Association between dietary polyphenols intake and an oxidative stress biomarker: interest of multiple imputation for handling missing covariates and outcomes

Table 1 Amount of covariates missing values among subjects with missing outcome and those without missing outcome

	Missing outcome (TBARS)			Available outcome (TBARS)
	n = 2,013			n = 2,116
Covariates introduced in the imputation models (n = 37)^a	Total of subjects included (covariates all measured)	Total of subjects excluded due to missing covariates	% of missing values	Total of subjects included (covariates all measured)	Total of subjects excluded due to missing covariates	% of missing values
Potential predictors^b
Gender	2,013	0	0	2,116	0	0
Age	2,013	0	0	2,116	0	0
Intervention group	2,013	0	0	2,116	0	0
Number of dietary records	2,013	0	0	2,116	0	0
Energy intake	2,013	0	0	2,116	0	0
Weight at inclusion	1,999	14	0.7	2,093	23	1.1
Height at inclusion	1,999	14	0.7	2,089	27	1.3
Zinc	1,952	61	3.1	2,024	92	4.5
Smoking status at inclusion	1,951	62	3.2	2,045	71	3.5
Selenium	1,944	69	3.5	2,024	92	4.5
Alcohol intake at inclusion	1,875	138	7.4	1,976	140	7.1
Alpha-tocophérol	1,729	284	16.4	1,824	292	16.0
Beta carotene	1,729	284	16.4	1,824	292	16.0
Retinol levels	1,729	284	16.4	1,824	292	16.0
Vitamin C	1,651	362	21.9	1,441	675	46.8
Total	1,279	734		1,112	1,004
Number of missing values
1		391	19.4		617	29.2
2		47	2.3		71	3.4
3		268	13.3		276	13.0
> 5		28	1.4		40	1.9
Auxiliary covariates
Years of inclusion	2,013	0	0	2,116	0	0
Physical activity at inclusion	1,975	38	1.9	2,081	35	1.7
Educational level	1,979	34	1.7	2,080	36	1.7
Smoking status at SU.VI.MAX2	2,000	13	0.7	2,095	21	0.01
Total	1,112	901

^aBeside the 15 predictive covariates included in the analysis (and thus in the imputation model) and the 4 auxiliary covariates only included in the imputation model, the imputation model also included the 17 subclasses of polyphenols (with no missing value) and the outcome (TBARS)
^bCovariates included in the analysis model (n = 15)

ISSN: 2055-0928