Table 3.

The association of the amounts of colorectal cancer-associated microbes in colorectal carcinoma tissue with the incidence of anastomotic leakage after resection of colorectal carcinoma

The amount of microbes in colorectal carcinoma tissue		Univariable OR (95% CI)	Multivariable OR (95% CI)^a
Model for anastomotic leakage after resection of colorectal carcinoma (n = 256, as an outcome variable)
*Fusobacterium nucleatum*	Negative	1 (reference)	1 (reference)
	Low	0.96 (0.34–2.52)	0.83 (0.27–2.34)
	High	1.44 (0.58–3.55)	1.23 (0.44–3.29)
	P_trend^b	0.45	0.73
pks-positive Escherichia coli	Negative	1 (reference)	1 (reference)
	Low	2.24 (0.77–6.15)	2.17 (0.66–6.76)
	High	1.92 (0.62–5.44)	2.71 (0.81–8.73)
	P_trend^b	0.13	0.039
*Enterococcus faecalis*	Negative	1 (reference)	1 (reference)
	Low	0.72 (0.26–2.02)	0.89 (0.30–2.74)
	High	0.96 (0.37–2.59)	0.88 (0.32–2.57)
	P_trend^b	0.99	0.83
Bifidobacterium genus	Negative	1 (reference)	1 (reference)
	Low	2.04 (0.67–5.62)	2.33 (0.72–6.95)
	High	4.18 (1.68–10.34)	3.96 (1.50–10.51)
	P_trend^b	0.002	0.004

Abbreviations: CI, confidence interval; OR, odds ratio.

The multivariable logistic regression analysis model initially included age, sex, body mass index, diabetes, serum albumin levels, tumor location, pT stage, lymph node metastasis, distant metastasis, surgical approach, operating time, intraoperative bleeding, and diverting ileostomy. A backward stepwise elimination with a threshold of P < 0.05 was used to select variables in the final models.

P_trend value was calculated by the linear trend test across the ordinal (negative, low, and high) categories of the amount of the intestinal microbes as a continuous variable in logistic regression model for the incidence of anastomotic leakage (a dichotomous outcome variable). All statistical tests were two-sided at α level of 0.005, considering the multiple comparisons and the consequent false positives.