Do structural ICU requirements reflect critical care demand in elective metabolic and bariatric surgery? A retrospective cohort study.

Abstract

Background: Metabolic and bariatric surgery (MBS) is increasingly delivered in lower-acuity settings, yet on-site intensive care unit (ICU) capability remains a structural requirement for the highest center-designation tier. Direct data on critical-care utilization after elective MBS without an on-site ICU are limited. We describe 30-day outcomes at such a center, using contemporaneous elective laparoscopic cholecystectomy (LC) in the same environment as a reference.

Methods: Retrospective cohort study of consecutive adults undergoing elective MBS (sleeve gastrectomy or Roux-en-Y gastric bypass) or LC (2022-2025) at a single center without an on-site ICU but with a pre-defined escalation and transfer protocol. The primary outcome was the 30-day complication rate (Clavien-Dindo); ICU utilization was grade ≥IV. Outcomes are reported descriptively with 95% confidence intervals (CIs); regression and propensity-score-matched analyses are exploratory (Supplementary Material).

Results: Of 1,065 patients (MBS n=351, LC n=714), 30-day complication rates were 4.27% (95% CI 2.39-6.94%) versus 3.64% (95% CI 2.37-5.27%). No MBS patient was admitted to intensive care (0/351), compatible with a true rate up to 1.05% (upper 95% CI); two LC patients required ICU admission (0.28%), both for retained-stone acute pancreatitis. All five MBS reoperations were for postoperative hemorrhage; none developed organ dysfunction or met ICU-transfer criteria, and none died. Because events were rare, this study cannot exclude clinically meaningful ICU demand, and absence of a detected difference does not establish equivalence.

Conclusions: In this single-center experience, elective MBS without an on-site ICU under a structured escalation protocol showed no observed ICU admissions over 30 days, though the confidence interval remains compatible with infrequent critical-care need. These hypothesis-generating findings demonstrate feasibility in one environment and should not be generalized to accreditation policy; prospective multicenter data are required.

Study registration: Research Registry (UIN: researchregistry11829); retrospectively registered; https://www.researchregistry.com.