Background: The computation of fluid balances (FBs) by subtracting fluid outputs from inputs is a common critical care practice. Limited information exists about the accuracy and consistency of nurse-registered cumulative FBs and regarding the value of suggested corrections for non-measurable losses.
Methods: From 147 ICU patients, we prospectively evaluated the cumulative FBs and their relationship to changes in body weight (BW). Standardised measurements of BW were performed on admission and discharge. FB charts were accurately reviewed and arithmetic errors corrected. Net cumulative FBs and adjusted cumulative FBs (considering sensible or insensible fluid losses/fever/liquid faeces) were analysed for all patients and 3 subgroups (cardiac-cerebral, septic, and others). Agreement between FBs and BW changes was calculated according to the defined subgroups and confounding variables.
Results: Cumulative FBs were inaccurate in 49 cases (33%) with errors ranging from -3606 mL to +2020 mL. The total (average daily) difference between measured BW and FBs (mean ± SD) was 0.185±1.874 kg (0.101±1.020 kg). Correlation (r(2)) and Bland-Altman agreement was poor between BW changes and net cumulative FBs (0.552 and -1.26±5.41 kg) and slightly better between BW changes and adjusted cumulative FBs (0.714 and +0.18±3.68 kg). Standard deviations of the average daily differences between BW changes and FBs were always >1 L. Correction of the net FBs as suggested in the literature were not useful. New multiple regression models only modestly improved correlation.
Conclusion: For a large portion of patients nurse-registered cumulative FBs are neither accurate nor do they agree with standardised BW measurements. Patient care and clinical decision-making should be based on more objective techniques.