Societal Costs of Dementia: 204 Countries, 2000–2019

INTRODUCTION

Currently over 57 million people are living with Alzheimer’s disease and other dementias (hereafter referred to as dementia) worldwide.¹ This population is expected to grow to over 150 million by 2050, with the largest increases in the Middle East, North Africa, and eastern sub-Saharan Africa. As the world’s population grows and ages, dementia is becoming a growing priority in many countries.²

Dementia represents a significant economic burden, and historically treatment options have been limited. The progressive nature of the condition and lack of treatment options lead to extensive demand for personal care. This care, whether provided at a nursing home, by formal (paid) caregivers at home, or by informal (unpaid) caregivers such as family or friends, leads to high societal costs of dementia. In countries with aging populations, health policy is often focused on keeping individuals out of nursing homes and in the community.^3,4 While it is important to support individuals who want to live independently, some studies have found that care in the community can be more expensive than nursing home care when the cost of informal care is taken into account.⁵ Understanding the full societal costs of dementia will be increasingly important as policymakers plan for the aging population and new disease-modifying treatments change the dementia care landscape.

A recent study by Wimo and colleagues⁶ estimated the global cost of dementia at US$1.3 trillion in 2019. This number includes all direct spending and caregiving provided to people living with dementia, rather than costs that are attributable solely to dementia. About half of the reported cost (US$651.4 billion) was due to informal care, with the remaining $661.9 billion going to direct costs, including the cost of long-term care. A prior study by Jia and colleagues found similar results, estimating that the global cost of dementia in 2015 was $958 billion and would increase to $2.5 trillion by 2030.⁷

Pedroza and colleagues⁸ focused on direct costs but used a detailed modeling framework that estimated separate costs for people who have diagnosed versus undiagnosed dementia, and those who live in nursing homes versus in the community. With this approach, their global estimates for direct spending on dementia were close to estimates from Wimo and colleagues. They estimated that $594 billion was spent worldwide on people living with dementia in 2019, with $263 billion being attributable to dementia. Other studies have found even higher economic burdens of dementia. Using the value of a statistical life to value the economic impact of dementia, Nandi and colleagues estimated global costs of dementia and found an economic burden of $2.8 trillion in 2019.⁹ While informative, these previous studies tend to be focused only on direct spending (Pedroza et al), ignore cost of informal care (Nandi et al), or are based on older data (Jia et al). ⁷ Of the existing global estimates for the cost of dementia care, our study is closest to the work by Wimo and colleagues.⁶ We add to their work in a few ways. First, we estimate direct spending attributable to dementia rather than direct spending on people living with dementia, i.e. we adjust the costs downward to account for the fact that many people living with dementia also have comorbidities that require care. We also explicitly model the percent of individuals with dementia symptoms who have a diagnosis. This allows us to estimate costs for diagnosed versus undiagnosed individuals separately, which is particularly important for a condition like dementia which often goes undiagnosed, even in high resource countries.¹⁰ We also allow this percentage to vary in future scenarios, given that improvements to the timeliness of dementia diagnoses is often listed as a goal in national dementia plans.¹¹

Our work creates estimates for direct and indirect costs of dementia for 204 countries from 2000 through 2019 and projects these costs to 2050 under multiple scenarios. We use the framework developed by Pedroza and colleagues⁸ to model the fraction of people living with dementia who are undiagnosed, as well as the fraction of people who are living in nursing homes. Modeling these parameters allows us to examine the impact of accelerated diagnosis rates or institutionalization rates on future health spending.

METHODS

RESULTS

From our systematic review of dementia diagnosis rates, we identified a total of 874 results for title and abstract screening, of which six proceeded to full article review and extraction and standardization. From our systematic review of dementia costs, we identified a total of 1,357 results for title and abstract screening, of which 30 proceeded to full article review and extraction. From our systematic review of dementia caregiving, we identified a total of 1,263 results for title and abstract screening, of which 77 proceeded to full article review and extraction (see the Supplementary Material for details of the three systematic reviews).

Our meta-regression estimated that 41% (95% uncertainty interval [UI] 29–54%) of the health care spending on patients receiving community-based care who are diagnosed with dementia can be attributed to dementia care (Supplementary Figure 5). For patients receiving nursing home– based care who are diagnosed with dementia, we estimated that 74% (95% UI 32–99%) of total spending was attributable to dementia. For patients with dementia who receive nursing home– based care, but who lack a dementia diagnosis, we estimated that 37% (95% UI 16–49%) of spending was attributable to dementia. For indirect cost, our meta-regression estimated that 58% (95% UI 51–65%) of caregiving hours were attributable to dementia (Supplementary Figure 6). Unless otherwise noted, all estimates herein have been adjusted to reflect these attribution rates.

We estimated that in 2019, health care spending attributable to dementia was $260.6 billion (95% UI 131.6–420.4) and the cost of caregiving attributable to dementia was $354.1.2 billion (95% UI 190.0–544.1), bringing the total societal cost to $614.8 billion (95% UI 403.1–861.3) (Table 1). Globally, 71% (95% UI 67–75%) of caregivers were female. This proportion was even higher for low– income countries, with 85% (95% UI 80–90%) being female. The costs of dementia care are not distributed globally according to dementia prevalence. While high-income countries accounted for 78.8% % of total spending (95% UI 74.8–82.8%), they accounted for just 43.8% of total prevalence (95% UI 43.1–44.5%) in 2019. From 2000 to 2019, our estimates indicate that direct costs grew faster than informal care costs. We estimated an annual growth rate of 4.15% (95% UI 1.81–6.57%) for direct costs and 3.65% (95% UI 0.00–6.83%) for informal care costs. Growth rates for both direct costs and informal care costs were highest in the Southeast Asia, East Asia, and Oceania GBD super-region and lowest in the high-income GBD super-region (Table 1). A complete set of modeled cost estimates reported at the country level are included in the Supplementary Material in both US dollars and purchasing power parity (Supplementary Tables 21 and 22). Findings from our sensitivity analyses were qualitatively similar to our baseline findings and are shown in Supplementary Tables 29–36.

The direct costs of dementia made up 3.0% (95% UI 1.5–4.9%) % of total health spending¹³ in 2019, varying from 3.5% (95% CI 1.8–5.6%) in high-income countries to only 0.1% (95% CI 0.1–0.2) in low-income countries. If past trends continue, dementia care will take up an increasing share of total health spending, climbing to 9.4% (95% CI 3.9–19.6%) globally by 2050 (Supplementary Table 26).

Our global estimate for societal costs attributable to dementia was $79 (95% UI 52–111) per capita in 2019 (Table 2). Of this cost, $34 (95% UI 17–54) was for direct medical spending and $46 (95% UI 25–70) was for informal care. Per capita costs of dementia increase with GDP per capita, ranging from $3.56 (95% UI 2.57–4.55) in World Bank– defined low-income countries to $401 (95% UI 262–558) in high-income countries (Table 1, Supplementary Figure 7). Table 2 also shows costs per prevalent case of dementia. The per-case direct costs increase with GDP, but the indirect costs follow a U-shaped pattern, with slightly higher costs in the low-income region than the lower-middle-income region.

The uneven distribution of costs is particularly pronounced for direct spending, with high-income countries accounting for 91.6% (95% UI 90.3–92.7%) of direct costs (Fig. 1A) despite only making up 37.3% (95% UI 35.9–38.5%) of prevalent cases. In terms of informal care, the distribution of costs is slightly more balanced, with high-income countries accounting for 69.2% (95% UI 64.7–73.5%) of indirect costs (Fig. 1A). Globally, we estimated that 85.4% (95% UI 80.8–89.2%) of people living with dementia are living in the community, with the remainder in nursing home care (Fig. 1B). Nursing homes have an outsized impact on the cost of dementia care, making up 69.6% (95% UI 49.4–80.4%) of the direct cost of dementia. However, when informal care costs are included, nursing home–based care drops to 25.7% (95% UI 11.7–41.7%) of global costs.

Fig. 1

Dementia prevalence and costs by World Bank income group and care setting/diagnosis status in 2019.

Fig. 2

Das Gupta decomposition by GBD region.

To examine the role of dementia prevalence and age structure in total cost, we decomposed the cost of dementia into five components— the age-standardized prevalence; the ratio of prevalence to age-standardized prevalence, herein denoted age ratio; the direct cost; the number of caregiving hours; and the expected wage (wage scaled by labor force participation) of the caregiver. Most regions with older populations also had higher direct costs and higher indirect costs due to higher expected wages for caregivers. Eastern Europe, Southern Latin America, East Asia, and the Caribbean were the only exceptions that had older than average populations but below average costs.

Informal caregiving is a large component of dementia care in all countries, but it plays an especially important role in some low-income countries (Fig. 3). While absolute costs for both direct medical spending and caregiving are higher in high-income countries, the relative contribution of caregiving to total costs decreases with GDP (Fig. 3). In many sub-Saharan African countries, the costs of dementia are almost entirely due to informal care.

Fig. 3

Informal care fraction of total dementia spending versus GDP per capita. Notes: In the linear regression equation shown, x represents the log base 10 of GDP and y represents the fraction of total spending on informal care.

Although informal care makes up a decreasing share of dementia care as GDP rises, the cost of informal care is increasing with GDP (Fig. 4A). This increase in cost occurs despite estimated per-case annual hours of caregiving declining with GDP (Fig. 4B), highlighting the large differences in wages across countries.

Fig. 4

A) Annual informal care cost per case versus GDP per capita. In the equation shown, x represents the log base 10 of GDP. and y represents the log base 10 of the cost of informal care. B) Annual informal caregiving hours per case versus GDP per capita. In the equation shown, x represents the log base 10 of GDP, and y represents the number of caregiving hours provided per year.

Under our baseline scenario, dementia costs will increase from $79 (95% UI 52–111) per person in 2019 (Table 2) to $268 (95% UI 132–490) in 2050 (Table 3). The scenario with accelerated diagnosis rates led to estimated dementia spending of $273 (95% UI 134–499) per person. Using the 85th percentile for the accelerated diagnosis rate did not lead to large changes, with the global average diagnosis rate reaching 20% (95% UI 16–24%) in the baseline scenario versus 23% (95% UI 19–28%) in the accelerated scenario (Table 4). The scenario with accelerated nursing home– based care rates yielded global spending estimates of $282 (95% UI 142–509 per person, while the scenario with decelerated nursing home– based care rates yielded spending estimates of $236 (95% UI 117–$434) per person. Of the parameters modified by these scenarios, accelerating the direct costs had the largest impact, leading to per-person costs of $336 (95% UI 177–575) in 2050 (Table 2).

DISCUSSION

We updated the work of Pedroza and colleagues⁸ and extended their models to include the cost of informal care. We estimated that the global costs of dementia were approximately US$600 billion in 2019 and are likely to rise to over $2 trillion by 2050. Despite a goal from the World Health Organization (WHO) Action Plan on the Public Health Response to Dementia 2017–2025⁴ for 50% of countries to have a 50% diagnosis rate for individuals living with dementia by 2025, we estimated that only 13% (95% UI 10–16%) of individuals living with dementia had a diagnosis in 2019. We also found that a large majority of people living with dementia live in the community, with only 12% (95% UI 10–14%) living in nursing homes.

Our estimate of US$600 billion for the worldwide cost of dementia in 2019 is substantially smaller than the estimate of US$1.3 trillion from Wimo and colleagues.⁶ The largest difference between our methodologies is that Wimo reported all spending on people living with dementia, while we estimated costs attributable to dementia by implementing a comorbidity adjustment for both direct and indirect costs. This approach avoids double-counting costs for people with more than one condition, which is particularly relevant for dementia as multimorbidity is known to increase with age.²³ Our global cost estimate without comorbidity adjustment was US$1.1 trillion, aligning more closely to Wimo’s estimate.

The additional sources found in our updated systematic reviews resulted in significant changes relative to the work by Pedroza and colleagues.⁸ Our diagnosis rate was smaller than the rate of 29% found by Pedroza and colleagues, due largely to a diagnosis rate of only 4% in Thailand that was added from our systematic review. Our rate of people living with dementia who are in nursing homes was also much smaller than the estimate of 23% from Pedroza and colleagues. High-income Asian Pacific countries had much smaller rates of nursing home residence in our model due to new data from South Korea. These large downward revisions highlight the sparsity of data underlying our models and the pressing need for data from a wider geographic representation.

We estimated that the indirect cost of informal care for people living with dementia makes up over half of the societal cost, echoing findings from Wimo and colleagues.⁶ While there are massive differences in costs between high-income countries and low-income countries, the inclusion of indirect costs narrows the gap somewhat. We estimated that direct costs in World Bank– defined high-income countries were over 7,000 times higher than direct costs in low-income countries, but when considering all costs, the ratio of high- to low-income countries drops to 186. Despite high-income countries providing fewer hours of informal care than low-income countries, the cost of informal care rises steeply with GDP, reflecting the higher wages in high-income countries. In many countries, particularly in sub-Saharan Africa, almost 100% of the costs of dementia are from informal care.

Under our baseline scenario which assumes a continuation of growth rates from 2000 through 2019, we estimated that the per capita cost of dementia care will more than triple by 2050. Under an accelerated direct cost scenario, the per capita cost in 2050 would be approximately 25% higher than under the baseline scenario. Scenarios that accelerate or decelerate the proportion of individuals who live in nursing homes did not have a large impact on expected cost. A decrease in the proportion of individuals who live in nursing homes is offset by an increase in the cost of informal care, leading to a small overall cost impact.

As the prevalence of dementia continues to rise, the projected increase in costs threatens to burden health care systems significantly. Our finding that informal care makes up over half of the total cost of dementia highlights the large amount of unpaid care provided by family and friends of those living with dementia throughout the world. The importance of caregivers has been stressed by WHO, as one of the action areas in their global plan is support for caregivers.⁴ Investing in caregiver support and training has been shown to decrease caregiver burden and improve quality of life for both caregivers and people living with dementia.²⁴ This reliance on informal care faces a potential crisis as demographic shifts lead to an older population structure with fewer younger caregivers available. Policymakers should develop sustainable strategies to fund medical costs of dementia and support caregivers.

There are several limitations to this analysis. Most importantly, primary data on dementia costs, diagnosis rates, and caregiving are sparse, and missingness is concentrated in low-income regions. We were only able to find one study from an African country. To deal with the large amount of missingness, our modeling approach relies on data that are available for all countries to make informed predictions for countries where no primary data are available. These data include economic and demographic indicators, but they do not include cultural norms. Cultural differences are known to affect caregiver motivation, wellbeing, and burden; however, there is limited research on whether these differences translate to more or fewer hours of care provided.^25–29 One channel through which cultural factors may impact the amount of care provided is through coresidency, as caregivers who live with their care recipient tend to provide more care.³⁰ However, even in cultures with high levels of familism, socioeconomic factors play a large role in determining the likelihood of adult children residing with their parents.³¹ Cultural factors will also likely impact the decision to transfer a loved one who is living with dementia to a nursing home, although government policies play a large role in these outcomes.³² Within the US, studies have shown that Hispanic and Black caregivers are more likely to provide high-intensity care (defined as over 50 hours per week) but did not control for the level of impairment of the care recipient. While cultural norms certainly affect caregiving decisions, we have no evidence to indicate that they are biasing our caregiving estimates in one direction.

Another limitation was the inability to estimate attributable fractions by age group for both direct spending and caregiving. The attributable fractions would likely vary with age due to an increasing number of comorbidities with older age, and estimating the attributable fraction by age group would allow us to adjust the overall attributable fraction for each country based on its age profile. Unfortunately, no studies in our systematic review reported attributable fractions of either direct spending or caregiving by age.

An additional limitation of the caregiving meta-analysis was the variation in how caregiving was estimated. Some sources include supervision time while some only include caregiving support for activities of daily living and instrumental activities of daily living. Due to the small amount of data available, we included all sources with an attributable fraction for caregiving despite these differences. Including sources that do not measure supervision time likely biases our attributable fraction downward, as supervision time is more often needed for dementia than for other comorbidities.

Another limitation of this work is the use of mean wages to value caregiving time. Using the mean wage covers up substantial heterogeneity in the demographics of caregivers and ignores the connection between a caregiver’s wage and their decision to provide care. This metric also does not include employer contributions that provide a monetary value, such as retirement plans or health insurance. Finally, we made strong assumptions around informal care hours, specifically that individuals living in nursing homes receive zero hours of care and that undiagnosed individuals living in the community receive the same number of hours of informal care as diagnosed individuals.

AUTHOR CONTRIBUTIONS

Amy Lastuka (Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Writing – original draft; Writing – review & editing); Elye Bliss (Formal analysis; Validation; Visualization; Writing – review & editing); Michael R. Breshock, MSE (Formal analysis; Investigation; Validation; Visualization; Writing – review & editing); Vincent C. Iannucci (Writing – review & editing; Systematic Review); William T. Sogge (Writing – review & editing; Systematic Review); Kayla V. Taylor (Project administration; Writing – review & editing); Paola Pedroza (Methodology; Writing – review & editing); Joseph L Dieleman (Conceptualization; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Writing – original draft; Writing – review & editing).