our population of interest included all ages, genders, and race/ethnicities.

Results
Figure 1 shows the selection flow chart of the studies included in the current review.The literature search identified a total of 5188 studies.After removing duplicates and articles not written in English, two co-authors (W.L. and K.R.K.) independently screened titles and abstracts of the remaining 5103 studies and removed 4681 studies that were not relevant to the topic.Then, two co-authors (W.L. and K.R.K.) independently conducted full-text reviews for the remaining 344 studies and excluded 33 systematics reviews, 57 protocols, 22 studies with no cost-effectiveness analysis, and 221 studies that did not meet the inclusion criteria.In the 23 years, only 11 studies were identified and included in the analysis [22][23][24][25][26][27][28][29][30][31][32].

Study design and reporting quality
Table 1 summarizes the study design and reporting quality of the 11 identified studies.8 studies were conducted in a European setting, 2 studies were conducted in Canada, and 1 study was conducted in Japan.8 studies were trial-based CEA, 2 were model-based CEA, and 1 was a "hybrid" CEA.Among the trial-based CEA, the sample size ranged from 52 to 494 individuals [23-28, 31, 32].All trial-based CEA investigated the short-term effect ranging from 16 weeks to 12 months.The longterm effect was studied in the model-based and hybrid CEA.One model-based CEA simulated 1000 individuals for 10 years, and the other simulated 1000 individuals for a lifetime [22,29].The "hybrid" CEA extended the results of a 2-year clinical trial and generated projections of lifetime outcomes for a simulated cohort of 1,000,000 individuals [30].All studies had good reporting quality as they reported 20 or more CHEERS Checklist items.The limitations of study reporting quality included not identifying the study as an economic evaluation in the title (n= 2) [24,29], not reporting study perspective (n= 2) [24,29], no justification for not discounting costs or outcomes (n= 2) [23,24], not addressing uncertainties (n= 1) [24], no descriptions on assumptions in the analytic solution (n= 1) [27], incomplete information on costs (i.e., unit costs and price date) (n= 1) [23], and no funding source information (n= 1) [32].All studies had good methodological quality as they scored 15-19 using the CHEC-list.The limitations of study methodology quality included not clearly describing competing alternatives [22,29], did not choose the most appropriate perspective [24,29], costs were not measured appropriately in physical units [23,25], did not perform an incremental analysis of costs and outcomes of alternatives [24], not discounting future costs and outcomes [23,24], did not conduct sensitivity analysis [24], and not including discussion of generalizability of the results [26].

Population characteristics
Among the 11 studies, 5 studies evaluated the cost-effectiveness of PA interventions as management strategies for patients already diagnosed with dementia, while the remaining studies evaluated PA interventions as prevention strategies in individuals with either MCI (n = 3 studies) or no dementia/MCI diagnosis (n = 3 studies).
Each study incorporated different inclusion criteria for participant recruitment.Examples of this include the studies by Sopina et [30].

Health economic properties
Table 2 summarizes the identified studies' health economic properties and cost-effectiveness outcomes.The selected studies were categorized as representing the healthcare sector perspective (i.e., include formal medical costs borne by third-party payers or paid for outof-pocket by patients) or societal perspective (i.e., also include time costs and effects on future productivity as well as relevant non-health-related impacts in other sectors) [33].Three studies were conducted from a healthcare sector perspective [26,27,32], 3 studies were conducted from a societal perspective [23,28,30], and 3 studies used both healthcare sector perspective and societal perspectives [22,25,31].Discounting reflects the loss in economic value that occurs when there is a delay in realizing a benefit or incurring a cost [34].Five studies reported that discounting was not needed due to the short time horizon used for the analysis, and 4 studies reported specific discount rates.Ten out of the 11 studies also reported sensitivity analyses to address uncertainties (i.e., changes in the results given changes in the input values).The most frequently used measure of health outcome was quality-adjusted life year (n = 8), which was measured by health-related quality of life using European

Cost-effectiveness results
Table 3 summarizes the results of the included studies.Eight of the 11 included studies found that PA interventions were cost-effective regarding at least one health outcome [22,[24][25][26][27][28][29][30], whereas the other 3 studies reported a lack of cost-effectiveness [23,31,32].Three studies evaluated the PA interventions implemented in individuals with no existing MCI or dementia and found PA interventions were cost-effective in increasing life year and quality-adjusted life year in the long term.Baal et al. investigated the relationship between increasing PA levels, the incidence of dementia, mortality, and the use of health care and social care in individuals with no dementia [29].In this study, a simulation model was used to project various scenarios with different assumptions on increasing PA by one level among the English population aged 40-65.Preventing dementia by increasing PA was projected to increase life expectancy and decrease overall spending on health and social care, even after adjusting for additional spending during the life years gained [29]. Kato et al. estimated the cost-effectiveness of the combined physical and cognitive program designed to prevent community-dwelling healthy adults aged 65 years old from developing dementia.This study used a simulation model and found that the combined physical and cognitive program was not only effective in increasing quality-adjusted life years but also cost-saving during the 10-year period [22].Wimo       terms of improving physical performance and qualityadjusted life year [28].
In the 5 studies evaluated the PA interventions that implemented in individuals with dementia, we found mixed results regarding the cost-effectiveness of PA interventions.D' Amico et al. studied a dyadic PA program, which delivered PA in the form of a 12-week individually tailored walking program lasting for 20-30 min daily and designed to become progressively more intensive [25].D' Amico et al. concluded that this dyadic PA program may be cost-effective for improving behavioral and psychological symptoms of dementia but not for improving quality-adjusted life year [25].Pitkälä et al. found that a 12-month home-delivered, personalized PA program delayed the expected deterioration in physical function among individuals with dementia without increasing total health and service costs of standard-ofcare [24].However, in the same study, Pitkälä et al. also found that the paralleled 12-month group-based PA intervention with nondistinctive PA sessions was not cost-effective compared with standard-of-care [24].Khan et al. found that a 12-month moderate-to-high-intensity aerobic and strength training PA program was not costeffective compared with standard-of-care for individuals with mild to moderate dementia [31].The results showed that this PA program did not significantly affect cognitive outcomes or quality-adjusted life years.Similarly, Sopina et al. found that a 16-week moderate-to-high intensity aerobic training PA program was not cost-effective in terms of participant-reported and proxy-reported healthrelated quality of life [32].Van Santen et al. also studied the cost-effectiveness of "exergaming, " an innovative form of PA that integrates physical activity with cognitive stimulation in a gaming environment, and found that exergaming was not cost-effective compared to standardof-care in improving quality-adjusted life year, physical function, and mobility for individuals with dementia [23].
Two studies evaluated a 12-week PA intervention but had different outcomes.One of the studies targeted individuals with MCI, while the other targeted individuals with dementia.Eckert et al. found that a 12-week homebased tailored exercise program was cost-effective in improving physical performance and quality-adjusted life year among individuals with MCI at a relatively low willingness-to-pay threshold (i.e., equate to £18,000/ quality-adjusted life year) [28].In comparison, delivery of a PA program with the same duration of time was found to be less cost-effective among individuals with dementia.D' Amico et al. found that walking at least 20 min daily for 12 weeks for individuals with dementia was costeffective when considering behavioral and psychological symptoms of dementia.However, it was not cost-effective in terms of quality-adjusted life year, even considering a high willingness-to-pay threshold (i.e., £30,000/qualityadjusted life year) [25].

Main findings
This review provides an expanded discussion on the effectiveness of PA interventions [9][10][11] by evaluating the economic impact and cost-effectiveness of these interventions as they relate to populations with or without existing MCI or dementia.Our results indicate that an intensive PA program implemented before MCI diagnosis may be cost-effective in the long term.Also, PA programs may be cost-effective in the short term among individuals with MCI.However, we did not find sufficient evidence on the long-term cost-effectiveness of PA interventions in individuals with MCI or existing dementia.
In evaluating evidence supporting PA intervention as a prevention strategy in individuals with no existing dementia diagnosis, it is important to note that two studies may provide an overly conservative estimate of the health benefit of the PA interventions in individuals with MCI, given that participants may also have experienced a positive health benefit from the control interventions that included PA [26,28].Therefore, the evidence on the short-term cost-effectiveness of prevention strategies for dementia among individuals with MCI is likely underestimated.Our review found that even a short PA program (i.e., 12-week PA intervention) showed cost-effectiveness for individuals with MCI.To be cost-effective, PA interventions may need to be implemented earlier in life before dementia diagnosis.Short-term PA interventions that begin after the onset of dementia are likely insufficient to provide beneficial effects when considering economic sustainability.Increasing PA levels among physically inactive adults was projected to be cost-saving as a prevention strategy for dementia over the simulated lifetime, whereas the cost of PA intervention was not considered in the study [29].Also, the dominance (i.e., both clinically superior and cost-saving) demonstrated in the two multidomain programs (i.e., the combined physical and cognitive program and the multidomain lifestyle intervention program) that addressed several dementia risk factors was intriguing [22,30].Although it was not possible to isolate the benefits attributable to the PA domain, given the complex relationship between interrelated factors that protect against cognitive decline and dementia, interventions addressing many risk factors at once might offer the best prevention strategy for individuals who already at higher risk for cognitive impairment or dementia.
In evaluating evidence on whether PA interventions are cost-effective in patients with current dementia, we found mixed results.The lack of cost-effectiveness documented in some studies may have been due to limited clinical benefits of the PA intervention, lack of compliance with the intervention among individuals with dementia, or the intervention being too costly.For example, it is possible that the clinical benefits in physical function and mobility did not translate into improvements in functional activities that may have been required to demonstrate improved cognitive outcomes.The two studies reporting the cost-effectiveness of PA interventions in current dementia patients showed the importance of participant compliance and intervention cost in determining outcomes.One study evaluated a 12-week walking program, and the other evaluated a 12-month home-based exercise program [24,25].Even with limited clinical benefits, the low financial investment of implementing a walking program may substantially contribute to the cost-effectiveness of the intervention [25].Additionally, adherence was reported to be exceptionally high in the 12-month homebased PA program with personalized training, ensuring high levels of activity that lasted for an intense and sufficient duration of PA [24].Thus, it is reasonable to expect that low cost is conducive to high compliance, which leads to higher benefits the participant reaps from the intervention.Although it was not possible to extrapolate these findings to other programs due to the substantial heterogeneity across studies and PA programs, it is possible that other 12-month PA programs (e.g., the 12-month PA program with aerobic and resistance exercise classes reviewed in this study [31]) would have been more costeffective if a higher adherence was achieved.

Current gaps and future directions
Among the reviewed studies, only the model-based and "hybrid" CEA assess the long-term cost-effectiveness of PA interventions.In contrast, other studies used trialbased data to evaluate short-term cost-effectiveness.Clinical trials are often limited by finite (and potentially short) follow-up duration, and pure trial-based CEA may not have sufficient data to report long-term (e.g., lifetime) costs and the consequences of PA interventions [18].Previous research demonstrates that PA, such as resistance training in older adults, has long-term health benefits and economic impact that a longer time horizon would ideally capture [26,35].Therefore, it is possible that the time horizons of the reviewed trial-based CEA were not sufficiently long to capture all the pertinent clinical and economic ramifications of the strategies under study, and the estimation of cost-effectiveness may be biased [18].Moreover, pure trial-based analyses tend not to incorporate data from external sources, exposing the results to potentially greater uncertainty than if evidence from other prospective studies or trials was considered.Further investigation on the long-term cost-effectiveness of PA intervention, both as a prevention strategy and a management strategy for dementia, is warranted.When clinical trials looking at long-term costs and consequences are not feasible or complete, simulation models can be applied to estimate likely cost-effectiveness outcomes by incorporating data from a wide variety of sources as inputs.In particular, "hybrid" studies can address the limitations of trial-based CEA-the issue of truncated follow-up-by extending the results of the study through time, generating a range of plausible projections of longer-term outcomes [18,30].Moreover, as most of the identified studies were confined to a European setting, CEA conducted in other settings is warranted.

Limitations
The interpretation of this systematic review may be limited by the nature of narrative synthesis.Additionally, a meta-analysis was not feasible due to the substantial heterogeneity among the small number of identified studies.First, there were multiple sources of heterogeneity, such as population variations and measurement of CEA outcomes.Second, the studies were conducted from different perspectives in different settings with various willingness-to-pay thresholds.Third, there is structural and methodological heterogeneity between model-based and trial-based CEA.While the sample size is often used to weigh the impact of each study included in a metaanalysis, it needs to be clarified how to assign weights to model-based studies [36].

Conclusions
Our review identifies population traits and intervention characteristics that trend toward the cost-effectiveness of PA interventions to prevent and manage cognitive decline and dementia.PA interventions administrated in middle-aged or older individuals prior to MCI or dementia diagnosis were generally cost-effective in the long term.We also found short-term cost-effectiveness of PA interventions among individuals with MCI, who may represent a target population in urgent need of cost-effective lifestyle-modified interventions, given that they have not yet crossed the dementia threshold.
Future studies should further explore the long-term cost-effectiveness of PA interventions among individuals with MCI, who are at increased risk for eventual dementia diagnosis.Also, more CEA should be conducted in settings other than Europe.Although we found mixed results on the cost-effectiveness of PA interventions for individuals with existing dementia, there may still be a benefit of implementing PA strategies in this population because it could also provide cognitive benefits, as has been shown for other chronic conditions (e.g., cardiovascular diseases) [37,38].A lower cost burden and implementation strategies to improve adherence might be key factors in achieving the cost-effectiveness of PA interventions in individuals with dementia.
• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Fig. 1
Fig. 1 PRISMA flowchart.PRISMA indicates for Preferred Reporting Items for Systematic Reviews and Meta-Analyses Quality of Life 5 Dimensions 3 Level Version (n = 6), Dementia Quality of Life instrument-proxy (n = 1), or both European Quality of Life 5 Dimensions 5 Level Version and European Quality of Life visual analog scales (n = 1).Other measures of health outcomes included behavioral and psychological symptoms of dementia measured by the Neuropsychiatric Inventory (n = 1), executive cognitive function measured by the Stroop Test (n = 1), physical performance measured by short physical performance battery only (n = 2) or with the functional independence measure (n = 1), life years (n = 1), and Alzheimer's Disease Assessment Scale (n = 1).
et al. estimated the potential cost-effectiveness of the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) program [30].The FINGER program was the first randomized control trial to show a statistically significant beneficial effect on cognition of a multidomain lifestyle intervention program (including diet, PA, and cognitive training) for older adults at risk of developing dementia.Wimo et al. used a simulation model and projected that a multidomain lifestyle intervention program was cost-saving and clinically superior in preventing dementia than standard-of-care.Three studies evaluated the PA interventions that implemented in individuals with MCI and found PA interventions were cost-effective in improving cognitive function, physical performance, and quality-adjusted life year in the short term.Davis et al. found that either a 6-month aerobic or resistance training was more costeffective in improving executive cognitive function than other exercises focused on improving balance and muscle tone in older adults with MCI [26].Davis et al. also found that a 6-month aerobic training PA program was more cost-effective in maintaining the health-related quality of life compared to standard-of-care in individuals with subcortical ischemic vascular cognitive impairment, a subtype of MCI [27].Eckert et al. found that a 12-week home-based personalized PA program (that included exercises for balance and strength as well as walking) was more cost-effective than a flexibility training program in Studies were grouped by the population and ordered by the severity of dementia." + " means cost-effective and "-" means not cost-effective.HRQOL indicates for Health-related quality of life MCI mild cognitive impairment, FINGER The Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability, BPSD behavioral and psychological symptoms of dementia, QALY quality-adjusted life years Study Population Intervention Cost-effectiveness Conclusion Baal et al. [29] (2016) No MCI or dementia Increase of physical activity level by 1 level (out of total 5 levels) vs. no increase Life years:+ If prevention is targeted at the physically inactive by increasing physical activity level, savings in dementiarelated costs outweigh the additional spending in life years gained Wimo et al. [30] (2022) No MCI or dementia 2-year multidomain program (nutritional counseling, multicomponent exercise, and cognitive training) vs. standard-of-care QALY: + The model provides support that programs like FINGER have the potential to be cost-effective in preventing dementia Kato et al. [22] (2022) No MCI or dementia 90-min weekly combined physical and cognitive session QALY: + A program targeting community-dwelling healthy adults aged 65 years old could be cost-effective Davis et al. [26] (2013) MCI 6-month resistance training or aerobic training vs. balance/tone classes (control) Executive cognitive function:+ Resistance training and aerobic training result in healthcare cost saving and are more effective than balance/ tone classes in older adults with MCI.Resistance training is a promising strategy to alter the trajectory of cognitive decline in seniors with MCI Davis et al. [27] (2017) MCI 6-month aerobic training + standard-of-care + education vs. standard-of-care + education QALY:+ Aerobic training represents an attractive and potentially cost-effective strategy for older adults with mild subcortical vascular cognitive impairment Eckert et al. [28] (2021) MCI 12-week home-based exercise (walking, balance, strength training) vs. flexibility training Physical performance and QALY:+ The home-based exercise intervention demonstrated high probability of cost-effectiveness in terms of improved physical performance in older adults with MCI following discharge from ward rehabilitation.The intervention had high probability of being costeffective in terms of QALY when using a high willingness to pay threshold Khan et al. [31] (2019) Dementia 12-month aerobic and resistance exercise classes vs. standard-of-care Cognitive outcomes and QALY:-Exercise is not cost-effective in slowing cognitive impairment in people with mild to moderate dementia Pitkälä et al. [24] (2013) Dementia 12-month group-based exercise or home-based exercise vs. standard-of-care Physical functioning and mobility: + (home-based exercise) -(group-based exercise) An intensive and long-term exercise program administrated at patient's home had beneficial effects on the physical functioning of patients with Alzheimer's disease without increasing the total costs of health and social services or causing any significant adverse effects Sopina et al. [32] (2017) Dementia 16-week aerobic exercises (on bicycle, cross trainer, and treadmill) vs. usual treatment QALY:-The exercise intervention is unlikely to be cost-effective within the commonly applied threshold values Van Santen et al. [23] (2021) Dementia 6-month exergaming (interacting cycling) + regular activity program (music listening, singing, arts and crafts, cooking, gymnastics, and outdoor walking) vs. regular activity program QALY, physical activity function and mobility:-Exergaming by participants with dementia in daycare center was nor cost-effective compared to care as usual for our primary outcome measures: QALYs, physical activity and mobility D' Amico et al. [25] (2016) Dementia 12-week daily walking program vs. standard-of-care BPSD: + QALY:-For individuals with dementia, exercise could potentially be a cost-effective intervention for outcomes measured by BPSD, but not when measured by QALYs

Table 1
Study designs of the identified studies

Table 1 (continued) Study Country Design Time horizon Sample size Intervention Inclusion criteria CHEERS items CHEC-list
[24,32] their sample with current dementia;[24,32]and the Davis et al. study only considered individuals diagnosed with mild subcortical vascular cognitive impairment as their criteria for a sample with MCI [27].The Baal et al. study included the whole population of England as a sample with no dementia [29].Wimo et al. targeted older adults at risk for dementia as determined by a risk score based on age, sex, education, blood pressure, serum cholesterol, obesity, and physical activity as predictors of increased dementia likelihood Studies were grouped by the study design and then ordered alphabetically within each group.Time horizon indicates the time over which the costs and effects are measured.CHEERS indicates for Consolidated Health Economic Evaluation Reporting Standards; CHEC-list Consensus on Health Economic Criteria list; UK United Kingdom, CEA cost-effectiveness analysis, MCI mild cognitive impairment, ICD-10 DCR International Statistical Classification of Diseases and Related Health Problems, 10th revision, Diagnostic Criteria for Research (1992), BPSD behavioral and psychological symptoms of dementia, RT resistance training, AT aerobic training, HE home-based exercise, GE group-based exercise, MoCA Montreal Cognitive Assessment, MMSE Mini Mental State Examination score, AD Alzheimer's Disease, DSMIV Diagnostic and Statistical Manual, 4th Edition, The CAIDE risk score provides an estimate of dementia risk based on several risk factors (age, sex, education, systolic blood pressure, total serum cholesterol, obesity, physical inactivity)

Table 2
Health economic properties and analytical details of the identified studies

Table 2 (continued) Study Perspective Discount rate Sensitivity analysis Data source of cost Measures of health outcome Main economic outcome and result
Studies were grouped by the study design and then ordered alphabetically within each group (i.e., aligned with Table1).Dominant means the study group was less costly and more effective than the control group.Not applied means the authors reported that discounting was not appropriate due to the short time horizon used for the analysis.NA means the information is not available in the study.Strooptest is a test of selective attention and conflict resolution PSA probabilistic sensitivity analysis, DSA deterministic sensitivity analysis, ICER incremental cost-effectiveness ratio, QALY quality-adjusted life-year, EQ-5D-3L European Quality of Life 5 Dimensions 3 Level Version, EQ-5D-5L European Quality of Life 5 Dimensions 5 Level Version, EQ-VAS European Quality of Life visual analog scales, NPI Neuropsychiatric Inventory, WTP willingness-to-pay, DEMQOL-proxy Dementia Quality of Life Instrument-proxy, CSRI Client Service Receipt Inventory, BPSD behavioral and psychological symptoms of dementia, RT resistance training, AT aerobic training, BAT balance and tone classes, HE home-based exercise, GE group-based exercise, SIVCI subcortical ischemic vascular cognitive impairment, NHS UK National Health Service, FIM Functional Independence Measure, SPPB Short Physical Performance Battery, ADAS-Cog Alzheimer's Disease Assessment Scale-Cognitive Subscale, FINGER Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability program

Table 3
Summary of cost-effectiveness findings of the reviewed studies