Table 2.

Causal inference results for ordered multi-categorical exposure.

Variable	IVa	nb	$\widehat{\beta }$	95% Cl (Basic)	pc
Drink	6	866	− 1.127	− 2.689–0.285	0.138
Diet*
Fish*	7	882	− 0.752	− 2.551–1.151	0.426
Fruit*	3	874	− 0.835	− 3.427–1.612	0.516
Garlic*	5	877	0.710	− 0.885–2.601	0.425
Legume*	8	858	0.830	− 1.501–2.671	0.435
Meat*	8	827	− 0.164	− 2.299–1.835	0.877
Sugar*	2	892	1.126	− 1.577–3.488	0.384
Vegetable*	4	869	0.291	− 1.523–2.425	0.773
Exercise status	7	855	1.883	0.182–3.512	0.027
Activity*
Housework*	6	884	− 1.282	− 3.676–1.175	0.300
Mahjong*	3	879	− 1.075	− 2.431–0.002	0.083
Open-air*	5	877	− 1.438	− 3.844–0.703	0.215
Pet ownership*	6	862	− 1.080	− 2.192–0.064	0.061
Read*	4	876	− 0.516	− 2.775–1.450	0.636
TV/radio*	4	885	− 0.251	− 2.435–1.726	0.813

“*” represents that the variable was reverse coded, with higher values indicating lower frequency, as described in detail in Additional file 1. "a" represents the number of instrumental variables, "b" represents the number of samples for each exposure (excluding individuals with missing information on the corresponding IVs), and "c" indicates that the p value is reverse derived from CI through a normal distribution.