Prediction equations of forced oscillation technique: the insidious role of collinearity

Respiratory Research

Table 2 Comparison of the effect of collinearity on the coefficients of the equation developed on 291 healthy children for airway resistance (Rrs5) using a multivariable linear regression model (AlBlooshi, unpublished), with resistance expressed as cm H₂O.s.L⁻¹

Constant	Variables (units)	Coefficient	95% CI^a	P value	VIF^b	SEE^c	R²	Collinearity adjusted for
23.51	Age	−0.19	−0.37 to − 0.006	0.04	3.6	1.54	0.41	No
	Height	−0.12	− 0.16 to − 0.08	< 0.001	5.4
	Weight	−0.06	0.03 to 0.09	< 0.001	2.6
12.78	Age	−0.64	−0.77 to − 0.50	< 0.001	1.74	1.65	0.32	Height excluded
12.78	Weight	0.002	−0.02 to 0.02	0.8	1.74	1.65	0.32	Height excluded
19.71	Age	−0.20	−0.39 to − 0.02	0.03	3.6	1.58	0.38	Weight excluded
19.71	Height	−0.08	− 0.11 to − 0.05	< 0.001	3.6	1.58	0.38	Weight excluded
25.45	Height	−0.15	− 018 to − 0.12	< 0.001	2.6	1.55	0.40	Age excluded
25.45	Weight	0.06	0.03 to 0.09	< 0.001	2.6	1.55	0.40	Age excluded
12.77	Age	−0.62	−0.73 to − 0.52	< 0.001	1.0	1.65	0.32	Yes
9.76	Weight	−0.07	− 0.09 to − 0.05	< 0.001	1.0	1.87	0.13	Yes
21.73	Height	−0.11	−0.12 to − 0.09	< 0.001	1.0	1.59	0.37	Yes^d

^aConfidence interval, ^b Variance inflation factor, ^c Standard error of the equation; ^d Model finally retained

ISSN: 1465-993X