Share on Facebook
Share on Twitter
Share on LinkedIn
Share by Email
Print
The use of a more contextual, personalized, probabilistic, outcomes-based approach for interpretation of laboratory values in patients in a hospital intensive care unit (ICU) may be beneficial, particularly because the range of these values tends to deviate significantly from the range in healthy individuals.
A cross-sectional study of a large critical care database was conducted and the results were published in JAMA Network Open.
Investigators sought to comprehend how the distribution of ICU laboratory values differs from the reference range and the ways in which these distributions are linked to patient outcomes. They evaluated the Medical Information Mart for Intensive Care database, which contains data from a large tertiary care medical center in Boston, Massachusetts, between January 1, 2001, and October 31, 2012. The data were collected from medical, surgical, neurologic, and cardiac ICUs. Common laboratory measurements were sampled, with the analysis performed from March to June 2017.
A total of 38,605 patients from the ICU were assessed. The mean patient age was 74.5 years. Overall, 56.6% (21,852 of 38,605) of the patients were men. Of the patients analyzed, 23% (8878 of 38,605) experienced the best outcome (ie, ICU survival, shortest quartile length of stay). In comparison, 8% (3090 of 38,605) experienced the worst outcome (ie, ICU nonsurvival).
The distribution curves based on ICU data were shown to differ significantly from the hospital standard range (mean overlapping coefficient, 0.51). All of the laboratory values for the best outcomes group differed significantly from those in the worst outcomes group, with both the best and worst outcome group curves displaying little overlap with and marked deviation from the reference range.
A limitation of the analysis was that it was a single-center study from one ICU database. In addition, the approach does not immediately lend itself to direct clinical translation nor did it explore interactions with other factors, such as a patient’s specific pathology (or pathologies) and his or her underlying comborbidities.
“This study represents a conceptual approach toward using a more personalized and outcome-based interpretation of data in a designated patient population,” the investigators concluded.
“We believe that the introduction of such an approach to clinical care and electronic health record design will prove to be beneficial, as the clinicians who manage data entry of medical records and who are accountable for patient outcomes must analyze, and make decisions based on, increasing amounts and types of clinical data.”
Reference
Tyler PD, Du H, Feng M, et al. Assessment of intensive care unit laboratory values that differ from reference ranges and association with patient mortality and length of stay. JAMA Netw Open. 2018;1(7):e184521.
