Just over a year ago, I wrote a column at The Upshot arguing about how frustrating I found the USPSTF's determination that universally screening children for autism could only receive an I rating. I felt that the I rating (while asking for more research) ignored the vast amount we already knew about autism. Specifically, I cited the four questions I usually ask when considering screening tests:

Whenever anyone is discussing a screening test, I usually have four questions, which I’ve adapted from David Sackett’s classic handbook on evidence-based medicine: (1) Is the condition prevalent and severe enough to warrant screening? (2) Do we have a cost-effective means to screen the general population? (3) Does early diagnosis make a difference (that is, do we have treatments available that are more successful when patients are diagnosed earlier?) And (4) Will an early diagnosis motivate people to use information gained from screening?

When it comes to autism, I argued that all four criteria were satisfied. In the last year, however, I have been convinced that - perhaps - the USPSTF might be right. We could use research to know that the actual screening - universally - of kids would make a difference.

But an area that I thought wasn't even close to debatable was diabetes. That is, until now. From the BMJ, "Efficacy and effectiveness of screen and treat policies in prevention of type 2 diabetes: systematic review and meta-analysis of screening tests and interventions":

Objectives To assess diagnostic accuracy of screening tests for pre-diabetes and efficacy of interventions (lifestyle or metformin) in preventing onset of type 2 diabetes in people with pre-diabetes.

Design Systematic review and meta-analysis.

Data sources and method Medline, PreMedline, and Embase. Study protocols and seminal papers were citation-tracked in Google Scholar to identify definitive trials and additional publications. Data on study design, methods, and findings were extracted onto Excel spreadsheets; a 20% sample was checked by a second researcher. Data extracted for screening tests included diagnostic accuracy and population prevalence. Two meta-analyses were performed, one summarising accuracy of screening tests (with the oral glucose tolerance test as the standard) for identification of pre-diabetes, and the other assessing relative risk of progression to type 2 diabetes after either lifestyle intervention or treatment with metformin.

Eligibility criteria Empirical studies evaluating accuracy of tests for identification of pre-diabetes. Interventions (randomised trials and interventional studies) with a control group in people identified through screening. No language restrictions.

This was a systematic review of studies that looked at the accuracy of tests which are used to identify prediabetes, and then the efficacy of the interventions we use to prevent diabetes in those with prediabetes. The latter accepted randomized controlled trials or interventional studies (that had a control group)  of people identified through screening.

This seemed like a no-brainer to me. Given the plethora of studies on diabetes tests, as well as programs for preventing diabetes, I thought this would clearly show positive results. 

I was wrong. The final analysis included 49 studies of screening tests. They found at A1c had an overall sensitivity of 49% and an overall specificity of 79% for identifying prediabetes. Different studies used different cut-off values, though, which made analysis tricky. Of course, this makes publishing guidelines on how to use the test to screen even trickier. Fasting plasma glucose had a sensitivity of 25% and a specificity of 94%. You have to remember, though, that when you're screening, you want sensitivity first.

The tests didn't really correlate, either. Almost half of the people with an abnormal A1C didn't have any other glycemic abnormalities.

The interventions, on the other hand, at least showed some good news. They were associated with a reduction in the relative risk of developing type 2 diabetes of more than a third over six months to six years, depending on the study. Even in the follow-up period after the studies were done, the relative risk reduction was 20%.

Here's the thing, though. You can't even get to the four questions about screening programs until you have a screening test you trust. We don't appear to have that. Fasting plasma glucose was specific, but not sensitive - making it a poor choice for a screening test. Plus, it's much harder to get people to do it. A1c, which is much easier to use, is neither sensitive nor specific. That means that it's both missing people who have prediabetes and putting people who don't into interventions they may not need.

Since more intervention programs try and get people to adopt healthier lifestyles, there's little harm in getting them to eat better and be more active. Some trials put people on metformin, though, and that does come with side effects. They all involve costs, too.

Diabetes is such a big public health issue, though, and the prevention measures are so holistic, that we might do better to focus on public health efforts that hit everyone. Collectively trying to get people to maintain a healthy weight, not smoke, and exercise may be cost-effective. Unfortunately, though, it appears that our current plans to screen and then treat may wind up doing much less good.