Mechanical and oral antibiotics bowel preparation for elective rectal cancer surgery: A propensity score matching analysis using a nationwide inpatient database in Japan

2.1 Data source

In this nationwide retrospective study, we used the DPC database from January 2014 to December 2017. It contains discharge abstracts and administrative reimbursement claim data from inpatient cases collected at participating hospitals, and it has been used in various studies^{16, 17} The data were collected by the DPC Research Institute (a nonprofit organization) in collaboration with the Ministry of Health, Labor, and Welfare of Japan. The database includes the following data: disease names, hospitalization costs, comorbidities at admission and during hospitalization, coded according to the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10), age, sex, length of hospital stay, medical procedures including surgery, names and quantities of medicines administered, and discharge status (including in-hospital deaths).¹⁸ Medical procedures are indexed with a Japanese code (K-code),¹⁹ assigned by the Ministry of Health, Labor, and Welfare of Japan. This study was approved by the Ethics Committee for Medical Care and Research at the University of Occupational and Environmental Health Japan (R02-007).

2.2 Patient selection

Inclusion criteria were as follows: inpatient status and admission for rectal cancer (ICD-10 code: C20), and underwent low anterior resection (LAR) (K-code: K7402, K740-22). Exclusion criteria were bowel obstruction on admission, emergency admission and incomplete data. Patients who were given polyethylene glycol (PEG) or magnesium citrate before LAR comprised the MBP group. Patients who were given metronidazole and kanamycin before LAR comprised the OABP group. Patients who were given neither MBP nor OABP comprised the no bowel preparation (NBP) group. Patients who were given both MBP and OABP comprised the MOABP group (Figure 1).

2.3 Endpoints

Endpoints were as follows: in-hospital mortality, 30-d postoperative mortality, necessity of reoperation with general anesthesia, rate of surgical wound dehiscence, rate of anastomotic leakage, rate of postoperative small bowel obstruction needing a trans-nasal long decompression tube, necessity of perioperative transfusion, rate of CD colitis, rate of MRSA colitis, length of hospital stay (LOS), duration of postoperative antibiotic administration and total hospitalization cost.

2.4 Statistical analysis

We conducted propensity score-matched (PSM) analysis to compare the efficacy of MOABP with that of MBP. We used the following variables for PSM: sex, age, body mass index (BMI), clinical TNM stage, rate of diverting stoma, rate of laparoscopic surgery, smoking, the use of antidiabetic drugs, oral corticosteroid drugs, oral antiplatelet drugs, and oral anticoagulant drugs in the admission and hospital volume. The hospital volume, which was used as a hospital-level factor, was defined as the annual number of LAR executed at each facility and categorized into quartiles (low volume: <54 cases, medium volume: 55–95 cases, high volume: 96–147cases, and very high volume: 148–581 cases per 4 y).

Nearest-neighbor matching was employed with a caliper width equal to 0.2 of the standard deviation of the logit of propensity scores. We subsequently compared in-hospital mortality, 30-d postoperative mortality, necessity of reoperation with general anesthesia, rate of surgical wound dehiscence, rate of anastomotic leakage, rate of postoperative small bowel obstruction needing trans-nasal long decompression tube, necessity of perioperative transfusion, rate of CD colitis, rate of MRSA colitis. Using chi-square tests, and LOS, duration of postoperative antibiotic administration and total hospitalization cost, using the Mann–Whitney U-test.

Stata Version 15.0 (Stata Corp, College Station, TX, USA) was used for all statistical analyses. A P < 0.05 was considered statistically significant.

3.1 The current status of bowel preparation

During the study period, 51 010 patients who underwent low anterior resection for rectal cancer were registered in the DPC database. Patients who had bowel obstruction on admission (1934 patients), or who admitted emergently (4450 patients), or who lacked sufficient data (7335 patients) were excluded, resulting in inclusion of 37 291 patients (1159 hospitals). In all, 37 291 patients were divided into four groups (NBP: 6301 patients, OABP: 248 patients, MBP: 28986 patients, MOABP: 1756 patients) (Figure 1). The ratio of the four groups is shown in Figure 2. The MBP group accounted for 77.7%. The other groups were as follows: The NBP group 16.9%, the OABP group 0.7%, and the MOABP group 4.7%.

3.2 Patient characteristics

Patient characteristics of each group are shown in Table 1. The NBP group tended to include advanced clinical TNM stage (Stages III or IV) patients and patients who underwent LAR at very high-volume hospitals. The MOABP group tended to be higher for the laparoscopic surgery rate than other groups and included patients who underwent LAR at high or very high-volume institutions.

3.3 Postoperative outcomes before propensity score matching

Table 2 shows the postoperative outcomes in each group. In-hospital mortality (0.48%), 30-d postoperative mortality (0.33%), necessity of reoperation with general anesthesia (6.89%), rate of surgical wound dehiscence (2.16%), rate of anastomotic leakage (9.67%), and necessity of perioperative transfusion (7.28%) were the highest in the NBP group among the groups. The mean of LOS (24.1 d) was the longest in the NBP group. On the other hand, in-hospital mortality (0.00%), 30-d postoperative mortality (0.00%), necessity of reoperation with general anesthesia (3.76%), rate of surgical wound dehiscence (0.85%), rate of anastomotic leakage, (4.84%), rate of postoperative small bowel obstruction needing a trans-nasal long decompression tube (1.54%), and necessity of perioperative transfusion (5.81%) were the lowest in the MOABP group among the groups. In addition, the mean of LOS (22.3 d) was the shortest in the MOABP group.

3.4 Comparison of postoperative outcomes between MOABP and MBP after propensity score matching

The patient characteristics after PSM are summarized in Table 3. A total of 1756 pairs were matched (Figure 1). We confirmed that the characteristics of both groups were similar after PSM. The postoperative outcomes after PSM are shown in Table 4. The rate of anastomotic leakage was significantly lower in the MOABP group than in the MBP group (4.84% vs 7.86%, P < 0.001). Also, the rate of postoperative small bowel obstruction needing a trans-nasal long decompression tube (1.54% vs 3.08%, P = 0.002) and the necessity of reoperation with general anesthesia (3.76% vs 5.98%, P = 0.002) were significantly lower in the MOABP group than in the MBP group. The duration of postoperative antibiotics medication was significantly shorter in the MOABP group than in the MBP group (mean 5.2 d vs 7.5 d, P < 0.001). No significant differences were found between the MOABP and the MBP group in terms of the rate of CD colitis (0.40% vs 0.68%, P = 0.250) and the rate of MRSA colitis (0.11% vs 0.17%, P = 0.654). There were no significant differences in the in-hospital mortality, 30-d postoperative mortality, the rate of surgical wound dehiscence, necessity of perioperative transfusion, LOS, and total hospitalization cost between the MOABP groups and the MBP groups.