Fatigue in chronically critically ill patients following intensive care - reliability and validity of the multidimensional fatigue inventory (MFI-20)

Table 4 MFI-20 models tested via Confirmatory Factor Analysis (CFA) in n = 113 patients three months post-ICU (t2)

Model	Number of free parameters	Chi-square (df)	p value	CFI	TLI	RMSEA (90% CI)
A: Original 5-Factor Model	64	401.231 (166)	<.001	.783	.751	.112 (.098–.127)
B: 1-factor model	40	503.517 (170)	<.001	.692	.656	.132 (.119–.146)
C: 2-factor model	84	345.654 (146)	<.001	.816	.760	.110 (.096–.126)
D: 3-factor model (PF/ GF/ RA summarized)	80	281.998 (150)	<.001	.878	.846	.089 (.073–.104)
Original five factors^a
GF	8	3.041 (2)	.219	.985	.956	.068 (.000–.212)
PF	8	.353 (2)	.838	1.000	1.050	.000 (.000–.107)
MF	8	5.491 (2)	.064	.984	.951	.125 (.000–.255)
RA	8	3.634 (2)	.163	.991	.973	.085 (.000–.224)
RM	8	8.161 (2)	.017	.769	.308	.166 (.060–.291)

RMSEA Root Mean Squared Error of Approximation, CFI Confirmatory Fit Index, CI Confidence Interval, TLI Tucker-Lewis Index, df degrees of freedom, MFI-20 Multidimensional Fatigue Inventory-20, GF General Fatigue, MF Mental Fatigue, PF Physical Fatigue, RA Reduced Activity, RM Reduced Motivation
^aeach factor of the original MFI-20 was analyzed in independent models. In models A, C and D, the mean values of each latent variable were fixed to 0 and variances to 1. The latent fatigue factors were intercorrelated. In models B mean values and variance of the latent factor were not specified

ISSN: 1477-7525