SEARCH

SEARCH BY CITATION

Keywords:

  • Alzheimer's disease;
  • donepezil;
  • functional neuroimaging;
  • PET

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Background:  Functional neuroimaging, including positron emission tomography (PET), has been proposed for use in diagnosing Alzheimer's disease-related dementia (AD).

Objective:  The objective of this study was identify the circumstances under which PET scanning for the diagnosis of AD maximizes health outcomes.

Methods:  A Markov-model-based decision analysis was conducted using estimates derived from the literature on AD epidemiology, the accuracy of PET, and donepezil treatment efficacy. The target population for the analysis was assumed to be US men and women who either have mild AD or are asymptomatic but at an elevated risk of developing AD owing to disease in a first-degree relative (parent or sibling). The time horizon was the patient lifetime. We compared treatment 1) based on an American Academy of Neurology (AAN) clinical evaluation either alone; 2) in combination with PET scanning; or 3) empirically based on a family history. Outcomes measures were life expectancy, quality-adjusted life-years (QALYs), and (severe) dementia-free life expectancy (SDFLE).

Results:  For both patient populations, treating all patients based on an AAN evaluation without further testing using PET resulted in the greatest gains in life expectancy, QALYs, and SDFLEs. PET-based testing was the second preferred strategy compared to no intervention. The rankings of the strategies were sensitive to severity of treatment complications: analyses of hypothetical treatments with the potential for severe complications indicated that testing was preferred if the treatment was effective but had moderate complications.

Conclusions:  These results suggest that current treatments, which are relatively benign and may slow progression of disease, should be offered to patients who are identified as having AD based solely on an AAN clinical evaluation. A clinical evaluation that includes functional neuroimaging based testing will be warranted, however, when new treatments that are effective at slowing disease progression but have the potential for moderate to severe complications become available.


Introduction

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Each year approximately 360,000 individuals manifest Alzheimer's disease (AD), the most common form of dementia [1]. The fact that most of these individuals are over 65 years of age (30–50% of individuals in the eighth to ninth decades have AD [2]) and that the proportion of the population that will be aged ≥ 65 years is projected to increase from approximately 12% to 20% during the next three decades highlights the need to effectively identify and treat these individuals.

The diagnosis of AD hinges on identifying patients with dementia, an acquired syndrome of decline in memory and other cognitive functions that impairs daily activities in an alert patient [3,4]. Because a definitive diagnosis involves a histopathological examination of the brain tissue, which usually does not occur until after the patient has died [5], the current standard for diagnosis is a clinical evaluation as recommended by the American Academy of Neurology (AAN), which includes a complete history, physical and neuropsychiatric evaluation, and structural imaging tests to rule out non-AD causes of dementia [6].

Functional neuroimaging tests, including single-photon emission tomography (SPECT) and positron emission tomography (PET) have been proposed for use in evaluating patients for AD. While neither is currently recommended in the routine evaluation of AD-related dementia [6], functional neuroimaging may be useful for demonstrating the characteristic anatomical deficits seen among patients with AD [7–10].

Recent advances in the treatment of AD, including use of acetylcholinesterase inhibitors and the recent development of newer, more effective, but potentially more toxic therapies, raise important questions about the current and future role of functional neuroimaging in the diagnosis of AD-related dementia [11–14]. The aims of this article are to perform a decision analysis that compares standard AAN clinical evaluation with functional neuroimaging to clinical evaluation alone and to determine which strategy maximizes health outcomes. For this analysis we used PET scanning as an example of a functional neuroimaging test and donepezil as an example of an AChE-I treatment for AD. The results of the decision analysis were then used to identify potential future situations that would favor the adoption of existing or improved versions of these tests.

Methods

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Decision Model Structure

We constructed a state-transition Markov model (Fig. 1) using DATA 3.5 (TreeAge Software, Inc., Williamstown, MA). The model was used to depict the natural history of dementia, AD and non-AD, in a cohort of men and women and determine how treatment, based on an AAN evaluation with or without further testing using PET, would influence the natural history of the disease. Details of the model including sources of data and assumptions used will be presented later in this article. It should be noted that if evidence was lacking, we chose assumptions that would bias the analysis in favor of using PET.

image

Figure 1. Markov model depicting allowed transitions between states. Patients start in either the mild dementia state or the asymptomatic, at elevated risk state, depending on the scenario being modeled.

Download figure to PowerPoint

The states in the model that were used to depict both AD and non-AD dementia in order of increasing severity were: 1) asymptomatic but at an elevated risk owing to disease in a first-degree relative [6,15]; 2) mild cognitive impairment (MCI), a clinical state for individuals who were cognitively impaired but did not meet the clinical criteria for dementia [16,17]; 3) mild dementia; 4) moderate dementia; 5) severe dementia; and (6) death [5].

Intervention strategies were evaluated for three patient populations—those with mild dementia, MCI, or who were asymptomatic, but at an elevated risk for developing AD. These patient populations were chosen because they represented either patients for whom therapy is currently recommended (mild dementia) [12] or a population who might be treated if advances in therapy offered the possibility of delaying or even preventing the onset of symptomatic AD or MCI, asymptomatic, but at an elevated risk. Because the findings for patients with MCI were similar to those with mild dementia (possible AD) we have omitted the results.

All patients with mild dementia were assumed to enter the model having already undergone a clinical evaluation as per AAN guidelines and thus were considered to have AD [6]. The strategy of treating all patients based solely on an AAN clinical evaluation was referred to as the “AAN-based treatment” strategy. The second strategy, referred to as the “AAN + PET-based treatment” strategy, included a further evaluation of patients with PET scanning with treatment of patients with positive PET results. The strategy of treating asymptomatic, at elevated risk, patients based on a family history of AD was referred to as “empiric treatment.” Finally, treatment based on a positive PET scan in this group of patients was referred to as the “PET-based treatment” strategy. For comparison, a no intervention/natural history arm, in which a patient received neither testing nor AD-specific therapy was included. All three patient populations were limited to age ≥65 years.

Data Sources and Assumptions

Although the current gold standard for diagnosing AD is an AAN-based clinical evaluation [6], we assumed that the prevalence of AD would be <100% if the clinical evaluation was performed in nonspecialized settings [18]. For the base case, a conservative prevalence estimate of 56% was assumed for mild AD dementia patients based on the prevalence of AD in patients presenting to a cognitive impairment clinic [19]. For the asymptomatic population, we used the cumulative lifetime risk of developing AD (50%) among individuals with a family history of AD [20–22].

Yearly transition probabilities for the natural history component were calculated by age (65–100 years) and state (asymptomatic, MCI, mild, moderate, severe) based on a literature search, Consortium to Establish a Registry for Alzheimer's Disease (CERAD) data, and expert opinion (details available from http://www.clinpol.mc.duke.edu/Pubs/Publications/publications.html) [23]. Transitions to more severe states were assumed to be independent of age and to depend only on symptoms of dementia and not on underlying AD [24]. Thus, patients with Alzheimer's and non-Alzheimer's dementia were assumed to progress to more severe states at the same rate. Age-specific mortality was calculated based on the severity of dementia with more severe dementia states associated with a higher mortality [23].

Estimates of PET sensitivity and specificity for patients with mild dementia were based on a comprehensive review and meta-analysis of the literature [25]. For the base case, sensitivity was estimated to be 86% (range 74–92%), and specificity, 87% (range 78–93%). Owing to lack of data, we applied the same estimates to the asymptomatic population. To account for differences in test performance or newer tests, sensitivity analyses were conducted over the full range, 0% to 100%.

Treatment efficacy was modeled as a transition probability multiplier, K, which influenced the probability that a treated patient with underlying AD would progress to a more severe state in 1 year, the cycle length for the model. K was estimated to be 0.72, corresponding to a risk ratio for progression of 0.72, using information from a representative clinical trial [12,23], and applied to transition probabilities for patients identified as having AD based on a clinical evaluation, PET scan, or family history of AD. Treatment was assumed to begin at age 65 for all patients. Treatment was discontinued once patients developed severe dementia. Patients with mild dementia were treated for the duration of the clinical trial, which lasted for 18 months [12]: this assumption was varied in sensitivity analyses. In the absence of clinical evidence, we assumed asymptomatic individuals would respond similarly to therapy. We assumed that treatment initiated in the asymptomatic population would continue until they developed severe dementia, but varied treatment duration widely in sensitivity analyses. Pa-tients who did not have underlying AD but were positive based on their AAN exam and/or PET scan were assumed to receive no benefit from treatment.

For the base case, we assumed that 15% of patients experienced treatment complications, resulting in patients discontinuing treatment, with no serious long-term side effects [12]. To account for new treatments that might potentially have serious side effects, the risk of complications was modeled as consisting of any combination of short and long-term disutility or decrement in quality of life, an increased risk of progression of symptoms, or an increased risk of death. The latter two complications were modeled as relative risks and applied to transitions to either a more severe state or death.

Outcomes

Strategies were compared by calculating the incremental benefit, defined as the additional clinical benefit, for one strategy compared to the next less effective strategy. The impact of the different strategies on mortality was measured using life ex-pectancy (LE). Morbidity was captured using three measures: quality-adjusted life-years (QALYs), severe dementia-free LE (SDFLE, for patients with mild dementia), and dementia-free LE (DFLE, for patients who are asymptomatic, but at an elevated risk for AD. For the base case, LE, QALYs, and (S)DFLEs were calculated. For one- and two-way sensitivity analyses, measures that captured morbidity were calculated except when complications were fatal or resulted in an increased risk of progression to a more severe state, in which case LE was calculated. All outcomes were discounted at an annual rate of 3%[26].

Weights for the quality adjustments were derived from the literature and applied to the dementia natural history states [27]. Complications associated with theoretical treatments were modeled as disutilities. Because our model uses a year-long cycle and any disutility associated with a donepezil-related complication would last for a few days at most with the patient discontinuing treatment immediately, we assumed that patients would discontinue using donepezil at the start of a given cycle and that there would be no disutility associated with complications for the base case [12]. In sensitivity analyses, we varied the disutility associated with complications to account for severity, with a decrement of 20% over a lifetime and a 1-year, 50% decrement. For comparison, an average disutility of 13% associated with erectile dysfunction among survivors of prostate cancer has been reported in the literature [28].

Analytic Strategy

Base-case analyses were conducted using the best-available estimates as inputs for the model. One-way sensitivity analyses were conducted using the plausible ranges for each model input. Variables that had the largest impact on the choice of strategy were compared using two-way sensitivity analyses.

Role of the Funding Source

This study was funded by the Agency for Healthcare Research and Quality. The authors had complete independence in the design, conduct and reporting of the study.

Results

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Base-Case Results

In the natural history strategy, patients with mild dementia were projected to live 7.82 years, 4.02 QALYs, on average. SDFLE was 3.86 years. The AAN-based treatment strategy was the preferred strategy using all three outcome measures followed by the AAN + PET-based treatment strategy. Compared to the AAN + PET-based treatment strategy, the AAN-based treatment strategy increased LE, QALYs, and SDFLE by 3.65, 3.65, and 7.30 days, respectively.

These results were similar for the asymptomatic, at elevated risk population (Table 1). Compared to the PET-based treatment, the empiric treatment strategy would be the preferred strategy, with an increment of 3.65 days in LE, 7.30 days in QALYs, and 10.95 days in DFLEs.

Table 1.   Base-case results
StrategyMild dementiaAsymptomatic, at elevated risk
QALYsLESDFLEQALYsLEDFLE
  • Abbreviations: DFLE, dementia-free LE; LE, life expectancy; QALYs, quality adjusted life-years; SDFLE, severe dementia-free LE.

  • *

    Empiric treatment;

  • **

    **PET-based treatment;

  • ***

    ***Natural history.

AAN-based treatment4.107.894.0212.2512.7912.21
AAN + PET-based treatment4.097.884.0012.2312.7812.18
Natural history4.027.823.8612.1112.7112.03

Sensitivity Analysis

Mild dementia.

The AAN-based treatment strategy remained the preferred strategy, despite changes in the sensitivity and specificity of PET, prevalence of underlying AD, relative risk of progression as a result of treatment, length of efficacy of treatment, percentage of patients experiencing complications, and discount rate (full results available from http://www.clinpol.mc.duke.edu/Pubs/Publications/publications.html; selected results presented in Table 2). Nevertheless, when complications were fatal, the natural history strategy was preferred, followed by the AAN + PET-based treatment strategy. AAN-based treatment was the least preferred strategy of the three.

Table 2.  Selected one-way sensitivity analyses *
Mild dementiaAsymptomatic, at elevated risk
StrategyQALYsLESDFLEStrategyQALYsLEDFLE
  1. *Results for complete one-way sensitivity analyses available online from http://www.clinpol.mc.duke.edu/Pubs/Publications/publications.html.

  2. Abbreviations: AAN, American Academy of Neurology; DFLE, dementia-free LE; LE, life expectancy; PET, positron emission tomography; QALYs, quality adjusted life-years; SDFLE, severe dementia-free LE.

Treatment complications = 0%
AAN-based treatment4.11 4.04Empiric treatment12.27 12.24
AAN + PET-based treatment4.10 4.02PET-based treatment12.25 12.21
Natural history4.02 3.86Natural history12.11 12.03
Treatment complications = 30%
AAN-based treatment4.09 3.99Empiric treatment12.22 12.17
AAN + PET-based treatment4.08 3.97PET-based treatment12.21 12.16
Natural history4.02 3.86Natural history12.11 12.03
Complications = death
AAN-based treatment 6.793.51Empiric treatment 11.0610.48
AAN + PET-based treatment 7.283.72PET-based treatment 11.9011.31
Natural history 7.813.86Natural history 12.7112.03

To determine the impact of treatment complications on the choice of strategy, we conducted additional sensitivity analyses using hypothetical treatments with the potential for moderate to severe complications (Table 3, Fig. 2). The AAN + PET-based treatment strategy was preferred over the AAN-based treatment strategy if treatment complications resulted in either a 50% 1-year or a 20% lifetime disutility. The AAN + PET-based treatment strategy was also preferred if there was an increased risk of progression to a more severe state or death as a result of a severe complication.

Table 3.  One-way sensitivity analyses of a hypothetical treatment with complications
Mild dementiaAsymptomatic, at elevated risk
StrategyQALYsSDFLEStrategyQALYsDFLE
  1. Abbreviations: AAN, American Academy of Neurology; DFLE, dementia-free life expectancy; PET, positron emission tomography; QALYs, quality adjusted life-years; SDFLE, severe dementia-free life expectancy.

50% short-term decrease in utility
AAN-based treatment4.05 Empiric treatment12.18 
AAN + PET-based treatment4.07 PET-based treatment12.19 
Natural history4.02 Natural history12.11 
20% lifetime decrease in utility
AAN-based treatment3.99 Empiric treatment11.90 
AAN + PET-based treatment4.03 PET-based treatment12.05 
Natural history4.02 Natural history12.11 
RR for progression = 2
AAN-based treatment4.043.90Empiric treatment12.2312.19
AAN + PET-based treatment4.063.93PET-based treatment12.1812.22
Natural history4.023.86Natural history12.1112.03
RR for death = 5
AAN-based treatment4.063.97Empiric treatment12.2212.18
AAN + PET-based treatment4.073.97PET-based treatment12.2212.17
Natural history4.023.86Natural history12.1112.03
image

Figure 2. Two-way sensitivity analysis illustrating the optimal strategy for a given combination of treatment efficacy and complications for mild AD dementia.

Download figure to PowerPoint

To further explore the impact of complications on choice of strategy, we conducted a two-way sensitivity analysis varying treatment efficacy and complications, under the assumption that one would only adopt a treatment that had the potential for severe complications if it was also more effective than the existing treatment. We simultaneously varied the relative risk of progression from 0, when treatment completely prevented progression, to 1, when treatment did not prevent progression, and the utility associated with complications from 0, when complications were fatal to 1, when complications had no health effect. As shown in Fig. 2, the choice of strategy depended on the combination of efficacy and complications, with the AAN-based treatment strategy preferred if the treatment was very effective, and the AAN + PET-based treatment strategy preferred if complications were increasingly severe. A maximal gain in QALYs of 0.03 (10.95 days) was attained with the AAN + PET-based treatment strategy when complications were assumed to be fatal.

Asymptomatic, at elevated risk.

The results of the one-way sensitivity analysis were similar to those for patients with mild dementia (data available from http://www.clinpol.mc.duke.edu/Pubs/Publications/publications.html). Although variability in the proportion of patients experiencing complications did not change the relative ranking of the strategies, if complications were fatal, empiric treatment became the least preferred strategy (Table 2).

The results of the hypothetical treatment sensitivity analyses for the asymptomatic population were not as consistent as they were for the patients with dementia. The PET-based treatment strategy was preferred if there was a short-term decrement in utility associated with complications. The empiric treatment remained the preferred strategy if complications resulted in a short-term increase in the relative risk of progression to a more severe state or death. Neither intervention strategy was preferred if there was a 20%, long-term decrease in the utility as a result of a complication.

The results for the two-way sensitivity analysis were similar to the results obtained for patients with mild dementia (Fig. 3). Note that the PET-based treatment strategy is the optimal strategy for a greater range of efficacy/complication combinations in the asymptomatic population than in the mild dementia population. This was in part because the rate of progression in asymptomatic patients was low, an annual rate of 2%, so that true positives, more likely in the empiric treatment strategy, became less important and true negatives, more likely in the PET-based treatment strategy, became relatively more important in this population.

image

Figure 3. Two-way sensitivity analysis illustrating the optimal strategy for a given combination of treatment efficacy and complications for asymptomatic, at elevated risk.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

We evaluated different strategies for identifying and treating AD in patients with mild dementia or who are asymptomatic, but at an elevated risk. For patients with mild AD, treatment based on an AAN evaluation was preferable to treatment based on additional testing with PET. The preference for treating patients identified as having AD based on an AAN evaluation was robust for a moderately effective, relatively benign treatment such as AChE-I’s. A treatment strategy that relied on PET would be preferred if a new treatment became available that had serious adverse effects, suggesting that the value of PET will depend on the efficacy and side-effect profile of the new treatment.

We extended our analysis to patients who are asymptomatic, but at an elevated risk for developing AD owing to diagnosis of the disease in a first-degree relative. To our knowledge, prior (cost) decision models of management strategies for AD have not included this population [19,29–31]. If treatment is effective in delaying the onset of symptoms in this population, our findings suggest that they should also be treated without further testing.

There were several important limitations to this analysis. The first is the lack of detailed information from otherwise excellent databases such as CERAD that allowed for calculation of transition probabilities by age and state. As a result, we had to assume that the relative risk of death owing to AD was constant  across  age  and  that  progression  rates  were the same for patients with and without AD in the absence of treatment, to calculate the transition probabilities. If accurate data become available in the future, we will be able to recalculate the transition probabilities.

Evidence suggests that donepezil acts to delay progression of symptoms rather than delaying the progression of pathological disease [13]. Because we did not distinguish symptoms of dementia from pathological disease, to the extent that treatment improves symptoms rather than disease, these results will overstate the relative improvement in LE for the treatment strategies compared to the no intervention strategy. The same is true for adherence to treatment: if patients do not adhere to treatment owing to issues other than complications, the impact of treatment in extending LE will be overestimated for the treatment-based strategies.

Our analysis of the asymptomatic population is hypothetical; as such there is no basis for comparison with other studies. Because our calculations for the population aged 65 years and older were based on sparse data and numerous assumptions, we did not explore the impact of extending screening and treatment to men and women younger than age 65 years although that would be a reasonable expectation. Another assumption relates to the testing characteristics for PET scanning; these were derived from studies of patients with mild dementia compared to normal controls. As such, test discrimination may be different, if not lower, in an asymptomatic population. To address this concern, we varied the estimates for sensitivity and specificity over a wide range; our conclusions were unchanged. Another important assumption is that treatment is effective in preventing the onset of AD in this population. Because there is no evidence to support this assumption any decision to treat these patients would require further study. Finally, we assume that all asymptomatic patients progress through MCI to AD. If a proportion of patients’ progress directly from being asymptomatic to dementia, the LE associated with each strategy would decrease but our conclusions would not change.

Prior analyses have primarily focused on the trial-based question of whether treatment of all patients with clinical AD with AChE-I's is preferable to doing nothing [29–31]. Recent analyses have evaluated the cost-effectiveness of using functional neuroimaging tests to diagnose AD [19,32]. The analysis of PET by Silverman et al. [32] differs from the current analysis in two significant ways. First, the outcome of their analysis was based on number of correct diagnoses, thus equating false-negative and false-positive results. This implies that failure to receive effective therapy is the same as receiving unnecessary and, in the case of donepezil, relatively benign therapy. Second, patients with MCI were considered together with those with dementia, yet empiric treatment for MCI patients was not permitted unless the patient had a positive PET. This is a significant limitation because “off-label” use of AChE-I therapy for MCI patients is common in clinical practice. Moreover, the current analysis supports treatment without additional testing as a reasonable strategy if one accepts that such therapy is effective in those MCI patients who have histopathological evidence of AD.

McMahon et al. [19] examined the cost-effectiveness of functional neuroimaging for AD focusing on SPECT and MR imaging. Although outcomes were calculated over a time horizon of 18 months, which was shorter than ours, they similarly concluded that it was not cost-effective to add functional neuroimaging to the standard diagnostic workup for AD, given the effectiveness of currently available AChE-I’s.

We extend these findings by showing that functional neuroimaging is also unlikely to contribute to patient outcomes in an asymptomatic population if therapy is shown to be effective in delaying progression to symptomatic disease without long-term side effects. Testing would be valuable if such a new treatment was associated with the risk of serious adverse effects. This conclusion is consistent with the general principle that the desirability of testing depends not only on test operating characteristics, sensitivity and specificity, and disease prevalence but also on the relative value of a true positive, benefit of correct treatment relative to incorrect nontreatment, compared to the relative value of a true negative, benefit of correct nontreatment relative to incorrect treatment [33]. If the treatment is relatively benign and beneficial, treatment without further testing is preferred over a wide range of circumstances.

In conclusion, our results suggest that current treatments, which are relatively benign and may slow disease progression, should be offered to patients who are identified as positive based solely on a currently recommended clinical evaluation. A clinical evaluation that includes functional neuroimaging-based testing will be warranted, however, when new treatments that are effective at slowing disease progression but have the potential for moderate to severe complications become available.

This work was sponsored by the Agency for Healthcare Research and Quality Contract No. 290-97-0014, Task Order 7.

References

  1. Top of page
  2. ABSTRACT
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  • 1
    Brookmeyer R, Gray S, Kawas C. Projections of Alzheimer's disease in the United States and the public health impact of delaying disease onset. Am J Pub Health 1998;88: 133742.
  • 2
    Clark CM, Trojanowski JQ, eds. Neurodegenerative Dementias: Clinical Features and Pathological Mechanisms. New York: McGraw-Hill, 2000.
  • 3
    American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (4th ed.). Washington: The Association, 1994.
  • 4
    Small GW, Rabins PV, Barry PP, et al. Diagnosis and treatment of Alzheimer disease and related disorders. Consensus statement of the American Association for Geriatric Psychiatry, the Alzheimer's Association and the American Geriatrics Society. JAMA 1997;278: 136371.
  • 5
    McKhann G, Drachman D, Folstein M, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's disease. Neurology 1984;34: 93944.
  • 6
    Knopman DS, DeKosky ST, Cummings JL, et al. Practice parameter: diagnosis of dementia (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2001;56: 114353.
  • 7
    Hoffman JM, Welsh-Bohmer KA, Hanson M, et al. FDG PET imaging in patients with pathologically verified dementia. J Nucl Med 2000; 41: 19208.
  • 8
    Bookheimer SY, Strojwas MH, Cohen MS, et al. Patterns of brain activation in people at risk for Alzheimer's disease. N Engl J Med 2000;343: 4506.
  • 9
    Silverman DH, Small GW, Chang CY, et al. Positron emission tomography in evaluation of dementia: regional brain metabolism and long-term outcome. JAMA 2001;286: 21207.
  • 10
    Small GW, Ercoli LM, Silverman DH, et al. Cerebral metabolic and cognitive decline in persons at genetic risk for Alzheimer's disease. Proc Natl Acad Sci U S A 2000;97: 603742.
  • 11
    Doody RS, Stevens JC, Beck C, et al. Practice parameter: management of dementia (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2001;56: 115466.
  • 12
    Mohs RC, Doody RS, Morris JC, et al. A 1-year, placebo-controlled preservation of function survival study of donepezil in AD patients. Neurology 2001;57: 4818.
  • 13
    National Institute on Aging National Institutes of Health. Progress Report on Alzheimer's Disease, 2000: Taking the Next Steps [Internet]. Washington: The Institute, 2000. Available from: http://www.alzheimers.org/pubs/prog00.htm.
  • 14
    Gurwitz D. Immunization for Alzheimer's disease: yet closer to clinical trials. Trends Mol Med 2001;7: 385.
  • 15
    Scinto LF, Daffner KR, eds. Early Diagnosis of Alzheimer's Disease. Totowa (NJ): Humana Press, 2000.
  • 16
    Morris JC, Storandt M, Miller JP, et al. Mild cognitive impairment represents early-stage Alzheimer's disease. Arch Neurol 2001;58: 397405.
  • 17
    Petersen RC, Stevens JC, Ganguli M, DeKosky ST. Practice parameter: early detection of dementia: mild cognitive impairment (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2001;56: 113342.
  • 18
    Willmer J, Mohr E. Evaluation of selection criteria used in Alzheimer's disease clinical trials. Can J Neurol Sci 1998;25: 3943.
  • 19
    McMahon PM, Araki SS, Neumann PJ,  et al. Cost-effectiveness  of  functional  imaging  tests  in the diagnosis of Alzheimer disease. Radiology 2000;217: 5868.
  • 20
    Lautenschlager NT, Cupples LA, Rao VS, et al. Risk of dementia among relatives of Alzheimer's disease patients in the MIRAGE study: what is in store for the oldest old? Neurology 1996;46: 64150.
  • 21
    Mohs RC, Breitner JC, Silverman JM, Davis KL. Alzheimer's disease: morbid risk among first-degree relatives approximates 50% by 90 years of age. Arch Gen Psychiatry 1987;44: 4058.
  • 22
    Green RC, Cupples LA, Go R, et al. Risk of dementia among white and African American relatives of patients with Alzheimer's disease. JAMA 2002;287: 32936.
  • 23
    Matchar DM, Kulasingam SL, McCrory DC, et al. Use of Positron Emission Tomography and Other Neuroimaging Techniques in the Diagnosis and Management of Alzheimer's Disease and Dementia [Internet]. Prepared for the Agency for Healthcare Policy and Research Contract No. 290-97-0014, Task Order 7. Durham (NC): Duke Evidence-Based Practice Center, Center for Health Policy Research, 2001 Dec 14. Available from: http://cms.hhs.gov/coverage/download/8b3-ww3.pdf.
  • 24
    Ballard C, O'Brien J, Morris CM, et al. The progression of cognitive impairment in dementia with Lewy bodies, vascular dementia and Alzheimer's disease. Int J Geriatric Psychiatry 2001;16: 499503.
  • 25
    Patwardhan M, McCrory DC, Matchar DB, et al. The operating characteristics of PET in the diagnosis and management of AD. Radiology 2003;in press.
  • 26
    Gold MR, Siegel JE, Russel LB, Weinstein MC, eds. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press, 1996.
  • 27
    Neumann PJ, Kuntz KM, Leon J, et al. Health utilities in Alzheimer's disease: a cross-sectional study of patients and caregivers. Med Care 1999;37: 2732.
  • 28
    Smith KJ, Roberts MS. The cost-effectiveness of sildenafil. Ann Intern Med 2000;132: 9337.
  • 29
    O'Brien BJ, Goeree R, Hux M, et al. Economic evaluation of donepezil for the treatment of Alzheimer's disease in Canada. J Am Geriatr Soc 1999;47: 5708.
  • 30
    Jonsson L, Lindgren P, Wimo A, et al. The cost-effectiveness of donepezil therapy in Swedish patients with Alzheimer's disease: a Markov Model. Clin Ther 1999;21: 123040.
  • 31
    Neumann PJ, Hermann RC, Kuntz KM, et al. Cost-effectiveness of donepezil in the treatment of mild or moderate Alzheimer's disease. Neurology 1999;52: 113845.
  • 32
    Silverman DH, Gambhir SS, Huang HW, et al. Evaluating early dementia with and without assessment of regional cerebral metabolism by PET: a comparison of predicted costs and benefits. J Nucl Med 2002;43: 25366.
  • 33
    Pauker SG, Kassirer JP. The threshold approach to clinical decision making. N Engl J Med 1980;302: 110917.