TABLE 4.

Association between fat mass and all-cause mortality using Cox proportional hazard regression.

	<60 years			P for trend	≥60 years			P for trend
	T1	T2	T3		T1	T2	T3
Female
Unadjusted model	1	0.62 (0.42, 0.90)	0.62 (0.42, 0.90)	0.012	1	0.89 (0.61, 1.30)	0.74 (0.50, 1.08)	0.276
Model Ab	1	0.62 (0.42, 0.90)	0.61 (0.41, 0.89)	0.010	1	1.01 (0.69, 1.49)	0.88 (0.59, 1.30)	0.725
Model Bb	1	0.74 (0.50, 1.10)	0.74 (0.49, 1.11)	0.217	1	1.23 (0.81, 1.85)	1.05 (0.70, 1.58)	0.579
Model Cb	1	0.90 (0.56, 1.44)	0.89 (0.55, 1.45)	0.866	1	1.28 (0.79, 2.07)	1.27 (0.79, 2.02)	0.528
Model Db	1	0.83 (0.44, 1.55)	0.69 (0.37, 1.30)	0.513	1	1.82 (0.98, 3.39)	0.95 (0.47, 1.92)	0.062
Male
Unadjusted model	1	0.94 (0.66, 1.35)	0.90 (0.62, 1.30)	0.846	1	0.73 (0.54, 0.98)	0.67 (0.49, 0.90)	0.017
Model Ab	1	0.94 (0.65, 1.34)	0.91 (0.63, 1.32)	0.877	1	0.71 (0.53, 0.97)	0.63 (0.47, 0.86)	0.009
Model Bb	1	0.99 (0.68, 1.43)	1.04 (0.71, 1.53)	0.961	1	0.68 (0.50, 0.92)	0.63 (0.46, 0.87)	0.009
Model Cb	1	1.13 (0.71, 1.80)	1.14 (0.72, 1.81)	0.824	1	0.71 (0.49, 1.01)	0.70 (0.49, 1.02)	0.097
Model Db	1	1.13 (0.62, 2.05)	1.33 (0.76, 2.34)	0.601	1	0.98 (0.62, 1.55)	0.91 (0.57, 1.45)	0.906

^aValues are models adjusted HR and 95% confidence interval, ranges for tertiles (T) 1 through 3. Linear trends (p for trend) were obtained with fat mass as continuous variables.

^bModel A: adjusted for age, education, smoking, alcohol drinking, family history of cancer, version of Body Composition Analyzer; Model B: as model A and additionally adjusted for Karnofsky performance scores, duration of hospital stays, NRS 2002 score, nutrition support, and handgrip strength; Model C: as Model B and additionally adjusted for commodities, previous treatments, cancer types, TNM stages, and quality of life; Model D: as Model C and additionally adjusted C-reaction protein.