Accuracy of predictive equations for resting energy expenditure estimation in mechanically ventilated Thai patients

Kongpolprom, Napplika

doi:10.2478/abm-2023-0041

Figures & Tables

Bland–Altman plots of differences between predictive REE calculated by 14 equations and measured REE (predictive REE – measured REE) in 24 mechanically ventilated Thai patients. The solid lines at the middle part of each picture represent mean differences of compared values and the upper and lower solid lines represent upper and lower limits of agreement (LOA) (±1.96 SD from mean). The points on these plots non-uniformly and widely scatter around mean differences, indicating poor agreement and bias. REE, resting energy expenditure; SD, standard deviation.

Accuracy of predictive equations

Equations	n	Accurate prediction (%)	Under estimation (%)	Over estimation (%)	Bias (%)	Difference(Predictive value – measured value) (kcal/d)
Equations	n	Accurate prediction (%)	Under estimation (%)	Over estimation (%)	Bias (%)	Mean	Minimum	Maximum
ACCP	30	23.3	60.0	16.7	−13.0	−320	−1257	617
HBE	30	30.0	60.0	10.0	−4.0	−99	−1063	930
1.2×HBE	30	31.8	21.3	40.9	3.2	−49	−742	631
1.5×HBE	30	18.2	9.1	72.7	29.0	322	−476	979
MSJ	30	33.3	50.0	16.7	−10.8	−200	−1172	919
Ireton-Jones 1992	30	6.7	23.3	70.0	22.2	233	−812	923
Ireton-Jones 2002	30	26.7	23.3	50.0	8.0	43	−1018	665
Penn State 2003 (HBE)	24	25.0	50.0	25.0	1.6	−88	−740	837
Penn State 2003 (MSJ)	27	37.0	22.2	40.7	9.6	100	−502	1056
Penn State 2010	27	48.1	14.8	37.0	11.6	98	−472	1089
Swinamer 1990	27	18.5	11.1	70.4	25.9	295	−336	1191
Faisy equation	27	25.9	0.0	74.1	31.8	366	−176	1397
Brandi 1999	22	22.7	27.3	50.0	13.6	74	−549	1115
25 kcal/kg	30	30.0	20.0	50.0	25.2	440	−933	4030

The ICC of 14 equations

Equations (n = 22)	ICC	95% Confidence interval		F test
Equations (n = 22)	ICC	Lower bound	Upper bound	F test
ACCP	0.324	−0.338	0.692	0.136
HBE	0.608	−0.023	0.845	0.003
1.2×HBE	0.764	0.430	0.902	<0.001
1.5×HBE	0.686	0.017	0.886	<0.001
MSJ	0.809	0.515	0.923	<0.001
Ireton-Jones 1992	0.723	0.204	0.894	<0.001
Ireton-Jones 2002	0.800	0.519	0.917	<0.001
Penn State 2003 (HBE)	0.672	0.217	0.863	0.007
Penn State 2003 (MSJ)	0.843	0.626	0.934	<0.001
Penn State 2010	0.814	0.561	0.922	<0.001
Swinamer 1990	0.715	0.155	0.893	<0.001
Faisy equation	0.643	−0.123	0.873	<0.001
Brandi 1999	0.608	−0.104	0.859	0.039
25 kcal/kg	0.701	0.295	0.875	0.002

Predictive equations for REE

Equations	Formula
ACCP equation	= 25 × weightIf BMI is in the range of 16–25 kg/m², use usual BWIf BMI > 25 kg/m², use IBWIf BMI < 16 kg/m², use existing BW for the first 7–10 d, and then use IBW
HBE	Men: kcal/d = 66.47 + 13.75 (w) + 5 (h) − 6.75 (a)Women: kcal/d = 655.1 + 9.56 (w) + 1.85 (h) − 4.68 (a)w = actual weight (kg); but if >125% IBW, use ABWABW = [(actual body weight − IBW) × (0.25 to 0.5)] + IBW
MSJ equation	Men: kcal/d = 10 (w) + 6.25 (h) − 5 (a) + 5Women: kcal/d = 10 (w) + 6.25 (h) − 5 (a) − 161
Ireton-Jones 1992 equation	1925 − 10 (a) + 5 (w) + 281 (S) + 292 (T) + 851 (B)
Ireton-Jones 2002 equation	1784 − 11 (a) + 5 (w) + 244 (S) + 239 (T) + 804 (B)
Penn State 2003 equation	Harris–Benedict: kcal/d = 0.85 (HBE) + 175 (Tmax) + 33 (Ve) − 6433Mifflin: kcal/d = 0.96 (MSJ) + 167 (Tmax) + 31 (Ve) − 6212
Penn State 2010 equation	MSJ (0.71) + Ve (64) + Tmax (85) − 3085
Swinamer 1990 equation	(945 × BSA) − (6.4 × age) + (108 × BT) + (24.2 × RR) + (817 × VT) − 4349
Faisy equation	8 (w) + 14 (h) + 42 (Ve) + 94 (BT) − 4834
Brandi 1999 equation	0.96 (HBE) + 7 (HR) + 48 (Ve) − 702
Equation commonly used in critical illness	25 kcal/kg actual body weight including 1–1.5 g protein/kg

Sensitivity analysis for the accuracy of predictive equations

Equations (n = 22)	Accurate prediction (%)	Under estimation (%)	Over estimation (%)	Bias (%)	Difference(Predictive value – measured value) (kcal/d)
Equations (n = 22)	Accurate prediction (%)	Under estimation (%)	Over estimation (%)	Bias (%)	Mean	Minimum	Maximum
ACCP	22.7	54.5	22.7	−10.2	−272.4	−1091.0	617.0
HBE	27.3	59.1	13.6	14.02	296.7	−987.1	399.7
1.2×HBE	31.8	27.3	40.9	3.18	−49.1	−742.1	631.3
1.5×HBE	18.2	9.1	72.7	29.0	322.3	−475.7	978.6
MSJ	36.4	50.0	13.6	−9.3	−183.3	−773.0	490.8
Ireton-Jones 1992	9.1	18.2	72.7	25.0	275.5	−421.0	923.0
Ireton-Jones 2002	22.7	27.3	50.0	8.8	53.4	−662.0	665.0
Penn State 2003 (HBE)	22.7	50.0	27.3	2.6	−76.1	−739.9	837.1
Penn State 2003 (MSJ)	27.3	27.3	45.5	9.6	73.0	−501.7	944.4
Penn State 2010	40.9	18.2	40.9	12.1	82.6	−471.9	1088.6
Swinamer 1990	18.2	13.6	68.2	25.9	276.6	−335.2	1190.8
Faisy equation	27.3	0.0	72.7	32.4	353.5	−176.0	1397.2
Brandi 1999	23.5	29.4	47.1	14.1	61.4	−548.6	1115.4
25 kcal/kg	31.8	18.2	50.0	20.6	318.5	−933.0	1867.0

Correlation and agreement between measured REE and predictive REE

Equations	Correlation Spearman rho	Bland–Altman
Equations	Correlation Spearman rho	Mean bias	SD	Lower LOA Mean – 1.96 (SD)	Upper LOA Mean + 1.96 (SD)
ACCP	0.445	−320	457	−1217	576
HBE	0.675	−99	464	−1008	810
1.2×HBE	0.566	−49	355	−744	646
1.5×HBE	0.566	322	358	−379	1023
MSJ	0.736	−200	429	−1041	641
Ireton-Jones 1992	0.624	233	427	−603	1070
Ireton-Jones 2002	0.666	43	402	−745	831
Penn State 2003 (HBE)	0.714	−88	367	−807	631
Penn State 2003 (MSJ)	0.767	100	369	−623	823
Penn State 2010	0.771	98	312	−513	709
Swinamer 1990	0.708	295	335	−363	952
Faisy equation	0.778	366	333	−287	1019
Brandi 1999	0.587	74	394	−698	846
25 kcal/kg	0.675	439	1019	−4377	2457

Baseline characteristics of the patients

Baseline characteristics
Sex n (%)
Male	12 (50)
Age
Median (years)	64.5 [43.50, 75.50]
Height
Median (cm)	160 [151.0, 167.0]
Median (kg)
Median (cm)	62.5 [47.75, 85.50]
BMI n (%)	27 (100)
Underweight (<18.5 kg/m²)	6 (25)
Normal (18.5–22.9 kg/m²)	7 (29.2)
Overweight (23–24.9 kg/m²)	0 (0)
Obese 1 (25–29.9 kg/m²)	1 (4.2)
Obese 2 (>30 kg/m²)	10 (41.7)
BSA (Dubois) median (m²)	1.7 [1.46, 1.86]
Cause of respiratory failure n (%)
Pneumonia	4 (16.7)
Post ARDS	3 (12.5)
COPD	2 (8.3)
Chronic lung disease	1 (4.2)
Sepsis	2 (8.3)
Cardiac arrest	1 (4.2)
Heart failure	2 (8.3)
Tetanus	1 (4.2)
Others	8 (33.3)
Ventilator mode n (%)
Volume control	3 (10)
Pressure control	6 (20)
Pressure support	20 (66.7)
Proportional assist ventilation	1 (3.3)
Measured REE median (kcal/d)
Total	1553.0 [1126.75, 2065.75]
Underweight (<18.5 kg/m²)	1044.0 [884.50, 1591.50]
Normal (18.5–22.9 kg/m²)	1545.0 [1125.75, 1981.00]
Overweight (23–24.9 kg/m²)	–
Obese 1 (25–29.9 kg/m²)	1475.0 [1475.00, 1475.00]
Obese 2 (>30 kg/m²)	2043.5 [1525.25, 2329.50]
Average measured calories per actual BW (kcal/kg)	23.5
Average measured calories per BW (kcal/kg)	27
Use actual BW if BMI <30 kg/m²
Use adjusted BW if BMI ≥ 30 kg/m²
Average measured calories per IBW (kcal/kg)	29.5
REE calculated by predictive equations median (kcal/d)
ACCP	1255.0 [1050.00, 1423.00]
HBE	1180.5 [1078.89, 1518.95]
1.2×HBE	1424.3 [1292.80, 1752.98]
1.5×HBE	1780.4 [1615.99, 2191.23]
MSJ	1250.0 [1012.25, 1522.50]
Ireton-Jones 1992	1688.0 [1585.00, 2042.25]
Ireton-Jones 2002	1479.5 [1373.25, 1957.00]
Penn State 2003 (HBE)	1523.5 [1270.24, 1734.94]
Penn State 2003 (MSJ)	1516.5 [1182.04, 1905.00]
Penn State 2010	1542.4 [1310.89, 1846.61]
Swinamer 1990	1740.0 [1502.28, 2115.50]
Faisy	1840.4 [1576.60, 2155.20]
Brandi 1999	1547.3 [1302.86, 1794.43]
Equation commonly used in critical illness (25 kcal/kg)	1562.5 [1225.00, 2400.00]

Accuracy of predictive equations for resting energy expenditure estimation in mechanically ventilated Thai patients

Figures & Tables

Figure 1.

Figure 2.

Figure 3.

Accuracy of predictive equations

The ICC of 14 equations

Predictive equations for REE

Sensitivity analysis for the accuracy of predictive equations

Correlation and agreement between measured REE and predictive REE

Baseline characteristics of the patients

Paradigm

My account