Table 3.

Univariate and multivariate Cox regression analysis of serum YKL-40 and all-cause mortality in octogenarians.

	Univariate analysis			Multivariate analysis
Variables	HR	95% CI	P-value	HR	95% CI	P-value
Serum YKL-40
Tertile 1
Tertile 2	1·22	0·57–1·19	0·6	1·05	0·46–2·38	0·9
Tertile 3	2·38	1·19–4·78	0·02	2·20	1·05–4·63	0·04
Log YKL-40 (continuous variable)	2·51	1·15–5·52	0·02	2·39	0·98–5·85	0·06

All models (including univariate) are stratified by gender. Multivariate models are adjusted for the effect of smoking, body mass index, anti-inflammatory medicine and comorbidity (se Table 1). n = 151; 51 deaths within the 6 years of follow-up. HR: hazard ratio.