We prospectively investigated the effect of conventional resuscitation
on gastric intramucosal pH and lactate over 5 days in a group of pati
ents with newly diagnosed severe sepsis. Lactate and gastric intramuco
sal pH were measured on entry into the study, as soon as resuscitation
end points were met, eight hourly for 48 h and daily for 5 days. Sixt
een of 18 patients had a low gastric intramucosal pH (mean (SD) 7.17 (
0.12)) at the time of diagnosis of severe sepsis. At no time did gastr
ic intramucosal pH or lactate distinguish between shocked and nonshock
ed patients. Lactate distinguished survivors from nonsurvivors over ti
me (p = 0.02). Gastric intramucosal pH did not distinguish survivors f
rom nonsurvivors over time (p = 0.72). At 48 h lactate was lower in su
rvivors (p < 0.01) and gastric intramucosal pH higher in survivors (p
< 0.05). Receiver operating characteristic curves at this time indicat
e that lactate is a better predictor of survival. It is likely, based
on the inability of gastric intramucosal pH to distinguish survivors f
rom nonsurvivors until 48 h, that it is not possible to use this measu
rement to guide resuscitation in patients who are severely ill and who
have gastric intramucosal acidosis.