Matilde Røgilds Soltau

Abstract

Objective: To clarify if leucocyte count progression can predict one-week mortality in unselected medical patients admitted to the emergency department. Additionally, we investigated the importance of leucocyte count progression for admission to the intensive care unit and length of stay.

Method: This retrospective cohort study uses data collected in two phases from patients admitted through the emergency department at Hospital of South West Jutland. Upon admission, a nurse recorded the primary complaint and vital signs and registered these along with demographic information. Blood test results were subsequently extracted from the hospital’s computer systems. Patients were separated into three groups according to the development in their leucocyte count from the day of admission to the day after.

Results: The total cohort was 5894 patients; 30.5% remained after exclusion. Median age was 71, and 50.3% were female. Using logistic regression, we found significantly lower one-week mortality with falling leucocyte count progression, even when controlling for confounders.
A decreasing leucocyte count had a sensitivity for one-week mortality of 65%, specificity of 62%, positive predictive value of 4%, and negative predictive value of 99%.
Difference in admission to the intensive care unit was non-significant between the three groups. Difference in length of stay was significant, but with one day difference, the clinical significance is questionable.

Conclusion: Leucocyte count progression is not sensitive enough to predict one-week mortality, nor specific enough to discount it. It is important for physicians to be aware of this to avoid faulty assessments based on imprecise assumptions.