This is a very interesting question. BMI (or any measure of W/H) is known to be strongly influenced by body shape. The typical Somali pastoralist body shape (i.e. long limbs and short trunk) leads to depressed BMI values. This is one reason why the prevalence of malnutrition is so high in Somalia even in children (i.e. W/H overestimates). The dwarf body shape is short limbs in relation to trunk. In this body shape, BMI will be high. There is a way of correcting for body shape using the sitting height to standing height ratio which is also known as the Cormic Index. I am not sure how well this will work with an extreme body shape such as that associated with dwarfism. Nick Norgan would be the best person to advise on this. A summary of the issue can be found at: [url]http://www.unscn.org/Publications/RNIS/adultrnissupplement.pdf[/url] Which has examples of how to correct BMI for body shape. If you are using BMI to assess undernutrition then you may want to consider using MUAC. If overweight / obesity is the issue then waist to hip ratio might be useful. Sorry not to be of more help.

If the objective of assessing BMI is for admission to nutrition programmes, I would suggest going back to basics: clinical status and report of acute weight loss in the last 4 weeks are still the best indicators of presence of active acute malnutrition. Weakness (objective if possible, or at least reported weakness), in the absence of alternative explanation, may be a very good indicator too. On the other hand, if the objective of assessing BMI is for survey or surveillance, I would simply exclude the individual from the sample (in the same way that severe spine deformation and other similar are excluded from BMI studies)

Or ... use the CHANCES model which is outlines in the RNIS reference above (this is a simple clinical score).