Answers are grounded in this paper and cite their sources.
Background Patients who reach the same Sequential Organ Failure Assessment (SOFA) score after 24 hours of intensive care may have different preceding score changes. We examined the prognostic information in these changes and their relationship to organ components, recording rules and early care in older adults with sepsis. Methods We analysed selected cohorts of adults aged ≥ 65 years who were alive and in the ICU at 24 hours in MIMIC-IV and eICU. Locally adapted SOFA scores were reconstructed in four six-hour windows. We compared recorded mortality across score-change groups and matched patients with a three-point decrease versus no change at the same attained score. Component changes, timestamp restrictions and adjustment for early sedative and ventilation records were evaluated. Models developed in MIMIC-IV were tested in eICU for ICU death during hours 24–72. Results The cohorts included 3,689 MIMIC-IV and 1,395 eICU stays. In MIMIC-IV, recorded mortality by day 365 was 16.3%, 37.4% and 42.3% in the decrease, net-stable and increase groups. Among 116 matched pairs, mortality was 22.4% versus 38.8% (paired difference, − 16.4 percentage points; 95% CI, − 27.6 to − 5.2; nominal P = 0.0094). In all 600 strict-comparison candidates, the odds ratio changed from 0.49 (95% CI, 0.32–0.76) to 0.86 (0.52–1.42) after adding early care records. Opposing component changes occurred in 50.3% of MIMIC-IV and 33.1% of eICU stays with zero total change. Timestamp restrictions changed broad classifications in 3.1% and 2.7%, respectively. Adding the first score to the attained-score model yielded an external AUC difference of − 0.0062 (95% CI, − 0.0191 to 0.0033). Conclusions Early SOFA decreases were associated with lower recorded mortality, including among older patients reaching the same attained score. This association varied with early care records, while unchanged totals frequently concealed opposing component changes. Interpretation of first-day SOFA change should include the component scores and the care context in which they were measured.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: