Table 3.

Binary logistic regression analysis for outcome postoperative NRS at 24, 48, 72 h.

	Outcome: NRS ≥2 at postoperative 24 h
Predictors		Model 1
		OR (95% CI)	P value
Total gastrectomy		1.823 (1.094–3.040)	0.021*
AJCC TNM stage
I	(reference)
II		0.232 (0.062–0.872)	0.031*
III		0.185 (0.049–0.698)	0.013*
IV		0.369 (0.102–1.332)	0.128
	Outcome: NRS ≥2 at postoperative 48 h
Predictors	Model 2		Model 3
	OR (95% CI)	P	OR (95% CI)	P
BMI, kg/m2	1.699 (0.995–2.900)	0.052	1.75 (1.029–2.976)	0.039*
Duration of operation, min	1.565 (0.933–2.625)	0.090	1.587 (0.95–2.652)	0.078
Diabetes	2.205 (1.02–4.765)	0.044*	2.09 (0.977–4.473)	0.057
Lauren's histology
Intestinal type	(reference)
Diffuse type	0.841 (0.487–1.454)	0.536
Mixed type	1.558 (0.788–3.081)	0.202
Intraoperative blood loss, ml	0.223 (0.048–1.042)	0.056	0.234 (0.051–1.076)	0.062
	Outcome: NRS ≥2 at postoperative 72 h
Predictors	Model 4		Model 5
	OR (95% CI)	P	OR (95% CI)	P
BMI, kg/m2	1.663 (0.957–2.890)	0.071	1.697 (0.992–2.905)	0.054
Diabetes	1.791 (0.909–3.528)	0.092	1.939 (0.997–3.771)	0.051
Hypertension	1.209 (0.732–1.996)	0.459
Pre-operative hemoglobin, g/L	0.767 (0.454–1.297)	0.322
Distant metastasis	2.821 (0.932–8.535)	0.066	3.054 (1.019–9.155)	0.046*
Intraoperative transfusion, ml	1.876 (0.983–3.581)	0.056	2.246 (1.267–3.982)	0.006*

BMI, body mass index; NRS, Numerical Rating Scale; AJCC, American Joint Committee on Cancer; TNM, Tumor Node Metastasis.

*P <0.05.

Explanation for models Binary logistic regression models 1,2,4 were constructed using predictors found to be significant in the univariate analysis (p <0.05). Models 3,5 were derived from models 2,4 respectively with non-significant predictors eliminated in stepwise process called backward conditional. The resulting models include only significant predictors (p <0.05). The reported odds ratios (all significant ones are above 1) suggest that one unit increase in predictor score (or having categorical predictor) is associated with increase odds of pain.