Economic Burden of Dementia Caused by Cardiovascular Disease in Australia

Abstract

Background:

The established link between cardiovascular disease (CVD) and dementia may provide new insights into dementia prevention.

Objective:

It aims to quantify the burden of dementia attributable to people with CVD.

Methods:

A Markov microsimulation model was developed to simulate the lifetime cost and quality-adjusted life-years (QALYs) related to people with and without CVD in Australia. A de-novo systematic review was undertaken to identify all evidence around the association between CVD [i.e., stroke, myocardial infarction (MI), atrial fibrillation (AF), and heart failure (HF)] and the risk of developing dementia. Incremental costs and QALY losses were estimated for people by type of CVD compared to the general Australian population without CVD.

Results:

Of the comprehensive literature search, 19 observational studies were included in the qualitative synthesis. Patients who had CVD incurred both higher healthcare costs over their lifetime (ranging from $73,131 for patients with AF to $127,396 for patients with HF) and fewer QALYs gains (from –1.099 for patients with MI to –5.163 for patients with stroke), compared to people who did not have CVD. The total incremental economic burden of dementia from patients aged 65 years and over with CVD was $6.45 billion (stroke), $11.89 billion (AF), $17.57 billion (MI), or $7.95 billion (HF) over their remaining life expectancy.

Conclusion:

The results highlighted the importance of CVD prevention to reduce the CVD burden and decrease the prevalence of dementia. Interventions that target patients with dementia risk factors like CVD may prove to be effective and cost-effective strategies.

Keywords

Burden of disease cardiovascular disease dementia meta-analysis systematic review

INTRODUCTION

Dementia is a major health issue causing disability and dependency in the aged population and is associated with significant personal, social, and economic burden [1, 2]. Dementia can be devastating not only for people who experience it, but also for their families and carers [3]. In Australia, it was the second-leading cause of death for Australians in 2018 and is the first-leading cause of death among persons 85 years and over [4]. Based on projections of population aging and growth, it was estimated that 399,800 Australians had dementia in 2020 of which Alzheimer’s disease and vascular dementia are the two dominant disease entities accounting for approximately 85% of all dementia cases [5]. Total direct health and care expenditure on dementia was at least $4.8 billion and $3.4 billion respectively in 2017, and is projected to reach $24.1 billion by 2056 [6]. Despite substantial research effort [3], there is currently no cure for dementia. Available medications can alleviate symptoms and improve quality of life in some people, but they cannot prevent the progression of the disease. While research into treating or curing dementia is critical, it is equally imperative to investigate strategies for preventing dementia by modifying its risk factors, such as certain types of cardiovascular disease (CVD) [7].

There is strong evidence establishing the association between certain CVD conditions and an increased risk of developing dementia [8]. Myocardial infarction (MI) [9], heart failure (HF) [10], atrial fibrillation (AF) [11], and stroke [12, 13] are known risk factors contributing to the development of dementia. Mechanisms that may relate CVD as a risk factor for dementia include chronic hypoperfusion of the brain that is caused by low cardiac output (e.g., common in patients with post-MI, AF, HF) that contributes to impaired vascular autoregulation, and white matter injury; neurohormonal activation that may cause inflammation and cerebral microvascular dysfunction; and subclinical cerebral infarcts (e.g., patients with stroke) that adversely impact the frontal cortex, basal ganglia, and thalamus which are regions critical for executive function and memory [8 , 15]. Preliminary evidence from the UK indicate that early dementia intervention and health services can lead to a reduction in care home/institutional bed days and overall societal costs, as well as be deemed to be cost effective in reducing the cost to the society [16]. Therefore, targeted preventive interventions in patients with certain CVD conditions may potentially contribute to the decreased burden of dementia.

The strong link between CVD and development dementia often prompts the question about the economic consequences of such an association. Research examining the economic impact of dementia in people with CVD is scarce. Ascertaining such economic impacts could assist policy makers to better formulate and prioritize national health policy for both dementia and CVD. This study aims to conduct a model-based simulation study to quantify the overall economic burden of dementia caused by patients with CVD to highlight the importance of holistic healthcare for the CVD population.

METHODS

Modelled population

The modelled population comprised Australian patients aged 65 years and over, with a history of at least one CVD condition including MI, HF, AF, or stroke. Their counterparts with no history of these CVD conditions were employed as controls.

Model structure

A Markov microsimulation model (Tree Age Pro 2020, Tree Age, Williamstown, MA) was developed to simulate a hypothetical cohort of 10,000 patients over the 25-year time horizon with a yearly cycle length. The model structure is illustrated in Fig. 1. The model structure was informed by two earlier studies [17, 18].

Fig. 1

Markov model structure for the long-term simulation.

All patients enter the model from a no dementia state, and then in each cycle may transit to any of the following states: no dementia, mild dementia, moderate dementia, severe dementia or dead. The severity of dementia was determined using the Clinical Dementia Rating scale (a 5-point scale used to characterize cognitive and functional performance in dementia) [19]. Living arrangement was captured according to the dementia stages (i.e., mild, moderate, and severe), and change in living arrangements was also simulated (see Fig. 1). Given dementia is a non-reversible condition with no cure as yet, the model assumed that once patients progressed to a more advanced stage, they will remain in that health state until progression to the next stage (e.g., individuals can only move from a normal cognitive state to mild cognitive impairment state, but not in the reverse direction). The probability of dementia progression was varied, supported by evidence from a large-scale population study that showed that the condition of dementia patients could deteriorate quite quickly between stages. Epidemiological modelling showed a jump stage in dementia within one year cycle (e.g., from mild to severe without experiencing a moderate stage) [20].

It was assumed that people with no dementia remained in the community while people who had developed dementia had a probability of receiving either community or institutional care. However, once people with dementia transitioned to institutional care, they cannot move back to community care due to the irreversible course of the disease. Patients with CVD, regardless of dementia status and living arrangement, were assumed to require general medical management of their CVD conditions during each cycle.

The Markov model structure for the long-term simulation is provided in Fig. 1.

Model inputs

Literature search

A comprehensive systematic review was conducted in EBCOHOST Platform (Academic Search Complete, CINAHL Complete, Health Source: Nursing/Academic Edition, MEDLINE Complete) and Embase to identify evidence (e.g., systematic review of observation studies or single observational study/population studies) around the increased risk of dementia relating to certain CVD conditions. Complete search terms could be found in the Supplementary Table 1.

Systematic reviews with meta-analysis, or single studies that reported on the time independent indices including risk ratio and odds ratio between CVD and incidence of dementia were included (i.e., studies deriving hazard ratios from Cox-proportional hazard regression model were excluded). We only selected studies that 1) included cohort of patients who had not developed cognitive impairment and/or dementia at baseline; and 2) published in English. Similar criteria were applied for reviews; however, additional criteria, such as number of included studies (reviews with greater number of studies were considered better), pooled sample size (reviews with larger pooled sample size were given more weight), dementia type (a study that reported on all dementia types was given a better overall estimate than one that reported certain subtypes), and magnitude of heterogeneity was utilized (i.e., I² < 75% was considered as may indicate heterogeneity, I² > 75% was defined as considerable heterogeneity) [21]. The detailed inclusion and exclusion criteria are summarized in Supplementary Table 2.

Cochrane systematic review guidelines were followed [22]. As the systematic review aimed to inform the model structure and model parameters, we did not register our review in PROSPERO. Stata SE16 was employed for calculating the risk ratio (RR) from reported odds ratio (OR) with a number of different confidence intervals (CIs) when applicable, using mean function with common effect inverse variance model, 95% CI. Subgroup analysis was performed where considerable heterogeneity was detected. RR with 95% CI of each exposure CVD condition was then employed to inform the long-term simulation. RR was computed using the formula of transformation provided in the Cochrane handbook for systematic review [23].

Studies that did not meet the inclusion criteria were excluded (Supplementary Tables). The reference lists of excluded reviews were searched to identify any additional relevant studies.

All identified studies were screened by one author (DN) as follows: 1) title and abstracts were assessed for potential study relevance; 2) full text of all potentially relevant articles were assessed against the inclusion/exclusion criteria. Data extraction was conducted by DN for all included studies, using predefined data fields (authors, year, title, study type, population, context (CVD/CVD risk factor), reported measure/analysis and outcome). Included studies were checked by the second reviewer (LG).

The Cochrane data base of systematic reviews, PubMed, Google and Google scholar, and the grey literature were also searched for other model parameters including transitional probabilities (e.g.,, probability of mild dementia progressing to moderate dementia), utility weights (e.g., strength of preference for living without CVD, with CVD/dementia), and cost (e.g., costs of management for CVD, dementia, institutional care, etc.).

Transition probabilities

Probabilities related to dementia progression among simulated dementia health states were identified from a longitudinal study (i.e., the National Alzheimer’s Coordinating Centre (NACC) study in USA consisted of a population of 7,736 patients from 32 different NACCs) and the severity of dementia was defined by Clinical Dementia Rating score [24].

All patients with dementia had a probability of remaining in the community or moving into institutional care. The transition probability to move into institutional care was estimated based on 2012 Australian Institute for Health and Wellbeing (AIHW) dementia report [2]. The mortality rate for patients with dementia was calculated using the Australian lifetable adjusted by increased mortality due to dementia (i.e., hazard ratio, HR) [25] (Supplementary Tables).

Utility weights

Utility weight associated with the general Australian population, CVD, and dementia respectively were sourced from published literature. CVD type-specific (i.e., MI, HF, or stroke) utility weights were applied for patients who had not developed dementia in the model, while dementia-related utility weights were applied to patients who developed dementia in accordance with severity (mild, moderate, and severe). The utility decrements associated with CVD and dementia were not assumed to be additive, and the lower utility weight was applied when patients with CVD developed dementia.

Cost

All costs were valued in Australian dollars for the 2019 reference year. Costs were identified from relevant literature for a limited societal perspective. A half-cycle correction was applied to adjust for both costs and benefits. Costs and benefits were discounted at a rate of 3% to be consistent with available literature that reported on dementia prevention [26, 27]. The corresponding health price index (HPI) was used to inflate the costs to the year 2018 [28, 29]. Due to the absence of HPI for the 2018–19 financial year, consumer price indices (CPI) from the health sector were used as a proxy [28].

All relevant costs from the healthcare system perspective were incorporated in the model including costs of CVD management, and dementia care (e.g., health care utilization including out of hospital attendants, medical services, pharmaceuticals, and hospital related admissions). The health care services cost was derived from relevant literature using Aged Care Funding Instrument (ACFI) approach. ACFI is an assessment that provides a level of sensitivity to the core individual care needs in the domains of daily living, behavior, and complex health care of dementia population [30]. Therefore, the health services cost employed in this model is not affected by facility-level factors such as type of provider (e.g., public or private).

Costs related to institutional care were considered from a limited societal perspective given other costs for informal care at home and travel (to receive healthcare) were not incorporated. For patients who did not move into institutional care and were assumed to live in their own homes, no residential cost was assigned.

Key transition probabilities, costs, and utility weights are shown in Table 1.

Table 1

Transition probabilities, costs and utility weights for the simulation model

Variable	Base case value	Range/SD	Source
Relative risks
Develop dementia in non-CVD	1
Develop dementia in Stroke	1.63	1.43–1.86	[33 –45]
Develop dementia in AF	1.29	1.02–1.62	[36]
Develop dementia in MI	1.34	1.06–1.70	[36]
Develop dementia in HF	1.60	1.13–2.25	[46]
Transition probability of dementia stage
Dementia free to mild	0.01693	[68]
Mild to moderate	0.158	0.139–0.176	[24]
Mild to severe	0.013	0.009–0.017	[24]
Moderate to severe	0.214	0.186–0.243	[24]
Hazard ratios
Mortality with no dementia	1
Mortality with mild dementia	2.92	1.55–2.14	[25]
Mortality with moderate dementia	3.85	2.94–5.05	[25]
Mortality with severe dementia	9.52	6.6–13.74	[25]
Living arrangement
Transfer to institutional care_mild dementia	0.012	0.003–0.028	[24]
Transfer to institutional care_moderate/severe dementia	0.63		[25]
Utility weights
No dementia (65+)	0.87	0.16	[69]
No dementia (75+)	0.83	0.16	[69]
Mild dementia (65+)	0.59	0.26	[70]
Mild dementia (75+)	0.59	0.24	[70]
Moderate dementia (65+)	0.38	0.28	[70]
Moderate dementia (75+)	0.38	0.25	[70]
Severe dementia (65+)	0.22	0.28	[70]
Severe dementia (75+)	0.26	0.29	[70]
Stroke	0.47	0.36	[71]
Atrial fibrillation	0.7381	0.6–1	[72]
Myocardial infarction	0.80	0.23	[73]
Heart failure	0.69	0.2	[74]
Costs
Institutional care (dementia)	A$89807	80826–98787	[75]
Disease management costs
Stroke	A$2969	2672–3266	[76]
Atrial fibrillation	A$1672	1505–1839	[72]
Heart failure	A$5087	4578–5596	[74, 77]
Myocardial infarction	A$3424	3081–3766	[78]
Dementia	A$6363	5726–6999	[75]

AF, atrial fibrillation; HF, heart failure; MI, myocardial infarction; CVD, cardiovascular disease; SD, standard deviation.

Long-term simulation

The long-term economic costs and QALY gains related to patients with certain types of CVD and the corresponding general Australian population without those CVD conditions were estimated. The incremental economic cost and loss in QALYs due to the presence of CVD were reported by type of CVD in the base case results.

Sensitivity analysis

Deterministic sensitivity analyzes were performed to test the robustness of base case results. A series of one-way sensitivity analyzes was undertaken by varying key model parameters within a plausible range, either by drawing on inputs from published literature reported as means and standard deviations (Mean (SD)) or by assuming a range of ±10%. Tornado diagrams were used to graphically summarize the incremental costs from the deterministic sensitivity analyzes.

Similarly, probabilistic sensitivity analyzes were run to test the overall impact of uncertainty in the model by defining distributions for the key parameters (i.e., transition probabilities, utilities, and costs). A total of 5,000 iterations (i.e., second-order Monte Carlo simulation) were run to establish a mean and 95% CI for the incremental cost and reduction in QALYs. The difference in costs and QALY loss were presented on the cost-effectiveness plane. Distribution of parameters examined are summarized in Supplementary Table 3.

Model validation

Extensive literature searches in key databases such as PubMed and Google Scholar were conducted to identify relevant literature to validate the modelled outcome. In particular, population studies that reported on dementia prevalence, QALY gains in patients with dementia or CVD conditions (such as stroke, MI, HF, and AF), life expectancy in the dementia population, and dementia care cost were searched.

Ascertaining the national impact

The prevalence of each type of CVD in Australia was used to estimate the total national impact. Adapting approaches used in the AIHW 2012 dementia report and Brown et al. 2017, the AF prevalence rate by age reported for the Australian population study and the Australian Bureau of Statistics (ABS) 2017-2018 National Health Survey long term condition data [31, 32] were used to estimate the national impact of dementia caused by people with CVD. The prevalence of CVD by type is summarized in Table 2.

Table 2

Prevalence of certain types of CVD in Australia, 2020

Age	ABS	AF		Stroke		HF		MI
Group	population	Prevalence	Cohort	Prevalence	Cohort	Prevalence	Cohort	Prevalence	Cohort
(2020)	(2020)	rate (%)		rate (%)		rate (%)		rate (%)
65–69	1,258,176	4.04	50,830	2.1	26,422	1.5	18,873	5.6	70,458
70–74	1,104,991	6.13	67,736	2.1	23,205	1.5	16,575	5.6	61,879
75–79	773,742	9.57	74,047	2.9	22,439	2.8	21,665	9.0	69,637
80–84	528,443	13.79	72,872	2.9	15,325	2.8	14,796	9.0	47,560
85+	527,530	17.64	93,056	3.5	18,464	3.6	18,991	9.9	52,225

AF, atrial fibrillation; HF, heart failure; MI, myocardial infarction; ABS, Australian Bureau of Statistics.

RESULTS

Systematic review

A PRISMA statement for standard reporting was employed to report the review process and findings (Supplementary Figure 1). Of a total of 4,145 titles identified, 3,089 titles were screened after removing duplicates, 69 articles were selected for full text screening and 17 additional articles (seven additional reviews and 10 individual studies were identified from reviews’ reference lists) were added from a citation search. None of the identified systematic reviews met the inclusion criteria, while 19 observational studies were included in the qualitative synthesis. Of the latter, five studies reported on multiple CVD conditions. The number of studies for each condition was stroke (n = 14), MI (n = 2), AF (n = 5), and HF (n = 4).

Quantitative synthesis of 13 studies that reported on the association between stroke and dementia was conducted (Supplementary Figure 2). Due to inconsistency in reporting (e.g., no CI reported, included studies had different outcome measure, etc.), we excluded Bunch et al. [33] from the quantitative synthesis for stroke; meta-analysis was not performed for any other CVD conditions. Characteristics of these studies are summarized in Supplementary Table 4.

RR with 95% CI of each exposure CVD condition showed increased risk of dementia due to the presence of the CVD condition compared to people without such exposure: RR(stroke) 1.63 (1.43–1.86) (meta-analysis of 13 studies from seven countries) [34 –46]; RR(MI) 1.34 (1.06–1.70) [37]; RR(AF) 1.29 (1.02–1.62) [37]; RR(HF) 1.6(1.13–2.25) [47].

Long-term simulation

Patients who had CVD as an underlying risk factor incurred both higher healthcare costs (ranging from $73,131 for patients with AF to $127,396 for patients with HF) and fewer QALYs (varying from –1.099 for patients with MI to –5.163 for patients with stroke), compared to people who did not have CVD (Supplementary Table 5). It was estimated that an increased 1,512 CVD patients per 10,000 population developed dementia along the remaining course of their lives (Table 3).

Table 3

Results of long-term simulation

Condition	Cost	Incr. Cost	QALYs	Loss in	% of QALY	LYs	Loss	% of LY loss	Total health	Total QALY	No. of dementia	No. of cases
				QALYs	loss due to CVD		in LYs	due to CVD	care cost	loss	cases*	in Inst. Care*
Base case
No CVD	$39,983	–	11.252	–		13.72	–		–	–	0.28	0.13
Stroke	$100,877	$60,895	6.089	5.163	45.9%	13.08	0.639	4.7%	$6,445,921,268	546,521	0.41	0.21
AF	$73,131	$33,149	9.484	1.768	15.7%	13.427	0.292	2.1%	$11,885,221,977	633,902	0.34	0.17
MI	$98,196	$58,213	10.153	1.099	9.8%	13.369	0.35	2.6%	$17,566,385,067	301,759	0.35	0.17
HF	$127,396	$87,414	8.595	2.657	23.6%	13.103	0.617	4.5%	$7,945,884,962	331,634	0.41	0.20
Sensitivity analysis: Probabilistic
No CVD	$29,369	–	11.245	–		13.713	–		–	–	–	–
Stroke	$85,320	$55,952	5.785	4.072	36.2%	13.096	0.616	4.6%	$5,922,710,668	431,035	–	–
AF	$60,041	$30,672	9.010	1.777	15.8%	13.428	0.285	2.1%	$10,997,198,684	637,129	–	–
MI	$84,746	$55,377	9.645	1.094	9.7%	13.378	0.334	2.5%	$16,710,534,281	330,125	–	–
HF	$112,555	$83,187	8.165	2.614	23.2%	13.125	0.587	4.4%	$7,561,678,751	237,612	–	–

AF, atrial fibrillation; HF, heart failure; MI, myocardial infarction; QALY, Qualify-adjusted life year; LY, life year; Incr., incremental; Inst., institutional care; Attributable %, Attributable fraction. Italic formatted number: expressing the incremental in outcome (differences between condition, e.g., dementia with no CVD condition versus dementia with CVD as a risk factor (Stroke, AF, MI, or HF). ^*Number of cases for 10,000 patients.

The analysis estimated a higher proportion of patients with CVD required institutional care compared to the non-CVD population, varying from 374 to 805 per 10,000 population respectively. By the end of the simulated time horizon, 27.6% of the simulated non-CVD population had developed dementia; this contrasts with the higher proportion of patients affected by dementia among patients with established CVD (ranging from 34.2% to 41.2%). The prevalence of dementia in persons aged 90 years and over is 28% in Australia according to the ABS [48]; the estimation from the simulation model was highly comparable.

At the national level, the total incremental economic burden of dementia from patients with CVD (aged 65 years and over in 2019) was $6.45 billion (stroke), $11.89 billion (AF), $17.57 billion (MI), or $7.95 billion (HF) respectively over their remaining life expectancy. The corresponding total QALY loss was 546,504 (stroke), 634,077 (AF), 331,567 (MI), or 241,516 (HF) for the same simulated population (Table 3).

Sensitivity analyzes

Deterministic sensitivity analyzes indicated that the RR of developing dementia by CVD types, costs of managing CVD conditions and HR of mortality in severe dementia were most sensitive to the incremental cost for dementia management for all four CVD conditions (Supplementary Figures 3–6).

Probabilistic sensitivity analysis incorporating key model parameters showed consistent results with the base case (Table 3 and Supplementary Figures 7–10).

Model validation

The prevalence of dementia was estimated when the hypothetical non-CVD patients reached age 90+ years in this model. The prevalence of dementia predicted by our model (27.6%) was highly comparable to the national figure of 28% reported by the ABS [48].

Our model predicted a mean cost of $39,983 for non-CVD patients in the base case. The result falls well within the range of reported costs internationally (between AU$9,073 and $43,668; all cost was converted to 2019 Australian dollar value using relevant CPI, HPI, and the AUD to USD average exchange rate in 2013) that was discussed in a 2015 systematic review [49]. The review was conducted to investigate the cost of illness of dementia in different setting (community, institutional, or mixed setting). The estimation for the average dementia management cost was comparable to that reported in an Australian study estimating the economic impact of dementia ($8,050 versus $7,801) [5].

The predicted life expectancy (13.7 years) for dementia patients (including years from non-dementia to dementia onset/diagnosis) was similar to the reported literature. A 2014 review identified a median survival of 12.6, 95% CI (11.7–13.4) in which the survival accounted for all dementia phases (from the onset of symptom to severe) [50]. In addition, the authors also reported a mean of 12.3 for survival which comprised 5.6 (mild dementia), 3.5 (moderate dementia), and 3.2 (severe dementia) [50]. Given the time between symptom onset and clinical diagnosis (mean = 2.7–3.3 years), the mean survival will probably extend beyond 12.3 years [50].

Our study results were also consistent with the literature in term of QALY estimations. The estimated QALY loss associated with stroke, HF, AF, and MI were comparable with the QALYs reported in an US study for several chronic conditions, including heart disease, stroke, diabetes, and cancer over the lifetime [51]. Overall, our estimation of QALY for dementia with non-CVD conditions were slightly higher than QALY for patients with diabetes or cancer (11.25 versus 9.0 and 10.7 respectively) [51]. The QALYs of patients with CVD conditions in our simulation were similar to the QALYs reported in the real world. For example, the mean QALY in our model was 6.089 for stroke patients compared to 7.0 (1.3) from a US study [51]. The predicted mean QALYs of AF, MI, and HF were 9.484, 10.153, and 8.585, respectively. These were highly comparable to an average QALY of 9.5 (1.2) for all heart disease patients (consisted of three conditions including coronary heart disease, angina, or heart attack) [51]. In addition, the QALY loss in stroke patients from our study was similar to the QALY loss of stroke from a 2010 Australian study that calculated the lifetime health loss’ for first-ever stroke (5.163 versus 5.09 (0.20) [52].

DISCUSSION

The simulation results suggest that patients with certain types of CVD contributed to the increased cost burden of dementia and significant loss in QALYs when comparing to the non-CVD population. The higher economic burden was mainly driven by more patients developing dementia and the greater chance of utilizing institutional care. The potential savings from institutional care is worth mentioning. In 2017–2018, the Australian government spent over $21.5 billion on aged care with 63% of total expenditure being on residential care ($13.5 billion) [53]. Our modelled study indicated that 805 additional patients care per 10,000 people with stroke would enter institutional care because they developed dementia later in life. The burden from healthcare and long-term care posed by dementia, along with the absence of disease-modifying treatment for dementia, make preventive interventions for dementia particularly important. Studies have reported that delaying dementia onset for five years could halve the prevalence of the disease; this further strengthens the value of dementia prevention [54, 55].

The results from this study suggesting potential savings from dementia prevention by targeting patients with CVDs carry significant implications for public policy formulation. The substantial economic burden and loss in QALYs not only stresses the importance of optimal management for patients with certain types of CVD (as risk factors for dementia) to tackle the disease burden of dementia, but also underlies the value of primary prevention of CVD in the real-world. As reported in an earlier study, the incorporation of the dementia disease state in a health economics model for evaluating a lifestyle intervention that aims to prevent type 2 diabetes and CVD, was associated with QALY gains and cost savings if the intervention directly modified the incidence of dementia [56]. It is highly possible that with the incorporation of a dementia module in the long-term economic modelling of CVD primary prevention interventions, the cost-effectiveness of existing interventions might become more favorable given the additional gains in QALYs and reduction in costs arising from dementia prevention.

The relationship between CVD and the development of dementia was informed by our de novo comprehensive systematic review that followed Cochrane standard methods for systematic review and meta-analysis. This up-to-date systematic review holistically examined the evidence regarding the risk of dementia in the CVD population. Out of 19 studies included, the statistically significant association between the presence of stroke and the incidence of dementia was informed by our meta-analysis of 13 observational studies. According to the analysis, patients with a history of stroke (including mini-stroke) had a mean odds ratio of 1.65 (95% CI 1.44–1.88) of developing dementia in later life compared to those with no stroke history. The results of our meta-analysis are highly comparable to previously published meta-analyzes that were conducted to quantify the risk between stroke and dementia [13] and to predict dementia in primary care [12].

Evidence supporting the association between other CVD conditions and dementia is relatively scarce. Inconsistent findings were demonstrated in relation to the evidence available for HF and AF. Five (AF) and four (HF) studies reported on the causal relationship between these two CVD conditions and dementia. Qualitative synthesis captured HF and AF as protective factors to prevent Alzheimer’s disease [57, 58], while consistent with the previous finding, AF and HF were reported as risk factors for dementia (especially vascular dementia) [10, 11]. However, given the study that reported on the inverse risk factor was a case control study, and did not report all cause dementia, no meta-analysis was conducted. Of note, only one prospective study with the longest follow up for each CVD type was selected for the model input upon consideration of the high probability for selection bias and information bias in case control studies [59]. Similarly, only two observational studies reported MI as a risk factor for patients to develop dementia in later life [35]; results from Bejot et al. were selected because of the low risk of bias and longer follow up [37]. A recently published systematic review and meta-analysis of the prevention of Alzheimer’s disease reported that coronary artery bypass grafting (CAGB) surgery was associated with an increased risk of Alzheimer’s disease (RR 1.71, 95% CI 1.04 to 2.79) [60]. The RR of MI applied in the current simulation model was 1.34 (95% CI, 1.06–1.70) [37]. The significantly increased risks of dementia in patients with coronary heart disease (i.e., MI, CABG) were generally consistent, whereas patients who underwent CABG surgery were more likely to be cases with severe coronary artery blockage and thus characterized by a higher risk of dementia due to possibly further reduced cardiac output [61 –64].

Of note, the incremental economic burden was highly proportional with the increased risk of dementia by the type of CVD which could externally validate the modelled results. For example, the RR of developing dementia was lowest for patients with AF; correspondingly, the per capita incremental costs (i.e., related to dementia only) and QALY loss for dementia were lowest in the same cohorts. Moreover, the simulated results from the non-CVD population showed comparable prevalence of dementia as reported by the Australian national statistics (27.6% versus 28%) [48].

Our study did not evaluate the cost-effectiveness of interventions modifying dementia risk factors (i.e., certain types of CVD) but signified the importance of optimal management of CVD. An economic evaluation that assessed the cost-effectiveness of hypothetical dementia prevention measures reported that a 5% reduction in dementia risk factors at the national level (e.g., midlife obesity, physical inactivity, smoking, hypertension, depression, etc.) was cost-effective [26]. To the best of our knowledge, although evidence suggests that using therapeutic agents such as statins to treat hypercholesterolemia and other CVD symptoms has a significant role in reducing dementia risk, there is a lack of evidence around potential interventions to reduce the risk of dementia for CVD patients due to the unresolved underlying mechanism between CVD and dementia. However, a recent study suggests an association between higher cardio health score at age 50 and lower risk of developing dementia and better brain volume when they get older [65]; there is a lack of evidence around potential interventions for dementia prevention for CVD patients due to the unresolved underlying mechanism between CVD and dementia. Generally, it is hypothesized that reduced cerebral blood flow caused by heart disease (HF, AF, MI) worsens the vascular homeostasis of the brain, and magnifies any cognitive problems from the build-up of tau and Aβ proteins [66]. For example, in AF, there is a growing body of evidence supporting AF as a risk factor for dementia without stroke [67, 68], and the possible mechanism is the irregular rapid ventricular rates leading to decreased cerebral perfusion and covert cerebral infarction [69, 70]. A prior modelled study reported that a 10% reduction in the prevalence of the modifiable risk factors for CVD (including stroke and other condition in the real-world prevention program, e.g., Diabetes Prevention Program) would save US$37 billion for both Medicare and Medicaid population over a lifetime horizon [66]. The study also indicated a statistically significant reduction in the risk of developing dementia and the duration of living with dementia (p < 0.05) for this cohort [66]. Therefore, optimizing the CVD management could potentially contribute to dementia prevention.

Other than heart disease, there is strong evidence around post-stroke dementia. It was reported that 10% of new dementia cases developed soon after a first-ever stroke and increased to over 30% after recurrent stroke. The strong association of post stroke dementia with multiple strokes stresses the essential causal relationship of stroke as opposed to the other underlying vascular risk factors. In addition, the prognostic value of stroke characteristics (i.e., factors indicative of a greater stroke lesion burden and complications of stroke like left hemisphere stroke, stroke severity, infarct volume, incontinence, early seizures, etc.) highlight the importance of optimal acute stroke care and secondary stroke preventions that are likely to reduce the burden of post-stroke dementia [71]. Our results echo the findings from this study— that patients with a stroke history incur the biggest burden of dementia compared to those other cardiac conditions.

This is the first study to quantify both the economic and QALY loss burden due to dementia from patients with CVD. The relationship between CVD and dementia was informed by an up-to-date systematic review of all the existing evidence. The limited societal perspectives were taken to measure the costs and QALY loss. We also undertook extensive sensitivity analyzes to examine the robustness of base case results. However, this study is not without limitations. First of all, the RR of CVD on the incidence of dementia was derived from an observational study (comprising nine cohort, seven case-control and three cross-sectional studies) that were of relatively short duration (5/9 cohort studies had less than 8 years of follow up) (Supplementary Table 3). There may be confounding factors not controlled for in the estimation of these statistics while the risk of information bias and selection bias was high in 7/19 studies (Supplementary Table 3). Second, the long-term modelling was realized through a Markov microsimulation in which the inputs were sourced from published studies that may have limited face validity. Nevertheless, the validation of the modelled output against national statistics alleviates this concern to some extent. Lastly, first-ever CVD was not accounted for in the comparator arm (i.e., people in the controls were assumed to be CVD-free over the entire modelled time horizon, thus the incidence of dementia for the general population was applied constantly) which may have resulted in an underestimation of the burden of dementia in the comparator arm.

Conclusions

Patients with certain CVD conditions generated a substantial economic and QALY loss burden due to dementia. The results highlighted the importance of both primary and secondary prevention of CVD to not only to reduce the CVD burden, but also decrease the occurrence of dementia. The absence of effective disease-modifying treatment for dementia makes preventive interventions of critical value. Targeting patients with dementia risk factors like CVD may prove to be effective and cost-effective.

Footnotes

ACKNOWLEDGMENTS

This study was supported by a small grant from School of Health & Social Development, Deakin University.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Australian Institute of Health and Welfare (2016) Contribution of vascular diseases and risk factors to the burden of dementia in Australia: Australian Burden of Disease Study 2011. J Home Econ Inst Australia 23, 33.

Australian Institute of Health and Welfare (2012) Dementia in Australia. Cat. no. AGE 70. Canberra: AIHW. Available from: https://www.aihw.gov.au/getmedia/199796bc-34bf-4c49-a046-7e83c24968f1/13995.pdf.aspx?inline=true

Prince

, Bryce

, Ferri

(2011) World Alzheimer Report 2011. The benefits of early diagnosis and intervention. Alzheimer’s Disease International, London.

Australian Institute of Health and Welfare, Deaths in Australia, Leading causes of death, https://www.aihw.gov.au/reports/life-expectancy-death/deaths-in-australia/contents/leading-causes-of-death.

Cations

, Lang

, Ward

, Crotty

, Whitehead

, Maddison

, Inacio

(2021) Cohort profile: Dementia in the registry of senior Australians. BMJ Open 11, e039907.

Brown

, Hansnata

, La

(2017) Economic cost of dementia in Australia. Alzheimer’s Australia, Canberra.

Patterson

(2018) World Alzheimer Report 2018. The state of the art of dementia research: New frontiers. Alzheimer’s Disease International, London.

Wanleenuwat

, Iwanowski

, Kozubski

(2019) Alzheimer’s dementia: Pathogenesis and impact of cardiovascular risk factors on cognitive decline. Postgrad Med 131, 415–422.

Sundbøll

, Horváth-Puhó

, Adelborg

, Schmidt

, Pedersen

, Bøtker

, Henderson

, Toft Sørensen

(2018) Higher risk of vascular dementia in myocardial infarction survivors. Circulation 137, 567–577.

10.

, Wu

, Zhang

, Nie

(2020) Associations between heart failure and risk of dementia: A PRISMA-compliant meta-analysis. Medicine (Baltimore) 99,, e18492.

11.

Islam

, Poly

, Walther

, Yang

H-C

, Wu

, Lin

M-C

, Chien

S-C

, Li

Y-C

(2019) Association Between Atrial Fibrillation and Dementia: A Meta-Analysis. Front Aging Neurosci 11, 305.

12.

Ford

, Greenslade

, Paudyal

, Bremner

, Smith

, Banerjee

, Sadhwani

, Rooney

, Oliver

, Cassell

(2018) Predicting dementia from primary care records: A systematic review and meta-analysis. PLoS One 13, e0194735.

13.

Kuźma

, Lourida

, Moore

, Levine

, Ukoumunne

, Llewellyn

(2018) Stroke and dementia risk: A systematic review and meta-analysis. Alzheimers Dement 14, 1416–1426.

14.

Tariq

, Barber

(2018) Dementia risk and prevention by targeting modifiable vascular risk factors. J Neurochem 144, 565–581.

15.

Konno

, Meyer

, Terayama

, Margishvili

, Mortel

(1997) Classification, diagnosis and treatment of vascular dementia. Drugs Aging 11, 361–373.

16.

Banerjee

, Wittenberg

(2009) Clinical and cost effectiveness of services for early diagnosis and intervention in dementia. Int J Geriatr Psychiatry 24, 748–754.

17.

Michaud

TA-O

, Kane

, McCarten

, Gaugler

, Nyman

, Kuntz

(2018) Using cerebrospinal fluid biomarker testing to target treatment to patients with mild cognitive impairment: A cost-effectiveness analysis. Pharmacoecon Open 2, 309–323.

18.

Nguyen

KA-O

, Sellars

, Agar

, Kurrle

, Kelly

, Comans

(2017) An economic model of advance care planning in Australia: A cost-effective way to respect patient choice. BMC Health Serv Res 17, 797.

19.

Khan

(2016) Chapter 2 - Clinical diagnosis of Alzheimer’s disease. In Biomarkers in Alzheimer’s Disease, Khan

, ed. Academic Press, pp. 27–48.

20.

Davis

, O’Connell

, Johnson

, Cline

, Merikle

, Martenyi

, Simpson

(2018) Estimating Alzheimer’s disease progression rates from Normal cognition through mild cognitive impairment and stages of dementia. Curr Alzheimer Res 15, 777–788.

21.

Deeks

, Higgins

, Altman

(2019) Analysing data and undertaking meta-analyses. In Cochrane Handbook for Systematic Reviews of Interventions, Higgins

JPT

, Thomas

, Chandler

, Cumpston

, Li

, Page

, Welch

, eds. Wiley-Blackwell, pp. 241–284.

22.

Cochrane Training Cochrane Handbook for Systematic Reviews. Available from: https://training.cochrane.org/handbook

23.

Schünemann

, Vist

, Higgins

, Santesso

, Deeks

, Glasziou

, Akl

, Guyatt

(2019) Interpreting results and drawing conclusions. In Cochrane Handbook for Systematic Reviews of Interventions, Higgins

JPT

, Thomas

, Chandler

, Cumpston

, Li

, Page

, Welch

, eds. Wiley-Blackwell, pp. 403–431.

24.

Spackman

, Kadiyala

, Neumann

, Veenstra

, Sullivan

(2012) Measuring Alzheimer disease progression with transition probabilities: Estimates from NACC-UDS. Curr Alzheimer Res 9, 1050.

25.

Andersen

, Lolk

, Martinussen

, Kragh-Sørensen

(2010) Very mild to severe dementia and mortality: A 14-year follow-up–the Odense study. Dement Geriatr Cogn Disord 29, 61–67.

26.

McRae

, Zheng

, Bourke

, Cherbuin

, Anstey

(2021) Cost-effectiveness of dementia prevention interventions. J Prev Alzheimers Dis 8, 210–217.

27.

Handels

, Wimo

(2019) Challenges and recommendations for the health-economic evaluation of primary prevention programmes for dementia. Aging Mental Health 23, 53–59.

28.

Australian Bureau of Statistics, Consumer Price Index, Australia, https://www.abs.gov.au/statistics/economy/price-indexes-and-inflation/consumer-price-index-australia/latest-release#key-statistics.

29.

Australian Institute of Health and Welfare, Health expenditure Australia 2017–18, https://www.aihw.gov.au/reports/health-welfare-expenditure/health-expenditure-australia-2017-18

30.

Gnanamanickam

, Dyer

, Milte

, Harrison

, Liu

, Easton

, Bradley

, Bilton

, Shulver

, Ratcliffe

(2018) Direct health and residential care costs of people living with dementia in Australian residential aged care. Int J Geriatr Psychiatry 33, 859–866.

31.

Ball

, Thompson

, Ski

, Carrington

, Gerber

, Stewart

(2015) Estimating the current and future prevalence of atrial fibrillation in the Australian adult population. Med J Australia 202, 32–35.

32.

Australian Bureau of Statistics, National Health Survey: First Results, 2017–2018, ABS Canberra.

33.

Bunch

, Weiss

, Crandall

, May

, Bair

, Osborn

, Anderson

, Muhlestein

, Horne

, Lappe

, Day

, Bunch

, Weiss

, Crandall

, May

, Bair

, Osborn

, Anderson

, Muhlestein

, Horne

(2010) Atrial fibrillation is independently associated with senile, vascular, and Alzheimer’s dementia. Heart Rhythm 7, 433–437.

34.

Tatemichi

, Desmond

, Paik

, Figueroa

, Gropen

, Stern

, Sano

, Remien

, Williams

, Mohr

, Mayeux

(1993) Clinical determinants of dementia related to stroke. Ann Neuro 33, 568–575.

35.

Yip

, Brayne

, Matthews

(2006) Risk factors for incident dementia in England and Wales: The Medical Research Council Cognitive Function and Ageing Study. A population-based nested case-control study. Age Ageing 35, 154–160.

36.

Smith

, Flicker

, Dwyer

, Atkinson

, Almeida

, Lautenschlager

, LoGiudice

(2010) Factors associated with dementia in Aboriginal Australians. Aust N Z J Psychiatry 44, 888–893.

37.

Béjot

, Aboa-Eboulé

, Durier

, Rouaud

, Jacquin

, Ponavoy

, Richard

, Moreau

, Giroud

(2011) Prevalence of early dementia after first-ever stroke: A 24-year population-based study. Stroke 42, 607–612.

38.

Burton

, Campbell

, Jordan

, Strauss

, Mallen

(2013) The association of anxiety and depression with future dementia diagnosis: A case-control study in primary care. Family Practice 30, 25–30.

39.

Bruce

, Davis

, Starkstein

, Davis

(2014) Mid-life predictors of cognitive impairment and dementia in type 2 diabetes mellitus: The Fremantle Diabetes Study.S. J Alzheimers Dis 42, 63–70.

40.

Hendrie

, Hake

, Lane

, Purnell

, Unverzagt

, Smith-Gamble

, Murrell

, Ogunniyi

, Baiyewu

, Callahan

, Saykin

, Taylor

, Hall

, Gao

(2015) Statin use, incident dementia and Alzheimer disease in elderly African Americans. Ethnicity Dis 25, 345–354.

41.

Yang

, Jin

, Yan

, Jin

, Yu

, Wu

, Xu

(2016) Prevalence of dementia, cognitive status and associated risk factors among elderly of Zhejiang province, China in 2014. Age Ageing 45, 707–711.

42.

Zahodne

, Schupf

, Brickman

, Mayeux

, Wall

, Stern

, Manly

(2016) Dementia risk and protective factors differ in the context of memory trajectory groups. J Alzheimers Dis 52, 1013–1020.

43.

Ikeda

, Yamagishi

, Tanigawa

, Cui

, Yao

, Noda

, Umesawa

, Chei

, Yokota

, Shiina

, Harada

, Murata

, Asada

, Shimamoto

, Iso

(2008) Cigarette smoking and risk of disabling dementia in a Japanese rural community: A nested case-control study. Cerebrovasc Dis 25, 324–331.

44.

Cations

, Draper

, Low

, Radford

, Trollor

, Brodaty

, Sachdev

, Gonski

, Broe

, Withall

(2018) Non-genetic risk factors for degenerative and vascular young onset dementia: Results from the INSPIRED and KGOW studies. J Alzheimers Dis 62, 1747–1758.

45.

Chiang

, Yip

, Wu

, Lu

, Liou

, Liu

, Chu

, Hwang

, Sung

, Hsu

, Chen

, Liu

, Yan

, Fong

, Chang

, You

, Chen

(2007) Midlife risk factors for subtypes of dementia: A nested case-control study in Taiwan. Am J Geriatr Psychiatry 15, 762–771.

46.

Chen

, Hu

, Wei

, Ma

, Liu

JR C (2011) Incident dementia in a defined older Chinese population. PLoS One 6, e24817.

47.

Witt

, Rotter

, Stearns

, Gottesman

, Kucharska-Newton

, Richey Sharrett

, Wruck

, Bressler

, Sueta

, Chang

(2018) Heart failure and cognitive impairment in the Atherosclerosis Risk in Communities (ARIC) study. J Gen Intern Med 33, 1721–1728.

48.

Australian Bureau of Statistics, Dementia in Australia, https://www.abs.gov.au/articles/dementia-australia

49.

Schaller

, Mauskopf

, Kriza

, Wahlster

, Kolominsky-Rabas

(2015) The main cost drivers in dementia: A systematic review. Int J Geriatr Psychiatry 30, 111–129.

50.

Kua

, Ho

, Tan

, Tsoi

, Thng

, Mahendran

(2014) The natural history of dementia. Psychogeriatrics 14, 196–201.

51.

Jia

, Lubetkin

(2016) Impact of nine chronic conditions for US adults aged 65 years and older: An application of a hybrid estimator of quality-adjusted life years throughout remainder of lifetime. Qual Life Res 25, 1921–1929.

52.

Cadilhac

, Dewey

, Vos

, Carter

, Thrift

(2010) The health loss from ischemic stroke and intracerebral hemorrhage: Evidence from the North East Melbourne Stroke Incidence Study (NEMESIS). Health Qual Life Outcomes 8, 1–8.

53.

Australian Institute of Health and Welfare (2021) Spending on aged care. Available from: https://www.gen-agedcaredata.gov.au/Topics/Spending-on-aged-care

54.

Brookmeyer

, Gray

, Kawas

(1998) Projections of Alzheimer’s disease in the United States and the public health impact of delaying disease onset. Am J Public Health 88, 1337–1342.

55.

Jorm

, Dear

, Burgess

(2005) Projections of future numbers of dementia cases in Australia with and without prevention. Aust N Z J Psychiatry 39, 959–963.

56.

Breeze

, Thomas

, Thokala

, Lafortune

, Brayne

, Brennan

(2020) The impact of including costs and outcomes of dementia in a health economic model to evaluate lifestyle interventions to prevent diabetes and cardiovascular disease. Med Decis Making 40, 912–923.

57.

Imfeld

, Bodmer

, Jick

, Meier

(2015) Benzodiazepine use and risk of developing Alzheimer’s disease or vascular dementia: A case-control analysis. Drug Saf 38, 909–919.

58.

Imfeld

, Toovey

, Jick

, Meier

(2016) Influenza infections and risk of Alzheimer’s disease. Brain Behav Immun 57, 187–192.

59.

Sutton-Tyrrell

(1991) Assessing bias in case-control studies. Proper selection of cases and controls. Stroke 22, 938–942.

60.

Moon

, Kim

, Choi

J-W

, Park

J-B

, Hwang

H-Y

, Kim

H-K

, Kim

Y-J

, Kim

K-H

, Kim

K-B

, Sohn

D-W

(2020) Early Surgery versus watchful waiting in patients with moderate aortic stenosis and left ventricular systolic dysfunction. Korean Circ J 50, 791–800.

61.

Zuccala

, Onder

, Pedone

, Carosella

, Pahor

, Bernabei

, Cocchi

, GIFA-ONLUS Study Group [Grupo Italiano di Farmacoepidemiologia nell’Anzanio] (2001) Hypotension and cognitive impairment: Selective association in patients with heart failure. Neurology 57, 1986–1992.

62.

Breteler

, Claus

, Grobbee

, Hofman

(1994) Cardiovascular disease and distribution of cognitive function in elderly people: The Rotterdam Study. BMJ 308, 1604–1608.

63.

Aronson

, Ooi

, Morgenstern

, Hafner

, Masur

, Crystal

, Frishman

, Fisher

, Katzman

(1990) Women, myocardial infarction, and dementia in the very old. Neurology 40, 1102–1106.

64.

Gharacholou

, Reid

, Arnold

, Spertus

, Rich

, Pellikka

, Singh

, Holsinger

, Krumholz

, Peterson

, Alexander

(2011) Cognitive impairment and outcomes in older adult survivors of acute myocardial infarction: Findings from the translational research investigating underlying disparities in acute myocardial infarction patients’ health status registry. Am Heart J 162, 860–869.e861.

65.

Sabia

, Fayosse

, Dumurgier

, Schnitzler

, Empana

J-P

, Ebmeier

, Dugravot

, Kivimäki

, Singh-Manoux

(2019) Association of ideal cardiovascular health at age 50 with incidence of dementia: 25 year follow-up of Whitehall II cohort study. BMJ 366, I4414.

66.

Justin

, Turek

, Hakim

(2013) Heart disease as a risk factor for dementia. Clin Epidemiol 5, 135–145.

67.

Ratcliffe

, Wilcock

(1985) Cerebrovascular disease in dementia: The importance of atrial fibrillation. Postgrad Med J 61, 201–204.

68.

Sabatini

, Frisoni

, Barbisoni

, Bellelli

, Rozzini

, Trabucchi

(2000) Atrial fibrillation and cognitive disorders in older people. J Am Geriatr Soc 48, 387–390.

69.

Ben Morrison

, Jared Bunch

, Gersh

(2009) Pathophysiology of concomitant atrial fibrillation and heart failure: Implications for management. Nat Clin Pract Cardiovasc Med 6, 46–56.

70.

Zito

, Muscari

, Marini

, Di Iorio

, Puddu

, Abate

(1996) Silent lacunar infarcts in elderly patients with chronic non valvular atrial fibrillation. Aging (Milano) 8, 341–346.

71.

Pendlebury

, Rothwell

(2009) Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: A systematic review and meta-analysis. Lancet Neurol 8, 1006–1018.