Table 2

Multivariate Cox regression analysis for the associations between different patterns of BMI trajectories and the risk of self-reported type 2 diabetes

BMI trajectory change pattern	Model 1		Model 2		Model 3		Model 4		Model 5
BMI trajectory change pattern	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Loss	1.01 (0.70 to 1.45)	0.965	1.07 (0.75 to 1.53)	0.721	1.05 (0.72 to 1.54)	0.792	1.06 (0.72 to 1.55)	0.768	0.99 (0.67 to 1.45)	0.949
Stable	Reference		Reference		Reference		Reference	Reference
Moderate gain	1.18 (0.97 to 1.43)	0.103	1.16 (0.96 to 1.41)	0.125	1.16 (0.94 to 1.42)	0.169	1.15 (0.94 to 1.41)	0.182	1.15 (0.94 to 1.42)	0.174
Substantial gain	1.38 (1.02 to 1.87)	0.036	1.36 (1.01 to 1.84)	0.047	1.49 (1.09 to 2.03)	0.012	1.49 (1.09 to 2.03)	0.012	1.50 (1.10 to 2.05)	0.010

Model 1 includes only the category of BMI trajectory change pattern.
Model 2 adds age and gender to model 1.
Model 3 includes covariates in model 2 plus smoke, drink, and physical activity.
Model 4 includes covariates in model 3 plus dietary energy intake.
Model 5 IPCW method adjusted for potential bias from loss to follow-up.
Bold font indicates significant association.
BMI, body mass index; IPCW, inverse probability of censorship weighting.