Effects of rapid fluid infusion on hemoglobin concentration: a systematic review and meta-analysis

Abstract

Abstract Background Rapid fluid administration may decrease hemoglobin concentration (Hb) by a diluting effect, which could limit the increase in oxygen delivery (DO2) expected with a positive response to fluid challenge in critically ill patients. Our aim was to quantify the decrease in Hb after rapid fluid administration. Methods Our protocol was registered in PROSPERO (CRD42020165146). We searched PubMed, the Cochrane Database, and Embase from inception until February 15, 2022. We selected studies that reported Hb before and after rapid fluid administration (bolus fluid given over less than 120 min) with crystalloids and/or colloids in adults. Exclusion criteria were studies that included bleeding patients, or used transfusions or extracorporeal circulation procedures. Studies were divided according to whether they involved non-acutely ill or acutely ill (surgical/trauma, sepsis, circulatory shock or severe hypovolemia, and mixed conditions) subjects. The mean Hb difference and, where reported, the DO2 difference before and after fluid administration were extracted. Meta-analyses were conducted to assess differences in Hb before and after rapid fluid administration in all subjects and across subgroups. Random-effect models, meta-regressions and subgroup analyses were performed for meta-analyses. Risk of bias was assessed using the Cochrane Risk of Bias Assessment Tool. Inconsistency among trial results was assessed using the I2 statistic. Results Sixty-five studies met our inclusion criteria (40 in non-acutely ill and 25 in acutely ill subjects), with a total of 2794 participants. Risk of bias was assessed as “low” for randomized controlled trials (RCTs) and ‘low to moderate’ for non-RCTs. Across 63 studies suitable for meta-analysis, the Hb decreased significantly by a mean of 1.33 g/dL [95% CI − 1.45 to − 1.12; p < 0.001; I2 = 96.88] after fluid administration: in non-acutely ill subjects, the mean decrease was 1.56 g/dL [95% CI − 1.69 to − 1.42; p < 0.001; I2 = 96.71] and in acutely ill patients 0.84 g/dL [95% CI − 1.03 to − 0.64; p = 0.033; I2 = 92.91]. The decrease in Hb was less marked in patients with sepsis than in other acutely ill patients. The DO2 decreased significantly in fluid non-responders with a significant decrease in Hb. Conclusions Hb decreased consistently after rapid fluid administration with moderate certainty of evidence. This effect may limit the positive effects of fluid challenges on DO2 and thus on tissue oxygenation.