Development and validation of a prediction model for in-hospital mortality in patients with sepsis.
Wen Shi, Mengqi Xie, Enqiang Mao and 4 others
PMID 40189929WHAT IT FOUND
In 2389 ICU sepsis patients, a model predicted in-hospital mortality using infection source, malignancy, Gram-negative and Gram-positive bacteria, ventilation days, vasopressors, SOFA and APACHE II scores.
It was checked only in split hospital data, not in care.
Key findings
01Eleven independent predictors of in-hospital mortality were digestive tract infection (HR 0.431), urinary tract infection (HR 0.182), other infections (HR 0.634), malignancy (HR 1.648), Gram-negative bacterium (HR 0.616), Gram-positive bacterium (HR 0.533), days of mechanical ventilation (HR 0.737), vasopressor use (HR 1.680), minimum SOFA (HR 1.109), maximum SOFA (HR 1.101), and minimum APACHE II (HR 1.215).
02At 4 weeks, AUROC was 0.882 in the training set and 0.851 in the validation set; at 6 weeks it was 0.836 and 0.820; at 8 weeks it was 0.843 and 0.794.
03An online calculator was developed to input the 11 predictive variables and calculate survival probability and its 95% CI at 4, 6 and 8 weeks.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The paper first states 4435 sepsis patients were included, then states 2389 study subjects were ultimately identified after excluding 2041 patients and five with missing critical data. The data came from a single centre, and the authors say future multicentre studies are needed to validate stability and applicability. The study was retrospective, so the authors note missing data or measurement errors could affect predictive accuracy.
Declared interests
Funded by National Natural Science Foundation of China and Shanghai Shenkang Hospital Development Center.
The easy way to misread this
Do not read the model as proven to improve care or survival. It was developed from retrospective data at a single centre, and the authors say future multicentre studies are needed.