Hydration During Thoracic Surgery

Hospital Length of Stay; Adverse Events

The purpose of this retrospective study was to analyze the role of intraoperative crystalloid administration on postoperative hospital length of stay and on the incidence of previously reported adverse events following thoracic surgery.

The role of intraoperative crystalloid administration, when expressed as dose-response curves, on the duration of hospital length of stay (hLOS) and on previously reported incidences of surgical, cardiovascular, pulmonary, renal, other, and long-term AEs was evaluated with F-ratio statistics or with Chi-square tests set at a more stringent P\<.01 value for significance to reduce the incidence of false discovery rates. Loglinear variance was utilized to examine the role of postoperative hLOS residuals across the range of intraoperative crystalloid administration. Multivariable analysis screened those previously reported independent variables with P\<.1 with length of surgery, and the a posteriori variable of interest, rate of intraoperative crystalloid administration on duration of hLOS in a stepwise fashion with a maximum 5-fold R2 stopping rule, a cross validation technique that chooses the best model. The multivariable analysis was repeated with a regional instrumental variable based upon state residency to represent unmeasured confounders in this population that may have an association with the outcome of interest. Variance inflation factors calculations were conducted to determine the role of multicollinearity by the independent variables of interest. Sample size calculations for multivariable analysis require a minimum of 100 patients with ≥200 the preferred sample size. Recursive partitioning with 5-fold cross-validation, a measure of internal model validation, examined administration rates of intraoperative crystalloid administration by quantiles on hLOS and the role of intraoperative transfusion of pRBCs on hLOS. Tests to support the likelihood of causality of the intraoperative crystalloid administration on duration of hLOS was examined following analysis of seven likelihood tests; published findings consistent with prior research, biologic plausibility, dose-response relationship, strength of effect, and exclusions of effect-cause, confounding, and bias. The statistical program, JMP 13.2 (SAS Institute, Cary, NC) was utilized in this study.

No related papers found yet.