intraoperative intravenous (IV) fluid replacement is of concern as a means to reduce incidences of post-operative acute kidney injury (AKI). The amount of IV fluids suffi- cient to prevent this outcome during pancreaticoduode- nectomy (PD) is unknown. Methods: Records are from an ongoing retrospective cohort of all patients who underwent PD procedure by the principal investigator between 2012 and 2017. We defined acute kidney injury (AKI) and AKI stage based on KDIGO AKI criteria (if 72 hours post-operatively patients had a serum creatinine that was 1.5 times greater than pre-operative creatinine or their creatinine increased by greater than or equal to 0.3). For reporting, we normalized the total intraoperative fluids by weight. Low, intermediate, and high fluid thresholds were defined based on the 25th and 75th percentile of the normalized total intraoperative fluids. Univariate and multi- variate analyses were performed to look for association be- tween pre- and intra-operative factors on AKI. Results: 275 patients with complete records were included in this analysis. There were 81, 118, and 76 patients who received less than 1200 mL (low), 1200 mL to less than 2250 mL (intermediate), and 2250 mL and greater (high) normalized total intraoperative fluids respectively. The median fluid amount administered for the low fluid, inter- mediate and high fluid categories were 810 ml, 1746 ml and 2993 ml. In these three groups the rate of postoperative AKI development was 18.2%, 22.2%, and 16.2% (p = 0.58). Compared to the low fluid group (2250 mL)(OR: 1.71, 95% CI: 0.70 e 4.14, p = 0.778). Higher rates of intraoperative hypotension were correlated with receiving more fluids intraoperatively but there was no significant differences among groups in age, BMI, pre-op chronic kidney disease by GFR, pre-op creatinine, post-op AKI, or post-op ileus. There was a difference in mean length of stay between groups, with on average those receiving more fluids staying longer (p = 0.03). There was no association on univariate or multivariate logistic regression for fluid level, BMI, or CKD at baseline on postoperative AKI rates. Conclusion: Our results suggest that concern for post- operative AKI is not a reason to increase volume of IV fluids given during pancreaticoduodenectomy. Compared to published literature where restrictive fluid amounts are considered to be less than 4000 mL, the IV fluids admin- istered to this cohort of patients can be considered to be in the ultra-restrictive range. Nevertheless, AKI rates did not change between our most restricted (2250 mL)) groups. P 32 EARLY DRAIN REMOVAL AFTER HEPATECTOMY: AN UNDERUTILIZED MANAGEMENT STRATEGY A. Fagenson * , E. Gleeson, A. Karachristos and H. A. Pitt *Corresponding author. Alexander Fagenson, Lewis Katz School of Medicine at Temple University Hospital, United States Background: Randomized controlled trials and a Cochrane Review suggest that routine drainage is not necessary in patients undergoing hepatectomy. Contem- porary ACS-NSQIP data suggest that 50% of patients un- dergoing hepatectomy in North America receive abdominal drains, and early drain removal is not widely practiced. Recent single center analyses from Japan demonstrates that drain removal by POD 2 or 3 is safe and may result in fewer organ space infections as well as shorter length of stay. Thus, the aim of this analysis is to compare the outcomes of North American patients who have early (POD 0-3) versus routine (POD 4-7) drain removal. Methods: Patients undergoing major hepatectomy ( 3 segments) or partial hepatectomy ( 2 segments) were identified in the 2014e16 ACS-NSQIP Procedure Targeted Participant Use File. Patients undergoing concomitant biliary reconstruction or colon resection were excluded. Only patients who had one or more drains placed at surgery were analyzed. Patients who had drains in place for more than 7 days also were excluded so that early drain removal (POD 0-3) could be compared to routine drain removal (POD 4-7) in patients without early bile leaks or other complications. Patients were stratified by extent of hepa- tectomy and propensity score matched for multiple de- mographic, comorbidity, laboratory, procedure and pathologic variables. Outcomes of early and routine drain removal patients were compared by Mann-Whiting U and chi-square test. Results: Of 2,599 patients, 1,654 (64%) underwent a par- tial (PH) and 945 (36%) had a major hepatectomy (MH). Early drain removal was performed in 661 (40%) of PH and 211 (22%) of MH patients. Multiple outcomes were significantly (p < 0.01) better in the early drain removal patients, but these patients were less likely (p < 0.01) to have hepatitis, a prior biliary stent, neoadjuvant therapy, a major hepatectomy, a Pringle maneuver, a concurrent ablation, perioperative transfusions, or malignant pathol- ogy. Therefore, 779 early drain removal patients were matched to 779 routine drain removal patients (Table). Early drain removal patients had significantly less (p < 0.02) overall and serious morbidity, fewer pneumo- nias and required fewer postoperative biliary drainage procedures. Length of stay was 2 days shorter, (p<0.001) and readmissions were 3% lower (p < .01) in early drain removal patients. Conclusion: Early drain removal (POD 0-3) is performed in only one-third of hepatectomy patients who receive drains. Routine drain removal (POD 4-7) is practiced in a variety of patients some of whom are perceived to be at higher risk for complications. After controlling for these factors, early drain removal is associated with lower overall and serious morbidity as well as shorter length of stay and fewer readmissions. Early drain removal after hepatecomy is an underutilized management strategy. HPB 2019, 21 (S1), S51eS91 S68 Mini Oral Abstracts