Population characteristics (given in Table 1, page 1579):
(1) The mean age was 60 years.
(2) The male to female ratio was 53%:47%.
(3) 45% Caucasian, 47% African American, 6.5% Hispanic
(4) Diabetes was present in 37%.
Parameters:
(1) age in years
(2) gender
(3) height in cm
(4) weight in kilograms
(5) diabetic status
(6) weight * weight
(7) age * gender
(8) gender * weight
(9) age * weight
(10) height * weight
Parameter
|
Finding
|
Points
|
gender
|
male
|
1
|
|
female
|
0
|
diabetic status
|
present
|
1
|
|
absent
|
0
|
total body water in kilograms =
= (-0.07493713 * (age in years)) - (1.01767992 * (points for gender)) + (0.12703384 * (height)) - (0.04012056 * (weight)) + (0.57894981 * (points for diabetes)) - (0.00067247 * (weight) * (weight)) - (0.03486146 * (age in years) * (points for gender)) + (0.11262857 * (points for gender) * (weight)) + (0.00104135 * (age in years) * (weight)) + (0.00186104 * (height) * (weight))
where:
• The first equation in Table 2, page 1580 uses 0.0186104 as the factor for (height) * (weight). The beta value in the listing above is 0.00186104, which is required to get valid results.
Performance:
• The results of the equation were comparable to findings from body impedance analysis.
• Equations developed in healthy adults (Watson, Hume-Weyer) tended to underestimate the total body water in this population by 3-4 kilograms.
Limitations:
• Regression equations may sometimes not transport to groups that differ significantly from the original population.