It stands to reason that the relationship should incorporate a component reflecting the BMI, which in itself is a less than precise measure (but better than ignoring the obvious error).

Dear Dr Tausif, thank for your observation. But to use BMI instead of BSA drives to the same  kind of errors, because in these models of indexation both there is a correction of the height, not subjected  per se to fluctuating variations, by the weight, a variable very often fluctuating in increasing and sometimes decreasing direction in short time. And BMI too does not have any rational relationship with the physiologic functions eventually indexed for it .  

Dear Giancarlo, thanks for bringing up this issue! As we well know, nothing is perfect, but in my mind BSA ir better that body weight alone and also better than BMI. All of them has shortages and limitations and are not applicable in some cases (e.g. amputees). Nevertheless, we need one universal standard and I have no idea, what may replace the BSA at the moment. Actually, while analyzing pediatric renal functions, I've found that BSA-based results are much more precise in comparison to standard pediatric weight-based ones. More important, pediatric BSA-based results are compatible with adult ones, e.g. normal diuresis values, etc.

In short - BSA is not perfect, but I do not know anything better for now. Any ideas, what can substitute it in a futrue? And, surely, 1,73 vs 1,97 issue should be addressed by the renal community.

Dear Colleagues

An interesting debate. I would like to make four comments.

First, it seems physiologically plausible and indeed likely that the GFR might parallel the basal metabolic rate (BMR). Since there happens to be a reasonably good (though obviously not perfect) relationship between BMR and body surface area (BSA), it is reasonable that our predecessors looked at ‘normalising’ or linking GFR measurements to BSA as well, and that it fits pretty well. At the end of the day, the only purpose of doing this is to minimise the size-dependent differences in GFR between individuals, and this method works well in practice. Don’t suggest ditching it before you come up with a better clinical tool!

Second, as a paediatrician, it is of interest to me that when you look at very small babies you find that the BMR varies very closely with body weight; much more closely than it does with BSA at that size (read Edmund Hey’s work on this). And it becomes critically important the smaller you get; the BSA to weight relationship is essentially one of height (BSA varies with height-squared, while weight varies with height-cubed), so those weeny 28-weekers have a BSA/kg about 4 or 5 times greater than you or I. Not surprisingly, it is also true that GFR varies much more closely with weight than BSA at that size too. It makes it all a bit messy (but that is life!), but the best measures to use are weight in small infants, and BSA in older children (certainly above the age of 2). I refer you to a publication I produced with Ed Hey about this a long time ago [Coulthard MG, Hey EN. Weight as the best standard for glomerular filtration in the newborn. Archives of Disease in Childhood 1984;59:373-75].

Third, it really doesn’t matter whether you ‘normalise’ or ‘standardise’ your GFR measurements to 1.73 m2 or to 1.97, or even to a more comfortable-feeling 1 m2. All it means is defining what the norm would be for a subject of that size. It would have been much more sensible to use 1 m2, then we wouldn’t be having this debate, because it would be obvious to all that this was a normalised value.

Fourth, I commend you to use the Boyd self-adjusting-power equation which estimates BSA from weight alone, which she recommended after having tested a very sophisticated equation which had self-adjusting powers for both height and weight with no advantage [Boyd E. The growth of the surface area of the human body. Minneapolis: University of Minnesota Press, 1935]. It is prone to far less error, both in not needing to precisely measure height, and in calculation terms [Coulthard MG. Surface area is best estimated from weight alone: pocket calculators and nomograms are unnecessary. Archives of Disease in Childhood 1994;71:281]. It is now adopted by the UK paediatric oncology group to calculate drug dosages per unit surface area for these very reasons.

Best regards

Dr Malcolm Coulthard

Body surface area does not form urine. Even if it did, the BSA estimating formulas would not be BSA, but functions of weight to a power and height to a power. When some of us tested the statistical validity of using height in such a formula for correlation to a one exponential term model of 99m-Tc-DTPA clearance, we found that including height to a power and weight to a power decreased the correlation to GFR that would be achieved using a power of weight alone. That is, BSA is spuriously correlated to GFR (Look up "spurious correlation' as a search term, if you have not seen this before) . Much more accurate body scaling can be obtained using fractal geometry for which GFR is proportional to the 99m-Tc-DTPA volume of distribution to the 2/3 power and weight to the 1/4 power.

[See Wesolowski CA, Babyn PS, and Puetter RC. An improved method for determining renal sufficiency using volume of distribution and weight from bolus 99mTc-DTPA, two blood sample, paediatric data. Nuclear Medicine Communications 27(12): 961-968, 2006 and

Wesolowski CA, Babyn PS, and Puetter RC. A Power Law for Determining Renal Sufficiency Using Volume of Distribution and Weight from Bolus 99mTc-DTPA, Two Blood Sample, Pediatric Data. Nuclear Science Symposium Conference Record, 2006 IEEE, pp. 1986-1994, 2006.]

For those interested in body scaling, there are papers and books on the subject that have gone unnoticed by many. I might mention the work of West GB, Brown JH, Enquist BJ. The fourth dimension of life: fractal geometry and allometric scaling of organisms. Science 1999;284:1677-9, among many other papers. And the extensive review Agutter PS, Wheatley DN. Metabolic scaling: consensus or controversy? Theor Biol Med Model. 2004;1:13.

Now if one uses Kleiber's law namely GFR is proportional to weight to 3/4 power, better results are to be had (see publications above). However, GFR is not endogenous creatinine clearance. Because muscle mass increases as roughly the 1.08 power of weight, the creatinine clearance should use a lower power of weight for body scaling than GFR, and endogenous creatinine clearance is only roughly related to GFR, and, to make matters more confusing, the experimentally obtained power of weight for scaling creatinine is 0.69, and that for scaling inulin 0.77.

[See Adolph EF. Quantitative relations in the physiological constitutions of animals. Science. 1949;109:579-85.]

Finally, what do you want to use body scaling for? Errors in GFR measurement are proportional, Thus, the logarithm of GFR for a patient should be compared to a later logarithm of GFR of the same patient as logarithm of GFR errors are homoscedastic. However, taking logarithms does not normalize GFR (That is, convert GFR values into a normal distribution). For such, a different GFR transform is needed (information withheld, unpublished).

Normalization for drug dosimetry is an entirely different problem. For example, blood volume is approximately the first power of weight. That is, to calculate a dose of a vasoactive drug administered by intravenous infusion (example dipyridamole) one uses weight because independent of the body size of mammals, blood volume is roughly 5% of weight. This contrasts to dosing furosemide that I would dose by metabolism, that is, using for example

Pediatric dose = Adult dose * (W/70)^0.75

Next problem, how does one measure GFR? The best method is not affected by alterations in body fluid and would require few samples drawn over the shortest time possible. That is, as far as some of us can tell, the Tk-GV method

[Wesolowski CA, Puetter RC, Ling L, Babyn PS. Tikhonov adaptively regularized gamma variate fitting to assess plasma clearance of inert renal markers. J Pharmacokinet Pharmacodyn. 2010;37:435-74.

Wesolowski CA, Ling L, Xirouchakis E, Burniston MT, Puetter RC, Babyn PS, Giamalis IG, Burroughs AK. Validation of Tikhonov adaptively regularized gamma variate fitting with 24-h plasma clearance in cirrhotic patients with ascites. Eur J Nucl Med Mol Imaging. 2011;38:2247-56. doi:10.1007/s00259-011-1887-9.

Wickham, Fred; Burniston, Maria Teresa; Xirouchakis, Elias; Theocharidou, Eleni; Wesolowski, Carl Adam; Hilson, Andrew J W; Burroughs, Andrew Kenneth. Development of a modified sampling and calculation method for isotope plasma clearance assessment of glomerular filtration rate in patients with cirrhosis and ascites. Nucl Med Commun 34:1124–1132, doi:10.1097/MNM.0b013e32836529ab, 2013

And shortly, Surajith N. Wanasundara, Michal J. Wesolowski1, Mark C. Barnfield, Michael L. Waller, Anthony W. Murray, Maria T. Burniston, Paul S. Babyn, Carl A. Wesolowski1, Accurate and precise plasma clearance measurement using four Tc-99m-DTPA plasma samples over four hours, In press in Nuclear Medicine Communications, Sept. 9, 2015]

In summary, BSA is appropriate for estimating percentage damage of body surface area for burn victims as a guide to fluid replacement, but, there are few other situations for which I would not use a better estimator.

I carefully read the interventions by Dr Coulthard and Dr. Wesolowsky , and I have to do some observations induced by their notes. The first one is to remind  the history of the scientific events finally resulting in the current method to normalize the measures of the renal functions and of many other physiologic functions.  it has to be remembered that in 1923  Taylor and Al.  noted that the maximum urea clearance showed a  best correlation with BSA than with the only weight : this result is reported by McIntosh et Al. (McIntosh JF, Möller E,Van Slyke DD. Studies on Urea Excretion. III. The influence of body size on urea output . J Clin Invest. 1928; 6(3): 467- 483) : from this Taylor proposed that the indexation of the maximum urea clearance would based on the ratio “the mean body surface [of that time]/ BSA of the studied subject”.  Mcintosh in 1928  proposed as standard BSA 1,73 m2 , the average BSA resulted in a study of 1927. It must now to note that a method of indexation proposed to normalize maximum urea clearance  was successively  used to normalize the glomerular filtration rate, a function quite different  from urea clearance.. It has also to strongly emphasize that the same Authors warmed to the risk of erroneous conclusions if the ratio 1,73/BSA would result quite different from 1, and they suggested to use in the overweight or underweight  subjects the ideal weight according to their height, a recommendation totally neglected in all the following years, at least as a correction to be universally adopted on a formal general consensus. On these bases It is very questionable  to agree on the concept to use as the best chance  the formulae for BSA estimation based only on the weight., if in the indexation on BSA the weight was evaluated  to be a variable possible cause of erroneous conclusions already from the researchers introducing this methodology .  The second observation concerns the relationships between weight and GFR and between BSA and GFR : it has now to state on what kind of weight and what kind of BSA we are discussing. There are no doubts that if the concerned weight would be within the ideal weights range for each height according to the different body frames, we certainly will  have an high correlation between GFR and weight and between GFR and a BSA estimate built on that weight, because the organs mass and their descending functions developed until the adulthood according  to height and correlated to a normal weight : in a normal body structure the relationship between muscle mass and fat mass and water will be in reciprocal correct relationship within the limits expected according to that specific body frame  But in case of excessive weight due to an abnormal increase of body fat or water, or both, or due to a decrease of muscle mass,  the mentioned relationships will progressively weaken, how much the body weight will be characterized by an increasing abnormal composition, a condition very frequent in many diseases, as in renal and cardiac and liver insufficiency   The third observation concerns the effects of the frequent and easy change of the weight  in time, frequently  due in normal condition of health as well as in disease condition to an increasing of fat deposition. It seems useful to fit an explanatory example  It is possible to assume that  a subject underwent two successive GFR measurings at a distance of six months, both of them resulting in the same value. His height was 170 cm and his weight, at first measuring, was 69 kg, the  measured GFR was 100 ml/min at the first and the second GFR measuring .  GFR indexed to 1;73/BSA according to the suggested formula by Boyd  resulted in 91,76 ml/min/1,73 m2 . But six months later at second measuring his weight was increased to 76 kg, and the indexation performed according to the same formula and the modified weight resulted in 87,49 ml/min/1,73 m2 , with an estimated reduction of GFR of 4,66 % . Successively, the subject decreased his weight to 68 kg and repeated the measuring of GFR, this resulting to be the same,100 ml/min, and the indexation on the changed weight resulted in 92 ml/min/1,73, a value greater than the first, while any actual modification of GFR occurred in time.  Assuming that a same situation could happen in case of a subject with an indexed GFR at the lowest limit of normality, the indexation operated in case of increased weight would induce to diagnose a first phase of renal insufficiency, this causing  possible therapeutic interventions and increased frequency of controls, with any real need and many problems for the patient .The fourth observation concerns the problem to have a suitable  BSA, not only as estimated BSA but also as measured BSA, taking into account the many criticisms on this topic, within them the concrete impossibility to have an actual measure of  BSA, because a  fractal magnitude, according to the Mandelbrot geometry. See for this :

 Heaf JG. The origin of 1,73 m2 body surface area normalization : problems and implications.:

Clin Physiol Funct Imaging ,2007; 27: 135–137

Benoît Mandelbrot. The Fractal Geometry of Nature, 1982 : W. H. Freeman & Co

Beyond this full limitation, the reading  of the papers on which are shown the methodologies used to measure BSA and to create the formulae for BSA estimation, will point out the casual and inadequate  methodology used  in selecting and recruiting the populations on which to perform the measures of body surface on whose results to elaborate the equations for BSA estimation. Furthermore it has to emphasize the current inadequate and casual  methodology   in selecting and using  the suggested equations. What above mentioned  justifies any doubt on the suitability of this method of normalization. See for this :

Ruggieri G, Rocca AR - An Analysis of Past and Present Methods of Measuring and Estimating Body Surface Area and the Resulting Evaluation of Its Doubtful Suitability to an Universal Application, Blood Purif 2010; 30 : 296-305 

All the considerations above concern the normalization of the measured functions in youth and adulthood , while a very different aspect has what concerning the newborn subjects and the childhood where the body surface is characterized by the presence of many large curved areas, this driving to a reduction of the actual and possibly measured surfaces notwithstanding high weights respect to the heights, this last relationship often inducing in adulthood  large BSAs, even in many of adult subjects  the actual measured BSA would result lower than its estimation on base of the used equations  See for this the above mentioned paper.

WE PREFER TO USE BODY WEIGHT : REGISTERED (ESTIMATED WITH NORMOHYDRATION STATUS).

WE BELIEVE THAT CALCULATION OF BSA INTRODUCES ANOTHER NOT MEASURED VARIABLE WITH MANY DIFFERENT FORMULAE AND ITS SUBJECT TO A THIRD SOURCE OF INDIVIDUAL ERROR

Dear Dr Repetto, thank you for your reply to my question. I have to admit to be perplexed by the many researchers using the weight for normalizing  their measured data.: the existence of a normal  hydration let to exclude that the weight is not modified by a water excess, but certainly not excludes the excessive fat deposition, whose variation in time  would modify the normalization of measured function even this is quite independent from the  fat fluctuation and even in presence of the same body water. To normalize  by a variable subjected to casual variation and very often with frequent variations, will have the same effect that indexing on BSA : in fact BSA fluctuates exactly because the weight variation, height being a variable strongly stable during the years..

Giancarlo Ruggieri

Interesting text you provided. When comparing longitudinally the results for the same patient, one probably should not body scale irrespective of changes in body morphology, and pay attention to the raw GFR. If one does body scale, one should use a formula that accounts for changes in volume of distribution, which is particularly important status post chemotherapy. The formula I listed above is such a formula, k V2/3W1/4 where V is volume of distribution in liters, W in weight in kilograms, and k is a constant that varies slightly depending on the method of measuring GFR , and k is selected such that the formula predicts 100% normal function scaled for whichever curve fit model and fit method is used. Rewriting that formula we get k V11/12 / (V/W)1/4 where V/W adjusts for chemotherapeutic and tumor induced relative increases in body fluids. Indeed, it is a fractal that can only be observed when there are enough patients with fluid disturbances to make the confidence interval of the exponent of weight sufficiently narrow to achieve significant partial probability (ANOVA). It is a fractal because V is Euclidean (2/3 = 2 of 3 dimensions) and W is fractal (1/4 = 1 of 4 dimensions in the language of West, Enquist, and Brown), which makes for 3 of 3 Euclidean dimensions. Using this formula from human children with a large weight range (8 fold, if memory serves) allowed it to be used unmodified in 2 kg cats as well as overweight adult patients. No other body scaling function has yet accomplished that feat. Think of it, suppose you increase fluid volume with cytotoxic chemotherapy or have a patient with 5 liters of ascetic fluid and do a GFR study. What happens then is that the redistribution of any bolus GFR marker takes longer, and the AUC increases, which causes the calculated GFR to decrease even when there is no actual change in renal clearance. This is particularly problematic for one compartment models, and less so for two-compartment models and least for the Tk-GV method. Finally, there is another problem, and that is that polyexponentials underestimate AUC and overestimate CL, which worsens significantly in renal failure. In anuria, if one injects a bolus GFR marker, it will redistribute and concentration will decrease in time, so if we fit a polyexponential to that, we will obtain a clearance value that is greater than zero even when the clearance is known to be zero. Tk-GV is a lot less affected by renal failure because the Tikhonov (Tk) regularization will flatten the curve fit markedly, in fact, it can then find a flatline with zero clearance to be a least error of GFR fit to a monotonically decreasing concentration curve.

Concerning fat; the presence of fat increases catabolism and BMR, which increases GFR. That is what GFR is, auto-regulated elimination of metabolic waste. It scales the same as oxygen consumption, that is, when renal filtration is in physiologic balance with respiration. Fat people can have obviously increased GFR values. Also, there is renal reserve that one can see if the patient eats a hamburger prior to a GFR study, GFR can be markedly increased in that case in its attempt to return the body to homeostasis by eliminating the products of metabolism, in lock step with increase oxygen consumption.

That does not imply that the full power of weight is a proper scaling. Far from it. If used in isolation, and without volume of distribution information, the 3/4 power of weight, or Kleiber's Law, provides an accurate, if excessively imprecise, scaling for catabolism, e.g., GFR. That it is common practice to scale by the full power of weight in animals, which has the particular undesirable trait that the normal range for each species is then unique to that animal species, and then only for a particular age range. Within that range, for each fairly young but still adult animal, the scaling errors are hard to characterize (because of not testing a large weight range), and yet, they are very different between species. What a monumental waste of time, and an opportunity to scale bacteria entirely differently than whales, and a major source of futile "scientific" literature.

With respect to creatinine, it may well be that formulas of weight and height are predictive, now if only creatinine elimination was predictive of GFR, that would solve something. Unfortunately, as creatinine concentration increases, tubular excretion does as well, so that that particular metabolite has a rather nonlinear relationship with GFR and respiration. Not a best choice marker. There are much better ones.