Table 3.

IPW-Adjusted Ordinal Logistic Regression Analysis to Assess the Association of the Amount of Bifidobacterium DNA (Predictor) with the Extent of Signet Ring Cells (Outcome)

Variable	Extent of signet ring cells
	Univariable OR (95% CI)∗	Multivariable OR (95% CI)∗†
Amount of Bifidobacterium DNA in colorectal cancer tissue
Negative	1 (referent)	1 (referent)
Low	1.26 (0.75–2.10)	1.29 (0.74–2.24)
High	1.99 (1.28–3.09)	1.87 (1.16–3.02)
Ptrend‡	0.003	0.01

IPW, inverse probability weighting; OR, odds ratio.

^∗IPW was applied to reduce a bias attributable to the availability of tumor tissue after cancer diagnosis (Statistical Analysis).

^†The multivariable ordinal logistic regression model initially included age, sex, year of diagnosis, family history of colorectal cancer, tumor location, microsatellite instability, CpG island methylator phenotype, KRAS, BRAF, and PIK3CA mutations, and long-interspersed nucleotide element-1 methylation level. A backward elimination with a threshold of P = 0.05 was used to select variables for the final model. The variables that remained in the final models are shown in Supplemental Table S5.

^‡P_trend was calculated by the linear trend across the ordinal categories of the amount of Bifidobacterium DNA (negative, low, and high) in the IPW-adjusted logistic regression model for the extent of signet ring cells (absent, low, and high, as an ordinal outcome variable).