New metabolic health definition might not be a reliable predictor for mortality in the nonobese Chinese population

Table 2 Associations between MUH and all-cause mortality in the crude analysis, multivariable analysis and propensity-score analyses

Analysis	All-cause mortality
No. of deaths/no. of participants at risk (%)^a
MH	17/920 (1.85%)
MUH	13/237 (5.49%)
Crude analysis	3.04 (1.47, 6.25), 0.003
Propensity-score analyses
With overlap weighting (univariable)	1.45 (0.50, 4.19), 0.490
With overlap weighting (multivariable)^b	1.42 (0.49, 4.17), 0.519
With PSM (univariable)	0.66 (0.11, 3.97), 0.652
With PSM (multivariable)^c	0.69 (0.11, 4.20), 0.685
Adjusted for PS^d	1.18 (0.39, 3.61), 0.766
Adjusted for PS^e	1.16 (0.39, 3.45), 0.790
Multivariable analysis^e	1.09 (0.38, 3.13), 0.875

Values are n (%) or HRs (95% CI) with p values
^aBinary event rates
For the relatively small number of all-cause mortality, multivariable models only adjusted for some basic variables to ensure the convergence of the model: ^badjustment for age and sex; ^cadjustment for age, sex, DBP, TC and HDL-C; ^dadjustment for propensity score; ^eadjustment for propensity score plus age and sex. ^fAdjustment for age, sex, smoking, drinking, exercise, cardiovascular diseases, DBP, TC, HDL-C, triglycerides, LDL-C and BMI; for including many covariates, the convergence of the model may be poor, and the results were exploratory
Abbreviations: MH Metabolic health, MUH Metabolically unhealthy, PSM Propensity score matching, PS Propensity score

ISSN: 1471-2458