Table 2. Results of applications of joint models with shared random effects (JM-SRE) to measurements of LPC 15:0/0:0 and mortality data in the LLFS: Estimates of the association parameters for the random intercepts and random slopes of the metabolite in the survival sub-model.

Model	Sex	Variable	α0 (α1)	HR	95% CI for HR	SD of Variable
int	F+M	b0i	−0.436	0.786	(0.728, 0.840)	0.553
intslope	F+M	b0i	−0.452	0.775	(0.726, 0.846)	0.563
intslope	F+M	b1i	−0.765	0.998	(0.994, 1.004)	0.002
int	F	b0i	−0.428	0.783	(0.709, 0.858)	0.572
intslope	F	b0i	−0.408	0.778	(0.694, 0.865)	0.615
intslope	F	b1i	−0.084	0.999	(0.973, 1.041)	0.011
int	M	b0i	−0.461	0.785	(0.711, 0.854)	0.525
intslope	M	b0i	−0.439	0.780	(0.698, 0.855)	0.568
intslope	M	b1i	−0.238	0.997	(0.956, 1.006)	0.011

Model – type of joint model (int – random intercept of LPC in survival sub-model, intslope – random intercept and slope of LPC in survival sub-model), see section Joint models: General specifications; Variable – b_0i: random intercept of the metabolite, b_1i: random slope of the metabolite; α₀ (α₁) – estimates of the regression parameters for b_0i (b_1i) in respective models; HR – hazard ratios for an increase by a standard deviation of Variable; 95% CI for HR – respective 95% confidence intervals for HRs; SD of Variable – standard deviation of Variable. Highlighted in bold are cases where confidence intervals do not contain one. The JM were estimated using R-package joineR. LPC values were transformed (see section Data).