Author + information
- Received July 28, 2006
- Revision received September 12, 2006
- Accepted September 19, 2006
- Published online May 15, 2007.
- George A. Diamond, MD, FACC⁎ (, )
- Sanjay Kaul, MD, FACC and
- Prediman K. Shah, MD, FACC
- ↵⁎Reprint requests and correspondence:
Dr. George A. Diamond, 2408 Wild Oak Drive, Los Angeles, California 90068.
It has recently been proposed that asymptomatic diabetic patients undergo routine screening for subclinical atherosclerotic disease using myocardial perfusion scintigraphy. We herein analyze the expected cost and benefit associated with such a conditional test-treatment strategy (scintigraphic testing followed by statin treatment in positive test responders) in comparison to an unconditional treatment strategy (no testing and statin treatment in all). This analysis shows that unconditional treatment is the dominant strategy, costing 24% less ($3.2 billion) and preventing 25% more (16,800) atherosclerotic events annually on a national basis.
All right, Mr. DeMille, I’m ready for my close-up.
Norma Desmond in Sunset Boulevard(1950) (1)
Two recent perspectives argue the pros (2) and cons (3) of using myocardial perfusion scintigraphy (MPS) as the basis of a screening strategy to prevent atherosclerotic events in asymptomatic diabetic patients. However, although both essays allude to the importance of assessing the strategy’s cost effectiveness, neither offers such an assessment. The resultant analyses are therefore less than thorough, akin to reporting the sensitivity of a diagnostic test, but not its specificity or the efficacy of a new drug, but not its safety.
Accordingly, our purpose is to take a more thorough “close-up” picture of this proposal. Toward that end, we shall perform a simple back-of-the-envelope calculation of the expected cost and benefit associated with the proposed conditional test-treatment strategy (screen everyone with the test, and treat only those with an abnormal response), and compare these expectations to those associated with an alternative unconditional treatment strategy (test no one, and treat everyone). Our analysis addresses only the proposal at hand, and is not intended as a comprehensive discourse on the general process of epidemiologic screening, the principles of which are discussed elsewhere (4,5).
The results of our calculations are summarized in Table 1.Assume that the treatment is a preventive drug (a statin), at a cost of about $2 per day ($720 per year) based on 2006 average wholesale prices (6), and assume further that the test (MPS) has a one-time technical and professional cost of $809—the national average Medicare reimbursement for current procedure terminology codes 78465, 78468, 78480, and 93015, comprising 22.37 relative value units (3) at a conversion factor of $36.177 per unit under the 2006 Deficit Reduction Act (7). The cost of testing outside of the Medicare population, although generally higher, cannot be estimated reliably.
In the absence of testing or treatment, we expect about 280,000 atherosclerotic events per year in the target adult population of 14 million asymptomatic persons with diabetes not already known to have cardiovascular (CV) disease (3)—an event rate of 2% (twice that of the general population) (8). If we treat every one of these adults, as prescribed by the unconditional strategy, we can expect to reduce these events by about 30% based upon available randomized clinical trials (9,10). We thereby expect to prevent 84,000 events (30% of 280,000) at a total cost of $10.1 billion annually ($720 per patient per year times 14 million patients). Assuming that each event is equivalent to an average loss of 13 life-years (11), this represents a gain of 1,092,000 life-years (84,000 events times 13 years per event) and a cost-effectiveness ratio of $9,249 per life-year ($10.1 billion divided by 1,092,000 life-years)—well within the $5,063 to $23,792 range reported in a recent formal analysis (12).
The alternative conditional strategy would have us test all 14 million individuals at $809 per patient and a cost of $11.3 billion. Although it has not been explicitly validated in this context, Pareto’s 80/20 rule (13) is consistent with the available data regarding screen testing of asymptomatic diabetic patients (14–16). Based on this rule, we can project that our test will identify approximately 20% of the population (2.8 million patients) among whom 80% of the events will occur (224,000 events)—an event rate of 8% (4 times that of the untested diabetic patients). This leaves 11.2 million patients with a negative test, among which the remaining 56,000 events will occur (an event rate of 0.5%). These values are consistent with those in a number of empirical studies regarding the screening of asymptomatic diabetic patients (2,3).
If we now treat only the higher-risk population with a positive test, we can expect to prevent 67,200 events (30% of 224,000) at an overall cost of $13.3 billion ($11.3 billion for testing plus $2 billion for treatment)—a gain of 873,600 life-years (67,200 events times 13 years per event) and a cost-effectiveness ratio of $15,224 per life-year ($13.3 billion divided by 873,600 life-years). The latter figure probably represents an underestimation as it ignores the added costs associated with referral of some proportion of this population for coronary angiography and myocardial revascularization as a consequence of testing, procedures not known to provide any preventive benefit in asymptomatic diabetic patients (17).
Thus, the unconditional treatment strategy dominates the conditional test-treatment strategy because it costs at least $3.2 billion (24%) less and prevents at least 16,800 (25%) more events. A sensitivity analysis performed by varying the cost of the statin from $0.50 to $2.50 per day and the cost of MPS from $500 to $2,000 shows the annual cost differential to range from $3 billion in favor of test treatment (when testing costs $500 and treatment costs $2.50) to $26 billion in favor of treatment (when testing costs $2,000 and treatment costs $0.50). No matter how we vary the costs, however, unconditional treatment is still more effective than conditional testing, preventing 16,800 more events per year. However, despite favorable cost-effectiveness ratios, both strategies cost upwards of $10 billion annually (Table 1).
Less expensive alternative tests such as electron beam computed tomography (18) and high-sensitivity C-reactive protein (19) have been proposed as the basis for CV screening, but would fare no better, even if they were more accurate. Thus, a screening strategy based on an imaginary test costing only $20 and capable of identifying 10% of the 14 million diabetic patients experiencing 90% of the 280,000 events in the year after testing (an event rate of 18% in those testing positive vs. 0.2% in those testing negative, and an unprecedented risk ratio of 80:1) would still prevent 8,400 fewer events than an unconditional treatment strategy, even though it would cost $8.8 billion less. Under these circumstances, the most compelling argument one could mount against the unconditional treatment strategy is that the additional benefits might not be economically justified (having a marginal cost-effectiveness ratio of $80,586 per life-year). In the final analysis, although calls to perform “outcome studies” (2) and to somehow “enrich” the process of screening (3) are well-intentioned, we cannot change the simple fact that testing, per se, cannot prevent events. Only treatment can do that. Because screen tests identify a fraction of the target population for treatment, they can lower the cost of prevention, but because they fail to identify all who might benefit from treatment, they also lower the effectiveness of prevention.
In any case, it would be rather daunting to conduct a clinical trial of these strategies. Given the small difference in projected outcome (7.0% for unconditional treatment vs. 7.6% for conditional testing over 5 years), a trial to prove the superiority of the former over the latter would require the randomization of 80,000 subjects followed for 5 years (assuming no dropouts). It is doubtful, then, that strategies such as these will ever be prospectively validated.
Accordingly, we posit 3 questions based on accepted principles of consumer protection that any responsible advocate for preventive screening should be prepared to answer:
1. How many people will it help?What is the additional number of life-years saved, or events prevented, in the target population? The greater its effectiveness, the better the strategy.
2. How much will it cost?What is the additional cost of the strategy referenced to the entire target population? The lower this cost, the better the strategy.
3. How do you plan to pay for itgiven competing needs and existing budgetary constraints?
These considerations are fully consistent with the letter and the spirit of operative federal law governing Medicare’s regulatory policies, specifically, with Executive Order 12866 enacted by President Bill Clinton on September 30, 1993 (20):
In deciding whether and how to regulate, agencies should assess all costs and benefits of available regulatory alternatives, including the alternative of not regulating. Costs and benefits shall be understood to include both quantifiable measures (to the fullest extent that these can be usefully estimated) and qualitative measures of costs and benefits that are difficult to quantify, but nevertheless essential to consider. Further, in choosing among alternative regulatory approaches, agencies should select those approaches that maximize net benefits…
Proponents of global CV screening of asymptomatic individuals, diabetic patients or not, would be well advised to take note: risk stratification, as currently practiced, could be tough on your pocketbook and hazardous to your health (21).
- Abbreviations and Acronyms
- myocardial perfusion scintigraphy
- Received July 28, 2006.
- Revision received September 12, 2006.
- Accepted September 19, 2006.
- American College of Cardiology Foundation
- Wilder B.
- Sackett D.L.,
- Haynes R.B.,
- Tugwell P.
- Sackett D.L.,
- Haynes R.B.,
- Tugwell P.
- ↵HVPA Cost Comparison—Statins/Statin Combos June 2006. Available at: http://www.hvpa.com/hvpa_rx_drug_cost_tables_files/statin_cost_comparisons_0606upd.pdf. Accessed July 27, 2006.
- ↵Physician Fee Schedule. Available at: http://www.cms.hhs.gov/PhysicianFeeSched/. Accessed July 27, 2006.
- American Heart Association
- ↵(2006) Statistical Abstract of the United States. Table 116. Deaths—Life Years Lost and Mortality Costs by Age, Sex, and Cause: 2000 and 2002 (U.S. Department of Commerce, Washington, DC), p 92.
- Grover S.A.,
- Coupal L.,
- Zowall H.,
- Alexander C.M.,
- Weiss T.W.,
- Gomes D.R.J.
- Koch R.
- Wackers F.J.T.,
- Young L.H.,
- Inzucchi S.E.,
- et al.
- Rajagopalan N.,
- Miller T.D.,
- Hodge D.O.,
- Frye R.L.,
- Gibbons R.J.
- Zellweger M.J.,
- Hachamovitch R.,
- Kang X.,
- et al.
- Clinton W.J.
- Diamond G.A.,
- Kaul S.