Unmeasured anions as predictors of mortality in medical intensive care unit

Resumo

BACKGROUND AND OBJECTIVE: The present study aimedto evaluate the utility of variables that represent unmeasuredanions (anion gap, anion gap corrected for albumin, anion gapcorrected for albumin and lactate, base excess and modifiedbase excess) to predict mortality in medical intensive care unitpatients. METHODS: This prospective study included 156consecutivepatients admitted to a medical intensive care unitin a Medical school hospital between August 2006 and June2007. Serum levels of potassium, sodium, chloride, C-reactiveprotein, albumin, and lactate were measured. Variables that representunmeasured anions and APACHE II score were calculated. RESULTS: Among the studied patients, 60.9% were male,and mean age was 59.2±17.2 years. Mortality rate was 31.4%.Spearman’s test showed correlation among unmeasured anionsand lactate. Comparison between survivors and non-survivorsshowed differences in length of intensive care unit stay, APACHE II score, albumin, C-reactive protein, lactate, anion gap correctedfor albumin, base excess, and modified base excess. Anion gapcorrected for albumin, base excess, and modified base excess arepredictors of medical intensive care unit mortality; however,theirs areas under the ROC curves are smaller than APACHE II score and C-reactive protein. CONCLUSION: Variables thatestimate unmeasured anions, such as anion gap corrected foralbumin, base excess, and modified base excess, can be used topredict mortality in medical intensive care unit patients.