Call for a Uniform Strategy of Collecting Alzheimer’s Disease Costs: A Review and Meta-Analysis

Abstract

Background:

There is now a general attempt in developed countries to implement strategic plans to fight against Alzheimer’s disease and other dementia disorders. Among others, attention is paid to the issues of registers and calculations of economic burden. Currently available calculations of costs are difficult to compare. The problem is a different breakdown of cost categories and non-unified monitoring of cost types.

Objective:

The aim of this paper is to note the problem of poor availability and inconsistencies in cost monitoring. Furthermore, the intersection of cost items that are comparable and consistently monitored in expert studies are specified.

Methods:

The Web of Science, Elsevier Science Direct, PubMed, and Scopus databases are used in a systematic review. Two independent reviewers screened the identified records and selected relevant articles published in the period from 2010 to 2016. A meta-analysis of costs is performed in four categories related to patients suffering from Alzheimer’s disease.

Results:

The resulting estimation of total costs per patient per month through meta-analysis is € 3,896, with 95% CI [2078, 5713]. The highest costs arise from informal care following non-medical and medical care.

Conclusion:

The results confirm assumption that inconsistencies in cost monitoring of the treatment and care of people with dementia exists in Europe. Homogeneity could be assumed only in the medical costs of severe patients. Heterogeneity is assumed in non-medical costs, informal costs. Cost items should be defined and collected more precisely for future more precise monitoring of the economic burden.

Keywords

Alzheimer’s disease data collection uniformity developed countries meta-analysis

INTRODUCTION

In the 1990s, numerous studies [1 –5] began to highlight the problem of the increasing number of people with dementia in developed countries, especially in the context of demographic progress. This fact represents a major burden for the social and economic systems of developed countries [6 –10]. Developed countries therefore seek to devise strategic plans to deal with Alzheimer’s disease (AD) and other dementia disorders. The first stage emphasizes early diagnostics, and the following stages then add stress registers and calculations of economic burden. It is problematic, however, to compare available calculations because of differences in the breakdown of cost categories and non-unified monitoring of cost types. For example, Gervès et al. [11] state that average direct cost in France is € 8,892 per year. Schwarzkopf et al. [12] have published a study in which medical and non-medical costs were € 9,408 per year. Reese at al. [13] reported service use and costs for patients with AD, and the total cost was estimated to be € 13,080 per patient. The most important cost component was (long-term) care, which constituted 43% of total costs. Dodel et al. [14] identified the main factors associated with societal costs of AD in community-dwelling patients across three European countries. Mean monthly costs per patient differed for France (€ 1,881), Germany (€ 2,349), and the UK (€ 2,016), with informal care costs accounting for 50% to 61%.

Within the calculation of the economic burden, similar inaccuracies occur. Calculations are only based on global estimates of the number of people with dementia, and they are then applied to individual countries. More current estimates indicate that there are a higher number of people with dementia and rising costs. Wimo et al. [15] described the prognosis of dementia costs, which is only based on a prevalence prognosis that is based on the UN’s demographic forecast and combined with Eurocode’s prevalence figures for Europe. The demographic forecast of costs will result in an increase in Europe by approximately 43% between 2008 and 2030 to over € 250 bil. [16]. Costs of dementia in 2010 and 2015 according to the World Alzheimer Report (2015) [17] were as follows: Central Europe, € 14.2 bil.; Eastern Europe, € 14.3 bil.; and Western Europe, € 210.1 bil. The total cost was € 238.6 bil. in 2010. For the year 2015, they forecasted the total value for Europe to be € 301.2 bil. [17] Maresova et al. [18] estimated € 343 bil. in 2050 for AD. Costs will be considerably higher than the present prediction. However, with respect to the present data, these data are relatively imprecise, and the estimated burden calculated for the year of 2050 is calculated with a deviation of 100% of the specified average costs (range € 113, € 618–594, € 130 billion) [18]. The key issue is that there is no collection of epidemiologic data that would reliably map the prevalence and incidence of AD in Europe and other developed countries. Furthermore, there is no set model for the collection of at least partial costs that could map the economic burden of healthcare and the social system [19 –21].

The aim of this study is to note the problem of poor availability and inconsistencies in cost monitoring of the treatment and care of people with dementia in European countries, even at a time when the vast majority of European countries have adopted and approved strategic national plans to fight dementia. Furthermore, the intersection of cost items that are comparable and consistently monitored in expert studies are specified. The focus is on studies of AD.

The existence of accurate information will help to fulfil the purpose of the action plans of European countries, which is to improve the quality of life for patients with AD and similar diseases, which improves the quality of life for caregivers, raises awareness in the provision of health and social services as well as among the general public, and supports and develops education and research activities focused on dementia [22].

METHODS

Review and collection of information

Search strategy and eligibility criteria

The methodology followed the PRISMA guidelines published by Moher et al. [23]. Two investigators performed a systematic literature search of the Web of Science, Elsevier Science Direct, PubMed, and Scopus. The period covered the years from 2010 to 2016, and the electronic search included the following key words: “Alzheimer’s AND disease AND costs”. In the “Web of Science” database, 897 studies were identified. The keywords were searched within the “topic”. In the Scopus database, attention was paid to the type of output “articles.” In the “Elsevier Science Direct” database, the keywords were searched in the following sections: “Abstract”, “Title” and “KeyWords”.

A study was eligible for inclusion if the following criteria were met:

Track costs for persons suffering from AD according to the degree of dementia;

Specification of individual items of the cost groups in the study.

The monitoring of at least two cost categories and their 95% confidence intervals (CIs).

All of the studies with cost categories monitoring but without definition of the specific items included in each category were also excluded.

Data extraction and study quality evaluation

Two researchers working independently extracted from each publication the following data: author, country, study design, sample size, number of cases, age, disease ascertainment, exposure variable (stages of dementia, type of care), risk estimates with CIs. The most adjusted estimate was included in cases where a study stated more than one risk estimate. Two researchers assessed the quality of each study, using the recommended Newcastle-Ottawa Scale [24].

Altogether, 99 studies were assessed for eligibility. However, only seven studies could be used for a detailed specification of costs. Seven studies were submitted to a detailed analysis of cost groups, and these studies were expected to be included in the meta-analysis. Costs were universally divided in these studies according to levels of dementia and there were three comparable groups of costs monitored. Upon closer examination of the cost group content, it became apparent that each group monitors diverse cost items that would distort the results. Four studies were disqualified for this reason, with the three remaining studies ultimately being analysed. The specific procedure for the selection process is shown in Fig. 1.

The process provides the three studies for the meta-analysis (Darba and Kaskens [25] Gerves et al. [11], Rapp et al. [26]) The three studies use specific classifications of dementia level as presented in Table 1.

Fig.1

Results of the review.

Table 1

Cost groups and their various appellations in individual studies

Rapp et al. [26]	Gerves et al. [11]	Darbà and Kaskens [25]
Informal costs	Informal non-medical	Indirect cost
surveillance, activities of daily living (ADL), and instrumental ADL) – replacement wage method	care carried out ‘for free’ by relatives of patients – opportunity cost method	lost productivity of the caregiver – were calculated from estimates of reduced working hours per month and the loss of full and half working days per month
Non-medical costs	Formal non-medical care	Informal care costs
use of day care, alarm services, home help, and day/night surveillance	home assistance and respite care encompassing both day care and memory workshop, provided by professionals	total number of hours dedicated to basic activities of daily living and instrumental activities of daily living over the previous month as well as supervision of the patient
Medical costs	Direct sources	Direct medical
inpatient and outpatient hospital visits, physician visits, speech therapist visits, physical therapist visits, nurse visits, and medication use	medical care, encompassing AD-related drug consumption, hospital stays, physicians and nursing services	hospitalizations, emergency visits, diagnostic and monitoring tests, outpatient specialist visits, health and social care professional consultations, and health materials.

Another analysis uses the terms informal costs, non-medical costs, and medical costs.

Data unification

All data unification, statistical calculations, and meta-analyses are performed using Microsoft Excel 2016 for Windows 10 Enterprise. All statistical tests use the following sample characteristics: mean, standard deviation, and sample size. A significance level of p = 0.05 is always used. Monitored variables in meta-analysis are total costs and costs in the subcategories: informal costs, nonmedical costs, and medical costs, with respect to Mini-Mental State Exam scores (MMSE). Two studies (Gerves et al. [11] and Rapp et al. [26]) use MMSE “≤20” and “>20” as the classification of AD severity, while the study by Darba and Kaskens [25] uses Clinical Dementia Rating (CDR) score to determine mental state. A conversion scale between MMSE and CDR is described by Balsis et al. [27] However, for different types of costs measurement, Darba and Kaskens [25] use Disability Assessment for Dementia (DS DAT) and the conversion could only be approximate. In this case, some patients with moderate dementia (according to conversion between MMSE and CDR) could ultimately be included in the meta-analysis in the MMSE >20 group.

The scores “0–6” and “7-8” are grouped in the category of MMSE “>20,” while the scores “9-10” and “11–15” are grouped in the category of MMSE “<20” here (see the fourth column in Table 2) in the third study [22]. The results of the meta-analysis for total costs are without such uncertainty.

Table 2

Sample description and costs of dementia in relation to MMSE

	Gerves et al. [11]	Rapp et al. [26]	Darba and Kaskens [25]
N	54	815	338
Mean age (SD)	79(6)	79.77 (5.74)	78.9 (7.4)
Sex (% of female)	51%	68%	67%
Years since diagnosis (SD)	X	1.35 (1.65)	2.3 (2.3)
Mean (CDR) MMSE score	19(5)	19.69 (4.02)	5.5 (4.0) (CDR)
n >20 (mild)	26	375	132
n ≤20 (severe)	28	440	205
Total costs Mean (SD)
n >20 (mild)	1823.9 (3971.59)	2377.31(3380.66)	3488.03 (5028.54)
n ≤20 (severe)	4288.32 (4623.95)	3375.95 (4138.33)	8327.83 (10238.49)
p-value^*	0.04	0.001	0.001
Informal costs Mean (SD)
n >20 (mild)	1308.8 (3848.9)	1794.45 (2927.92)	70.64 (290.33)
n <20 (severe)	3321.99 (4327.52)	2792.02 (3683.85)	85.24 (279.80)
p-value^*	0.08	0.001	0.64
Non-medical costs Mean (SD)
n >20 (mild)	105.37 (134.74)	178.36 (647.59)	3303.06 (4883.49)
n ≤20 (severe)	343.9(500.4)	219.02 (806.57)	8002.23 (9738.94)
p-value^*	0.02	0.43	0.001
Medical costs Mean (SD)
n >20 (mild)	409.73 (1163.49)	404.5 (1309.98)	114.33 (197.18)
n ≤20 (severe)	622.43 (1668.94)	364.91 (1162.23)	240.36 (434.30)
p-value^*	0.59	0.65	0.001
Total costs – all severities AD Mean (SD)	3101.75 (4458.40)	2916.45 (3838.66)	5693.15 (8194.51)

p-value^* represents a two-sided p-value in a two-sample t-test in which the means in the mild group and in the severe group do not differ in the upper two rows. The last row contains the results calculated using the characteristics of Total costs in the two categories “mild” and “severe”. The last column from the study by Darba and Kaskens [25] represents the total results of the two dementia subcategories.

First, the standard weighted arithmetic mean is calculated using the specific means and the specific sample sizes in the subsets. The between-group variation and the within-group variation are calculated using sample size as a weight. Furthermore, the specific standard deviation, the differences between specific group means, and the total mean are calculated. Standard deviations of total costs for all categories of severity of AD together are not presented in the three studies. Consequently, they are calculated for each study using the same method (means, standard deviations, and sample sizes in specific group are used; see the last row in Table 2). The two categories “≤20” and “>20” represent the two groups of different patients in every study. It is also valid, for example, for the two categories “0–6” and “7-8” in the Darba and Kaskens [25] study if they are combined. If such categories are combined, then the resulting statistical characteristics (sample size, mean, and standard deviation) could be calculated using the algebraic decomposition of total variation to between-group and to within-group variation (data describes patients in the same “experiment-study” and is not a meta-analysis). Additionally, the Darba and Kaskens [25] study used the category “Social costs”, which is contained in the “Total costs” presented in the study. Consequently, the means of Total cost are calculated here as the sum of “Direct medical costs”, “Indirect care costs”, and “Informal care costs” only. Standard deviations are used without recalculations (they could not be calculated using the algebraic decomposition of total variation because the same patients are in the combined categories). The category “Social costs” represents less than 3% of Total costs in the study, and, consequently, the error of the standard deviation could be negligible.

For the purposes of comparison of cost groups, the terminology of their division had to be unified.

Meta-analysis

At first, the homogeneity of variation between the three studies is evaluated for every category of cost through a standard Q-test [29, 30]. It is tested for every individual row in Table 2 with specific arithmetic means, standard deviations, and sample sizes. The homogeneity of variations is rejected in 11 of 12 cost categories (p < 0.01). It is not rejected for medical costs in the patient category “severe” (p > 0.09). The p-values of the Q-test are in the second to last columns in Figs. 2–5. Some theoretical studies have recommended that “the Q-test only informs meta-analysts about the presence versus the absence of heterogeneity, but it does not report on the extent of such heterogeneity” [29, 30]. Furthermore, Higgins and Thompson recommended to use an I² index as a complement to the Q-test [30]. The I² index is also calculated here with the 95% confidence interval (CI), and it is in the last column in Figs. 2–5 [29]. The calculation also confirmed heterogeneity in all cases with the exception of medical costs in the severe category. Consequently, the random effects model is selected here to estimate means in meta-analysis with 95% confidence interval in 11 of the 12 cases [34, 35]. On the other hand, the fixed effects model is used for medical costs in the “severe” category [34, 35].

Fig.2

Meta-analysis of Total costs in the three groups of patients according to dementia level. p-value^* corresponds to the Q-test of homogeneity, and the random effects model is used in the three cases.

Fig.3

Meta-analysis of Informal costs in three groups of patients categorized according to dementia level. p-value^* corresponds to the Q-test of homogeneity and the random effects model is used in the three cases.

Fig.4

Meta-analysis of Non-medical costs in the three groups of patients according to dementia level. p-value^* corresponds to the Q-test of homogeneity, and the random effects model is used in the three cases.

Fig.5

Meta-analysis of Medical costs in three groups of patients categorized according to dementia level. p-value^* corresponds to the Q-test of homogeneity, ^**Medical costs in the “severe” category are homogenous, and consequently, the mean, its confidence interval and standard deviation are calculated in a fixed-effects model. The random-effects model is used for the two remaining cases.

The random effects model assumes that the observed variance could be decomposed into its two component parts, the within-studies and the between-studies components. Both parts are used when assigning the weights of individual study [34, 35]. The following weighted mean $\bar{X}$ is used in the fixed effects model: $\bar{X} = \frac{\sum_{j = 1}^{3} w_{j} . {\bar{X}}_{j}}{_{\sum_{j = 1}^{3}} w_{j}}$ (1) where $\bar{X}$ j is mean in the j-study and weights w_j are defined as the reciprocal of variation v_j inside the j-study:

$w_{j} = \frac{1}{v_{j}}$ (2)

On the other hand, weight under the random effects model $w_{j}^{*}$ in the-j study is: $\begin{matrix} w_{j}^{*} = 1 /_{(V_{j} + T^{2})} \end{matrix}$ (3) where symbol tau squared T² is the between-studies variance. The resulting mean $\bar{X}$ * in the random effects model is defined in the relationship: $\begin{matrix} {\bar{X}}^{*} = \sum_{j = 1}^{3} w_{j}^{*} \cdot {\bar{X}}_{j} /_{\sum_{j = 1}^{3} w_{j}^{*}} \end{matrix}$ (4)

Furthermore, two helping statistics Q and C are used: $\begin{matrix} Q = \sum_{j = 1}^{3} w_{j} \cdot {({\bar{X}}_{j} - \bar{X})}^{2} \\ = \sum_{j = 1}^{3} w_{j} {\bar{X}}_{j}^{2} - {(\sum_{j = 1}^{3} w_{j} \cdot {\bar{X}}_{j})}^{2} /_{\sum_{j = 1}^{3} w_{j}} \end{matrix}$ (5) $\begin{matrix} C = \sum_{j = 1}^{3} w_{j} - \sum_{j = 1}^{3} w_{j}^{2} /_{\sum_{j = 1}^{3} w_{j}} \end{matrix}$ (6)

The degree of freedom df is equal to the number of studies minus one. If Q >df (it is observed in all cases here), then the between-studies variance T² is: $\begin{matrix} T^{2} = \frac{Q - df}{C} = \frac{Q - 2}{C} \end{matrix}$ (7) where Q-df is the observed minus the expected variance and C is a scaling factor. Sample characteristics calculated in a specific subgroup in the same study are known in some cases, while the arithmetic mean and sample variance in all subgroups together are not presented. For example, two means and two standard deviations are known in the two patient groups “mild” and “severe”. The number of cases N in the category “all steps of AD” is simply the sum of cases Nj in subgroups. The arithmetic mean could be calculated using equation (1) if the weight equals the number of cases Nj in subgroups. Sample standard deviation in the combined group could be obtained using algebraic decomposition of total sample variance to between-group and within-group variance. Generally, within-group sample variance A in m groups is: $\begin{matrix} A = \sum_{j = 1}^{m} (N_{j} - 1) \cdot {SD}_{j}^{2} /_{(N - 1)} \end{matrix}$ (8) where N is the number of cases in all groups and SDj are sample standard deviations. The between-group sample variance B is: $\begin{matrix} B = \sum_{j = 1}^{m} N_{j} \cdot {({\bar{X}}_{j} - \bar{X})}^{2} /_{(N - 1)} \end{matrix}$ (9)

The resulting sample standard deviation SD in the whole sample is: $\begin{matrix} SD = \sqrt{A + B} \end{matrix}$ (10)

If the total sample characteristics are not in a specific study, then relationships (8–10) are used. For example, the two categories “0–6” and “7–8” in the Darba and Kaskens [22] study are combined in a single group, and m equals 2 here. It is also valid if all levels of dementia are combined in the single group “All steps of AD”.

RESULTS

Differences in the cost of Alzheimer’s disease in studies in Europe: Meta-analysis

Three articles met the inclusion criteria and were included in the meta-analysis. A total of 1,207 patients and their caregivers were recruited for the study: 338 in Spain [25], 815 in France by Rapp et al. [26], and 54 by Gerves et al. [11]. All community-dwelling patients were, with few exceptions, approximately 79 years old on average. The age of onset of disease increased with dementia severity and was relatively higher for patients in Spain. Most patients at the severe stages and in residential care settings were female. The mean MMSE in the studies by Gerves et al. [11] and Rapp et al. [26] was approximately 19. In the study by Darba and Kaskens [25], the CDR score was 5.5 on average (Table 2).

The proportions of female are higher than male in Table 2 in the three studies. The proportions calculated in the whole population using data from Eurostat in the two countries and in the convenient calendar years are similar. Namely, the relevant figures are: 58.7% in the whole population of France 2009–2010, while 51% in the study by Gerves et al. [11], 59.3% in the whole population of France 2003–2005 while 68% in the study by Rapp et al. [26] and 57.3% in the whole population of Spain 2011–2012, while 67% in the study by Darba and Kaskens [25]. The proportion in the whole population is higher than in the study by Gerves et al. [11] and it is not significant (p > 0.16). By contrast, the proportion in the whole population is less than in the studies by Rapp et al. [26] and by Darba and Kaskens [25] and it is significant (p < 0.0002). It could show that AD is moderately more frequent in female but the differences between studies and the whole population are not so big.

In terms of the two categories defined by disability level, the costs are generally higher in the “severe” patient category. These results are shown in Table 2. In the category of “Total costs,” this difference is statistically significant in all studies (with p < 0.04; see p-values^* in the 12th row in Table 2). The trend is similar in the “Informal costs” category and is significant in the study by Rapp et al. [26] (p < 0.001), whereas it is not significant in the study by Gerves et al. [11] (p > 0.08) or in the study by Darba and Kaskens [25] (p > 0.64). In the “Non–medical costs” category, the costs of more disabled individuals are also higher, and the difference is significant in the studies by Gerves [11] (p < 0.02) and Darba and Kaskens [25] (p < 0.001), whereas it is not significant in the study by Rapp et al. [26] (p > 0.43). “Medical costs” are higher for more disabled individuals in just two studies by Gerves et al. [11] (non-significant, p > 0.59) and Darba and Kaskens [25] (significant, p < 0.001). The “Medical costs” are conversely higher in less disabled individuals only in the study by Rapp et al. [26], and the difference is not significant (p > 0.65).

In Darba and Kaskens [25], the average costs among all patients are significantly higher than the costs in Gerves et al. [11] and Rapp et al. [26] (see the last row in Table 2; p < 0.001). This difference is probably due to a higher proportion of patients with severe intellectual disabilities in Darba and Kaskens [25] compared to Gerves et al. [11] and Rapp et al. [26]. According to the transfer between CDR and MMSE score, it would mean that there are 205 patients in the category of <20 and 132 patients in the category of >20. Several cost items are also included in each group.

The results of the meta-analysis of the data for 1,207 patients are specified in Figs. 2–5. In the mild dementia group, there are 585 patients, while there are 622 patients in the severe category. Homogeneity could be assumed only in the Medical costs of severe patients. Heterogeneity is assumed in all other cases (see the last two columns in Figs. 2–5). Therefore, the fixed effects model is used only in a single case (Medical costs among severe patients). The random effects model is used in all other cases. The highest average costs are in the Informal costs group, followed by non-medical costs and, ultimately, medical costs. The costs are higher for a higher degree of disability in the Informal costs category and in the non-medical category, whereas it is almost the same in the Medical costs category. The highest deviations are in informal costs. Generally, € 3896 with 95% CI [2078, 5713] represents cost per patient per month (see the last row in Fig. 2).

The results of both Gerves et al. [11] and Rapp et al. [26] were lower than the meta-analysis results for total costs (see Fig. 2). The results are valid in the “mild” category and “severe” category, as well as in both categories together. Gerves et al. [11] and Rapp et al. [26] correspond to similar ranges regarding costs, and the study Darba and Kaskens [25] has a higher overall cost. In certain cases when the MMSE score >20, the results of the Gerves et al. [11] study are beyond the scope of the meta-analysis for the overall costs and achieve a greater range and significantly lower values. The fundamental difference in this item may be due to the small sample population.

Gerves et al. [11] and Rapp et al. [26] had lower “Non-medical costs” than the meta-analysis results (see in Fig. 4). Conversely, in the “Informal costs” and in the “Medical costs” categories, the trend is reversed (Darba and Kaskens [25] had lower values than Gerves et al. [11] and Rapp et al. [26], as seen in Figs. 3 and 5). The most significant finding is the high average of non-medical costs in Darba and Kaskens [25]. An explanation could be found in the spectrum of costs and the cost structure. The cost structure is different between the Gerves et al. [11] and Rapp et al. [26] studies and Gerves et al. [11] and Rapp et al. [26] The main difference is that Non-medical costs account for 95% in the “mild” category and 96% in the “severe” category in Darba and Kaskens [25], whereas those values in both Gerves et al. [11] and Rapp et al. [26] are below 8%.

The difference between Darba and Kaskens [25] studies and Gerves et al. [11] and Rapp et al. [26] study is significant in Informal costs too. Informal costs represent the major item in Gerves et al. [11] and Rapp et al. [26] studies (more than 70% for both mild and sever cases) while it is the smallest item in Darba and Kaskens [25] study (less than 0.02% for both mild and sever cases).

DISCUSSION

The estimated total cost per patient per month through meta-analysis is 3,896€ with a 95% CI [2078, 5713]. The above-mentioned results show large differences in the accounting of costs that are incurred based on the monitoring period, sample size, number of patients with respect to the degree of dementia, and the scale at which dementia is measured. The main differences, however, lie in the monitored cost items.

The problem in many studies is the terminological disunity at the very beginning. Somewhere it is broken down into direct and indirect costs, and these costs are described in sub-groups as the costs of formal and informal care as well as medical and non-medical care. In terms of content, the largest match is in the group frequently referred to as direct medical costs. These costs include the cost items directly related to the treatment of patients, including medications themselves. Indirect costs are often neglected in the studies or are represented by calculating the cost of informal care.

That is also the reason why a small number of studies are included in the meta-analysis. In the monitored period the authors searched more studies but they have non-existent classifications for direct and indirect costs [36], no specification of cost items in monitored groups [37, 38], no cost classification in relation to disease states (MMSE) or found studies were outside Europe (studies were excluded due to a different healthcare system). Including these studies would mean distortions that the authors could not exactly identify.

Other studies confirm that during conducting meta-analyses based on economic variables, it is common that only few original are included. For example, Zimlichman et al. [39] has included five studies in its Health Care-Related Infectious Diseases (HAIs) meta-analysis and has already highlighted the great heterogeneity of data in such small numbers. The authors therefore analysed only those studies where they were sure with the content of the cost groups and data compatibility is ensured.

In theory, there are different kinds of cost classifications. Within strategic approaches, no methods were recommended for cost evidence that could help to unify their monitoring and reporting. For example, Rascati [20] specifies four main groups: Direct medical costs, Direct non-medical costs, Indirect costs, and Incalculable costs. Drummond [21] describes consumption sources in healthcare, a client and his family’s consumption of sources, consumption sources in other fields, and consumption sources with respect to care, and based on those result, the authors do not recommend the use of the term “incalculable costs” because they do not meet/suit/fit the definition of costs (resource consumption) and are measured by value changes in quality of life (i.e., utility) or by the willingness to pay (i.e., willingness-to-pay).

Previous studies discovered that if different methods of monitoring costs are used on the same patients, the results vary significantly [32, 33].

It is further apparent that the specification of daily life activities may prove important, particularly whether these activities are a necessary support of the patient’s health needs.

Limitations should be considered when interpreting the present data:

Establishing the amount of alternative salaries, which includes costs corresponding to the activity being performed by professionals,

Establishing the costs corresponding to ‘lost opportunity’ from partially performed activities using the ‘opportunity method’ [32].

The greatest variety and range of costs is in the area related to patient assistance services. With the appropriate terminology, these services are mostly known as social care and informal costs [40, 41]. A comparison with other studies would provide more relevant results in a cost meta-analysis, but in terms of the initial comparability of cost groups, it would be problematic. Mesterton et al. [41] distinguishes between direct medical costs, non-medical care, and informal care. Wimo et al. [15] specifies the groups of patient health care costs, patient social care costs, and caretaker healthcare costs. Jones et al. [42] in the United Kingdom only analysed direct medical costs, and Schwarzkopf et al. [12], in a German study, focused on costs of formal healthcare and costs of informal care. However, more detailed information listed in each study does not establish whether it is a valuation of the same operations.

The analysed studies show that in practice, the breakdown into direct and indirect costs is used. Within these two groups, the terminology then differs slightly, and the reference sub-groups then differ considerably. Insight into these studies can be found for the following:

medical direct costs (inpatient care, outpatient treatment, medication)

non-medical direct costs (day care centres, community health services, respite care, accommodation costs for patients)

indirect costs (time that the caretakers dedicate to the patient)

The availability of information concerning the above cost groups is currently very complicated in European countries because there is no single database [43, 44], not even at the level of individual countries themselves. A large part of direct costs relates to the healthcare sector. These costs are recorded and can be traced. The problem arises with the cost of social service providers in which health insurance companies do not cover the entire costs of nurses’ salaries or operations.

Conclusion

Care service providers in most European countries do not follow the distribution of provided care tasks, especially for people with AD or other forms of dementia. Therefore, it is not possible to ascertain the overall cost of care. Indirect costs primarily include an overview of the patient and the caretakers’ lost productivity. The number of hours is estimated to be typically in the range of 8–17 hours. In this regard, it should be sufficient to follow the studies conducted in Europe.

Gaining information about direct medical costs can be achieved through health insurance companies, although legal access to such information is restricted by the relevant laws on free access to information in individual countries. However, these restrictions often do not apply to aggregate data and to the possible creation of new information.

Presently, when many developed countries do not have a certain budget allocated for a strategic plan to fight dementia, it is also more difficult to accomplish objectives in the area of data collection, and acquisition of these data is currently reliant on the willingness of individuals in health and social facilities. As mentioned in one of the action plans in Europe, [22] the existence of more precise data would definitely contribute to the improvement of the quality of life of patients with AD and their relatives.

Footnotes

ACKNOWLEDGMENTS

This study was supported by the Alzheimer’s Endowment Fund, Czech Republic and project Excellence 2018 (University of Hradec Kralove, Faculty of Informatics and Management).

Authors’ disclosures available online ().

References

Maurer

, Volk

, Gerbaldo

(1997) Auguste D and Alzheimer’s disease. Lancet 349, 1546–1549.

Mega

, Cummings

, Fiorello

, Gornbein

(1996) The spectrum of behavioral changes in Alzheimer’s disease. Neurology 46, 130–135.

Boada

, Peña-Casanova

, Bermejo

, Guillén

, Hart

, Espinosa

, Rovira (1999) Costs of health care resources of ambulatory-care patients diagnosed with Alzheimer’s disease in Spain. Med Clin (Barc) 113, 690–695.

Ernst

, Hay

(1994) The US economic and social costs of Alzheimer’s disease revisited. Am J Public Health 84, 1261–1264.

Ernst

, Hay

, Fenn

, Tinklenberg

, Yesavage

(1997) Cognitive function and the costs of Alzheimer disease. An exploratory study. Arch Neurol 54, 687–693.

Marešová

, Zahálková

(2016) The economic burden of the care and treatment for people with Alzheimer’s disease - the outlook for the Czech Republic. Neurol Sci 37, 1917–1922.

Maresova

, Mohelska

, Dolejs

, Kuca

(2015) Socio-economic aspects of Alzheimer’s disease. Curr Alzheimer Res 12, 903–911.

Riggs

, Millecchia

(1992) Using the Gompertz-Strehler model of aging and mortality to explain mortality trends in industrialized countries. Mech Ageing Dev 65, 217–228.

Riggs

(1992) Rising mortality due to Parkinson’s disease and amyotrophic lateral sclerosis: A manifestation of the competitive nature of human mortality. J Clin Epidemiol 45, 1007–1012.

10.

Strehler

, Mildvan

(1960) General theory of mortality and aging. Science 132, 14–21.

11.

Gervès

, Chauvin

, Bellanger

(2014) Evaluation of full costs of care for patients with Alzheimer’s disease in France: The predominant role of informal care. Health Policy 116, 114–122.

12.

Schwarzkopf

, Menn

, Kunz

, Holle

, Lauterberg

, Marx

, Mehlig

, Wunder

, Leidl

, Donath

, Graessel

(2011) Costs of care for dementia patients in community setting: An analysis for mild and moderate disease stage. Value Health 14, 827–835.

13.

Reese

, Hessmann

, Seeberg

, Henkel

, Hirzmann

, Rieke

, Baum

, Dannhoff

, Müller

, Jessen

, Geldsetzer

, Dodel

(2011) . Cost and care of patients with Alzheimer’s disease: Clinical predictors in German health care settings. J Alzheimers Dis 27, 723–736.

14.

Dodel

, Belger

, Reed

, Wimo

, Jones

, Happich

, Argimon

, Bruno

, Vellas

, Haro

(2015) Determinants of societal costs in Alzheimer’s disease: GERAS study baseline results. Alzheimers Dement 11, 933–945.

15.

Wimo

, Reed

, Dodel

, Belger

, Jones

, Happich

, Argimon

, Bruno

, Novick

, Vellas

, Haro

(2013) The GERAS study: A prospective observational study of costs and resource use in community dwellers with Alzheimer’s disease in Three European Countries – study design and baseline findings. J Alzheimers Dis 36, 385–399.

16.

Wimo

Cost of illness and burden of dementia in Europe - Prognosis to 2030. Alzheimer Europe. http://alzheimer-europe.org/Research/European-Collaboration-on-Dementia/Cost-of-dementia/Prognosis-to-2030, Accessed on July 9, 2015.

17.

World Alzheimer Report 2015. The Global Impact of Dementia an analysis of prevalence, Incidence, cost nd trends. https://www.alz.co.uk/research/WorldAlzheimerReport2015.pdf, Last updated 2016, Accessed on April 9, 2016.

18.

Maresova

, Klimova

, Novotny

, Kuca

(2016) Alzheimer’s and Parkinson’s diseases: Expected economic impact on Europe—a call for a uniform European strategy. J Alzheimers Dis 54, 1123–1133.

19.

Czech Alzheimer Society (2015) http://www.alzheimer.cz/, Last updated 2016, Accessed on June 9, 2016.

20.

Rascati

(2014) Essentials of pharmacoeconomics. Wolters Kluwer Health/Lippincott Williams, Philadelphia.

21.

Drummond

, O'Brien

(2005) Methods for the economic evaluation of health care programmes. 3rd ed. Oxford University Press, New York.

22.

The Czech National Action Plan for Alzheimer’s Disease (2016) http://www.mzcr.cz/Odbornik/dokumenty/narodni-akcni-plan-pro-alzheimerovu-nemoc-a-dalsi-obdobna-onemocneni-na-leta-201_11429_2785_3.html, Last updated 2016, Accessed on April 9, 2016.

23.

Moher

, Liberati

, Tetzlaff

, Altman

(2009) The PRISMA Group. Preferred reporting items for systematic review and meta-analysis: The PRISMA statement. PLoS Med 6, e1000097.

24.

Wells

, Shea

, O’Connell

, Robertson

, Peterson

, Welch

, Losos

, Tugwell

(2011) The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analysis. [Website]. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp

25.

Darba

, Kaskens

(2015) Relationship between patient dependence and direct medical-, social-, indirect-, and informal-care costs in Spain. Clinicoecon Outcomes Res 7, 387–395.

26.

Rapp

, Andrieu

, Molinier

, Grand

, Cantet

, Mullins

, Vellas

(2012) Exploring the relationship between Alzheimer’s disease severity and longitudinal costs. Value Health 15, 412–419.

27.

Balsis

, Benge

, Lowe

, Geraci

, Doody

(2015) How Do Scores on the ADAS-Cog, MMSE, and CDR-SOB Correspond? Clin Neuropsychol v29, 1002–1009.

28.

Hurley

, Volicer

, Hanrahan

, Houde

, Volicer

(1992) Assessment of discomfort in advanced Alzheimer patients. Res Nurs Health 15, 369–377.

29.

Higgins

JPT

, Thompson

, Deeks

, Douglas

(2003) Measuring inconsistency in meta-analyses. BMJ 327, 557–560.

30.

Medina

, Meca

, Martines

, Bottela

(2006) Assessing heterogeneity in meta-analysis: Q statistic or I² index? Psychol Methods 11, 193–206.

31.

Higgins

JPT

, Thompson

(2002) Quantifying heterogeneity in a meta-analysis. Statist Med 21, 1539–1558.

32.

Berg

, Brouwer

, van Exel

, Koopmanschap

, Bos

, Rutten

(2006) Economic valuation of informal care: Lessons from the application of the opportunity costs and proxy good methods. Soc Sci Med 62, 835–845.

33.

Posnett

, Jan

(1996) Indirect cost in economic evaluation: The opportunity cost of unpaid inputs. Health Economics 5, 13–23.

34.

Borenstein

, Rothstein

(2007) Meta-Analysis Fixed effect vs. random effects, www.Meta-Analysis.com, https://www.meta-analysis.com/downloads/Meta-analysis%20fixed%20effect%20vs%20random%20effects.pdf

35.

Fang

(2005) Medical statistics and computer experiments, World Scientific, New Jersey.

36.

Gillespie

, O’Shea

, Cullinan

, Lacey

, Gallagher

, Ni Mhaolain

, Lawlor for the Enhancing Care in Alzheimer’s Disease (ECAD) Study Team (2013) The effects of dependence and function on costs of care for Alzheimer’s disease and mild cognitive impairment in Ireland. Int J Geriatr Psychiatry 28, 256–264.

37.

Reed

, Happich

, Argimon

, Haro

, Wimo

, Bruno

, Dodel

, Jones

, Vellas

, Belger

(2017) What drives country differences in cost of Alzheimer’s disease? An explanation from resource use in the GERAS study. J Alzheimers Dis 57, 797–812.

38.

Lenox-Smith

, Reed

, Lebrec

, Belger

, Jones

(2016) Resource utilisation, costs and clinical outcomes in non-institutionalised patients with Alzheimer’s disease: 18-month UK results from the GERAS observational study. BMC Geriatr 16, 195.

39.

Zimlichman

, Henderson

, Tamir

, Franz

, Song

, Yamin

, Keohane

, Denham

, Bates

(2013) Health care–associated infections: A meta-analysis of costs and financial impact on the US health care system. JAMA Intern Med 173, 2039–2046.

40.

Rattinger

, Schwartz

, Mullins

, Corcoran

, Zuckerman

, Sanders

, Norton

, Fauth

, Leoutsakos

, Lyketsos

, Tschanz

(2015) Dementia severity and the longitudinal costs of informal care in the Cache County population. Alzheimers Dement 11, 946–954.

41.

Mesterton

, Wimo

, Langworth

, Winblad

, Jönsson

(2010) Cross sectional observational study on the societal costs of Alzheimer’s disease. Curr Alzheimer Res 7, 358–367.

42.

Jones

, Romeo

, Trigg R Knapp

, Sato

, King

, Jones

, Romeo

, Trigg R Knapp

, Sato

, King

(2014) Dependence in Alzheimer’s disease and service use costs, quality of life, and carer burden: The DADE study. Alzheimers Dementia 11, 280–290.

43.

Maresova

, Mohelska

, Kuca

(2015) Social and family load of Alzheimer’s disease. Appl Econ 48, 1936–1948.

44.

Mohelska

, Maresova

, Valis

, Kuca

(2015) Alzheimer’s disease and its treatment costs: Case study in the Czech Republic. Neuropsychiatr Dis Treat 11, 2349–2354.