Of the 1863 patients enrolled in this study, 246 developed primary PPOI during their hospital stay. The incidence of primary PPOI was 13.2% among patients who underwent open abdominal surgery. The times to first flatus (median 4 vs. 3 days, p < 0.001) and first bowel movement (median 5 vs. 4 days, p < 0.001) were significantly longer in the PPOI group than in the non-PPOI group (Table 1). A total of 71.5% of patients in the PPOI group received patient-controlled analgesia (PCA), which mainly consisted of opioid drugs, compared with 63.4% of patients in the non-PPOI group. The other surgery-related characteristics, such as the surgical organ, duration of surgery, and intraoperative blood transfusion, were comparable between the non-PPOI group and the PPOI group. The patient status-related characteristics, including age, sex, body mass index (BMI), and National Nosocomial Infections Surveillance (NNIS) score, were balanced between the two groups (Table 1). The median hospital LOS was 11.00 (9.00–14.00) days, and the median healthcare cost was 68,087 (58,304–81,703) CNY among all included patients.

The median hospital stay was significantly longer in the PPOI than non-PPOI group (12 vs. 11 days, p < 0.001). The median healthcare cost was significantly higher in the PPOI than non-PPOI group (70,672 vs. 67,597 CNY, p = 0.016).

Poisson regressions were used to identify the predictive factors for hospital LOS (Table 2). Age and BMI showed an IRR of 1.004 (95% CI, 1.003–1.005) and 1.009 (95% CI, 1.004–1.014) for prediction of the LOS, indicating that older age and higher BMI were risk factors for a longer LOS. Patients with an NNIS score of 2 and 3 tended to have a longer LOS than those with an NNIS score of 0 [IRR, 1.087 (95% CI, 1.002–1.178) and IRR, 1.134 (95% CI, 1.011–1.271)]. The other patient status-related characteristics, including sex, were not associated with LOS. The surgery-related characteristics prominently affected the LOS. Patients who underwent pancreaticoduodenal surgery had a significantly longer LOS than those who underwent gastric surgery [IRR, 1.184 (95% CI, 1.146–1.222)]. Blood transfusion after surgery was associated with a significantly increased LOS [IRR, 1.158 (95% CI, 1.106–1.214)]. Patients who received PCA consisting of an opioid drug had a longer LOS [IRR, 1.123 (95% CI, 1.089–1.158)]. The development of postoperative complications was another considerable factor that resulted in a longer LOS. More severe complications with higher Clavien–Dindo grades had higher IRRs for LOS. For complications, the p for trend was < 0.001.

Primary PPOI was a risk factor for prolonged LOS [IRR, 1.139 (95% CI, 1.099–1.182)] in the univariate Poisson regression. To control for all these confounders, multivariate Poisson regression analyses were performed to depict the contribution of PPOI on LOS. All significant factors in the univariate analysis were included in the multivariate analysis. Primary PPOI was an independent risk factor for prolonged LOS [IRR, 1.122 (95% CI, 1.081–1.165)]. It contributed to 12% of prolonged LOS in patients who underwent major abdominal surgery.

Linear regression for log-transformed healthcare cost was performed to identify the factors that affected healthcare cost (Table 3). Patients’ age and NNIS scores were predictors of healthcare cost in the multivariate regression [SE, 1.003 (95% CI, 1.001–1.004) and SE, 1.209 (95% CI, 1.119–1.306), respectively]. The surgery-related characteristics were also significant predictive factors for healthcare costs. Compared with gastric surgery, pancreaticoduodenal surgery was higher in cost [SE, 1.034 (95% CI, 1.001–1.069)], and colorectal surgery was lower in cost [SE 0.918 (95% CI, 0.89–0.947)]. Blood transfusion either during or after surgery was significantly associated with increased expense [SE, 1.110 (95% CI, 1.068–1.154) and SE, 1.121 (95% CI, 1.068–1.175), respectively]. Patients who received PCA consisting of an opioid drug also had higher healthcare costs [SE, 1.084 (95% CI, 1.046–1.124)]. As expected, postoperative complications were also prominent factors that increased healthcare costs, but they did not reach statistical significance.

After adjusting for all these significant characteristics in the univariate analysis, primary PPOI was an independent risk factor for increased healthcare cost [SE, 1.046 (95% CI, 1.007–1.088)]. However, 4.6% of healthcare costs were due to primary PPOI.