Design: Retrospective single-centre cohort
Population: 1,926 adults undergoing pancreatoduodenectomy, 2015–2023
Primary outcome: Clavien-Dindo ≥III complications and 90-day mortality
Effect: Severe frailty mortality OR 2.8 (mFI-5) to 4.1 (mFI-11); AUC 0.757 vs 0.761

This Verona cohort, published 23 September, analysed 1,926 adults undergoing pancreatoduodenectomy for periampullary lesions between 2015 and 2023, comparing the 11-item and 5-item modified Frailty Indices.

Major complications occurred in 18.2% and 90-day mortality was 2.7%. Severe frailty was independently associated with death on both indices (mFI-11 OR 4.1, 95% CI 1.6 to 10.3; mFI-5 OR 2.8, 1.4 to 5.6). Discrimination for mortality was good and nearly identical (AUC 0.761 vs 0.757), but only modest for major complications (about 0.63).

The five-item index takes a minute from the history. Frailty seemed to predict failure to rescue from complications more than complications themselves, which argues for using it to plan escalation and monitoring as well as to counsel.

Record the five-item modified Frailty Index before every pancreatoduodenectomy.
Use it in consent: severe frailty was associated with about three to four times the odds of 90-day death.
Plan closer post-operative monitoring and early escalation for severely frail patients.
Do not rely on frailty scores to predict complications — discrimination for morbidity was modest.

Why it matters

Moves frailty from a vague impression to a one-minute score that shapes consent and post-operative vigilance.

The statistics, in plain English

An AUC of 0.76 means the score ranks a patient who dies above one who survives about three times in four. With only 52 deaths, the odds ratios have wide confidence intervals.