| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Health Policy and Practice |
1 From the Department of Radiology, Decision Analysis and Technology Assessment Group (P.M.M., G.S.G.) and the Computer-Assisted Diagnostics Laboratory (G.J.H.), Massachusetts General Hospital, Zero Emerson Pl, Ste 2H, Boston, MA 02114 and the Program on the Economic Evaluation of Medical Technology, Harvard School of Public Health, Boston, Mass (S.S.A., P.J.N.). Received October 20, 1999; revision requested November 19; revision received December 16; accepted January 12, 2000. Address correspondence to G.S.G. (e-mail: gazelle@nmr.mgh.harvard.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: A decision model was used to calculate costs and benefits (in quality-adjusted life-years [QALYs]) that accrued to hypothetical cohorts of patients at presentation to an Alzheimer disease center. Sensitivity analysis was performed to examine the effects of diagnostic test characteristics, therapeutic efficacy, disease severity, and costs on cost-effectiveness.
RESULTS: The incremental cost-effectiveness ratio of dynamic susceptibility contrast material–enhanced magnetic resonance (MR) imaging was $479,500 per QALY (compared with the usual diagnostic work-up), while visual or quantitative single photon emission computed tomography (SPECT) was dominated (higher costs, lower effectiveness) by the usual diagnostic work-up. These results depend critically on the sensitivity and specificity of the standard diagnostic work-up, the effectiveness of drug treatment, and the disease severity. Varying these parameters resulted in estimates of incremental cost-effectiveness for dynamic susceptibility contrast-enhanced MR imaging of $24,680 to $8.6 million per QALY. SPECT either was dominated by the usual diagnostic work-up or had cost-effectiveness ratios of $180,200 to $6 million per QALY.
CONCLUSION: The addition of functional neuroimaging to the usual diagnostic regimen at Alzheimer disease clinics is not cost-effective given the effectiveness of currently available therapies.
Index terms: Alzheimer disease, 10.83 • Cost-effectiveness • Magnetic resonance, (MR), utilization
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
The current standard diagnostic strategy at an Alzheimer disease center generally includes a detailed history, an assessment of cognition and functional status, laboratory testing, and a brain imaging examination such as nonenhanced computed tomography (CT) or magnetic resonance (MR) imaging to identify structural abnormalities caused by other conditions (13). Alzheimer disease thus essentially is a diagnosis of exclusion; it cannot be diagnosed definitively by using currently available imaging tests.
The standard diagnostic strategy has been reported to have a positive predictive value (the probability of disease given a positive test result) of 90% for Alzheimer disease at major referral centers where most patients receive diagnoses (10,13), but its sensitivity (the probability of a positive test result given disease) is unknown. Because of the progressive nature of the disease and the lack of a practical standard of reference (autopsy currently is the only definitive diagnostic test), it follows that positive predictive values and sensitivities for specific stages (mild, moderate, and severe) cannot be determined.
Functional imaging examinations such as positron emission tomography or single photon emission computed tomography (SPECT) also can aid in the diagnosis of dementia by helping to measure abnormalities in cerebral blood volume or flow that are characteristic of Alzheimer disease (14,15). Recently, quantitative dynamic susceptibility contrast-enhanced MR imaging was shown to be a promising alternative to nuclear medicine imaging for the assessment of Alzheimer disease (16,17). The purpose of this study was to compare the cost-effectiveness of a diagnostic work-up strategy that involves such a functional neuroimaging examination with the standard diagnostic strategy in the setting of a specialized Alzheimer disease center. The analysis was performed by using the techniques of decision analysis, with which we modeled the diagnosis, drug treatment, and care costs in patients with dementia symptoms at presentation to an Alzheimer disease center.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
The study was conducted from a societal perspective and incorporated all costs and benefits regardless of who incurred them. All costs were converted to 1998 dollars and were adjusted for inflation by using the medical care component of the Consumer Price Index, except where noted (19). Future costs and QALYs were discounted at 3% annually (18). The time horizon of the base-case analysis was 18 months. The Appendix lists all model inputs and data sources. A schematic drawing of the decision tree is shown in Figure 1.
|
|
Patients progress through disease states in accordance with transition probabilities that were derived previously (8). Initially, all patients in the cohort begin in the community setting and have a probability in each cycle, conditional on disease state, of making a transition to nursing home care. The probability of death for the patients in the no–Alzheimer disease group was obtained from the National Center for Health Statistics (26); we used the annual probability of death at age 76 years, the mean age of patients with Alzheimer disease at presentation to our institution. The model uses 6-week transition probabilities, which are equal to the 6/52 root of the matrix of annual probabilities shown in the Appendix.
The model assumes that all patients who receive a diagnosis of probable Alzheimer disease receive treatment with donepezil or with a hypothetical higher-efficacy drug. Once a patient progresses to severe Alzheimer disease, no further drug treatment is given, since existing trial data do not support the efficacy of donepezil in patients with severe Alzheimer disease (4). Since donepezil is a well-tolerated drug (4), we assume that treatment would not be discontinued unless the patient progressed to severe Alzheimer disease or died. Thus, for patients with severe Alzheimer disease, the model calculates no further drug-related costs or benefits. To our knowledge, long-term controlled data on the effectiveness of donepezil are unavailable, so for the base-case analysis, we estimated the duration of drug treatment and effectiveness at 18 months on the basis of the results of a previously described survey of clinical experts (8).
Donepezil treatment has been shown to reduce the annual decline in cognition in patients with Alzheimer disease when compared with patients in a placebo group (4). On the basis of these data, we modeled the beneficial effects of donepezil treatment as a 50% reduction in the probability of transition from mild to moderate Alzheimer disease (a risk ratio of 0.50) and as a 2.36-fold increase in the probability of progression from moderate to mild Alzheimer disease (a risk ratio of 2.36) (8). These effects were assumed to be constant throughout the duration of treatment, with no residual effect at drug discontinuation (4). In addition to drug costs, two follow-up visits per year were included in the cost of treatment for patients with a diagnosis of Alzheimer disease, on the basis of data on the frequency of follow-up visits at our institution.
Prevalence
In the population of patients who presented to an Alzheimer disease center, the prevalence of Alzheimer disease was estimated at 56% (derived from Massachusetts General Hospital data, in which, in 1996, 64% of patients with dementia had a diagnosis of Alzheimer disease and 90% of Alzheimer disease diagnoses were confirmed at autopsy). Thus, for the base-case analysis, we estimated that approximately 44% of patients would have no Alzheimer disease and assumed that the remaining patients would have either mild or moderate disease. We assumed that a vast majority of patients with severe Alzheimer disease would have sought medical attention earlier in the disease progression; thus, we limited our initial patient population to patients with mild or moderate dementia.
For the base-case analysis, we estimated the ratio of mild to moderate Alzheimer disease cases to be 1.5:1. Investigators in a recent meta-analysis (2) of estimates of Alzheimer disease incidence rates reported 0.9 million mild cases versus 1.0 million moderate to severe cases in the United States. Investigators in a recent Danish study (27) estimated that mild cases outnumbered moderate to severe cases. In neither study did the investigators further subclassify patients to the moderate and severe categories, which made it impossible to directly establish a ratio of mild to moderate cases; nevertheless, unless the prevalence of severe cases is very low, it seems reasonable to conclude that the results of both studies imply that there are more mild cases than moderate cases.
Costs and Effectiveness
The components of the initial diagnostic work-up at a specialized Alzheimer disease center were estimated on the basis of the literature (10,13) and a detailed assessment of resource use at Massachusetts General Hospital by using TRANSITION SYSTEMS software (Eclypsis, Atlanta, Ga), which interfaces with the hospital’s cost accounting system. This diagnostic work-up generally consists of two physician consultations (internal medicine and neurology), a series of laboratory tests, and a structural imaging examination (CT or MR imaging). The average series of laboratory tests for the initial work-up was determined by reviewing our institution’s patient records from 1998 and by compiling a list of laboratory tests common to initial work-ups. The cost for this laboratory series was estimated at $70 on the basis of resource use data from Massachusetts General Hospital.
The costs of CT and MR imaging were based on Medicare reimbursement rates, which were determined by using CODEMANAGER software (American Medical Association, Chicago, Ill).
For visual SPECT, we estimated cost as the Medicare reimbursement for SPECT (which includes the provision of radioelements) at our institution, which was $699 in 1998. We were unable to use Medicare reimbursements that were based on CPT4 codes, since the appropriate level II code A4641 (provision of radioelement) does not have an assigned reimbursement value, and the cost of the radioelement therefore would be excluded. As of 1998, for outpatient tests without an assigned reimbursement value, Medicare reimbursed a percentage (specific to the institution) of charges. For the cost of computed SPECT, we included an additional cost for computer-aided data manipulation that was derived from Medicare reimbursement for CPT4 code 76375. We estimated the cost for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging, a relatively new procedure, as equal to the Medicare reimbursements for MR imaging with and without contrast material plus a computerized three-dimensional reconstruction of the image data.
We estimated that completion of the standard diagnostic work-up would take 1 day (8 hours, including travel). Travel expenses (gas, parking, and lunch) were estimated at $40 for the day. If the patient returned for SPECT, we estimated that this visit would require a half day, plus travel costs of $25. As described above, dynamic susceptibility contrast-enhanced MR imaging is performed concurrently with conventional MR imaging, so no second trip is required. At the time this article was written, the time cost for the caretaker was estimated as the U.S. national mean wage rate of $12.78 per hour (22) or $102 per 8-hour day. The time cost for the patient was estimated at $50 per day; this cost was derived from the median income of persons aged 65 years and older (23). While this method of estimating the value of the time of elderly patients may be inaccurate, it has been suggested as a practical alternative to ignoring the value completely (18).
The costs of caring for patients with Alzheimer disease vary by disease severity and care setting (8,20) and were modeled as the cost of care plus the annual cost of living for an average age-matched individual. As a baseline cost of living for patients who did not have Alzheimer disease, the annual expenditures of persons older than 65 years of age from the Consumer Expenditure Survey (21) were used; included were mean out-of-pocket health care costs for the age group. The cost of donepezil treatment was estimated at $4.13 per day (8).
Quality-of-life weights for patients without Alzheimer disease were estimated at 0.826 on a scale of 0 (dead) to 1 (perfect health) on the basis of the mean of the time trade-off scores for men and women 65–84 years of age, which was derived from a study of community preferences (24). Quality-of-life weights for patients with Alzheimer disease at each disease stage and care setting (nursing home or community) were based on Health Utilities Index Mark 2 (HUI:2) scores that were published previously (8,28,29).
Diagnostic Test Characteristics
Base-case estimates of the sensitivity and specificity of dynamic susceptibility contrast-enhanced MR imaging and visual and computed SPECT (17) are shown in the Appendix. To our knowledge, there were no data available on the number of false-negative diagnoses from the standard examination, so the estimation of sensitivity was difficult. We used 0.75 as the base-case estimate. The specificity of the initial examination and of structural imaging can be high for some diseases (15) but not for others. We estimated the specificity of the standard examination at 0.9 for the base-case analysis.
Sensitivity Analyses
The values and data sources used for sensitivity analyses are summarized in the Appendix.
Test characteristics and strategy.—We modeled a hypothetical "treat-all" strategy in which no imaging or laboratory tests were performed and all patients received donepezil. The treat-all strategy included a consultation for the initial diagnosis of dementia and for the prescription of drug therapy, plus follow-up visits to monitor dementia progression and potential side effects.
We did not include a "no test-no treat" strategy, a common strategy in cost-effectiveness analysis, because our research was intended to determine the relative cost-effectiveness of different functional imaging tests that might be added to the conventional diagnostic work-up of patients treated at a specialized Alzheimer disease center. Viewed in this manner, the standard diagnostic work-up is, in essence, a "do-nothing" reference strategy.
We also modeled strategies in which all patients underwent CT as the structural imaging examination and underwent either visual or computed SPECT and were treated according to the results of SPECT, as described in Table 1 (note that we are not aware that these strategies are used currently in clinical practice; therefore, we did not examine them extensively).
The estimate of sensitivity for computed SPECT was modeled on the reported value of 0.90 for mild to moderate Alzheimer disease (17) for the base case, but it is likely that the true sensitivity is less for mild Alzheimer disease and is greater for moderate Alzheimer disease. We performed sensitivity analysis by using estimates of sensitivity of 0.88 for mild Alzheimer disease and 0.92 for moderate Alzheimer disease.
Sensitivity analyses were performed by using lower estimates of sensitivity (0.5) and specificity (0.8) of the standard examination to reflect the uncertainty that surrounded those values.
To examine the effects of improved sensitivity and specificity in future tests, we postulated the existence of a hypothetical perfect test, with a sensitivity and specificity of 1.0 and with a cost equal to that of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging.
Drug effects and duration.—To further investigate the relationship between the cost-effectiveness of the diagnostic strategies and treatment, we postulated the availability of additional therapeutic agents for Alzheimer disease (30). A hypothetical drug X was assumed to have the same costs, treatment duration, and duration of effectiveness as donepezil but to have a risk ratio of 0.1 as compared with no treatment (as compared with 0.5 in the base case) for the transition from mild to moderate Alzheimer disease and to have a risk ratio of 10.0 (as compared with 2.36 in the base case) for the transition from moderate to mild Alzheimer disease. An additional hypothetical drug, drug Y, also was postulated and was intermediate in effectiveness between donepezil and drug X. The risk ratio of drug Y was 0.25 for the transition from mild to moderate Alzheimer disease and was 5.0 for the transition from moderate to mild Alzheimer disease; both risk ratios for drug Y were within the confidence intervals of the risk ratios reported for donepezil (8).
While the treatment duration (the length of time the risk ratio is in effect) and the treatment effectiveness (risk ratio) were separate values that could have been varied independently, we assumed constant treatment effectiveness with no benefits after drug discontinuation, as described earlier. We compared donepezil treatment durations of 6, 12, 24, 36, and 48 months; in each case, costs and benefits were calculated for the duration of treatment. Since each strategy was compared incrementally with the standard diagnostic work-up, and costs and benefits accruing to the cohorts were the same after the discontinuation of donepezil treatment, in each case, the time horizon was set equal to the drug treatment duration, and drug effectiveness was constant during treatment.
Disease progression.—To model faster or slower underlying disease progression for the patients with Alzheimer disease, we increased transition probabilities by 10% for the mild-to-moderate, moderate-to-severe, and severe-to-dead transitions ("faster progression") and reduced the same values by 10% ("slower progression") relative to base-case probabilities. Probabilities for mild-to-mild, moderate-to-moderate, and severe-to-severe transitions were adjusted as required to compensate.
The base-case estimates of the probability of death in the patients without Alzheimer disease and in the patients with Alzheimer disease were derived from different sources, so that the patients without Alzheimer disease had a higher annual risk of death than those with mild Alzheimer disease. We performed an additional sensitivity analysis in which the probability for the no– Alzheimer disease-to-dead transition was set at 0.018, a value slightly lower than the base-case probability for the mild Alzheimer disease-to-dead transition.
Prevalence, costs, and quality of life.—Additional sensitivity analyses were performed to investigate the effect of different assumptions of disease prevalence, or the ratio of mild to moderate Alzheimer disease at presentation to the Alzheimer disease center. The ratios of mild to moderate Alzheimer disease cases examined were 2:1, 1:1, and the reverse of the base-case estimate, 1:1.5. In addition, separate analyses were performed in which the overall prevalence of Alzheimer disease was increased to 65% to reflect the possibility of completely undiagnosed Alzheimer disease cases that might not have been included in our base-case estimate of 56%.
Sensitivity analysis was also performed on the cost estimates for the consultations, follow-up visits, and diagnostic tests. Since our base-case estimate for SPECT costs was derived from Medicare reimbursements specific to our institution, we examined a range of cost estimates. For the lower bound, we based the cost of visual SPECT on the Medicare reimbursement of CPT4 code 78607 ($358), with an additional cost for code 76375 ($133) for computed SPECT. This lower bound did not include the cost of provision of the radioelement. The upper bound of the cost estimate for SPECT was represented by our institution’s resource cost values, equal to $847 for visual SPECT and $935 for computed SPECT (these include the resource use cost of the radioelement). The resource use costs for other imaging examinations and physician visits are shown in the Appendix.
Since dynamic susceptibility contrast-enhanced MR imaging is a relatively new procedure with uncertain costs, we varied the base-case cost estimate by ±25%. We also compared the different work-up strategies under the assumption of no costs for patient or caregiver time and no travel costs for trips to the Alzheimer disease center for diagnostic tests. The substitution of MR imaging for structural imaging was modeled and altered only cost in the model. No sensitivity analyses were performed on the discount rate, since the time horizon for the base-case analysis was only 18 months, and any effect of different discount rates was likely to be minimal.
The quality-of-life weights for patients with Alzheimer disease in each disease state and care setting were varied by ±0.1 to reflect uncertainty in the base-case estimates. The quality-of-life weight for the no–Alzheimer disease or other state was varied from the base-case assumption. We derived a QALY weight of 0.796 for the age cohort of 70–79 years (which excluded patients with a diagnosis of Alzheimer disease) from a study in which investigators administered the Health Utilities Index Mark 3 (HUI:3) to a large sample of community-dwelling Canadians (25).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
|
|
Under the more pessimistic set of assumptions for the sensitivity and specificity of the standard examination, strategy 3 (visual SPECT) was dominated, and strategy 5 (all patients undergoing CT and visual SPECT) had an ICER of $6.8 million per QALY. Computed SPECT had an ICER of $977,100 per QALY. Strategies 6 and 2, in which all patients underwent imaging with a quantitative functional examination, had similar ICERs under assumptions of lowered sensitivity and specificity of the standard examination. Strategy 6 (all patients examined with CT and computed SPECT) had an ICER of $180,200 per QALY, while strategy 2 (MR imaging plus dynamic susceptibility contrast-enhanced MR imaging) had an ICER of $195,000 per QALY.
Drug effects and duration.—The results of the sensitivity analyses of drug effects are presented in Table 4. In the scenario of treatment with the hypothetical superior drug X, the ICER of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging compared with the standard diagnostic examination was $174,470 per QALY. As expected, the results of the drug Y scenario are between those of the base case and those of the drug X scenario; MR imaging plus dynamic susceptibility contrast-enhanced MR imaging had an ICER of $224,470 per QALY compared with the standard examination. If the more pessimistic set of assumptions described earlier were used for the characteristics of the standard examination and were combined with a scenario of drug X, the ICER of the MR imaging plus dynamic susceptibility contrast-enhanced MR imaging strategy decreased to $24,680 per QALY. By using pessimistic assumptions for the standard examination characteristics plus treatment with drug Y, the ICER of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging was $72,950 per QALY.
|
Under the assumption that donepezil treatment was continued for only 6 or 12 months, with the assumption of constant effectiveness, MR imaging plus dynamic susceptibility contrast-enhanced MR imaging was either dominated or had an ICER of more than $3 million per QALY. As the duration was extended to 48 months, the ICER of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging decreased relative to the base case to $58,930 per QALY (results of 6- and 48-month scenarios are shown in Table 4 for illustrative purposes). Strategies 3 and 4 (visual and computed SPECT, respectively) were dominated by the standard examination in all scenarios.
Disease progression.—The results of varying the transition probabilities to simulate faster or slower disease progression are presented in Table 4. In both scenarios, assuming treatment with donepezil, both SPECT strategies were dominated by the standard examination, and the ICER for dynamic susceptibility contrast-enhanced MR imaging was similar to the base-case estimate. When the probability of death in the patients without Alzheimer disease was set lower than the probability of death in patients with mild Alzheimer disease, the ICER for dynamic susceptibility contrast-enhanced MR imaging was $2.1 million per QALY compared with the standard examination.
Prevalence, costs, and quality of life.—The results of varying the ratios of mild Alzheimer disease, moderate Alzheimer disease, and no–Alzheimer disease or other dementia cases in the patient population are summarized in Table 4. The ICER of the MR imaging plus dynamic susceptibility contrast-enhanced MR imaging strategy was $388,820–$585,710 per QALY for the scenarios in which the prevalence of mild cases was equal to or greater than the prevalence of moderate cases (including the base-case result). When the prevalence of moderate Alzheimer disease exceeded the prevalence of mild Alzheimer disease, the ICER for dynamic susceptibility contrast-enhanced MR imaging increased an order of magnitude to $8.6 million per QALY. In all cases, visual and computed SPECT strategies were dominated by the standard examination.
Table 4 summarizes the results of sensitivity analyses of cost estimates. When Medicare reimbursements based on CPT4 codes (which do not include provision of the radioelement) were used for the lower bound of cost estimates for SPECT, visual and computed SPECT again were dominated by the standard examination. The substitution of our institution’s resource use values for the Medicare reimbursement rates resulted in a cost-effectiveness estimate for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging similar to that of the base case, or $394,370 per QALY. Varying the base-case cost estimate for dynamic susceptibility contrast-enhanced MR imaging by ±25% resulted in ICERs of $357,500–$763,700 per QALY. The elimination of travel costs and any opportunity cost for the patient or caregiver resulted in a decrease in the ICER for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging to $328,830 per QALY. The elimination of only patient time costs results in an estimate of $773,850 per QALY for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging. The substitution of MR imaging for CT as the structural imaging component of the standard examination increased the cost of the standard examination and reduced the cost of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging relative to the base case, for an ICER of $335,740 per QALY. In all the scenarios, visual and computed SPECT were dominated by the standard examination.
The results of varying the quality-of-life weights by ±0.1 in patients with Alzheimer disease are shown in Table 4. Both SPECT strategies were dominated by the standard examination, and MR imaging plus dynamic susceptibility contrast-enhanced MR imaging had an ICER of more than $1 million per QALY in both scenarios.
The results of the sensitivity analysis of the quality-of-life weight for the no– Alzheimer disease or other state are presented in Table 4. The ICER for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging was $332,860 per QALY and for computed SPECT was $1.9 million per QALY.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
While visual SPECT was dominated in all the scenarios examined, the relative cost-effectiveness of the quantitative functional imaging strategies of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging and computed SPECT varied over a wide range and were heavily dependent on several assumptions: the drug effectiveness and the duration of effectiveness, estimates of the sensitivity and specificity of the standard diagnostic work-up, and the ratio of mild to moderate Alzheimer disease cases at presentation. The hypothetical strategy 6 (examination of all patients with CT and computed SPECT) was most similar to the MR imaging plus dynamic susceptibility contrast-enhanced MR imaging strategy in that all patients received diagnoses in accordance with a quantitative functional imaging examination; its ICER actually was lower than that of dynamic susceptibility contrast-enhanced MR imaging under either base-case or pessimistic assumptions of characteristics of the standard diagnostic examination. This was due primarily to strategy 6’s lower cost when estimated by using Medicare reimbursement rates. This advantage is lost when resource use costs are used instead, as shown in Table 4.
In the scenario of low sensitivity and specificity of the standard examination, the visual and computed SPECT strategies (strategies 3 and 4, respectively) appear less effective, since any false-negative diagnoses result in no treatment on the basis of the findings of the standard examination before SPECT. In contrast, strategies that include functional imaging tests for all patients (strategies 2, 5, and 6) appear relatively more cost-effective, since they are compared with a poor reference test. Comparing Tables 2 and 4, we observe that if the sensitivity of the standard examination were estimated at 0.5 instead of at the base-case assumption of 0.75 and if a newer, more effective drug became available, MR imaging plus dynamic susceptibility contrast-enhanced MR imaging would have an ICER of $24,680 per QALY, nearly 20-fold less than the base-case estimate for MR imaging plus dynamic susceptibility contrast-enhanced MR imaging and well within the range of commonly funded health care interventions in the United States (31,32).
The improved performance of functional imaging tests was modeled by including a "perfect" dynamic susceptibility contrast-enhanced MR imaging test. The perfect test was found to be more cost-effective than dynamic susceptibility contrast-enhanced MR imaging, but improvements in drug effectiveness or duration had a more dramatic effect on cost-effectiveness. Under our base-case assumptions, a perfect test adds only 0.004 QALY to the standard approach. Thus, any new diagnostic test would have to be inexpensive to meet commonly cited thresholds of cost-effectiveness.
Investigators in a study (9) in which the cost-effectiveness of donepezil treatment was examined modeled a 5-year treatment duration, an optimistic duration based on the results of a 24-week trial (4). We modeled longer time horizons to demonstrate a trend of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging becoming more cost-effective as the treatment duration and effectiveness increase, but these data should be viewed with some caution, since, with longer time horizons, the validity of a model that uses a one-time diagnostic examination decreases. We did not model the transition of no Alzheimer disease or other to Alzheimer disease, nor did we have any relevant data on the frequency of repeat testing after a false-negative diagnosis. Nonetheless, it is clear that if the results of future studies support a longer effectiveness duration of donepezil, the cost-effectiveness of dynamic susceptibility contrast-enhanced MR imaging will improve.
In the scenario in which moderate Alzheimer disease cases outnumber mild cases at diagnosis, the ICER of MR imaging plus dynamic susceptibility contrast-enhanced MR imaging increases by an order of magnitude, since the opportunity to treat mild cases, in which donepezil has the greatest proved role, is lost. It has been suggested previously (15) that earlier detection of Alzheimer disease has implications for the cost of Alzheimer disease.
Limitations
Several limitations were evident in this study. In this analysis, we modeled only a one-time diagnostic examination strategy for patients suspected to have Alzheimer disease. Therefore, these results cannot be extrapolated to a community-wide screening program or to repeated rounds of testing. In settings other than tertiary Alzheimer disease centers, the sensitivity and specificity of any standard diagnostic work-up may be much lower than those assumed here, and the functional imaging strategies might be more cost-effective. Also, this study was conducted from a societal perspective and did not address situations in which individual patients with uncertain diagnoses might benefit greatly from confirmation by using functional imaging. In this study, we did not attempt to quantify the benefits of a correct diagnosis of early Alzheimer disease to the patient or family; the ability to plan for the eventual decline in cognition, such as by preparing estate plans or caretaker arrangements, could be of great value. Any benefit from the avoidance of radiation exposure (required for SPECT) from dynamic susceptibility contrast-enhanced MR imaging was not quantified.
There are numerous causes of dementia in older patients, some of them reversible with interventions such as reducing drug doses or treating depression or alcohol abuse. It would be difficult to explicitly model all possible combinations of causes of dementia and potential treatments and outcomes, so a major simplification was made that treated all patients in the no–Alzheimer disease or other group as the same. Unless they mistakenly received a diagnosis of Alzheimer disease, the patients in the no–Alzheimer disease group in the model were calculated as incurring no costs of care beyond the normal expenditures for an age-matched healthy individual. This simplification underestimates the cost of caring for patients with other dementias and may have resulted in an inflated estimate of the cost-effectiveness ratio for the diagnosis of Alzheimer disease. In a similar fashion, the patients in the no-Alzheimer disease group were assumed to have a quality-of-life weight of age-matched controls.
The results of this analysis relied on the model and assumptions of disease progression reported in a previous study (8), in which investigators derived Alzheimer disease stage transition probabilities from data from the Consortium to Establish a Registry for Alzheimer Disease (33). These data might not be representative of all Alzheimer disease populations.
Issues for Future Research
Further research is necessary to establish a reasonable estimate of the sensitivity of the standard examination for the diagnosis of Alzheimer disease. The lack of a standard-of-reference test for Alzheimer disease prior to autopsy means that there is no practical way to derive a solid estimate forthe sensitivity of diagnostic tests. Also, more data on the long-term effectiveness of drug therapies are needed.
In conclusion, given our assumption of an 18-month duration of the effectiveness of donepezil, it does not appear cost-effective to include functional imaging in the diagnostic strategy for Alzheimer disease. However, if the sensitivity and specificity of the standard examination are shown to be less than our base-case assumption, and/or treatment effectiveness or the duration of effectiveness improves, the relative cost-effectiveness of functional imaging examinations such as dynamic susceptibility contrast-enhanced MR imaging can be expected to improve, and the inclusion of a quantitative functional imaging examination in the diagnostic work-up for Alzheimer disease may be better justified.
|
|
FOOTNOTES |
|---|
P.J.N. is the Deputy Director of the Program on the Economic Evaluation of Medical Technology (PEEMT) at the Harvard School of Public Health. PEEMT receives unrestricted funds for research support from Pfizer, manufacturer of donepezil (Aricept) for Alzheimer disease treatment modeled in the submitted publication.
Author contributions: Guarantors of integrity of entire study, P.M.M., G.S.G.; study concepts, G.J.H., P.M.M., G.S.G.; study design, P.M.M., G.S.G., S.S.A.; definition of intellectual content, P.J.N., G.S.G., P.M.M.; literature research, P.M.M., S.S.A.; data acquisition, P.M.M., S.S.A.; data analysis, P.M.M., S.S.A., G.S.G.; manuscript preparation, editing, and review, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Petrou and E. Kupek Estimating Preference-Based Health Utilities Index Mark 3 Utility Scores for Childhood Conditions in England and Scotland Med Decis Making, May 1, 2009; 29(3): 291 - 303. [Abstract] [PDF]
|
|
|
|
 |
|
 C. A. G. Wolfs, C. D. Dirksen, A. Kessels, J. L. Severens, and F. R. J. Verhey Economic Evaluation of an Integrated Diagnostic Approach for Psychogeriatric Patients: Results of a Randomized Controlled Trial Arch Gen Psychiatry, March 1, 2009; 66(3): 313 - 323. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. S. Gazelle, P. M. McMahon, U. Siebert, and M. T. Beinfeld Cost-effectiveness Analysis in the Assessment of Diagnostic Imaging Technologies Radiology, May 1, 2005; 235(2): 361 - 370. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. M. McMahon, S. S. Araki, E. A. Sandberg, P. J. Neumann, and G. S. Gazelle Cost-Effectiveness of PET in the Diagnosis of Alzheimer Disease Radiology, August 1, 2003; 228(2): 515 - 522. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Petrella, R. E. Coleman, and P. M. Doraiswamy Neuroimaging and Early Diagnosis of Alzheimer Disease: A Look to the Future Radiology, February 1, 2003; 226(2): 315 - 336. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. H.S. Silverman, S. S. Gambhir, H.-W. C. Huang, J. Schwimmer, S. Kim, G. W. Small, J. Chodosh, J. Czernin, and M. E. Phelps Evaluating Early Dementia With and Without Assessment of Regional Cerebral Metabolism by PET: A Comparison of Predicted Costs and Benefits J. Nucl. Med., February 1, 2002; 43(2): 253 - 266. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. J. Neumann, J. K. Hammitt, C. Mueller, H. M. Fillit, J. Hill, N. A. Tetteh, and K. S. Kosik Public Attitudes About Genetic Testing For Alzheimer's Disease Health Aff., September 1, 2001; 20(5): 252 - 264. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
= standard examination, 
= visual SPECT, 
= computed SPECT, 
= dynamic susceptibility contrast-enhanced MR imaging, and * = treat all).