Latest / Interprofessional Critical Care Network (ICCN) Podcast / Same patient, different SOFA: SOFA-2 versus the original organ failure score
Transcript
- 0:00SPEAKER_00: One patient scored twice, day one of septic shock,
- 0:03SPEAKER_00: norepinephrine at 0.15 micrograms per kilogram per minute with vasopressin added,
- 0:09SPEAKER_00: an antipsychotic started overnight for agitated delirium,
- 0:12SPEAKER_00: creatinine 1.1 milligrams per deciliter,
- 0:15SPEAKER_00: and urine output under 0.5 milliliters per kilogram per hour
- 0:18SPEAKER_00: for the last eight hours.
- 0:20SPEAKER_00: The original SOFA gives her cardiovascular row four points and her brain and kidneys nothing.
- 0:25SPEAKER_00: SOFA 2 gives the cardiovascular row three points and adds a point each
- 0:29SPEAKER_00: for the delirium drug and the low urine output.
- 0:32SPEAKER_00: Same patient, different SOFA.
- 0:35SPEAKER_00: This is the ICCN Daily Brief for Monday, November 9, 2026.
- 0:39SPEAKER_01: If you follow the show, each morning's brief is waiting for you.
- 0:42SPEAKER_01: Today's article looks at a 2026 study of more
- 0:46SPEAKER_01: than 134,000 adult ICU admissions scored both ways,
- 0:51SPEAKER_01: and at why the two versions disagree in ways that matter on rounds.
- 0:58SPEAKER_00: Why is there a SOFA to it all?
- 1:00SPEAKER_01: The original score had not been revised in 30 years,
- 1:03SPEAKER_01: and it was built before vasopressin, high-flow oxygen, non-invasive ventilation,
- 1:08SPEAKER_01: and ECMO were routine.
- 1:10SPEAKER_01: An international group combined a modified Delphi process with data from 1,319 ICUs
- 1:17SPEAKER_01: across nine countries, with an ICU mortality of 8.1%.
- 1:21SPEAKER_01: The six organ systems stay, with a total of 0 to 24.
- 1:26SPEAKER_01: What changed inside the rose?
- 1:28SPEAKER_01: SOFA-2 scores combine norepinephrine and epinephrine doses
- 1:31SPEAKER_01: and counts any second vasopressor, inotrope, or mechanical circulatory support.
- 1:37SPEAKER_01: It treats high-flow oxygen, CPAP, and noninvasive ventilation as advanced respiratory support.
- 1:43SPEAKER_01: Adds urine output windows of 6 to 24 hours and kidney replacement therapy criteria,
- 1:48SPEAKER_01: scores drug treatment for delirium, and moves the bilirubin and platelet threshold.
- 1:53SPEAKER_01: Across its development and validation cohorts,
- 1:55SPEAKER_01: it discriminated ICU mortality slightly better than the original,
- 1:59SPEAKER_01: an area under the curve of 0.79 against 0.77.
- 2:04SPEAKER_00: A slightly better average.
- 2:06SPEAKER_00: What did the anchor study add?
- 2:08SPEAKER_00: Ono and colleagues used one ICU, a Japanese registry from 15 hospitals,
- 2:14SPEAKER_00: with data at one-minute resolution.
- 2:16SPEAKER_01: They included adults admitted from April 2012 to December 2025,
- 2:21SPEAKER_01: removing patients younger than 18 years, stays shorter than 12 hours,
- 2:26SPEAKER_01: and admissions without a recorded ICU outcome,
- 2:29SPEAKER_01: which left 134,528 ICU admissions.
- 2:34SPEAKER_01: Both scores were calculated for the first ICU day by rules set before the analysis,
- 2:40SPEAKER_01: and a difference of two or more points in either direction counted as discordant.
- 2:45SPEAKER_01: Two mirror image analyses then matched patients exactly on one score
- 2:50SPEAKER_01: and compared ICU mortality across the three disagreement groups,
- 2:53SPEAKER_01: so only the other score differed.
- 2:56SPEAKER_01: Give me the headline numbers.
- 2:57SPEAKER_01: Mean day one totals were close, 5.64 for the original score and 5.81 for SOFA2.
- 3:04SPEAKER_01: Under SOFA2, the brain, respiratory, and cardiovascular rows mostly shifted upward,
- 3:10SPEAKER_01: liver and hemostasis followed the new thresholds,
- 3:13SPEAKER_01: and the kidney row moved in both directions.
- 3:16SPEAKER_01: Now, hold SOFA2 fixed.
- 3:18SPEAKER_01: Among patients matched on SOFA2, a mean of 7.41 in every group,
- 3:23SPEAKER_01: ICU mortality was 9% when the original score was at least 2 points higher,
- 3:29SPEAKER_01: 5.7% when the two roughly agreed, and 5% when SOFA 2 was at least 2 points higher.
- 3:35SPEAKER_01: And the mirror image?
- 3:37SPEAKER_01: Among patients matched on the original SOFA, a mean of 7.3 in every group,
- 3:42SPEAKER_01: ICU mortality was 3.4% when the original score ran higher,
- 3:47SPEAKER_01: 5% when the two roughly agreed, and 9.5% when SOFA2 ran higher.
- 3:54SPEAKER_01: Whichever score you hold fixed, the other one still separates patients.
- 3:58SPEAKER_01: Organ support drove much of the disagreement.
- 4:00SPEAKER_01: Back in the cohort matched on SOFA2,
- 4:03SPEAKER_01: half of the patients whose original score ran higher were on norepinephrine,
- 4:08SPEAKER_01: against 4.8% of those whose SOFA2 ran higher.
- 4:13SPEAKER_01: And kidney replacement therapy went the other way, 3.7% against 7.4%.
- 4:19SPEAKER_01: Do other cohorts agree?
- 4:21SPEAKER_01: They point the same way.
- 4:23SPEAKER_01: In 118,542 adults at three Chinese centers,
- 4:28SPEAKER_01: SOFA-2 discriminated ICU mortality better, 0.809 against 0.767.
- 4:35SPEAKER_01: And only 39.1% of patients had the same total under both versions.
- 4:40SPEAKER_01: In two Korean Emergency Department septic shock registries of 2,669
- 4:45SPEAKER_01: and 1,443 adults, SOFA-2 again discriminated in-hospital mortality better,
- 4:52SPEAKER_01: 0.734 against 0.702 and 0.719 against 0.671.
- 5:00SPEAKER_01: And SOFA-2 totals ran lower than the original in both.
- 5:03SPEAKER_00: What does this not tell us?
- 5:06SPEAKER_01: A better average is not a better bedside decision.
- 5:09SPEAKER_01: The gains are real and small, two to five hundredths of area under the curve,
- 5:15SPEAKER_01: and the anchors' authors note that a score does not need to predict death better
- 5:18SPEAKER_01: to describe organs better.
- 5:21SPEAKER_01: Neither version tells you what to do for the patient in front of you,
- 5:23SPEAKER_01: and neither should decide who receives care.
- 5:26SPEAKER_01: And the score itself?
- 5:28SPEAKER_01: Much of SOFA-2 now scores what we do to the patient.
- 5:31SPEAKER_01: Vasopressin, high-flow oxygen, non-invasive ventilation, ECMO,
- 5:36SPEAKER_01: and kidney replacement therapy all move points.
- 5:39SPEAKER_01: So the score can change because a treatment started or stopped, while the organ did not.
- 5:44SPEAKER_01: Kidney replacement therapy earns the maximum renal score for as long as it runs.
- 5:49SPEAKER_01: So improving kidney function can stay invisible until the machine comes off.
- 5:54SPEAKER_00: Has the sepsis definition moved?
- 5:56SPEAKER_01: Not yet. Sepsis 3, the 2016 consensus,
- 5:59SPEAKER_01: made a rise of two or more points the operational sign of organ dysfunction,
- 6:03SPEAKER_01: and it used the original score.
- 6:05SPEAKER_01: The same patient could cross a two-point rise under one version and not the other.
- 6:10SPEAKER_01: And none of the studies here measured how often that happens.
- 6:13SPEAKER_01: So units adopting SOFA2 for sepsis screening or quality reporting should check it
- 6:18SPEAKER_01: in their own data first.
- 6:20SPEAKER_01: The anchor is a retrospective analysis of a single country's registry.
- 6:25SPEAKER_01: Unmeasured components were scored as zero,
- 6:27SPEAKER_01: and it did not test whether either score changes decisions or outcomes.
- 6:32SPEAKER_01: The article also left out two anchor figures that differ between its text
- 6:35SPEAKER_01: and tables, and gives no study period for the Chinese cohort,
- 6:39SPEAKER_01: which reports two different start years.
- 6:41SPEAKER_00: Bedside.
- 6:43SPEAKER_00: What does each discipline own?
- 6:45SPEAKER_01: Name the version on rounds.
- 6:46SPEAKER_01: SOFA 2 of 8, Original SOFA 9 takes 4 seconds.
- 6:51SPEAKER_01: Pharmacists, report vasopressors as base and as a sum because SOFA 2 grades
- 6:56SPEAKER_01: the combined norepinephrine dose and adds a point for any second vasopressor
- 7:01SPEAKER_01: or inotrope.
- 7:02SPEAKER_01: Nurses, the hourly urine output and the delirium drug now count,
- 7:07SPEAKER_01: so keep urine output in clean 6-12 hour windows Record the last GCS before sedation
- 7:13SPEAKER_01: and make any drug given for delirium visible to whoever scores.
- 7:17SPEAKER_00: Respiratory Therapists and the Quality Teams
- 7:20SPEAKER_01: Document the support type alongside
- 7:22SPEAKER_01: the PaO2 to FiO2 ratio.
- 7:24SPEAKER_01: The SpO2 to FiO2 ratio counts only
- 7:27SPEAKER_01: when no blood gas is available and SpO2 is below 98%.
- 7:32SPEAKER_01: Quality in Research Teams, re-baseline before you compare because mixing versions
- 7:37SPEAKER_01: in one treadline will manufacture a change in severity that never happened,
- 7:42SPEAKER_01: and never use either version alone to decide escalation, limitation of treatment,
- 7:48SPEAKER_01: or eligibility for a scarce resource.
- 7:50SPEAKER_01: The teaching pearl?
- 7:52SPEAKER_01: SOFA II scores the treatments as well as the organs.
- 7:56SPEAKER_01: When the number jumps, check whether the patient changed or the plan did.
- 8:00SPEAKER_01: What changed?
- 8:02SPEAKER_01: SOFA2 is a better-fitted version of a score that defines sepsis,
- 8:05SPEAKER_01: grades organ dysfunction, and selects patients for trials.
- 8:09SPEAKER_01: Switching versions moved individual patients across categories,
- 8:13SPEAKER_01: mostly through how each version counts vasopressors, ventilation,
- 8:16SPEAKER_01: and kidney support.
- 8:18SPEAKER_01: And patients with the same score under one version still differed in ICU mortality
- 8:22SPEAKER_01: under the other.
- 8:24SPEAKER_01: Use SOFA-2 where it has been adopted.
- 8:26SPEAKER_01: Name the version every time and read a disagreement as a reason to look
- 8:30SPEAKER_01: at which organ and which treatment moved.
- 8:33SPEAKER_00: The full article with the bedside takeaways by discipline and the five things not
- 8:37SPEAKER_00: to over-assume is on the ICCN substack.
- 8:40SPEAKER_00: The link is in the show notes.
- 8:42SPEAKER_00: ICCN publishes one article a day for ICU clinicians.
- 8:45SPEAKER_00: $8 a month or $80 for the year with a seven-day free trial.
- 8:49SPEAKER_00: This has been the ICCN Daily Brief.
- 8:51SPEAKER_00: Back tomorrow at 6.30.