Background: Non-steroidal anti-inflammatory drugs (NSAIDs) are used commonly for postoperative analgesia but can potentially impair healing. Their effect on pancreaticoduodenectomy (PD) outcomes is unknown. We sought to examine the impact of early postoperative NSAIDs on pancreatic fistula (PF) after PD.
Methods: We reviewed our prospective pancreatectomy database supplemented by medication administration records, including all PDs from 2002 to 2012. Primary outcome was occurrence of clinically significant (grade B-C) PF. Secondary outcomes included major morbidity (Clavien grade III-V) and 90-day mortality. Patients were compared based on early postoperative NSAID use (first 3 days following surgery) using univariate and multivariate analyses. Subgroup analyses were conducted based on NSAID type (COX-2 inhibitors and non-selective inhibitors).
Results: We included 251 PDs, of whom 127 (50.6%) patients received NSAIDs postoperatively (35.5% COX-2 inhibitors, 18.3% non-selective inhibitors, and 4.4% both). Use of any NSAIDs was associated with a non-significant increase in PF (16.5 vs 11.3%%; p = 0.23), and no difference in major morbidity and mortality. Use of non-selective inhibitors was not associated with an increase in PF (8.7 vs 15.1%; p = 0.256). COX-2 inhibitors were associated with increased PF (20.2 vs 10.5 %; p = 0.033), but no difference in major morbidity or mortality. After adjusting for Charlson comorbidity and estimated blood loss, use of COX-2 inhibitors was independently associated with PF (odds ratio 2.12; p = 0.044).
Conclusions: COX-2 inhibitors are associated with PF in the early postoperative period. While non-selective inhibitors appear safe in this setting, caution is warranted with the use of COX-2 inhibitors.