Pancreaticoduodenectomy, often called the Whipple procedure, is one of the most technically demanding operations in gastrointestinal surgery. Common complications include postoperative pancreatic fistula, intra-abdominal abscess, delayed gastric emptying, and postoperative hemorrhage, any of which can prolong hospitalization or lead to death. Because outcomes depend on a patient’s overall health, perioperative management, the condition of the pancreas, and surgical technique, researchers have long sought to identify which preoperative factors independently worsen results.

Chronic steroid therapy is widely prescribed for autoimmune diseases, inflammatory disorders, allergic conditions, and immunosuppression after organ transplantation. While essential for managing these diseases, steroids can cause central obesity, adrenal insufficiency, electrolyte imbalance, osteoporosis, and diabetes, contributing to frailty and a higher risk of postoperative complications. Previous studies in colorectal and gastric surgery have linked preoperative steroid use to anastomotic leakage and wound infections, but evidence in pancreatic resection has remained limited, with earlier single-center studies including too few steroid-treated patients to draw firm conclusions or to adjust adequately for confounding factors.

To address this gap, researchers turned to the NCD, a nationwide registry established in 2011 that covers more than 95% of surgeries performed across Japan, with over 5,500 participating institutions and more than 14 million registered cases over the past decade. The team identified 47,155 patients who underwent pancreaticoduodenectomy between 2019 and 2022, a period chosen because registration formats and data specifications were fully standardized. After excluding 293 emergency operations, 5,970 procedures involving combined resection of other organs, and 14 patients with missing data, the final analytical cohort comprised 40,878 patients. Of these, 521 patients, or 1.3%, were classified as chronic steroid users.

The study defined preoperative chronic steroid use as regular oral or intravenous administration of medications such as prednisolone or dexamethasone within 30 days before surgery for the treatment of chronic diseases. Topical, inhaled, and rectal steroid use were excluded, as were short courses of 10 days or less. The outcomes examined were postoperative pancreatic fistula of Grade B or greater under the 2016 International Study Group of Pancreatic Fistula criteria, severe complications of Grade IIIb or greater under the Clavien-Dindo classification, a composite of infectious complications including surgical site infections, pneumonia, urinary tract infection, and sepsis, re-operation, and surgery-related mortality, defined as death within 30 days or in-hospital death up to 90 days after surgery.

The baseline comparison revealed striking differences between the groups. Postoperative pancreatic fistula occurred in 30.7% of steroid users versus 22.1% of non-users. Severe complications affected 7.7% versus 4.4%, and infectious complications 28.4% versus 17.9%. Most notably, surgery-related mortality was more than three times higher among steroid users, at 4.6% compared with 1.5%. Steroid users were also more likely to be elderly, to have diabetes mellitus, dyspnea, and impaired activities of daily living, and to show laboratory abnormalities such as elevated white blood cell counts, low hemoglobin, low albumin, and reduced kidney function. Nearly 28% of steroid users had an American Society of Anesthesiologists Physical Status grade of 3 or higher, compared with about 15% of non-users.

Because these crude differences could simply reflect the poorer baseline health of steroid-treated patients, the researchers built three nested mixed-effects logistic regression models with a random intercept for each hospital to account for clustering. Model 1 adjusted for age, sex, and facility-level effects. Model 2 added body mass index and fourteen comorbidities, including diabetes, hypertension, cardiac disease, cerebrovascular disease, and body weight loss. Model 3 further added laboratory values and physical status. Encephalopathy was excluded as a covariate because no steroid-group patients had this condition. Rather than relying on a single final model, the team tracked how risk estimates shifted across the adjustment levels to test the robustness of the association.

The adjusted results were consistent. The odds ratio for postoperative pancreatic fistula rose slightly from 1.71 in Model 1 to 1.80 in Model 3, with confidence intervals excluding 1 throughout. Severe complications showed odds ratios of 1.96, 1.79, and 1.71 across the three models, remaining significant at every level. Infectious complications were similarly stable, with odds ratios of 1.87 in Model 1 and 1.79 in Model 3. The strongest association was with surgery-related mortality: the odds ratio of 3.61 in Model 1 decreased to 3.14 after comorbidity adjustment and remained markedly elevated at 2.86 in the fully adjusted model. In contrast, re-operation showed only slightly elevated point estimates that never reached statistical significance in any model.

The authors contrast their findings with a previous NSQIP study by Aziz and colleagues, which evaluated preoperative steroid use in pancreaticoduodenectomy but found no significant associations with outcomes. The current cohort was substantially larger, 40,878 versus 5,976 patients, providing greater statistical power. Interestingly, the prevalence of steroid use was markedly lower in the Japanese database, 1.3% versus 10.4% in NSQIP, suggesting that Japanese patients requiring preoperative steroids may represent a distinct, highly vulnerable subgroup. The authors also distinguish their work from a related NCD-based study by Mizuma and colleagues, which focused on building an overall prediction model for pancreatic fistula, whereas the present analysis specifically evaluated the association of chronic steroid therapy across multiple short-term outcomes while deliberately avoiding potential intermediate variables on the causal pathway.

Biologically, the authors note that steroids impair collagen synthesis, suppress the early inflammatory responses needed for tissue repair, and compromise systemic immune function, which may explain the persistently high rates of adverse events. The study has limitations: the database did not record the primary disease requiring steroid therapy, the exact daily or cumulative dose, the duration of therapy, or concomitant immunosuppressive medications, so the steroid-exposed group was heterogeneous. Pancreas-specific risk factors for fistula, such as pancreatic texture and main pancreatic duct diameter, are not routinely captured. As a retrospective analysis, the study demonstrates associations rather than proving causation. The authors conclude that surgical teams should approach chronic steroid users undergoing pancreaticoduodenectomy with strict risk awareness and meticulous perioperative management, and that the objective risk estimates can help clinicians give patients a clearer understanding of perioperative risks during informed consent.

Subject of Research: Association between preoperative chronic steroid use and short-term outcomes after pancreaticoduodenectomy in a nationwide Japanese database cohort

Article Title: Association Between Preoperative Steroid Use and Short‐Term Outcomes Following Pancreaticoduodenectomy: A Nationwide Analysis Using the National Clinical Database

Article References: Tomimaru, Y., Endo, H., Yamamoto, H., Ishigame, T., Ueno, H., Shirabe, K., & Eguchi, H. (2026). Association Between Preoperative Steroid Use and Short‐Term Outcomes Following Pancreaticoduodenectomy: A Nationwide Analysis Using the National Clinical Database. Annals of Gastroenterological Surgery, Article ags3.70286. https://doi.org/10.1002/ags3.70286

Image Credits: AI Generated

DOI: 10.1002/ags3.70286

Keywords: pancreaticoduodenectomy, preoperative steroid use, postoperative pancreatic fistula, National Clinical Database, surgery-related mortality, severe complications, infectious complications, Clavien-Dindo classification, retrospective cohort study, Japan, perioperative risk, pancreatic surgery