This is a fresh way to think about things.

The "Slim People Change Proposal"

By Steven B. Halls, MD FRCPC

Using slim people’s BMI changes with aging to define "normal" change for all adults.

The Body Mass Index BMI calculator offers an alternative criterion to define "overweight", called "halls.md v2" criteria. Part of the criteria is a gradually increasing "curve shape" in adults. This page describes how the following curve-shape was derived.

10th percentiles composite curve

There’s something on this chart to pay attention to?

Hi. Yes, notice the pink line goes slowly upward as people get older.

And that’s a chart of just the skinny slim people.

This graph shows how very slim people, who remain slim all their lives, tend to increase their BMI with age. It is based on combining data from several large studies. Those studies are:

American NHANES III study (1988-1994), using the 10th percentile BMI values for men and women.
American NHANES I study (1971-1974), using the 15th percentile BMI values for men and women.
French survey data, using the 10th percentile BMI values for men and women.
German survey data, using the 10th percentile BMI values for men and women.
American survey of self-reported slimness, stable over time, and corresponding BMI values.

You did this data meta-analysis yourself?

Yes indeed. That’s a very good combined dataset.

For details, click on each link above and look for the 10th or 15th percentile graphs. From each survey’s data, the "slim" percentile data points were shifted up or down until they clustered together, and then a trendline was curve-fitted, using Second (2nd) order polynomial, One-phase exponential, and one-site binding curve formulas. The best choice is the "One-site Binding formula": BMI = (23 * Age) / (4 + Age).

curve-fitting choices

Yes, I can see, the One-Site-Binding is definitely the best curve fit.

Having determined a suitable curve-shape and a formula to describe it, the position of the curve is shifted upward, by adding a constant, until the line crosses several key points, such as a BMI=25 at Age=18.5. This key point is designed to approximately match the criteria suggested by the WHO^1,2. It is not overly important to match this key point exactly however, since the CDC growth charts cross BMI=25 at about age=17, and a suggested UK chart³ crosses BMI=25 at age=19.5.

You anchored the pink line at BMI=25 at age 18.5, that’s very standard.

Additional validation of other key points in BMI for middle aged and older adults was also performed.

References

Poskitt, EM. Body mass index and child obesity: are we nearing a definition? Acta Paediatr, May 2000;89(5):507- 509
Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000 May 6;320(7244):1240-3
Chinn S, Rona RJ. International definitions of overweight and obesity for children: a lasting solution? Ann Hum Biol 2002 May-Jun;29(3):306-13

Back to the halls.md v2 criterion description.