Crystalloid administration among patients undergoing liver surgery: Defining patient- and provider-level variation.

Background: Fluid administration among patients undergoing liver resection is a key aspect of perioperative care. We sought to examine practice patterns of crystalloid administration, as well as potential factors associated with receipt of crystalloid fluids.

Method: Patients who underwent liver resection between 2010 and 2014 were identified. Data on clinicopathologic variables, operative details, and perioperative fluid administration were collected and analyzed using univariable and multivariable analyses; variation in practice of crystalloid administration was presented as coefficient of variation (COV).

Results: Among 487 patients, median crystalloid administered at the time of surgery was 4,000 mL. After adjusting for body size and operative duration, median corrected crystalloid was 30.0 mL kg(-1) m(2) h(-1), corresponding with a COV of 35%. Patients who received <30 mL kg(-1) m(2) h(-1) crystalloids were more likely to be younger (58 vs 60 years), white (79% vs 74%), and have a higher body mass index (BMI; 28.2 vs 25.4 kg/m(2); all P < .001). On multivariable analysis, increasing Charlson comorbidity index, BMI, estimated blood loss, and each additional hour of surgery were all associated with increased crystalloid administration (all P < .05). Corrected crystalloid administration varied among providers with a corrected COV ranging from 14% to 61%. When overall variation in crystalloid administration was assessed, 80% of the variation occurred at the patient level, and 20% occurred at the provider level (surgeon, 3% vs anesthesiologist, 17%).

Conclusion: There was wide variability in crystalloid administration among patients undergoing liver resection. Although the majority of variation was attributable to patient factors, a large amount of residual variation was attributable to provider-level differences.