Table 7 Comparison of manual medical record review for identification of criterion-standard community-acquired sepsis and severe sepsis, and of Martin and colleagues’ and Angus and colleagues’ discharge diagnoses for sepsis identification [1 , 3 ]
From: Discharge diagnoses versus medical record review in the identification of community-acquired sepsis
Characteristic | Manual medical record review for identification of community-acquired sepsis and severe sepsis | Martin and colleagues’ and Angus and colleagues’ discharge diagnoses for identification of sepsis and severe sepsis |
|---|---|---|
Data source | Manual review of initial hospital records (emergency department and admission notes and laboratory test results from the first 28 hours of hospitalization) | Hospital discharge diagnosis codes |
Criteria for sepsis or severe sepsis | Sepsis [infection + ≥2 SIRS criteria] | Sepsis: ICD-9 discharge diagnoses for sepsis (Martin and colleagues; Additional file 1) |
Severe sepsis [sepsis + ≥1 SOFA organ dysfunction] | Severe sepsis: ICD-9 discharge diagnoses for [infection + organ dysfunction] (Angus and colleagues; Additional file 2) | |
Timeframe/horizon | 28 hours | Entire hospital stay |
Strengths | Based upon structured review of medical records | Can utilize existing hospital discharge data |
Focused on initial hospitalization (community-acquired sepsis) | ||
Verified connection between infection and sepsis (infection must be major reason for hospitalization) | ||
Extensive data on pre-existing comorbid conditions | ||
Limitations | Limited to the REGARDS cohort | Limited information on pre-existing comorbid conditions |
Requires manual review of medical records | Cannot differentiate initial (community-acquired) from later (hospital-acquired) sepsis | |
Limited to initial hospitalization presentation and records – cannot detect later (hospital-acquired) sepsis | Depends upon accuracy of coded discharge diagnoses | |
Assumes connection between coded infection and organ dysfunction (Angus and colleagues) |