https://diabetes.acponline.org/archives/2013/02/08/7.htm

A single screening for type 2 diabetes in high-risk adults did not reduce mortality over 10 years

A trial in the United Kingdom randomized 33 primary care practices and 20,000 of their high-risk patients to screening for diabetes or usual care.


A trial in the United Kingdom randomized 33 primary care practices and 20,000 of their high-risk patients to screening for diabetes or usual care. Three percent of the screened patients were diagnosed with diabetes, and screening for type 2 diabetes did not reduce all-cause, cardiovascular (CV), cancer, or diabetes-related mortality compared with no screening.

The study was published by The Lancet on Nov. 17. The following commentary by Jeffrey Mahon, MD, MSc, was published in the ACP Journal Club section of the Jan. 15 Annals of Internal Medicine.

The unique feature of this well-done trial by Simmons and colleagues, and one that self-declared experts doubted was possible, was randomization of participants to diabetes screening or not. This is the best way to control biases that overinflate estimates of benefit in observational studies of screening. Other strengths include testing the question in a way that approximates how screening works in the messier circumstances of real-life clinical practice and near-complete ascertainment of the primary outcome.

If screening for type 2 diabetes reduces premature all-cause mortality, most of this probably occurs through earlier use of therapies proven or strongly suspected to prevent premature CV mortality in persons with unrecognized diabetes (e.g., statins; angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers, especially in patients with nephropathy including microalbuminuria; smoking cessation; and metformin). The study by Simmons and colleagues had the power to reliably exclude a modest effect (about 25% relative risk reduction) of screening on CV mortality over 10 years. However, detection of smaller, but still important, reductions in CV deaths requires a larger sample, longer follow-up, or both. The authors also acknowledge that unmeasured factors could have eroded study power to detect mortality differences, including a lower risk for death in control participants, perhaps through improved vascular risk management (e.g., statin use), and opportunistic diabetes screening.

The lack of effects on mortality and CV mortality does not strengthen current recommendations by most major practice guideline groups to screen for diabetes. At the same time, the authors note that study limitations make it difficult to be sure that there were not (or will not be) clinically important benefits that justify the costs of early detection of diabetes under some circumstances. Diabetes screening will not disappear, although the standard for the type of evidence needed to endorse widespread screening may have just been reset.