Predicting Late Anemia in Critical Illness
Author Information
Author(s): Eric B Milbrandt, Gilles Clermont, Javier Martinez, Alex Kersten, Malik T Rahim, Derek C Angus
Primary Institution: University of Pittsburgh School of Medicine
Hypothesis
Can a predictive model be constructed to identify ICU patients at risk of late anemia?
Conclusion
A decision support tool using early ICU admission data may enable timely and appropriate use of blood-sparing strategies.
Supporting Evidence
- The predictive model showed good discrimination with ROC curve areas ranging from 0.72 to 0.77.
- A five-item point system predicted late anemia effectively, allowing for earlier intervention.
- The study identified significant predictors of late anemia, including abnormal lactate levels and surgical admission.
Takeaway
Doctors can use information from the first few hours in the ICU to figure out which patients might need help with anemia later on.
Methodology
The study used logistic regression to develop predictive models based on demographic and physiologic variables from ICU patients.
Potential Biases
Potential biases may arise from relying on data from computerized databases that could miss unrecorded predictors.
Limitations
The study was retrospective and conducted at a single center, which may limit the generalizability of the findings.
Participant Demographics
The cohort included adult ICU patients with a mean age of 60.5 years, with 47.1% aged 65 or older.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website