PURPOSE: Fluid resuscitation is a key intervention in patients with sepsis and circulatory impairment. The recommendations for continued fluid therapy in sepsis are vague, which may result in differences in clinical practice. We aimed to evaluate associations between hospital and patient characteristics and fluid resuscitation volumes in ICU patients with severe sepsis.

METHODS: We explored the 6S trial database of ICU patients with severe sepsis needing fluid resuscitation randomised to hydroxyethyl starch 130/0.42 vs. Ringer's acetate. Our primary outcome measure was fluid resuscitation volume and secondary outcome total fluid input administered from 24 hours before randomisation until the end of day 3 post-randomisation. We performed multivariate analyses with hospital and patient baseline characteristics as covariates to assess associations with fluid volumes given.

RESULTS: We included 654 patients who were in the ICU for 3 days and had fluid volumes available. Individual trial sites administered significantly different volumes of fluid resuscitation and total fluid input after adjusting for baseline variables (P<0.001). Increased lactate, higher cardiovascular and renal SOFA subscores, lower respiratory SOFA subscore and surgery were all independently associated with increased fluid resuscitation volumes.

CONCLUSIONS: Hospital characteristics adjusted for patient baseline values were associated with differences in fluid resuscitation volumes given in the first 3 days of severe sepsis. The data indicate variations in clinical practice not explained by patient characteristics emphasizing the need for RCTs assessing fluid resuscitation volumes fluid in patients with sepsis.