Table 2.
Comparison of total body water (TBW) measured by 2H dilution and predicted from bioelectrical impedance analysis by using the published equations, in the eighty-eight measurements performed in thirty-four HIV-infected African outpatients†
Published equation | Correlation coefficient between TBW measured and predicted* | Bias (kg)‡ |
P value | 95 % limits of agreement (kg)§ | Pure error (kg)∥ | Correlation between mean and difference of the two methods | P value | |
---|---|---|---|---|---|---|---|---|
Mean | sd | |||||||
A | 0·94 | −1·70 | 2·38 | <0·0001 | −6·36, + 2·96 | 2·92 | 0·35 | 0·0006 |
B | 0·93 | −0·97 | 2·71 | 0·0011 | −6·28, + 4·34 | 2·86 | 0·11 | 0·31 |
C | 0·91 | −3·06 | 3·01 | <0·0001 | −8·96, + 2·84 | 4·28 | 0·29 | 0·26 |
D | 0·95 | −4·34 | 2·21 | <0·0001 | −8·67, 0 | 4·87 | 0·092 | 0·52 |
E | 0·94 | +1·32 | 2·47 | <0·0001 | −3·52, + 6·16 | 2·79 | 0·10 | 0·35 |
F | 0·94 | −1·69 | 2·47 | <0·0001 | −6·53, + 3·15 | 2·98 | 0·22 | 0·050 |
G | 0·94 | −1·51 | 2·56 | <0·0001 | −5·53, + 3·05 | 2·96 | -0·017 | 0·89 |
H | 0·91 | −2·34 | 3·59 | <0·0001 | −9·38, + 4·70 | 4·27 | -0·33 | 0·0052 |
I | 0·94 | +0·11 | 2·34 | 0·67 | −4·48, + 4·70 | 2·33 | 0·12 | 0·24 |
J | 0·91 | −1·85 | 3·16 | < 0·0001 | −8·04, + 4·34 | 3·65 | −0·032 | 0·79 |
K | 0·56 | −3·21 | 10·39 | 0·0057 | −23·57, + 17·15 | 10·81 | 0·59 | 0·021 |
L | 0·95 | −4·05 | 2·28 | < 0·0001 | −8·52, + 0·42 | 4·64 | 0·10 | 0·34 |
M | 0·93 | −0·36 | 2·75 | 0·23 | −5·75, + 5·03 | 2·76 | 0·055 | 0·62 |
N | 0·94 | +1·25 | 2·44 | < 0·0001 | −3·53, + 6·05 | 2·73 | 0·035 | 0·75 |
O | 0·90 | +1·34 | 3·22 | < 0·0001 | −4·97, + 7·65 | 3·47 | 0·10 | 0·39 |
P<0·05 for all.
For details of subjects and procedures, see Subjects and Methods. For details of the published equations, see Table 1.
Measured - predicted value (a negative value of the bias reflects an overestimation by the prediction equation).
95 % CI of the individual differences.
Pure error = √Σ(measured - predicted)2/n, where n is the number of measurements. The smaller the pure error the greater the accuracy.