Background
The syndrome of dementia is characterized by a loss of memory and other mental abilities that are severe enough to interfere with daily life [
1]. Most often, dementia is caused by a chronic and progressive neurodegenerative disease, usually Alzheimer’s disease. Worldwide, there are more than 46.8 million persons living with dementia. This number is expected to double every 20 years, reaching 74.7 million in 2030 and 131.5 million in 2050 [
2‐
4]. In Germany, the current number of persons with dementia is estimated to be over 1.6 million, with an annual incidence of over 300,000 new cases [
5]. From an economic point of view, dementia is the main cause of long-term institutional care in the older population and is therefore associated with substantial healthcare costs [
6]. Specifically, the total worldwide cost of dementia was estimated at 784 billion € (US$ 818 billion; 1 € = US$ 1.043, exchange rate as of 16 December 2016) in 2016. Thus, dementia is one of the most expensive diseases in old age [
2,
7].
Persons with dementia require a timely diagnosis as a basis for adequate and cost-effective drug and nondrug treatments to delay the progression of the disease and diminish increasing healthcare costs [
8,
9]. Diagnosing dementia involves determining the presence of dementia as well as nosological diagnosis of the specific causes of the syndrome.
The criteria for diagnosing Alzheimer’s disease, the most frequent cause of dementia in older people, can serve as a blueprint for dementia diagnosis procedures. In agreement with international guidelines [
10‐
12], the national guidelines on dementia care of the German Association for Psychiatry, Psychotherapy and Psychosomatics and the German Association for Neurology [
13] recommend using the clinical history, medical and neurological examinations, and assessments of cognitive functions (e.g., Mini-Mental State Examination (MMSE) [
14,
15], DemTect [
16], Clock Drawing Test [
17]) in the first step. If the syndrome of dementia or mild cognitive impairment (MCI) is detected, further investigations include blood sampling and cranial computer tomography (CT) or magnetic resonance imaging (MRI) on a regular basis. Under certain circumstances, biomarker-based diagnosis is recommended, including neurodestruction markers from cerebrospinal fluid puncture (CSF) and metabolic and molecular markers from positron emission tomography (PET). Complete adherence to these recommendations will typically only be possible in specialized care settings, such as in memory clinics. Primary care guidelines are oriented around these recommendations and allow the diagnosis of dementia diseases in primary care as well (e.g., the guidelines of the German College of General Practitioners and Family Physicians – DEGAM [
18]), but they are rarely implemented in clinical routine [
19‐
21]. Approximately 50–80% of persons with dementia do not receive a dementia diagnosis [
19‐
21], and between 45 and 55% of persons with a dementia diagnosis receive a diagnosis of unspecific dementia in primary care [
22,
23].
Only two studies have so far evaluated the diagnostic work-ups and cost of diagnosing dementia in Sweden [
24,
25]. The cost per diagnosed patient has been estimated to be between 477 € (US$ 497) at the primary care level and 1115 € (US$ 1163) at the specialist level. Wimo et al. [
24] further identified age and cognitive impairment as crucial cost drivers. However, little is known about cost differences in diagnosing dementia referring to specific dementia disorders, such as Alzheimer’s diseases or cerebrovascular disease. The costs of dementia diagnosis in different settings and related to different underlying causes of dementia are important for assessing the feasibility of a more rigorous implementation of diagnostic procedures in routine care, including primary care.
Aims of the study
The objectives of this study were to analyze the cost of diagnosing dementia in patients with suspected dementia at a German memory clinic in specialized care, to evaluate the differences in the cost of diagnosing dementia in relation to different dementia disorders, and to determine sociodemographic and clinical factors associated with the cost of diagnosing dementia.
Discussion
This study demonstrates the costs of diagnosing different dementia disorders for patients with suspected dementia in a specialized memory clinic on the basis of assessed time-related clinical consultations and the specific technical procedures used to identify relevant biomarkers. The total cost per patient and process was 501 €, and was 659 € for patients who were diagnosed with dementia. The cost of diagnosing different dementia disorders varied between 649 € for patients diagnosed with Alzheimer’s disease and 705 € for patients diagnosed with unspecific dementia. The cost of a true dementia case (nondementia case) was 1134 € (897 €).
These costs compare with the slightly higher costs reported in two previous studies that amounted to 1334 € and 1298 €, respectively, for diagnosing a true dementia case and 1115 € per patient with suspected dementia at a specialized care level [
24,
25]. Jedenius et al. [
25] evaluated the total costs associated with diagnosing dementia diseases from the beginning of the diagnostic process to the time when a dementia diagnosis was established or rejected using a prospective time-related and resource-related study, which is comparable to the methods used in our analysis. However, only 40 patients with suspected dementia at the age of 83 were included. The sample of Wimo et al. [
24] was at a comparable mean age of 82. Thus, these two previous samples were older than those in our analysis (mean age 73). Because of the exponential increase of dementia prevalence with age, the inclusion of an older sample with suspected dementia will likely lead to a higher rate of finally diagnosed dementia patients. In the study by Jedenius et al. [
25], two-thirds of patients were diagnosed with dementia diseases at the end of the diagnostic process. This rate is higher compared with those reported in another longitudinal trial [
37]. This trial reported that, on average, 50% of patients with suspected dementia had a confirmed diagnosis, which is comparable to the 44% in our sample. This lower rate of dementia cases should lead to a higher cost for one true dementia case. However, the cost of diagnosing dementia in our study was favorable compared to previous studies. The major reason for the lower evaluated costs could be the variability in the technical procedures used and in the cost of diagnostic procedures between different countries. The unit costs for the staff involved in the diagnostic processes as well as for the diagnostic procedures used, such as CT, CSF, and MRI, were slightly higher in Sweden [
24]. For example, the reimbursement rate for a CSF (CT) was 130 € (217 €) in specialized care in Sweden, but 118 € (172 €) in Germany, if the purchasing-power parity in both countries remains unconsidered. Furthermore, based on a subsample of only 10 patients, Jedenius et al. [
25] estimated the costs for diagnosing dementia by adding the entire diagnostic cost at the primary care level (general practitioner based) to the cost of diagnosing dementia at the specialist level. In this analysis, primary care consultations are not taken into account and primary care technical procedures conducted previously are solely included. These could be the reasons for the demonstrated higher costs compared to the study of Jedenius et al. [
25]. However, it seems that the demonstrated costs are estimated under uncertainty, and thus further research is needed to confirm these estimates with larger samples in different settings.
The wide variation of cost associated with different dementia disorders is related to differences in the complexity of diagnostic processes as well as to higher frequencies of imaging or biomarker-based diagnostic procedures. The diagnosis of “unspecific dementia” was given in five cases and was associated with the highest costs. Compared to patients diagnosed with Alzheimer’s diseases, vascular dementia, or mixed dementia, the unspecific dementia diagnosis after a comprehensive examination was associated with longer process times, higher utilization of diagnostic procedures, and enhanced neuropsychiatric tests. This was due to the fact that those with dementia disease presented with more atypical features compared with the other patients.
In Germany, reimbursement for a diagnostic process in a memory clinic is neither standardized nor adjusted for different dementia diagnoses and their associated costs. The financing of memory clinics could therefore differ tremendously. First of all, “memory clinic” is not a protected term, leading to several possibilities of institutional connections. Thus, each memory clinic has an individual basis for the reimbursement. In most cases, reimbursement is based on flat-rate payments or special agreements with health insurers, and the financial means provided by those insurers can be insufficient to fully cover the cost of the diagnostic process of each patient. Therefore, most clinics use additional funds for compensation, such as donations [
38]. However, reimbursement is not based on procedures that are necessary to reach a final diagnosis. It could be possible that this represents an incentive toward fewer diagnostic procedures and declaring a diagnosis with greater remaining uncertainty, especially for untreatable forms of dementia. Paying by the final diagnosis would seem to create an incentive to identify specific dementia diseases. This may be informed by the sequence of diagnostics that are typically performed. However, it may be open to manipulation in the future. Thus, designing a reimbursement system that aligns the incentives of physicians, patients, and the public payer could be very difficult. Even though sufficient funding by health insurance would enable annual planning of diagnostic processes and sustainable reimbursement, memory clinics are actually not standardized according to their medical services or their connection to hospital structures. The development of homogeneous structures could help to establish memory clinics more often and to initiate sustainable standardized funding. The evaluation of cost differences when diagnosing dementia diseases is very important to standardizing a reimbursement system for memory clinics.
However, the currently unstandardized financing structure represents a risk for diagnostic facilities, such as memory clinics. Diagnostic processes for atypical cases are by far the most expensive in a specialized memory clinic, even if only a small proportion of the total sample (4%) is diagnosed with unspecific dementia. In primary care, between 45 and 55% of persons with dementia receive an unspecific dementia diagnosis by general practitioners [
22,
23]. However, Wucherer et al. [
39] revealed that a dementia diagnosis is not a prerequisite for a guideline-based medication treatment in primary care: 38% of cognitively impaired GP patients without dementia diagnosis are treated with anti-dementia drugs. Thus, it is unclear why there are many undiagnosed and unspecific cases in GP practices. Given that cases with unspecific dementia in memory clinics are few (4% in this analysis) and complex (highest average cost in this analysis), it can be assumed that cases of unspecific dementia in primary care are probably clear but seemingly not fully clarified with the risk of inadequate treatment. However, this is neither clear nor established.
The shift of diagnostic criteria from phenomenological to cost-intensive biomarker-based procedures could increase the number of specific dementia disorders and reduce cases with unspecified dementia, leading to adequate treatment. Therefore, reimbursement for such procedures has the potential to improve dementia treatment and reduce overall dementia costs per person, for example, due to an initiated anti-dementia drug treatment resulting in delayed progression and finally institutionalization [
40]. The many non-clarified cases in primary care are relevant to determining excess costs for dementia diagnosis in primary care.
According to different cost drivers of diagnostic procedures in dementia, Wimo et al. [
24] revealed that in newly diagnosed dementia patients, their age and cognitive impairment were significantly associated with a higher cost of diagnosing dementia diseases. Their retrospective analysis includes a sample of diagnosed patients with dementia and no patients with suspected dementia. For patients with suspected dementia, we found that the final dementia diagnosis (dichotomous: yes, no) was the only factor associated with a higher cost of diagnosing dementia diseases after adjusting for age, sex, comorbidity, medication, and cognitive impairment. If the categorization of the MMSE was included and the final dementia diagnosis excluded from the multivariate model, MMSE score 20–26 was significantly associated with a higher cost for the entire diagnostic process (
b = 214, CI
– 46, CI
+ 382,
p = 0.01). Thus, this observation demonstrates the nonlinear correlation between the costs of diagnosing dementia and patients’ cognitive impairment, leading to a peak in patients with MCI. This finding, however, seems plausible and comparable to the finding of a higher cost for patients with unspecific dementia. Whereas fewer diagnostic processes are needed after detecting severe or no cognitive impairment, mild cognitive cases have to be clarified using further comprehensive neuropsychological assessments and technical procedures, resulting in higher cost for the total diagnostic process.
Furthermore, we found evidence for a negative association between diagnostic costs and patients’ age (
b = – 17, CI
– – 34, CI
+ 0,
p = 0.054) if only newly diagnosed patients were included in the multivariate model. This is comparable with the findings of Wimo et al. [
24]. A possible explanation could be that, in older patients, a decision must be made between the strain and stress imposed by the diagnostic procedure and a precise diagnosis. Because of the possibility of drug-related problems in older age, it is therefore possible that these patients were less likely to be treated with anti-dementia drugs, meaning that the distinction between Alzheimer’s disease and other dementia disorders may be regarded as less important [
24]. In addition, older patients are more comorbid, which often results in polypharmacy. Older patients are more likely to have medical conditions that lead to an adapted diagnostic work-up; for example, omitting CSF due to anticoagulation or CT instead of MRI due to the existence of an implanted pacemaker. Furthermore, the association between age and cost could occur due to the possibility that younger patients, who are mostly less cognitively impaired, receive more tests and more diagnostic procedures to clarify their deficits in cognition. However, this study indicated that sociodemographic and clinical factors do not have any impact on total diagnostic expenditures in a sample of patients with suspected dementia.
The overall cost of diagnosing the total sample of 120 patients was, on average, 60,000 €, meaning that the cost of a true dementia case was 1134 €. First, it is important to note that diagnosing dementia is not the only function of a memory clinic; its tasks also include the diagnosis of different neuropsychiatric disorders. However, if we assume that approximately one-third of the 300,000 new incident dementia cases in Germany are being referred to specialized memory clinics [
5], the national diagnostic cost would be more than 113 million €. However, Germany spent over 10.5 billion € annually for persons with dementia from a payer perspective [
41]. Thus, the cost of diagnosing dementia disease represents only a small proportion of 1% at the current rate of referral. However, a timely diagnosis allows prompt initiation of pharmacological and nonpharmacological interventions and prevents inappropriate treatment of patients with false-positive diagnoses [
42,
43]. These opportunities could lead to a reduction in healthcare costs, especially due to a delayed need for care and institutionalization [
44‐
46]. Thus, this small percentage of diagnostic costs among total expenditures on dementia diseases can result in a substantial reduction of lifetime patient costs, exceeding the cost of diagnosing dementia. Lee et al. [
42] underlined that the cost-effectiveness of biomarker analysis depends critically on the prevalence of Alzheimer’s disease in the tested population. Specifically, patients with suspected dementia referred to memory clinics have a higher pretest prevalence of Alzheimer’s disease (exceeding 15%) than patients with memory complaints in, for example, GP practices. This fact leads to potential cost savings and thus to cost-effectiveness [
42]. However, currently there is no curative treatment available for dementia diseases. For the upcoming disease-modifying therapies, biomarker-based diagnosis will be even more relevant because these new therapies will be very expensive and only effective in a small group of patients with specific clinical and biomarker characteristics. Cost-effectiveness of these therapies will be strongly related to the reliability and validity of diagnosis. The demonstrated results provide prerequisites for such analyses, especially the diagnostic costs identified for different dementia disorders.
Our study has some limitations. First, our data were derived from only one memory clinic in Germany. Furthermore, there were some missing values for each assessed clinical consultation. For two processes (test evaluation and preparation of the physician letter), 40–57% of the duration data were missing. To handle these missing values, we used univariate imputation by linear regression. This method has emerged as a principle method for dealing with missing data and is particularly useful for large imputation procedures. According to the estimated costs, the clinical consultations conducted by staff of the memory clinic represent only a small proportion of the total cost compared to the costs of the technical diagnostic procedures, such as imaging or blood and CSF testing. Therefore, any discrepancies in cost due to missing values for time-related clinical consultations should not be too large. In addition, test evaluation and preparing the physician letter are relatively uniform procedures so that the margin of possibly induced errors appears to be small. Second, the sample size was low, meaning that the demonstrated costs of diagnosing dementia are not representative for the entire population of patients with suspected dementia in Germany, especially for the demonstrated cost of diagnosing unspecific dementia (n = 5). However, the proportion of finally diagnosed patients with suspected dementia is comparable to that in other longitudinal studies. Therefore, it seems that the demonstrated costs of diagnosing dementia are representative for newly diagnosed patients with suspected dementia in specialized care.