Abstract
Objective: To establish a score to predict late-onset sepsis in very low birth weight infants.
Study Design: Two successive prospective multi center studies in very low birth weight infants were used: infants in the first served as calculation cohort, those in the second as validation cohort. Infants were divided into a sepsis group, a non-sepsis group, and an undetermined group by predefined criteria partially based on the sepsis definition by the Robert Koch institute. In the first study 39 parameters recorded daily were compared in the sepsis and the non-sepsis group between Day -3 and Day 0 of sepsis. The score was created combining those parameters with the highest area under the ROC curve.
Results: A score was generated consisting of five clinical and laboratory categories: peripheral circulation, respiration, gastrointestinal tract, general laboratory findings and infection-specific laboratory findings. Sepsis was defined as pathological findings in at least 2 out of these 5 categories. This scoring system reached a sensitivity of 84.0% and a specificity of 91.7% (PPV 87.5% and NPV 89.2%) in discriminating the sepsis group from the non-sepsis group in the validation cohort.
Conclusion: We created a score for early diagnosis of late-onset sepsis in very low birth weight infants. The score may be of value when comparing infants with sepsis in multi center studies.
Get full access to this article
View all access options for this article.
