American Association for Clinical Chemistry
Better health through laboratory medicine
Clinical Laboratory Strategies: June 24, 2010

Strategies logo

Gauging the Cost-Effectiveness of Diabetes Screening
New Mathematical Modeling Supports Earlier, More Frequent Testing
By Bill Malone

As the incidence of diabetes has grown nearly exponentially, physicians, laboratorians, and public health experts have concentrated their efforts on the best way to screen for and diagnose the disease in the hope of turning back the alarming trends. Agreement about what age screening should begin and how often it should occur has not come as easily, at least in part due to questions of cost. This issue of Strategies examines new research that used advanced mathematical modeling to explore different strategies of diabetes screening and their cost-effectiveness.

Currently, recommendations for diabetes screening rely on consensus statements from professional associations like the American Diabetes Association (ADA) or government bodies such as the U.S. Preventive Services Task Force (USPSTF). The most aggressive recommendations come from ADA, which in 2010 has advised that testing be considered in all adults who are overweight (BMI ≥25 kg/m2) and have additional risk factors such as physical inactivity or a first-degree relative with diabetes. In the absence of these criteria, ADA recommends testing begin at age 45 and, if results are normal, testing is repeated at 3-year intervals. USPSTF recommends screening adults for diabetes only if the person has hypertension.

A recently published mathematical modeling study lends further support to the ADA recommendations with findings that suggest several screening strategies of asymptomatic adults can be cost-effective (Lancet 2010:375;1365–1374). This paper is the first time that The Lancet has published research based on mathematical modeling. Called Archimedes, the model simulates human physiology, diseases, interventions, and healthcare systems with every variable estimated from empirical sources.

The researchers took person-specific data from the National Health and Nutrition Examination Survey 1999–2004 to create a simulated population of 325,000 30-year old people without diabetes, then ran eight simulated screening strategies for type 2 diabetes, including a control group, through Archimedes. The strategies spanned a starting age of 30–60 years, with intervals of repeated screening at 6 months and 1, 3, and 5 years. Two of the strategies simulated screening only those with hypertension, repeated annually or every 5 years. With the help of distributive computing, Archimedes produced detailed results for each screening scenario, finding that screening is cost2 effective when initiated between the ages of 30 and 45 years and repeated every 3–5 years.

 “I think the most significant finding is that there were a number of strategies that were cost-effective, as opposed to just one,” said the study’s lead author, Richard Kahn, PhD, professor of medicine at the University of North Carolina School of Medicine. “Another key finding is that the strategy recommended by the USPSTF, while also cost-effective, doesn’t win the day, because the absolute number of adverse outcomes prevented in the long run is less than the other strategies. Moreover, those strategies that we found to be cost-effective are truly cost-effective and come in at less than $10,000 per quality-adjusted life-year.”

 In the study, the difference between beginning screening at 45 years with a repeat every 3 years almost doubled the number of quality-adjusted life-years (QALY) gained in comparison to yearly hypertension-triggered screening—and at an equivalent cost per QALY. Years gained jumped to 5.33 years from 2.84 years at a cost of $6,106 per QALY compared to $6,287, respectively.

The strength of this modeling study lies squarely with the sophistication of the Archimedes model and the fact that it has been extensively validated, Kahn said. Most other cost-effectiveness studies are based on a Markov model that assumes people jump from one stage to another at yearly intervals. For example: a person is healthy, has had a heart attack, or is dead. Moreover, such models do not take into account all the variables that influence adverse outcomes and have not been well-validated. “Of course, the Markov model structure is not the way biology works,” Kahn said. “And because of that, a huge number of assumptions have to be made. The fact that these models have not been validated means that we don’t have any evidence to suggest that the predictions made are likely to be true.”

In contrast, the Archimedes model has been extensively validated, in that the model faithfully reproduced the results of more than 70 real-world clinical trials when the data were run through its equations. “The Archimedes model runs just the way that biology runs: it has underlying physiology imbedded in it so it’s a full virtual biology and virtual healthcare system just like the real thing,” Kahn said.

Advanced modeling techniques will become more important as evidence-based medicine seeks higher standards for screening recommendations, noted William Winter, MD, professor of pathology in the department of pathology, immunology, and laboratory medicine at the University of Florida College of Medicine. “Certainly, we’re going to see more computer simulations, because the reality of doing a prevention study—where we’re talking about preventing complications that are just not years, but decades in the future—is just not practical,” he said. “Since we are driven more and more by evidencebased research, one would ask, why is it better to screen at age 30 instead of age 45? Why would it be better to screen more often than less often? And this study can provide information now on cost-effectiveness and data on how many people need to be treated to avoid a single adverse outcome over a period of time.” Winter was not associated with the study.

While the study did show differences in QALYs for detecting type 2 diabetes with the various screening strategies, demonstrating a strong link to outcomes is still difficult. In fact, the study found that a delay of nearly 2 years in diagnosing diabetes did not make very much of a difference for early stage disease. “A widely held belief is that the earlier type 2 diabetes is detected, the greater the likelihood that complications will be prevented,” noted the authors. “This theory can be assessed in our analysis by comparing the three strategies in which screening was started at 45 years, with screening repeated every 1 year, 3 years, and 5 years, respectively. Compared with no screening, these three strategies detected type 2 diabetes a mean 6.0 years, 5.3 years, and 4.7 years earlier, respectively. All three strategies showed a decrease in rates of myocardial infarction and microvascular complications compared with no screening, but there were no significant differences in health outcomes between the three strategies, even in a population of 325,000 people followed up for 50 years.”

The virtue of combing screening for diabetes with screening for other diseases was highlighted in an accompanying editorial by Guy Rutten, MD, professor of diabetology in primary care at the Julius Center for Health Sciences and Primary Care, in Utrecht, Netherlands. “Today’s paper provides further evidence that screening for diabetes should be combined with screening for hypertension and lipid tests,” he wrote. “This recommendation is also in line with the current guideline for screening from the American Diabetes Association.”

Rutten also noted that the Archimedes model might be improved with more nuanced data about populations and the disease itself. “Further input into the model of information on screen-detected people with type 2 diabetes, and separate analyses of different populations or health-care systems, might strengthen the role of the Archimedes model to provide further useful information for future guidelines about screening for diabetes,” he wrote.

Join Us On LinkedIn Join us for a discussion of this article, and interact with your peers, on LinkedIn.


Rate this page:       

Clinical Laboratory Strategies Podcast
Gauging the Cost-Effectiveness of Diabetes Screening©