Plasma Ionized Calcium and Perioperative Bleeding in Cardiac Surgery.

PERIOPERATIVE TIME STAMPS: The perioperative phase is divided into the following time periods: the operating room phase before cardiopulmonary bypass (pre-CPB), cardiopulmonary bypass (on-CPB), the operating room phase after cardiopulmonary bypass (post-CPB), 0-6 hours in the intensive care unit (ICU-0-6h), 6-12 hours in the intensive care unit (ICU-6-12h), and 12-24 hours in the intensive care unit (ICU-12-24h). If the patient is transferred from the intensive care unit to the ward earlier than 24 hours after surgery, the final time period ends at the time of transfer.

PRIMARY ANALYSIS: The association between time-weighted mean plasma ionized calcium concentration during the post-CPB phase as a continuous variable and dichotomous UDPB (UDPB 3-4 vs. UDPB 0-2) in patients who did not receive blood products during the post-CPB phase. SECONDARY ANALYSIS: The association between time-weighted mean plasma ionized calcium concentration during the post-CPB phase as a continuous variable and the original five-grade UDPB in patients who did not receive blood products during the post-CPB phase. SENSITIVITY ANALYSES: The corresponding analyses to the primary and secondary analyses including all patients.

STATISTICAL ANALYSES: Differences in continuous variables between two groups will be tested using Student's t-test, and correlations between continuous variables will be assessed using Pearson's correlation test. The association between two dichotomous variables will be analyzed using Pearson's chi-square test. Temporal changes in plasma ionized calcium concentration across the entire study population will be analyzed using one-way analysis of variance (ANOVA). Post hoc comparisons will be performed using Dunnett's test, in which perioperative plasma ionized calcium concentrations will be compared with pre-CPB values. The association between plasma ionized calcium and UDPB class will be assessed with regression models. In addition to the primary, secondary and sensitivity analyses, corresponding multivariable analyses will be performed adjusted for clinical risk factors for bleeding. The clinical risk factors included in the model will be those previously identified by the research group among cardiac surgery patients at Meilahti Hospital. Multicollinearity among explanatory variables in the multivariable models will be considered present if the variance inflation factor (VIF) exceeds 5. In the presence of multicollinearity, the variable considered more clinically relevant for the model will be prioritized, and secondarily, the variable showing the stronger univariable association with the outcome will be selected. Otherwise, all predefined explanatory variables will be included in the multivariable model regardless of their statistical significance in univariable analyses. If needed, the five-grade UDPB will be modified to meet the assumption for proportional odds in ordinal regression models. Linearity will be tested. If needed, non-linear modeling will be used. The extent of missing data is not known at the time of writing the study protocol. The need for imputation of missing data and the specific imputation methodology will be evaluated after the study dataset has been obtained. Results will be presented as means and standard deviations, frequencies and proportions, or odds ratios with 95% confidence intervals (CI). A two-sided p-value < 0.05 will be considered statistically significant.