• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Clinical Outcomes and Cost-Effectiveness of Strategies for Managing People at High Risk for Diabetes

Posted on: Saturday, 20 August 2005, 03:00 CDT

Background: Lifestyle modification can forestall diabetes in high- risk people, but the long-term cost-effectiveness is uncertain.

Objective: To estimate the effects of the lifestyle modification program used in the Diabetes Prevention Program (DPP) on health and economic outcomes.

Design: Cost-effectiveness analysis using the Archimedes model.

Data Sources: Published basic and epidemiologic studies, clinical trials, and Kaiser Permanente administrative data.

Target Population: Adults at high risk for diabetes (body mass index >24 kg/m^sup 2^, fasting plasma glucose level of 5.2725 to 6.9375 mmol/L [95 to 125 mg/dL], 2-hour glucose tolerance test result of 7.77 to 11.0445 mmol/L [140 to 199 mg/dL]).

Time Horizon: 5 to 30 years.

Perspective: Patient, health plan, and societal.

Interventions: No prevention, DPP's lifestyle modification program, lifestyle modification begun after a person develops diabetes, and metformin.

Measurements: Diagnosis and complications of diabetes.

Results of Base-Case Analysis: Compared with no prevention program, the DPP lifestyle program would reduce a high-risk person's 30-year chances of getting diabetes from about 72% to 61%, the chances of a serious complication from about 38% to 30%, and the chances of dying of a complication of diabetes from about 13.5% to 11.2%. Metformin would deliver about one third the long-term health benefits achievable by immediate lifestyle modification. Compared with not implementing any prevention program, the expected 30-year cost/quality-adjusted life-year (QALY) of the DPP lifestyle intervention from the health plan's perspective would be about $143 000. From a societal perspective, the cost/QALY of the lifestyle intervention compared with doing nothing would be about $62 600. Either using metformin or delaying the lifestyle intervention until after a person develops diabetes would be more cost-effective, costing about $35 400 or $24 500 per QALY gained, respectively, compared with no program. Compared with delaying the lifestyle program until after diabetes is diagnosed, the marginal cost- effectiveness of beginning the DPP lifestyle program immediately would be about $201 800.

Results of Sensitivity Analysis: Variability and uncertainty deriving from the structure of the model were tested by comparing the model's results with the results of real clinical trials of diabetes and its complications. The most critical element of uncertainty is the effectiveness of the lifestyle program, as expressed by the 95% CI of the DPP study. The most important potentially controllable factor is the cost of the lifestyle program. Compared with no program, lifestyle modification for high- risk people can be made cost-saving over 30 years if the annual cost of the intervention can be reduced to about $100.

Limitations: Results depend on the accuracy of the model.

Conclusions: Lifestyle modification is likely to have important effects on the morbidity and mortality of diabetes and should be recommended to all high-risk people. The program used in the DPP study may be too expensive for health plans or a national program to implement. Less expensive methods are needed to achieve the degree of weight loss seen in the DPP.

Ann Intern Med. 2005;143:251-264. www.annals.org

Recent randomized, controlled studies have shown that diabetes can be prevented or delayed in high-risk individuals by intensive lifestyle modification programs (1, 2) or glucose-lowering drugs (2- 4). For example, in the Diabetes Prevention Program (DPP), the relative reductions in the 2.8-year incidence of diabetes were 58% in the lifestyle modification group and 31% in the metformin group (2). This raises hopes of substantially reducing the morbidity, mortality, and cost of this important disease. However, the trial was too short to observe the effects on microvascular or macrovascular outcomes, and the programs cost several hundred dollars a year (5). These findings generate obvious questions: What are the long-term effects of trying to prevent diabetes in high- risk people? Does lifestyle modification truly prevent or just postpone diabetes? Is such a prevention program "cost-effective"? What is the best strategy? A previous analysis has suggested that lifestyle modification would be cost-effective over 75 years from a societal perspective (6). We used a more thorough, clinically realistic, and independently validated model to estimate the short- and intermediate-term health and economic effects of different prevention programs for high-risk individuals and health plans, as well as for society.

METHODS

We conducted the analysis by using the Archimedes model, which has been described elsewhere (7-9). Briefly, it is a simulation model written at a relatively high level of anatomic, physiologic, clinical, and administrative detail. It uses object-oriented programming to create in the model objects that correspond to objects in reality, one-to-one. Among the hundreds of objects are people, pancreases, β cells, plasma glucose levels, coronary arteries, plaque, chest pain, emergency departments, electrocardiograms, aspirin, and angioplasties. Helpful analogies might be a flight simulator (in which the objects include the plane and its wings, airports, runways, buildings, and the wind), or the SimCity computer game. In the Archimedes model, each individual is simulated down to the level of hepatic glucose production, insulin resistance, β-cell fatigue, and similar biological variables. The core of the model is a set of differential equations that represent the anatomy and physiology pertinent to diseases and their complications. Currently, the model includes diabetes, congestive heart failure, coronary artery disease, stroke, hypertension, and asthma in a single integrated model. The structure and equations of the model pertinent to diabetes and its complications are described elsewhere (8, 9). The Appendix (available at www.annals.org) and a technical report available through our Web site (10) describe additional aspects of the model and its validations that are pertinent to this analysis. Calculations are performed by using a distributed computing network.

Clinical Events

The model includes the biological variables and outcomes relevant to diabetes and its complications. Examples are basal hepatic glucose production; insulin amount; insulin resistance; fasting plasma glucose; hemoglobin A^sub 1c^ (HbA^sub 1c^); 2-hour oral glucose tolerance; high-density lipoprotein (HDL) cholesterol, low- density lipoprotein (LDL) cholesterol, and total cholesterol; triglycerides; systolic and diastolic blood pressures and their determinants (for example, cardiac output, arterial compliance, peripheral resistance); weight and body mass index (BMI); stenosis of coronary arteries; retinopathy (assessed by the Early Treatment of Diabetic Retinopathy scale); urine protein; creatinine; peripheral neuropathy; foot ulcers of varying degrees of severity; and amputations. The use of differential equations preserves the continuous nature of biological variables as well as the interactions between them. Clinical outcomes are defined in terms of the underlying variables, as occurs in reality. For example, a person is said to have diabetes if his or her fasting plasma glucose level exceeds 6.9375 mmol/L (125 mg/dL) or results on a 2-hour oral glucose tolerance test exceed 11.0445 mmol/L (199 mg/dL). This enables the model to incorporate different definitions and changes in definitions. The model is continuous: Biological variables are changing and interacting continuously, the natural histories and severity of conditions progress smoothly, any clinical event can occur at any time, and the timing of events is as condensed or drawn out as occurs in reality. The model also includes a detailed representation of the processes and logistics of clinical care and their related costs.

Interventions, both to prevent diabetes and to manage it when it occurs, are modeled at the level of the underlying biology. Pertinent to this analysis is that in the model, diet and exercise reduce weight (2); reduce blood pressure (11); improve LDL cholesterol, HDL cholesterol, and total cholesterol levels (12); and decrease fasting plasma glucose levels (2). The effects of metformin in the model are to reduce fasting plasma glucose and 2-hour oral glucose tolerance test results (2) (by reducing basal hepatic glucose production), decrease LDL cholesterol and triglyceride levels (13), and retard weight gain.

Data used to build the model were derived from basic physiologic studies, surveys, epidemiologic studies, and clinical trials using methods described in the technical report (10). Every variable in the model is estimated from at least 1 empirical source; no variables are simply assumed. We identified specific sources by searching MEDLINE from 1970 to 28 February 2005 and by consulting textbooks and clinical experts. Because the model includes scores of continuously valued, interacting variables, it does not have simplified "states," transitions, or events at discrete time intervals that can be tabulated, as is commonly done for a Markov- type model. The equations themselves are in the technical re\port (10). For nonmathematical readers, we have calculated annualized rates of change of representative biological variables and annualized rates of occurrence of representative clinical events, and compared them with rates for comparable events observed in epidemiologic studies and clinical trials. The Appendix (available at www.annals.org) reports those results.

Costs

The DPP measured the direct medical costs of delivering the lifestyle and metformin interventions (for example, personnel, health education materials, medications, and laboratory tests). Compared with the placebo group's costs, costs in the lifestyle group were $1356 more per person in the first year, with approximately $672 in annual costs thereafter; for the metformin group, costs were $977 in the first year and averaged $742 per year thereafter (5). Following the completion of the DPP, metformin became generic. When this is considered, the cost of the metformin program is reduced to about $780 for 3 years, or about $260 a year. In the DPP study, costs apply to the year 2000.

To calculate the routine costs of providing health care to high- risk people before they develop diabetes, as well as to people with diabetes and its complications, the model includes a detailed mathematical representation of a health care system, including such elements as facilities, personnel, tests and treatments, protocols, and provider behaviors. For the base-case analysis, we obtained itemized costs from Kaiser Permanente, a nonprofit, group-practice, integrated managed care organization that provides comprehensive care (with no deductibles or copayments). The facilities, personnel, protocols, and costs in the model are based on that organization's records, at the level of detail at which actual accounts are kept (for example, 37 different kinds of office visits). The model calculates costs by keeping track of the occurrence of every event that has cost implications and adding them up. The costs assigned to any event or item were calculated by Kaiser Permanente's cost- accounting department using "micro-costing" methods (14), and they represent the real costs to the organization, not charges, reimbursements, or diagnosis-related groups. Because costs vary from setting to setting, the implications of different cost structures are examined in the sensitivity analysis. Calculation of costs applies to the year 2000. Indirect costs, such as lost time from work and decreased productivity, are included in the cost- effectiveness analysis through the Quality of Well-Being Index (14).

We calculated the effects of lifestyle and metformin interventions on quality of life. For people who do not yet have diabetes, we used utility weights reported for the participants of the DPP study (15). For people who have diabetes and its complications, we used the results of a published survey by Coffey and colleagues (16). Both surveys used the Quality of Well-Being Index. The decrements in quality of life were assumed to be additive for people who have 2 or more complications, with a limit that quality of life could not be less than 0. Use of an additive rule biases the calculation of cost/quality-adjusted life-year (QALY) in favor of a prevention program, making the program appear more cost- effective than would occur if a multiplicative model were used. We discuss the potential effects of lifestyle modification itself (for example, exercise and weight loss) in the sensitivity analysis section.

Definitions

We use the term high risk to describe people who meet the eligibility requirements of the DPP study, which include all of the following: BMI greater than 24 kg/m^sup 2^, fasting plasma glucose level of 5.2725 to 6.9375 mmol/L (95 to 125 mg/dL), and 2-hour oral glucose tolerance test result of 7.77 to 11.0445 mmol/L (140 to 199 mg/dL). Diabetes is considered to be present if the fasting plasma glucose level was greater than 6.9375 mmol/L (>125 mg/dL) or the 2- hour oral glucose tolerance test result was greater than 11.0445 mmol/L (>199 mg/dL) (17).

Validation of the Model

Internal Testing

We tested the model for internal consistency and bugs by a variety of methods, including face validity, use of inputs with known outputs, independent duplicate programming of parts, and simulation of studies and trials that have empirically known results.

Calibration

Each equation in the model was estimated by fitting functions to data. We confirmed the fits by comparing the resulting functions with the data from which they were fitted. To prevent overfitting of curves, we chose functional forms with the smallest higher-order derivatives (the smoothest curves), and we confirmed each fit visually. We also prevented overfitting of curves by conducting simulations that involved dozens of equations or spanned long durations of time and by conducting independent validations that involved data points never used to fit any equation.

Clinical Outcomes

The Archimedes diabetes model has been validated by simulating real epidemiologic studies and clinical trials at a high level of detail and by comparing the model's results with the results actually observed in the trials (9). Thus far, the model has been validated against 19 clinical trials that are pertinent to this application (2, 12, 18-34). An independent committee appointed by the American Diabetes Association chose 18 of the trials on the basis of their quality and ability to collectively span the full spectrum of the natural history of the disease, its complications, and its treatments. When the individual arms and outcomes of the different trials are counted, a total of 74 validation exercises have been published (9). Overall, the correlation between the model's results and the trials' results was 0.993. Ten of the 18 trials were not used to help build the model and thus provide independent validations. The correlation of the model's results and the trials' results for these independent validations was 0.99. The 19th trial (34) was the subject of a publicly announced, prospective, blinded prediction by the Archimedes model. Because more than half of the validations involve trials that were never used to help build the model, there was no "fitting" of any of the model's parameters to the results of these trials, and therefore no possibility of "overfitting" of the model to their results. The Appendix (available at www.annals.org) summarizes the methods and results.

The validations cover whatever outcomes were observed in the trials, including the incidence of diabetes, myocardial infarctions, strokes, retinopathy, and end-stage renal disease. The validations also cover whatever follow-up times were observed in the trials, ranging from about 3 years (2) to 15 years (18). Finally, the validations span several decades in the progression of the disease: from normal to high risk to newly diagnosed diabetes (which itself spans more than a decade, from the time a person first meets the biological definition at a fasting plasma glucose level > 6.9375 mmol/L [>125 mg/dL] to the occurrence of symptoms at fasting plasma glucose levels of about 9.99 mmol/L [180 mg/dL]), to late complications, such as myocardial infarctions and end-stage renal disease.

Costs

The direct and nondirect medical costs of the management strategies were obtained from the DPP study and are assumed to be valid. To validate the model's method for calculating the net costs associated with managing people with diabetes and their complications, we compared costs calculated by the model to the actual costs measured in an independently conducted study of patients with diabetes in Kaiser Permanente Northern California. The annual diabetes-related cost of an average patient with diabetes in this health care plan was $4241; the cost calculated by the model was $4683. For people with prediabetes, costs calculated by the model were compared with costs estimated for the people in the DPP study (5). In the trial, the average annual direct medical cost for the placebo group was about $1670. The comparable cost calculated by the model was $1552.

Figure 1. Effects of 4 programs on progression to diabetes.

This Analysis

To analyze the effects of different management strategies, we used the Archimedes model to simulate what would happen if 10 000 people who met the entry criteria of the DPP trial were exposed to 4 different management strategies and were followed for 30 years. In the first strategy, called the "DPP lifestyle program," people were immediately (while still high risk, before a diagnosis of diabetes) exposed to lifestyle modification such as that described for the lifestyle arm of the DPP study (35). Those who developed diabetes were maintained on the intensive lifestyle modification and followed for disease progression. Persons whose HbA^sub 1c^ level exceeded 7% were entered into an intensive diabetes treatment protocol designed to reduce their HbA^sub 1c^ level to below 7%. This treatment protocol was modeled after the intensive policy group of the United Kingdom Prospective Diabetes Study (UKPDS) (18) and reduced HbA^sub 1c^ levels to an average of 6.6%. The goal of 7% corresponds to the recommendations of the American Diabetes Association.

In the second strategy, called "baseline," no lifestyle or other intervention was given initially. If these people developed diabetes, they were given dietary advice (but were not entered into an intensive lifestyle program) and monitored for progression of the disease. If their HbA^sub 1c^ level exceeded 7%, they were entered into an intensive management program with a goal of controlling HbA^sub 1c^ level to below 7%, based on recommendations of the American Diabetes Association.

In a third strategy, called "lifestyle when FPG > 125" (that is, when fasting plasma glucose level > 125 mg/dL [>6.9375 mmol/L]), no lifestyle or other intervention was given initially. If these people developed diabetes, they were entered i\nto the intensive DPP lifestyle program and monitored. If their HbA^sub 1c^ level increased to greater than 7%, they began receiving intensive treatment to control their HbA^sub 1c^ levels to a goal of less than 7%, as recommended by the American Diabetes Association.

The last strategy, called "metformin," corresponded to the metformin arm of the DPP trial and involved putting patients on metformin as soon as they were determined to be at high risk. If diabetes was diagnosed in these patients, they continued receiving metformin, were given dietary advice, and continued to be monitored. If their HbA^sub 1c^ level exceeded 7%, their drug treatment was intensified to control the HbA^sub 1c^ level to less than 7%.

We assumed that both the effectiveness and the costs observed at the end of the DPP would persist as long as a person was receiving lifestyle intervention (2, 5). Specifically, after an initial weight loss of about 7%, the simulated persons' weights gradually increased to a 4% weight-loss level after 3 years, as occurred in the DPP trial, and that degree of weight loss persisted as long as they were receiving the lifestyle intervention. When a person's fasting plasma glucose level reached 9.99 mmol/L (180 mg/dL) or their HbA^sub 1c^ level reached 7%, the person was switched from the DPP lifestyle program to the intensive diet and exercise treatment protocol described above. Throughout the entire program, every individual in every group was at risk for other clinical events, such as angina, stroke, and heart attack.

Table 1. Expected Outcomes over Various Time Horizons for a Typical Person with Diabetes Prevention Program Characteristics*

Perspectives

We examined each management strategy from 3 different points of view: a high-risk person, a 100 000-member health plan (approximately twice the average size of managed care organizations), and a societal perspective. For the health plan and societal perspectives, we included the entire distribution of people with DPP characteristics, in the proportions seen in the DPP trial (2). The health plan point of view differs from the more traditional "health system perspective" in that it takes into account the size of the plan, the need to consider shorter time horizons, and member turnover. The societal perspective corresponds to the recommendations of the Panel on Cost-effectiveness Analysis (14). Specifically, we calculated the logistic events and clinical outcomes for each person in the population over a 30-year horizon, noting when each event and outcome occurred. We then assigned costs to each logistic event (for example, each test, visit, admission, and treatment) and assigned quality-of-life weights to each clinical outcome (for example, each heart attack, stroke, and amputation) to calculate the time stream of costs and quality-adjusted life-years (QALYs) for each person. We added all the costs and the QALYs for the entire population to obtain time streams of the aggregated costs and QALYs. We used standard methods (14) to convert the time streams of events and outcomes into their present values, and divided the former by the latter to determine the cost per QALY. Because the cost per QALY is a ratio, it is not necessary to do this for the millions of people in the United States who are at high risk but only for a statistically meaningful sample-in this case, 10 000 such people. Logistic costs experienced by individuals (for example, out- of-pocket medical costs, nonmedical costs such as travel time, or social costs such as lost productivity) were included in the denominator through the use of the Quality of Well-Being Index.

Table 2. Expected Number of Cases of Diabetes and Complications in a Health Plan of 100 000 Members*

Calculations

For the main comparison, the DPP lifestyle program was compared with the baseline program. We also calculated its incremental cost- effectiveness compared with the "lifestyle when FPG > 125" program and the metformin program. Results for each program were calculated for 10 000 people. All calculations were performed with full precision; results were rounded for presentation. For all calculations of cost per QALY in the base case, both costs and QALYs were discounted at an annual rate of 3%. For the reference comparison, the DPP lifestyle program was compared with the baseline program. We also calculated the incremental cost-effectiveness of the lifestyle program compared with the other programs.

Role of the Funding Sources

The funding sources did not influence the decision to analyze this topic; the design, conduct, or reporting of the study; or the decision to publish the manuscript.

RESULTS

Individual High-Risk Person Perspective

Figure 1 shows the effect of lifestyle modification and metformin on development of diabetes for different lengths of follow-up. The 30-year probability that a person with DPP characteristics will develop diabetes is about 72%. Lifestyle modification, if persistent, would reduce that risk by a relative 15%, to about 61%. This indicates that over 30 years, the disease would be prevented in about 11% of cases and postponed in about 61%. In absolute terms, the risk reduction would be 11 percentage points and the long-term number needed to treat for benefit would be about 9. For metformin, about 4% of cases of diabetes would be prevented over a 30-year period, a relative reduction of about 5.5%.

Table 1 shows the effects of a DPP lifestyle program on clinical outcomes for various periods of time for a typical high-risk person. For example, intensive lifestyle modification would decrease the chance that a high-risk person would have a myocardial infarction over the next 30 years by an absolute 1.7 percentage points (from 12% to 10.3%). The chance of having any severe complication would be decreased by an absolute 8.4 percentage points (from 38.2% to 29.8%).

Health Plan Perspective

The health and economic outcomes that health plans can expect from implementing a DPP-like lifestyle program depend on the size of the plan, the planning horizon, and the turnover rate. A health plan with 100 000 members can expect to have approximately 4% of its membership, or 4000 people, at high risk for diabetes (36). Table 2 shows the expected number of events for that cohort of people, for various lengths of follow-up, assuming no turnover.

Table 3 shows the effects of the DPP lifestyle program on a health plan's costs for several time horizons. For most events, they are multiples of the probabilities in Table 2. For myocardial infarction, however, the numbers for health plans include second and further events. To help health plans anticipate how the costs and savings will be distributed across different departments, Table 3 also shows the financial effects for the main components of a plan's budget. The last 2 rows of the table show the net cost translated into a per member per month cost for high-risk people and for the entire membership, averaged over the applicable time horizon.

Table 3. Effect of Lifestyle Program on Expected Costs in a Health Plan with 100 000 Members for 4 Time Horizons*

For calculating the cost-effectiveness of the DPP lifestyle program from the plan's perspective, we assumed a turnover rate of 10% per year. The expected value of the cost/QALY is about $143 000. The expected cost of gaining a QALY is higher for shorter time horizons. Over periods of 5, 10, or 20 years after the start of the program, the expected costs of gaining a QALY would be about $2.7 million, $1.2 million, and $180 000, respectively. The much higher costs required to gain a QALY in the earlier years of follow-up are due to the fact that in the first few years after the introduction of the program, the costs are high and the clinical benefits, with their related cost-savings, are relatively small.

Societal Perspective

Figure 2 and Table 4 show the cost-effectiveness of the DPP lifestyle program from a societal perspective. Compared with the baseline program, providing a nationwide DPP lifestyle program to all high-risk people would gain a QALY at a cost of about $62 600 (Table 4, average cost/QALY). As for health plans, the cost/QALY at the societal level depends on the time horizon: The cost/QALY after 5, 10, and 20 years would be about $492 000, $222 000, and $88 000, respectively (not shown in the table). Calculation of the expected cost/QALY of the DPP lifestyle program from the societal perspective differs from the cost/QALY calculated for the health plan because of the lack of turnover and because the recommended methods for calculating cost/QALY from a societal perspective do not include the size of a population.

Incremental Cost-Effectiveness

Table 4 and Figure 2 also show the overall effects of the other programs on the discounted 30-year costs and QALYs from the societal perspective. The "lifestyle when FPG > 125" program delivers the greatest increase in QALYs for the cost. Compared with the baseline program, this program would deliver a 30-year cost/QALY of approximately $24 500. If this strategy were made the reference point for calculating the cost-effectiveness of the DPP lifestyle program, then the incremental cost-effectiveness of that program would be about $202 000 per QALY gained, over a 30-year horizon.

Table 4. 30-Year Costs, Quality-Adjusted Life-Years, and Incremental Costs/Quality-Adjusted Life-Years for 4 Programs from Societal Perspective (Discounted 3%)*

Sensitivity Analysis

There are 2 main types of uncertainty about the results of any model. One is the structure of the model-the equations it uses. The other is the accuracy of the values that are assigned to the variables in the equations. Traditional sensitivity analysis addresses the second source of variability and uncertainty, and the results of our sensitivity analyses are described in this section. However, traditional sensitivity analysis does not address uncertainty about the structure of a model b\ecause it uses the very model about which the uncertainty exists and merely shows the effects of changing some variables within that structure. It does not make any comparisons with actual, empirically observed results. Far from exploring uncertainty about a model's structure, traditional sensitivity analysis assumes that the structure is correct. The best way to explore variability and uncertainty about the entire model-both its structure and its variables-is to compare the model's results with real-world results. The ability to do this depends on the available empirical evidence. The fact that the Archimedes model independently and prospectively predicted the results of the DPP provides a test of the parts of the model that address prediabetes, early diabetes, and the effects of metformin and lifestyle. The model's success in matching or predicting the results of 18 other clinical trials relating to diabetes and its complications tests the structure of those other parts of the model.

Figure 2. Quality-adjusted life-years (QALYs) versus cost for 4 programs.

Several sources of variation and randomness affect the results. Individuals vary widely with respect to such factors as their chances of developing diabetes, when they will develop it, the rate at which it will progress, and their chances and timing of complications. In addition to these individual variations are random factors that affect such things as test results and response to treatments. To address these factors, the Archimedes model represents them as random variables (10). Each simulated individual is created by simultaneously drawing a value for each variable from its respective distribution. The calculated results are subject to a range of uncertainty determined by the number of individuals in the simulation. In this case, the number of simulated individuals is so large that the effect is negligible. For example, the 95% CI for the 30-year chance of developing diabetes is 71.28% to 73.08%. The 95% CI for the decrease in chances of developing diabetes caused by the lifestyle program is 10.83% to 10.85%.

A much more important factor is the uncertainty in the results of the DPP itself. The observed relative reduction in incidence of diabetes after an average follow-up of 2.8 years was 58%, but the 95% CI for this relative reduction was 48% to 66% (2). This has a large effect on the estimated health benefits of the program. For example, incorporating this source of uncertainty in the model increases the 95% CI for the 30-year chance that a high-risk person will develop diabetes to 69.4% to 75%, and increases the 95% CI for the absolute decrease in the chance of developing diabetes caused by the lifestyle program to 7.3 to 14.4 percentage points. The 95% CIs for the chance of any complication and the effect of the lifestyle program on that chance are 35.2% to 41.3% and 4.9 to 12 percentage points, respectively.

An important factor affecting the expected clinical effects of the DPP for a health plan is its size. This is conceptually similar to the sampling variation that affects the outcomes of clinical trials. For example, with the baseline program, the number of myocardial infarctions (including repeated myocardial infarctions) that can be expected to occur over a 30-year period is 659 (Table 2), but the 95% CI for the number of myocardial infarctions ranges from 612 to 706, or a range of 7% from the expected value given in Table 2. The 95% CI for the difference in myocardial infarctions caused by the lifestyle program is 9.5%. The CIs are smaller for more frequent outcomes, such as the occurrence of diabetes (2%), or for aggregated outcomes, such as "total for all complications" (4.3%), but are larger for less frequent outcomes, such as those relating to nephropathy and neuropathy (>50%).

The size of the health plan has a much smaller effect on the costs shown in Table 3. For example, the 95% CI for the increase in cost at 30 years ($58.88 million in Table 2) is about $56.8 million to $60.8 million. However, because of the wide range of uncertainty about the clinical benefits in a plan of this size, uncertainty about the cost of gaining a QALY is considerable. Figure 3 shows a probability distribution for the cost/QALY for the DPP lifestyle program versus the baseline program. The most likely value (mode) is about $121 000 per QALY gained, but the expected value of the cost/ QALY, which considers the uncertainty about the magnitudes of the clinical benefits, is about $143 000. It is exceedingly unlikely (<0.1% chance) that the cost/QALY for a health plan of 100 000 members would be less than the arbitrary threshold of $50 000.

Several other factors that are typically known to a health plan but that can vary across plans affect the estimates of quality of life and cost/QALY. Table 5 summarizes these, including turnover (0%, 5%, 10% [reference], and 15%); discount rate (0%, 3% [reference], and 5%); cost of diabetes care (across-the-board 20% increase and 20% decrease); size of the health plan (50 000, 100 000 [reference], 200 000, and 500 000); and the cost of the lifestyle intervention (20% increase, 20% decrease). Table 5 also shows the effects of uncertainty or variability about utility weights (across- the-board 20% increase and 20% decrease) and includes several time horizons.

Figure 3. Probability distribution for cost/quality-adjusted life- year (QALY) for Diabetes Prevention Program lifestyle program versus no program.

Figure 4. Net cost to health plan (discounted) of 2 programs, as function of annual cost of lifestyle program.

Uncertainty in the results of the DPP itself also has a large effect on the cost/QALY calculated from the societal perspective. When this is considered, the uncertainty about the increase in QALYs (approximately 0.2 year) is greater than the increase in QALYs itself (0.125 year). This large range of uncertainty about the gain in QALYs renders the cost/QALY almost meaningless for practical planning.

The most important factor that is potentially controllable is the cost of the lifestyle intervention. Of particular interest is the possibility that fundamentally different, less expensive ways of modifying lifestyles to lose weight may be found. For example, the DPP Research Group has suggested that group-based programs might be just as effective in achieving sustained weight loss as the individual-based program used in the DPP study (15). According to their methods and assumptions, the annual cost of a lifestyle program would be reduced to about $650 over 3 years, or about $217 a year. Table 5 shows the cost implications of such a program, assuming the same degree of effectiveness on weight, blood pressure, and cholesterol levels, in the row labeled "Group ($217/y)."

Figure 4 shows the relationship between the annual per person cost of a lifestyle intervention and the 30-year per person increase in net cost of the DPP lifestyle program compared with no program (solid line) and the "lifestyle when FPG >125" program compared with no program (dotted line). Figure 5 shows the effect of the annual per person cost of a lifestyle intervention on the 30-year cost/ QALY for the following comparisons: DPP lifestyle versus baseline (solid line), "lifestyle when FPG > 125" versus baseline (dotted line), and DPP lifestyle versus "lifestyle when FPG > 125" (dashed line).

Table 5. Sensitivity Analysis: Cost/Quality-Adjusted Life-Year*

Figure 5. Cost/quality-adjusted life-year (QALY) of 3 comparisons as function of cost of lifestyle program.

DISCUSSION

Our analysis indicates that the 30-year probability that a person with DPP characteristics will develop diabetes is about 72%, and that lifestyle modification, if persistent, would reduce that risk by a relative 15%, to about 61%. This indicates that over 30 years, the disease would be prevented in 11% of cases (number needed to treat for benefit, 9) and postponed in 61%. This may appear to conflict with the results of the DPP study, in which the authors stated that "the lifestyle intervention reduced the incidence [of diabetes] by 58%" (2). The apparent conflict is explained by the fact that the 58% figure applies to the average follow-up time of the DPP study, which was only 2.8 years. It does not apply to the long-term effect. In fact, the relative effect of the lifestyle program on development of diabetes decreases over time (Figure 1). Two main factors diminish the relative effect. First, the relative effect of lifestyle on progression to diabetes (measured as percentage reduction) decreases over time. This is seen in the DPP study itself; the relative reduction was 68% at 1 year, 64% at 2 years, 51% at 3 years, and 46% at 4 years (Figure 2 in reference 1). Second, even if the rate of progression to diabetes is decreased by a constant proportion (for example, is assumed to be a constant 58%), as people progress from high risk to diabetes there are fewer people at high risk and therefore fewer people to be affected by the treatment. The simplest case, in which both the transition from high risk to diabetes and the relative effect of lifestyle on that transition are assumed to be constant, is illustrated in an analysis of this problem by the DPP Research Group. Their report illustrates the gradual diminution in relative effect (6). The Archimedes model shows the same phenomenon of a diminishing relative effect, but with greater accuracy because it incorporates the fact that neither the transition rate nor the relative effect of lifestyle is constant.

Even though the long-term effect of the lifestyle program is smaller than that observed in the short duration of the trial, and even though effects always appear smaller when presented in absolute rather than relative terms, the long-term absolute effects of lifestyle modification are still very important clinically. The most important messages for a person with DPP characteristics are the followin\g: 1) Your risk for developing diabetes is about 72%; 2) Your chance of having a serious complication of the disease is almost 40%, and your chance of dying of a complication of the disease is about 13%; 3) If you change your lifestyle and permanently lose just 4% of your weight (about 7 pounds for the typical high-risk person), you can reduce your risk for a serious complication or for dying of diabetes by about 20% (relative risk); and 4) Over the range of weight loss seen in the DPP, every pound you lose and keep off permanently will lower your risk for a serious complication more than 1 percentage point (absolute risk).

From the perspective of a health plan, the clinical benefits of implementing a DPP-like lifestyle modification program are clear. Reductions in cases of diabetes and its complications would be seen within 5 years and would gradually grow over time (Table 2). Over a 30-year horizon, such a program would be expected to prevent about 11% of cases of diabetes, about 22% of its most serious complications, and about 18% of diabetes-related deaths in people who are at high risk for this disease (Table 2).

Unfortunately, a DPP-like lifestyle program would considerably increase costs. Even for the most optimistic picture-a 30-year horizon and assuming no turnover-the net effect on diabetes-related costs would be an increase of about 25% (calculated from Table 3). The proportional increase would be even greater over shorter time horizons. For example, in the first 5 years the increase in diabetes- related costs would be about 50%. Cast in terms of per person costs and the immediate burden a health plan would face, the actual cost increase in the first 5 years would be about $60 per month per high- risk person, or about $2.30 per member (any person) per month. To put this in perspective, it would represent a permanent increase in total costs for all interventions for all conditions of about 1 percentage point. In an era when the United States is struggling to keep annual health care cost increases to single digits, this would represent a heavy burden.

For cost-effectiveness, from the perspective of a health plan of about 100 000 members and with a member turnover of about 10% per year, the expected 30-year cost/QALY of a DPP-like lifestyle intervention compared with no prevention program at all would be in the range of $143 000. The cost/QALY is higher for shorter time horizons and smaller plans (Table 5). From a societal perspective, the cost/QALY compared with no program would be approximately $63 000 over a 30-year horizon and would be higher for shorter time horizons (Table 5). From either perspective, and setting aside uncertainty about the effectiveness of the intervention itself, there is a less than 0.1% chance that the 30-year cost/QALY would be below $50 000.

Two additional factors make the argument for cost-effectiveness even more pessimistic. First, other options are available. The figures just given represent average cost/QALY of the DPP lifestyle intervention, in which the comparison is with no program at all. Another possible strategy is to delay implementing the lifestyle intervention until after a person actually gets diabetes. Compared with no program, this has a 30-year cost/QALY of about $24 500. If this is used as the reference for calculating the marginal cost/ QALY of the DPP lifestyle program (which is begun immediately upon determination that a person is at high risk, instead of waiting until diabetes develops), the expected cost-effectiveness of that program increases to about $202 000 from a societal perspective (Table 4). It would be even greater for a health plan. Indeed, even the metformin program has a lower average cost/QALY than the DPP lifestyle program.

Second, the cost/QALY is considerably higher in early years than in later years (Table 5). In fact, it is the outcomes in the early years (5 and 10 years) that will actually occur. Calculation of longer-term costs/QALY requires an assumption that the program will be in place for decades without change and that there will be no technological advancements in the management of diabetes, such as new drugs, tests, devices (for example, insulin pumps), or procedures (pancreas transplants). While the long-term cost/QALY is useful as a method of making comparisons across technologies or as a measure for determining the sensitivity of results to various factors, the long-term cost/QALY is an abstraction and does not represent an outcome that health plans or a national program would actually experience.

The results are sensitive to several factors, as described in the section on sensitivity analysis. The most important is uncertainty about the results of the DPP trial itself. On this point, it is encouraging that the Archimedes model independently and prospectively predicted results almost identical to those seen in the DPP. The Archimedes model has also been validated against a broad range of other clinical trials to help ensure that it accurately represents the best information available to date. The high concordance between the Archimedes prediction of the DPP results and the actual DPP results indicate that the DPP's results are consistent with preexisting information.

Given the high clinical value of improving the lifestyle program, finding ways to deliver it at a lower cost is critical. There are 2 main ways to do this: Delay starting the program until people actually develop diabetes, and find less expensive ways to modify lifestyle and maintain the changes. From the perspective of a health plan, implementing a lifestyle modification program at the time people develop diabetes (after fasting plasma glucose level is >9.9375 mmol/L [>125 mg/dL] or 2-hour oral glucose tolerance test result is >11.0445 mmol/L [>199 mg/dL]) would increase the net cost only 3% in the first 5 years, and 9% over 30 years (data not shown in the tables). This is considerably more feasible than the 50% and 25% 5-year and 30-year increases that occur with the DPP lifestyle program. In absolute terms, the increase in per member per month cost with the "lifestyle when FPG > 125" program would be about $0.37 in the first 5 years, and lower thereafter.

Finding less expensive ways to modify people's lifestyles and achieve the degrees of weight loss seen in the DPP study could have a more important effect on the desirability and cost-effectiveness of lifestyle modification. Less expensive methods would affect the net costs of both the DPP lifestyle program and the "lifestyle when FPG > 125" program. The relationship between the annual cost of achieving lifestyle changes and the net 30-year per-person cost of diabetes-related care can be determined from Figure 4. For example, lifestyle modification after the diagnosis of diabetes ("lifestyle when FPG > 125") would become cost-neutral over 30 years if the annual cost of implementing it could be reduced to about $220. To make the DPP lifestyle program cost-neutral, the annual cost of implementing the lifestyle changes would have to be approximately $100. As the cost of implementing the lifestyle program is reduced, the "DPP lifestyle when FPG > 125" program becomes even more attractive; at an annual implementation cost of $100, the "DPP lifestyle when FPG > 125" program is cost-saving.

Figure 5 shows the cost/QALY that would be achieved with various programs under different assumptions about the annual cost of a lifestyle modification program, for a societal perspective and 30- year horizon. For example, if the annual cost of a lifestyle program could be decreased to $217, compared with no program (baseline), a lifestyle intervention delivered as soon as a person develops diabetes ("lifestyle when FPG > 125") would gain a QALY at a cost of about $2000. Compared with waiting until the diagnosis of diabetes ("lifestyle when FPG > 125"), the marginal cost/QALY of starting the lifestyle program immediately (DPP lifestyle) would be about $49 500. Figure 5 can also be used to work backward to find the annual cost of lifestyle modification needed to make the DPP lifestyle program cost-effective, for any specified threshold for an acceptable cost/QALY. For example, if $50 000 is determined to be an appropriate threshold for a 30-year cost/QALY, the DPP lifestyle intervention should have an annual cost of about $210 to justify beginning the program as soon as a person is determined to be at high risk, instead of waiting until after the diagnosis of diabetes (read from Figure 5, solid line).

It is unlikely that the effects of prevention on specific complications of diabetes will ever be shown in a clinical trial. The DPP study involved approximately 1000 people in each of the 3 groups, and the average follow-up time was only about 3 years. The results in Table 1 imply that if a trial like the DPP were continued for an average follow-up of 10 years, the power for finding a statistically significant effect on the most frequent complication, myocardial infarction, at a significance level of a P value less than 0.05 would be about 7.5% (37). A 20-year follow-up would increase the power to about 14%. The prospects for showing an effect on a compound outcome are better, but still difficult and expensive. For example, if a combined outcome of any serious complication or death were used, extending the trial for 10 years would have a power of about 70%. The cost of the 3-year DPP study was on the order of $175 million; extending the trial for 10 years would at least double that cost. The implications are that it is unlikely a trial will be conducted to document the effects of prevention on the complications of diabetes. A corollary is that the evidence is unlikely to get any better, and it would be pointless to insist on such documentation before recommending prevention.

Our results differ from those of a recent analysis by Herman and colleagues, which showed a larger effect on the chance \of developing diabetes, larger increases in QALYs, lower costs, and a lower cost/QALY (6). Several factors explain the differences. First, the models have very different structures. Herman and colleagues used a Markov model, which represents diseases as a set of discrete clinical "states," represents the progression of a disease as annual transitions between states, and represents the effects of treatments as changes in the chances of transitions between states. The Archimedes model was designed to be considerably more thorough and clinically realistic. Appendix Table 1 (available at www.annals.org) provides additional details and examples from the 2 models. Second, because they had to fit a complex disease such as diabetes and its complications into the Markov structure, and because there are no direct data for many of the transition probabilities, Herman and colleagues had to make many simplifications and assumptions. These are described in their paper (6) and its accompanying technical report (available at www.annals.org). These assumptions differ considerably from the way the same issues are addressed in the Archimedes model. Some of the most important differences are compared in Appendix Table 2 (available at www .annals.org), which readers are encouraged to study. Third, Herman and colleagues' model has not been validated.

Our analysis has several limitations. First, there may be other populations for whom the natural history or response to lifestyle modification is substantially different from that seen in the DPP study or UKPDS. This analysis would not apply to them. To the extent that such populations can be identified, their rates of disease progression determined, and the effects of lifestyle changes determined, separate analyses must be done. Second, what happens in clinical trials might not represent what happens in actual practice, especially in terms of the effectiveness of the interventions. Although the Archimedes model has the ability to include such factors as provider and patient behaviors (for example, adherence to a guideline or treatment recommendation), for this analysis we assumed that the lifestyle and metformin interventions would be as effective as in the DPP study. The effect of this assumption is a bias in favor of the lifestyle intervention (reducing the calculated cost/QALY).

This analysis uses costs from a very large, nonprofit managed care organization that provides comprehensive care to a diverse population. Thus, our costs reflect true resource costs, as recommended for the societal perspective (14), and avoid many of the distortions that can occur because of deductibles, copayments, profits, discounts, or underpayment. However, in settings where charges reflect true costs less accurately, the costs incurred by insurers, individuals, employers, Medicare programs, and others who pay bills may be different. For example, market prices or charges will tend to be higher than true resource costs in for-profit settings and lower in settings where reimbursement rates fall short of true resource costs. Because financial arrangements vary so widely across settings, no single set of costs will reflect every setting or the United States as a whole. Sensitivity analysis is the best way to estimate the cost-effectiveness of the programs for settings where charges differ more widely from true resource costs. The range of values shown in Table 5 covers a wide variety of health plans. The cost-effectiveness calculated for the societal perspective is less vulnerable to the differences in cost structures seen across real settings because the societal perspective is intentionally designed to be an idealized representation of true resource costs.

Finally, although the Archimedes model has been rigorously validated against the pertinent clinical trials, there is no way to ensure that it is perfectly accurate for predicting events that have never been studied empirically with trials. The purpose of the validations is to confirm that the model faithfully represents the pathophysiology of the dis ease and the effects of treatments as they are currently understood through existing evidence. This builds confidence but does not guarantee that the model will be accurate for questions for which no evidence exists.

Context

A previous Markov model-based analysis estimated that use of the Diabetes Prevention Program diet and exercise intervention to forestall diabetes in high-risk people would be cost-effective from a societal perspective.

Contribution

Using a validated model designed to be more complete and realistic than previous models, the authors estimated that the intervention would cost society about $62 600 per quality-adjusted life-year saved. It would be cost-saving if the annual cost of the intervention decreased from $672 to $100.

Implications

This model suggests that the Diabetes Prevention Program intervention costs more per quality-adjusted life-year saved than previously estimated, and health plans and insurers may consider it too expensive to cover.

-The Editors

References

1. Tuomilehto J, Lindstrom J, Eriksson JG, Valle TT, Hamalainen H, Ilanne-Parikka P, et al. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med. 2001;344:1343-50. [PMID: 11333990]

2. Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346:393- 403. [PMID: 11832527]

3. Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M, et al. Acarbose for prevention of type 2 diabetes mellitus: the STOP-NIDDM randomised trial. Lancet. 2002;359:2072-7. [PMID: 12086760]

4. Buchanan TA, Xiang AH, Peters RK, Kjos SL, Marroquin A, Goico J, et al. Preservation of pancreatic beta-cell function and prevention of type 2 diabetes by pharmacological treatment of insulin resistance in high-risk Hispanic women. Diabetes. 2002;51:2796-803. [PMID: 12196473]

5. Herman WH, Brandie M, Zhang P, Williamson DF, Matulik MJ, Ratner RE, et al. Costs associated with the primary prevention of type 2 diabetes mellitus in the diabetes prevention program. Diabetes Care. 2003;26:36-47. [PMID: 12502656]

6. Herman WH, Hoerger TJ, Brandle M, Hicks K, Sorensen S, Zhang P, et al. The cost-effectiveness of lifestyle modification or metformin in preventing type 2 diabetes in adults with impaired glucose tolerance. Ann Intern Med. 2005;142: 323-32. [PMID: 15738451]

7. Schlessinger L, Eddy DM. Archimedes: a new model for simulating health care systems-the mathematical formulation. J Biomed Inform. 2002;35:37-50. [PMID: 12415725]

8. Eddy DM, Schlessinger L. Archimedes: a trial-validated model of diabetes. Diabetes Care. 2003;26:3093-101. [PMID: 14578245]

9. Eddy DM, Schlessinger L. Validation of the Archimedes diabetes model. Diabetes Care. 2003;26:3102-10. [PMID: 14578246]

10. Schlessinger L, Eddy DM. Equations for the Archimedes model of diabetes and its complications. Technical Report, Parts A and B. 2004. Accessed at http:// archimedesmodel.com.

11. Blumenthal JA, Sherwood A, Gullette EC, Babyak M, Waugh R, Georgiades A, et al. Exercise and weight loss reduce blood pressure in men and women with mild hypertension: effects on cardiovascular, metabolic, and hemodynamic functioning. Arch Intern Med. 2000;160:1947-58. [PMID: 10888969]

12. Frick MH, Elo O, Haapa K, Heinonen OP, Heinsalmi P, Helo P, et al. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-aged men with dyslipidemia. Safety of treatment, changes in risk factors, and incidence of coronary heart disease. N Engl J Med. 1987;317:1237-45. [PMID: 3313041]

13. DeFronzo RA, Goodman AM. Efficacy of metformin in patients with noninsulin-dependent diabetes mellitus. The Multicenter Metformin Study Group. N Engl J Med. 1995;333:541-9. [PMID: 7623902]

14. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost- Effectiveness in Health and Medicine. New York: Oxford Univ Pr; 1996.

15. Within-trial cost-effectiveness of lifestyle intervention or metformin for the primary prevention of type 2 diabetes. Diabetes Care. 2003;26:2518-23. [PMID: 12941712]

16. Coffey JT, Brandle M, Zhou H, Marriott D, Burke R, Tabaei BP, et al. Valuing health-related quality of life in diabetes. Diabetes Care. 2002;25:2238-43. [PMID: 12453967]

17. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care. 1997;20:1183- 97. [PMID: 9203460]

18. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352:837-53. [PMID: 9742976]

19. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin- dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977-86. [PMID: 8366922]

20. Parving HH, Lehnert H, Brochner-Mortensen J, Gomis R, Andersen S, Arner P, et al. The effect of irbesartan on the development of diabetic nephropathy in patients with type 2 diabetes. N Engl J Med. 2001;345:870-8. [PMID: 11565519]

21. MRC/BHF Heart Protection Study of antioxidant vitamin supplementation in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:23-33. [PMID: 12114037]

22. Yusuf S, Sleight P, Pogue J, Bosch J, Davies R, Dagenais G. Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med. 2000;342:145-53. [PMID: 10639539]

23. Effects of ramipril on cardiovascular and microvascular outcomes in people with diabetes mellitus: results of the HOPE study and MICRO-HOPE substudy. Heart Outcomes Prevention Evaluation St\udy Investigators. Lancet. 2000;355: 253-9. [PMID: 10675071]

24. Sacks FM, Pfefifer MA, Moye LA, Rouleau JL, Rutherford JD, Cole TG, et al. The effect of pravastatin on coronary events after myocardial infarction in patients with average cholesterol levels. Cholesterol and Recurrent Events Trial investigators. N Engl J Med. 1996;335:1001-9. [PMID: 8801446]

25. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD. The effect of angiotensin-converting-enzyme inhibition on diabetic nephropathy. The Collaborative Study Group. N Engl J Med. 1993;329:1456-62. [PMID: 8413456]

26. Lewis EJ, Hunsicker LG, Clarke WR, Berl T, Pohl MA, Lewis JB, et al. Renoprotective effect of the angiotensin-receptor antagonist irbesarran in patients with nephropathy due to type 2 diabetes. N Engl J Med. 2001;345:851-60. [PMID: 11565517]

27. Lewis SJ, Moye LA, Sacks FM, Johnstone DE, Timmis G, Mitchell J, et al. Effect of pravastatin on cardiovascular events in older patients with myocardial infarction and cholesterol levels in the average range. Results of the Cholesterol and Recurrent Events (CARE) trial. Ann Intern Med. 1998;129:681-9. [PMID: 9841599]

28. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. The Long-Term Intervention with Pravastatin in Ischaemic Disease (LIPID) Study Group. N Engl J Med. 1998;339:1349-57. [PMID: 9841303]

29. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). SHEP Cooperative Research Group. JAMA. 1991; 265:3255-64. [PMID: 2046107]

30. The Lipid Research Clinics Coronary Primary Prevention Trial results. I. Reduction in incidence of coronary heart disease. JAMA. 1984;251:351-64. [PMID: 6361299]

31. MRC trial of treatment of mild hypertension: principal results. Medical Research Council Working Party. Br Med J (Clin Res Ed). 1985;291:97-104. [PMID: 2861880]

32. Shepherd J, Cobbe SM, Ford I, Isles CG, Lorimer AR, MacFarlane PW, et al. Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia. West of Scotland Coronary Prevention Study Group. N Engl J Med. 1995;333:1301-7. [PMID: 7566020]

33. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet. 1994;344: 1383-9. [PMID: 7968073]

34. Colhoun HM, Betteridge DJ, Durrington PN, Hitman GA, Neil HA, Livingstone SJ, et al. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo- controlled trial. Lancet. 2004;364: 685-96. [PMID: 15325833]

35. The Diabetes Prevention Program (DPP): description of lifestyle intervention. Diabetes Care. 2002;25:2165-71. [PMID: 12453955]

36. Benjamin SM, Valdez R, Geiss LS, Rolka DB, Narayan KM. Estimated number of adults with prediabetes in the US in 2000: opportunities for prevention. Diabetes Care. 2003;26:645-9. [PMID: 12610015]

37. Eddy DM, Hasselblad V. FAST*PRO, Software for Meta-analysis by the Confidence Profile Method. Boston: Academic Pr; 1992.

David M. Eddy, MD, PhD; Leonard Schlessinger, PhD; and Richard Kahn, PhD

From the Archimedes Project, Kaiser Permanente, Oakland, California, and the American Diabetes Association, Alexandria, Virginia.

Note: The order of authorship for Drs. Eddy and Schlessinger is alphabetical.

Disclaimer: The views expressed herein are those of the authors and do not necessarily reflect those of Kaiser Permanente or the American Diabetes Association.

Grant Support: This analysis was funded by Kaiser Permanente. The validation of the Archimedes model was funded by a grant from the American Diabetes Association, supported in part by Bristol-Myers Squibb.

Potential Financial Conflicts of Interest: The American Diabetes Association, which provided funding for validation of the Archimedes model, has received grants from Bristol-Myers Squibb, the manufacturer of metformin.

Requests for Single Reprints: David M. Eddy, MD, PhD, 1426 Crystal Lake Road, Aspen, CO 81611; e-mail, eddyaspen@yahoo.com.

Current author addresses and author contributions are available at www .annals.org.

Copyright American College of Physicians Aug 16, 2005

Source: Annals of Internal Medicine

More News in this Category