Table 2. Refitting the regression model after excluding influential observations.

Covariatesa	Coefficient	Standard error	Lower limit of 95% CI	Upper limit of 95% CI	p
(Glucose/100)−2	10.865	2.589	5.789	15.941	<0.001
(Glucose/100)-1.96	92.193	0.985	90.261	94.125	<0.001
Insulin rate-2.85	−0.861	0.316	−1.481	−0.241	0.007
(Interval/100)3	−5.530	7.977	−21.168	10.109	0.488
(Interval/100)3 × ln(interval/100)	77.272	19.426	39.189	115.356	<0.001
[(Dextrose rate + 0.01)/100]0.5 − 9.3 × 10−6	5.584	1.703	2.247	8.922	0.001
[(Dextrose rate + 0.01)/100]3 ×ln[(Dextrose rate + 0.01)/100] + 3.6 × 10−5	−4.011	1.106	−6.180	−1.843	<0.001
PO volume	−0.013	0.008	−0.029	0.002	0.094
TPN volume	0.091	0.030	0.031	0.150	0.003
Lactate (mmol/l)	0.865	0.227	0.421	1.310	<0.001
Congestive heart failure	2.376	1.151	0.120	4.632	0.039
Renal failure	−2.983	1.817	−6.546	0.580	0.101
Liver disease	4.254	1.977	0.379	8.129	0.031
(Glucose/100)−2 × (insulin rate-2.85)	0.489	1.163	−1.792	2.769	0.674
[(Glucose/100)-1.96] × (insulin rate-2.85)	0.189	0.358	−0.513	0.891	0.598
Constant	185.875	0.910	184.091	187.659	<0.001

Notes.

Number of obs = 4585, F(15, 4569) = 1760.88, Prob >F = 0.0000, R-squared = 0.88525, Adj R-squared = 0.8520, Root MSE = 29.784.

PO

by mouth, orally (from the Latin “per os,” by mouth)

TPN

total parental nutrition

^aSome covariates were centered and transformed with fractional polynomials.