Factors associated with prolonged postoperative length of hospital stay after laparoscopic colorectal cancer resection: a secondary analysis of a randomized controlled trial

Table 4 Univariate and multivariate analysis to identify influencing factors for PPLOS

Factors^a	Univariate OR (95% CI)	P-value	Multivariate OR (95% CI)	P-value
Preoperative SpO₂		0.014		0.006
≥ 96%	1 (Reference)		1 (Reference)
< 96%	2.47 (1.20–5.05)		3.09 (1.38–6.92)
Distant tumor metastasis		0.004		0.031
No	1 (Reference)		1 (Reference)
Yes	0.26 (0.10–0.65)		0.34 (0.13–0.91)
Surgical approach		0.001		0.012
Approach C	1 (Reference)		1 (Reference)
Approach A	0.57 (0.31–1.06)	0.074	1.69 (0.86–3.35)	0.131
Approach B	3.27 (1.47–7.26)	0.004	4.51 (1.67–12.18)	0.003
Perioperative surgical event^#		0.001		0.009
No	1 (Reference)		1 (Reference)
Yes	2.61 (1.46–4.65)		2.44 (1.25–4.76)
Postoperative albumin infusion^*		0.003		0.018
No	1 (Reference)		1 (Reference)
Yes	2.49 (1.38–4.49)		2.19 (1.14–4.19)
Postoperative early ambulation^*		< 0.001		0.002
No	1 (Reference)		1 (Reference)
Yes	0.30 (0.16–0.56)		0.35 (0.18–0.68)

^*Modifiable factor. PPLOS, prolonged postoperative length of stay; SpO_2, pulse oxygen saturation; OR, odds ratio; CI, confidence interval. Approach A: Parks procedure, right hemicolectomy, or natural orifice specimen extraction surgery. Approach B: Miles procedure or left hemicolectomy. Approach C: Dixon procedure or sigmoidectomy
^#Intraoperative events (positive anastomotic leakage test, reoperation or intestinal anastomosis twice, duration of non-pneumoperitoneum ≥ 2 h, or abdominal drainage ≥ 200 mL in the postoperative anesthesia care unit) or abdominal drainage ≥ 300 mL in the first three postoperative days
^aTwo variables considered clinically relevant (ASA physical status and lung-protective ventilation) and nine variables with a P-value < 0.1 (preoperative SpO₂, distant tumor metastasis, surgical approach, chief surgeon, duration of surgery ≥ 3 h, blood loss ≥ 100 mL, perioperative surgical events, postoperative albumin infusion, and postoperative early ambulation) were introduced into the multivariate logistic regression model. A forward stepwise elimination with thresholds of input P < 0.05 and output P < 0.1 was used to select variables in the final model (goodness of fit: Hosmer–Lemeshow test, χ² = 3.463, P = 0.749)

ISSN: 1471-2482