In my field, a screening tool is used by staff to determine if a patient is at risk for a certain problem. If a screen determines that a patient may be at risk, then a consult is ordered for a diagnostic test administered by a qualified professional. This diagnostic test is much more specific and is typically norm-referenced.
As far as BMI is concerned, it is a screening tool. Just because a person has a BMI which states that they are overweight or obese does not mean that they necessarily have bad knees, or heart disease, or arterial plaque. It just means that they are more at risk--specific diagnostic tests would have to be conducted to determine if they truly have any of these diagnoses.
I'm tired. Does that make sense?

Pretty good! Screening tests cannot be used for diagnosis. Classic example. How many cases of breast cancer have been diagnosed with a mammogram? Zero. It's not diagnostic, and it's for statistical reasons.

If you think in terms of normal distributions, both diagnostic and screening tests try to discriminate between two populations, each with a normal distribution for a particular character and some degree of overlap in their distributions. Screening tests have a higher risk of either false positives or false negatives due to a greater degree of overlap. The point of discrimination for a screening test is set to effectively eliminate false negatives, but this leaves them open to false positives.

Consider the quad screen which is used to determine a woman's risk of trisomies like Down syndrome. Since a screening test is used to quickly assess a large population for a particular risk, we don't want false negatives (failing to catch a case of Down), but there is an elevated risk of a false positive (a pregnancy without Down that registers as high probability of Down). These types of screens are done more aggressively after age 35 due to elevated risk...but the average age of a woman with a Down syndrome pregnancy is actually around 24.

Diagnostic tests have a higher degree of specificity. These tests typically examine (often in invasive fashion) a variable in which the two populations (say women without and women with breast cancer) differ with little overlap in their normal distributions. (If you aren't familiar with the term "normal distribution," you may know the term bell curve...this is a normal distribution!) Thus, there is less risk of false positives or false negatives. Hence, the test is highly specific and can be used to clearly diagnose a specific condition.

To the BMI, now, which is a simple way of screening the population for weight related health risks. There is a lot of data which suggests that the majority of people with BMIs greater than 25 are at increased risk of weight-related health problems, including heart disease, T2DM, and stroke. The high risk of false positives comes in the form of body shape variances, athleticism, etc. So in the majority of cases, a score above 25 indicates risk for health problems just as the majority of suspicious mammograms indicate real cases of breast cancer and the majority of positive quad screens detect real cases of trisomic pregnancies. But 51% is a majority...basically, there are plenty of women with suspicious quad screens or mammograms who have no issues. The problem with BMI is that people (especially men) are too often willing to dismiss a high BMI because of body builders, athletes, etc. Frankly, the majority of people with BMIs above 25 are like me. They are fat and carrying health risks. Okay, I WAS fat but am thinner now. But I still have a BMI over 25 and a % body fat (more diagnostic but still screening) > 20%, so I'm definitely not the picture of perfect health.

Ultimately, too many doctors (and as noted by rebekah529 other health practitioners) fail to understand the limits on screening tests and leave patients with the impression from a quad screen, mammogram, BMI, or other screening test that they have a condition that that test cannot diagnose. You need an aminocentesis, biopsy, or blood work to diagnose conditions linked to the screens.

Bottom line is this. Screening tests are pretty good at giving us a useful snapshot. They catch problems early if considered properly. Basically, mammograms provide a chance to catch cancer early, quad screens let us know when a closer look at fetal chromosomes is justified, and keeping yourself at a healthy BMI is a smart way to provide for long-term health. If you go outside the "normal' range you then take a closer diagnostic look to determine if it's a point of concern.

Basically, if you're an adult and have never been a serious athlete...and your BMI is 28 or higher...you're probably not an exception who can ignore it...until your blood work and body fat analysis tell you otherwise.

Cheers!

In my opinion, that's a skewed statistic because of the difference in fertility rates:
The risk for Downs is 1:1500 between 20-24, 1 in 200 between 35-39 and above 45 is 1 in 20
Its just that women between 20 and 25 have most of the babies and hence you see Downs, even though there is a magnitude lower risk for it. That's compounded by women over 35 being screened and terminating pregnancies as well

This is not a skewed statistic. What you point out is the REASON this stat exists. The point is that the screening test is set up based on the risk being higher in one group (women over 35)...but that doesn't mean women outside that high risk group will be affected. The simple reason the screens aren't done more aggressively at earlier ages is that there would be a large number of false positives if you did. Bottom line is that screens need to be used judiciously because of their statistical limits.