Retrospective Japan cohort links liver-resection drains to lower mortality

A nationwide retrospective cohort study of 73,613 adults undergoing hepatic resection for liver tumors in Japan found that prophylactic abdominal drain placement was not associated with fewer intra-abdominal infections, but it was associated with lower in-hospital mortality. After statistical adjustment, intra-abdominal infection occurred in 5.1% of patients who received a drain and 5.5% of those who did not, a difference that was not statistically significant. The adjusted death rate was 0.5% with drains and 1.0% without them. These observational results identify an association rather than establishing that drains caused the mortality difference, and the authors cautioned that the findings may not extend to highly complex cases in which drainage was considered nearly certain.

The analysis drew on Japan’s Diagnosis Procedure Combination database, a nationwide inpatient system that included more than 1,200 acute-care hospitals as of April 2024, including all 82 university hospitals. The study population comprised adults who underwent an initial hepatic resection without biliary reconstruction between January 2014 and March 2022. Patients were assigned to the drain group when a prophylactic intra-abdominal drain was placed on the day of surgery and to the no-drain group otherwise. Drain use was widespread: 62,934 patients, or 85.5% of the cohort, received a drain, while 10,679 did not. Procedures included major and minor resections performed through open, laparoscopic or robot-assisted approaches.

Before adjustment, the groups differed in several clinically relevant ways. Major hepatectomy was more common among patients who received drains, accounting for 20.5% of that group compared with 11.1% of the no-drain group. Minimally invasive surgery was more frequent among patients without drains, at 54.0% versus 31.7%. The drain group also had longer anesthesia and more intraoperative transfusions. To address these imbalances, the researchers used propensity score overlap weighting with patient, procedural and institutional variables that included age, sex, body mass index, smoking history, functional status, comorbidity, liver disease, tumor category, extent of resection, surgical approach, anesthesia time, transfusion, teaching-hospital status, annual hepatectomy volume and fiscal year. After weighting, all measured covariates were considered balanced using an absolute standardized difference threshold below 0.10.

The primary endpoint was postoperative intra-abdominal infection, defined for the study as bile leakage, intra-abdominal abscess or pancreatic fistula. In the weighted comparison, infection occurred in 5.1% of the drain group and 5.5% of the no-drain group. The estimated risk difference was −0.3 percentage points, with a 95% confidence interval of −0.8 to 0.1, and was not statistically significant. Bile leakage was also similar after weighting, at 1.0% with drains and 0.9% without them, with a risk difference of 0.1 percentage points and a 95% confidence interval of −0.07 to 0.3. Neither postoperative transfusion nor postoperative ventilation for at least two days differed significantly between the groups.

In-hospital mortality showed a different statistical pattern. The weighted death rate was 0.5% among patients with drains and 1.0% among those without them, corresponding to a risk difference of −0.6 percentage points and a 95% confidence interval of −0.8 to −0.4. A sensitivity analysis used generalized estimating equations to account for patient clustering within hospitals and included 72,071 patients treated at 1,147 hospitals. This analysis again found no significant association between drain placement and intra-abdominal infection, with an odds ratio of 0.96 and a 95% confidence interval of 0.87 to 1.06. By contrast, drain placement remained associated with lower in-hospital mortality, with an odds ratio of 0.50 and a 95% confidence interval of 0.39 to 0.63.

Drain placement was associated with a slightly longer postoperative hospital stay, but the difference was small. Weighted mean postoperative stays were 14.1 days for patients with drains and 13.8 days for those without them, a difference of 0.3 days with a 95% confidence interval of 0.008 to 0.7. Total hospitalization costs did not differ significantly: the weighted means were US$10,859 with drains and US$10,848 without them, producing a difference of −US$11 and a 95% confidence interval of −US$141 to US$165. The database also showed no significant difference in postoperative drainage, which was recorded in 1.0% of the weighted drain group and 1.1% of the weighted no-drain group. Thus, the study detected no clear association between prophylactic drains and either lower costs or a shorter postoperative stay.

Exploratory subgroup analyses suggested that the statistical patterns varied with patient characteristics and institutional context. Intra-abdominal infection was significantly lower in the drain group among patients treated at high-volume hospitals, but no significant difference was detected in the low- or middle-volume categories. Mortality was significantly lower with drains among patients with a body mass index below 18.5 kg/m² or from 18.5 to 24.9 kg/m², but not among those at 25.0 to 29.9 or at least 30.0 kg/m². Lower mortality was also reported across the categories of resection extent and surgical approach, as well as in low- and middle-volume hospitals, but not in high-volume hospitals. These analyses were exploratory, did not identify which patients should receive drains and cannot establish that drain placement produced the observed subgroup differences.

One possible explanation advanced by the authors is that a drain may furnish clinicians with information that helps them identify and manage severe postoperative complications earlier, even if the device does not prevent the complication itself. The paper notes that bile leakage can progress to peritonitis or intra-abdominal sepsis and that some refractory cases require additional surgery. Established monitoring and rescue systems at high-volume hospitals might make the additional information less consequential, while differences in postoperative monitoring resources at lower-volume hospitals could make drain output more useful. These are interpretations rather than demonstrated mechanisms: the database did not record whether drains altered monitoring or treatment, and it could not identify the specific causes of death.

Several limitations constrain interpretation of the mortality association. The retrospective design permits residual confounding, particularly because the reasons for drain placement and the decisions of individual surgeons were unavailable. Information was also missing on the duration and type of drain management, which could influence postoperative outcomes. The database did not capture pain, discomfort, delayed mobilization, cosmetic impairment or injury to vessels or organs during insertion. Although the Diagnosis Procedure Combination database generally records diagnoses with high specificity, its sensitivity varies, and the study-specific algorithms used to identify bile leakage, intra-abdominal abscess and other infections had not been individually validated against medical records. Complication underestimation or misclassification therefore remains possible.

The timing data were also incomplete. The database did not provide the exact intraoperative moment when a drain was placed, so the temporal sequence among drain placement, anesthesia duration and intraoperative transfusion could not be established. Hospital type and hepatectomy volume were included in the adjustment, but unmeasured institutional policies, surgeon preferences and perioperative systems may have influenced both drain use and patient outcomes. In addition, the overlap-weighted estimates primarily represent patients for whom either drainage strategy was reasonably plausible on the basis of measured characteristics. The findings may not generalize to severely complex operations, unusual bile-leakage risk or other situations in which a clinician would almost certainly place a drain. Overall, the study found no association between prophylactic drainage and reduced intra-abdominal infection among adults undergoing hepatic resection without biliary reconstruction, while lower mortality among drain recipients remains an observational signal that requires prospective testing rather than evidence of a causal survival benefit.

Subject of Research: Associations between prophylactic abdominal drains and short-term outcomes after liver resection in Japan

Article Title: Association of Prophylactic Abdominal Drain Placement With Short‐Term Postoperative Outcomes After Hepatectomy: A Nationwide Inpatient Database Study in Japan

Article References: Kouzu, K., Aso, S., Hirano, Y., Matsui, H., Yoshimura, K., Kitagawa, H., Kato, T., Yamane, N., Hagiwara, O., Miyoshi, N., Matsuda, S., Maruyama, H., Morikane, K., Sasaki, J., Yasunaga, H., Kitagawa, Y., Watanabe, M., & Tsujimoto, H. (2026). Association of Prophylactic Abdominal Drain Placement With Short‐Term Postoperative Outcomes After Hepatectomy: A Nationwide Inpatient Database Study in Japan. Annals of Gastroenterological Surgery, Article ags3.70299. https://doi.org/10.1002/ags3.70299

Image Credits: AI Generated

DOI: 10.1002/ags3.70299

Keywords: liver resection, abdominal drains, hepatectomy, surgical infection, bile leakage, in-hospital mortality, retrospective cohort, Japanese inpatient database, hospitalization costs, patient outcomes, Association, Prophylactic