Inquiry Regarding Patient Extraction Using MIMIC-IV Data

[Yoshihiro-glitch]

Hi,

We are currently planning a study using MIMIC-IV data targeting patients diagnosed with sepsis at the time of ICU admission.

We would like to confirm whether the following approach would be appropriate to extract patients who were in sepsis at ICU admission. Additionally, if there are any recommended additional criteria or points of caution for this extraction method, we would greatly appreciate your guidance.

Proposed patient extraction method:

Inclusion criteria for target patients

Patients meeting Sepsis-3 determination within 48 hours after ICU admission (patients with delta SOFA ≥2 and suspected infection).

Sepsis-3 determination

Based on sepsis3.sql, we define the onset time as the first time when the SOFA score rises by 2 points or more from baseline (pre-admission or ICU admission) and infection is suspected (culture obtained + antimicrobial therapy administered).

SOFA score at ICU admission will be calculated using first_day_sofa.sql, assuming the pre-admission SOFA is 0.

Extraction of suspected infection

Using suspicion_of_infection.sql, we will identify cases in which both microbiological cultures and antimicrobial therapy are recorded within 48 hours of ICU admission.

We also plan to obtain sepsis diagnosis based on ICD codes for sensitivity analysis.

We would greatly appreciate your confirmation on whether this approach is appropriate and any additional recommendations for patient extraction.

Thank you very much for your support.