Table 3.

Results from the naive method and joint model trajq_s

LCSS Item	Naive model		Joint Model Trajq_s
	HR (LCSS)	HR (Trt)	β00	β10	β11	β20	e γ: HR (LCSS)	e α: HR (Trt)
Anorexia	1.011	0.67	28.8	5.93	−0.138	−1.15	1.014	0.66
Cough	1.009	0.72	14.7	−1.29	−0.410	0.193	1.015	0.72
Dyspnea	1.010	0.73	32.6	2.74	−1.280*	−0.327	1.011	0.73
Fatigue	1.012	0.70	36.4	6.62	−1.033	−1.032	1.014	0.70
Pain	1.016	0.70	26.8	0.63	−0.891	0.061	1.018	0.71
Interference	1.013	0.66	41.5	5.87	−0.881	−0.905	1.015	0.66
QoL	1.012	0.69	41.1	4.86	−0.873	−0.710	1.014	0.69
Symptoms	1.012	0.68	34.1	3.08	−0.805	−0.432	1.014	0.68
ASBI5	1.021	0.69	27.9	3.09	−0.866	−0.435	1.025	0.68
ASBI8	1.019	0.67	32.0	3.72	−0.883	−0.516	1.022	0.67

LCSS Lung Cancer Symptom Scale, ASBI5 the mean of the five symptom items (anorexia, cough, dyspnea, fatigue, and pain), ASBI8 the mean of the five symptom items and the three global items (interference, QoL, and symptoms), HR hazard ratio, Trt treatment

β₀₀: mean of intercepts b _0i in LCSS trajectory η_i(t) = b _0i + b _1i t + b _2i t ²

β₁₀: mean of slopes b _1i for patients treated with cisplatin

β₁₁: treatment effect on b _1i; *P-value <0.05

β₁₀ + β₁₁: mean of slopes b _1i for patients treated with pemetrexed/cisplatin

β₂₀: mean of quadratic coefficient b _2i

γ: the association between LCSS trajectory and time to progressive disease (TTPD)

P-values for γ and α were less than 0.01