Table 4. Cox regression models.

Prognostic marker (for outcome)	Outcome	Hazard ratio (CI)	p-value
Top quintile plasma betaine	HF	1.4 (0.7–3.0)	0.30
Top quintile plasma betaine	MI	1.2 (0.6–2.2)	0.64
Top quintile plasma betaine	Death	0.5 (0.2–1.5)	0.22
Bottom quintile plasma betaine	HF	1.2 (0.5–2.9)	0.63
Bottom quintile plasma betaine	MI	1.9 (1.1–3.5)	0.034
Bottom quintile plasma betaine	Death	0.9 (0.4–2.3)	0.90
Top quintile urine betaine excretion	HF	2.7 (1.05–2.7)	0.039
Top quintile urine betaine excretion	MI	1.0 (0.5–2.2)	0.99
Top quintile urine betaine excretion	Death	0.6 (0.2–1.8)	0.37
Bottom quintile betaine excretion	HF	2.7 (1.05–6.9)	0.046
Bottom quintile betaine excretion	MI	1.0 (0.4–2.3)	0.93
Bottom quintile betaine excretion	Death	0.2 (0.02–1.2)	0.077
Top quintile plasma DMG	HF	1.6 (0.8–3.1)	0.20
Top quintile plasma DMG	MI	1.7 (0.9–3.2)	0.076
Top quintile plasma DMG	Death	1.7 (0.5–3.4)	0.10
Top quintile plasma homocysteine	HF	3.0 (1.4–6.2)	<0.001
Top quintile plasma homocysteine	MI	2.6 (1.4–4.7)	0.002
Top quintile plasma homocysteine	Death	2.1 (0.9–4.8)	0.082

Cox proportional hazards regression models with admission for heart failure (HF, acute myocardial infarction (MI) or death (all causes) as the outcomes. Models included left ventricular ejection fraction (LVEF) and age as continuous variables. Categorical variables were gender, and high, middle or low plasma betaine concentration or urine betaine excretion (Table 2), or plasma homocysteine or N,N-dimethylglycine (DMG): the middle 60% group used as the reference group. Significant (p<0.05) markers in bold. DMG: N,N-dimethylglycine. CI: 95% confidence interval.