The SD diabetes model specifies how population groups accumulate in several states of health (e.g., prediabetes, uncomplicated diabetes, complicated diabetes) along with the rates at which people flow from one state to another. The full model contains many such states and rates (13); however, in Figure 2 we show only a simplified and generic view for explanatory purposes.

Figure 2 depicts a generic stock-and-flow structure that can be used to illustrate the diagnosed prevalence for any disease. One may think of the box labeled diagnosed prevalence as a bathtub in which the water level represents the number of people who have been diagnosed with a disease (23). The rate at which a condition is diagnosed, diagnosed onset, is analogous to the rate at which water flows into the bathtub. The rates of recovery or death for people with diagnosed disease are analogous to the rates at which water flows out of the bathtub through two separate drains. As Figure 2 illustrates, all changes in diagnosed prevalence must be accounted for by changes in these related flows. (For a complete accounting, the flows of births, migration, and recovery among the undiagnosed, as well as deaths among those without the disease and deaths among the undiagnosed, would be needed, but for clarity these are not depicted in Figure 2.)

Following is a summary of how the generic elements of Figure 2 relate specifically to diabetes:

The SD model employed in our simulations tracks the flows and accumulations of people with normal blood glucose levels, undiagnosed or diagnosed prediabetes, undiagnosed or diagnosed diabetes without complications, and undiagnosed or diagnosed diabetes with complications. In the model, we specify key factors — some of them potentially amenable to policy intervention — that may change over time and that affect the model's population flows. These variable policy factors include the prevalence of obesity (i.e., the leading modifiable risk factor for diabetes); the prevalences of glycemic screening, prediabetes management, and diabetes management; as well as the percentage of the population with access to preventive care (13). A scenario involves specified future values for each variable factor. The model can then simulate the consequences of any given scenario for future trajectories of diagnosed prevalence and other measures of disease burden.

Beginning in 2004, results from the first simulation experiment focus on a status quo future, in which we assumed no further changes in the scope or effectiveness of prevention, detection, or management efforts or in the prevalence of obesity. In Figure 1, the line marked status quo (from point F to G) shows the projected prevalence of diagnosed diabetes through 2010 under these assumptions. Diagnosed prevalence is projected to rise throughout this period because the inflow of people with newly diagnosed diabetes is projected to exceed the rate at which people are dying. Under this scenario, the prevalence of diagnosed diabetes is projected to increase 21%, from 48.9 per 1000 in 2003 to 59.1 per 1000 in 2010 (point F to G).

A straightforward comparison of the estimates of inflow (diagnosis) and outflow (death) explains why the upward trend in diabetes prevalence, which began around 1990, will not soon abate. If the diagnosed onset rate in 2000 of approximately 1.1 million cases per year and the death rate of about 500,000 per year were to stay the same, we projected that the diagnosed prevalence will continue to increase. Although the model suggests that this gap between inflow and outflow is gradually closing, the inflow of diagnosed diabetes onset would have had to drop substantially (e.g., by about 50% in 2006) just for diagnosed prevalence to stop increasing, let alone to begin decreasing.

Our SD model reveals certain insights about the long-term effects of interventions to reduce onset, boost detection, and better manage diabetes. For example, aside from reducing prevalence, another HP 2010 objective calls for the percentage of people with diagnosed diabetes to increase from 68% to 80% (Objective 5-4) (3). Such an increase in diagnoses would, in terms of Figure 2, increase the inflow of people diagnosed with the disease. Thus, the prevalence of diagnosed diabetes would also increase. Figure 1 quantifies the effect of this scenario as the difference between points H and G (i.e., 63.2 vs 59.1 in the year 2010).

Another HP 2010 objective calls for an 11% reduction in the diabetes-attributable death rate (Objective 5-6) (3), a result presumably to be achieved through improved disease management. As Figure 2 indicates, a reduction in deaths would also increase diagnosed prevalence because more people with the disease would remain alive. (Figure 1 does not display a curve for this scenario because it overlaps the status quo line [i.e., 59.3 vs 59.1 in the year 2010]).

The inconsistency in these HP 2010 objectives for diabetes is clear: meeting the objectives for increasing the diagnosis rate or decreasing the mortality rate would tend to increase the prevalence of diagnosed diabetes.

One type of public health intervention that could possibly reduce the diagnosed prevalence of diabetes would be one aimed at reducing the initial onset rate. HP 2010 calls for a 29% reduction in the number of new diabetes diagnoses per 1000 (Objective 5-2) (3), presumably to be achieved through enhanced efforts to detect and manage prediabetes. This effort would, perhaps, be combined with efforts to reduce the prevalence of obesity in the general population. A reduction in the inflow resulting from diagnosed disease onset is clearly a move in the right direction because it leads to a lower diagnosed prevalence than would be the case under the status quo (i.e., without an intervention to reduce disease onset). But a reduction in diagnosed prevalence relative to the status quo is not the same as an absolute reduction over time — an actual reversal of growth. We described previously how a straightforward comparison of the inflow and outflow rates in Figure 2 indicates that a reduction in the onset of diagnosed diabetes on the order of 50% would be required to halt the growth in diagnosed prevalence. However, the question still remains: to what extent could a significant reduction in onset at least slow the growth of diagnosed prevalence?

To address this question, we simulated an intervention begun in 2003 that would reduce the rate of diabetes onset 29% below its 1997 level by 2010. In this scenario, as shown in Figure 1, the prevalence of diagnosed diabetes increases by 7% (from F to I) per 1000 population from 2003 to 2010 as compared with increasing by 21% (from F to G) in the status quo scenario (i.e., 52.3 vs 59.1 in the year 2010). This slower growth in the number of people with diagnosed diabetes certainly would improve the overall disease picture, but it would not yield a decline in diabetes prevalence.

The simulation modeling thus helps quantify what the stock-and-flow logic of Figure 2 and previously described numerical analysis suggested: that the HP 2010 target of a 29% reduction in the rate of diabetes onset is too modest a reduction to achieve the desired reduction in prevalence and can only slow its growth.

The SD model can also be used to explore more extreme possibilities. For example, what would happen if initial onset of undiagnosed diabetes had dropped suddenly to zero during 2004? Even under this impossible-to-achieve scenario, diagnosed prevalence would decrease by only 14% from 2003 to 2010 (data not shown). This decrease is relatively small partly because cases of diabetes will continue to be diagnosed during this period even though initial onset has ceased, and partly because of the relatively low death rate among people with diabetes (about 4% per year). Thus, even this most optimistic scenario of a 14% reduction in diagnosed prevalence of diabetes during 2003 through 2010 falls far short of the 38% objective of HP 2010.

Findings from our study indicate that the HP 2010 objective for reducing diagnosed diabetes prevalence by 38% will not be achieved — not because of ineffective or underfunded health protection efforts but because the objective itself is unattainable. Moreover, if current investments in diabetes screening and disease management continue to succeed in diagnosing a greater number of people and in enabling people to live longer with the disease, then diagnosed prevalence will move still farther away from the HP 2010 target.

In setting long-range numerical targets for health objectives, particularly those that may be viewed as intervention outcomes, it is important to recognize that the diagnosed prevalence metric is subject to misinterpretation and to unrealistic expectations for two basic reasons:

If prevalence objectives are to be attainable within their specified time frame, it is important first to recognize what the disease trajectory would be under status quo assumptions and then to factor in the effects of any planned interventions, recognizing that measurable success in one area (e.g., an increase in the percentage of cases diagnosed) may reduce apparent progress in others (e.g., decreased prevalence). In the case of diabetes, we found in our simulation experiments that under current conditions, that is, without any new interventions, diagnosed prevalence would increase 21% from 2003 to 2010. Proposed detection and management initiatives, if successful, would increase that number even further. If disease prevalence is to serve as a benchmark for assessing the performance of national public health interventions, then prevalence-reduction goals must account for the compounding effects of successful disease detection and management interventions.

Simulation modeling helps improve our collective understanding of health and disease dynamics, and in turn supports the development of long-range objectives that are both achievable and mutually consistent. Such models enable planners and policy makers to explore for themselves the plausible short- and long-term consequences of historic trends and to compare the effects of alternative interventions before committing limited resources. For that reason, diabetes program planners in Vermont have worked with members of the CDC Diabetes System Modeling team to use the model described here as a support for their efforts to set plausible and internally consistent objectives for diabetes-related outcomes at the state level (28,29). Planners in Minnesota, California, Alabama, Tennessee, and Florida are currently exploring similar uses.

Without the reality checks available through formal simulation experiments, long-range target-setting may fall prey to the weaknesses of flawed and sometimes biased intuition (mental models) (17,23). Popular conceptions about how certain phenomena change over time may often fail to account for real-world sources of inertia and delay and may suggest that things can change more rapidly than is actually possible. The prevalence of a chronic disease like diabetes changes only gradually because, as noted above, people with such conditions die at a relatively slow rate, and there is currently no cure for these conditions. In this respect, chronic diseases are unlike many acute infectious diseases such as influenza or measles, whose victims do not linger in the disease condition for years but instead recover or die relatively quickly. For such acute diseases, the large outflow creates a close correlation between decreases in the rate of onset and in diagnosed prevalence. For chronic illnesses, however, decreases in onset rates do not correlate with immediate decreases in prevalence; instead, they correlate with prevalence increasing more slowly.

Those working to prevent and manage chronic diseases may use stock-and-flow diagrams to develop a clearer understanding of the characteristic dynamics of these diseases. In addition, simulation experiments may bring new insights to the task of setting realistic and achievable goals for the nation's health. That approach could help ensure that numerical objectives are mutually consistent and achievable within their stated time frames. The objectives may still be difficult to achieve in practice and in that sense may be aspirational. But even aspirational objectives can and should be crafted in a way that is consistent, logical, and feasible given the causal structure of the system and the historical processes under way, particularly those responsible for the pattern of increasing incidence and diagnosis, as well as declining mortality.

Although simulation models can help improve our understanding of chronic disease dynamics, they have several inherent limitations. All models are incomplete simplifications of reality, and their conclusions are affected both by structural boundaries and by the uncertainties of the data with which they are calibrated (29). Techniques such as boundary critique (30) and sensitivity testing (17,31) can be used to assess the extent to which models may be affected by these simplifications and uncertainties. In the case of the diabetes SD model, sensitivity testing suggests that the magnitudes of its simulated futures, such as those seen in Figure 1, are subject to some imprecision because of uncertainties about input parameters, but that the directions of change and thus our general findings are unaffected by these uncertainties.

Even with their inevitable imprecision and incompleteness, simulation models can enhance learning and decision making, and that is their primary purpose (17,32). These tools can improve our collective understanding about how interventions will affect health indicators over many years within the complex systems of cause and effect that shape the public's health.