Health Policy & Community Health: Faculty Publications

Respiratory-support phenotypes and in-hospital mortality in pediatric postprocedural respiratory complications

Document Type

Article

Publication Date

9-1-2026

Publication Title

Healthcare

DOI

10.3390/healthcare14172805

Abstract

Background/Objectives: Postprocedural respiratory complications encompass heterogeneous patterns of invasive respiratory support. We examined whether hospital-course respiratory-support phenotypes were associated with in-hospital mortality among operative pediatric discharges. Methods: We analyzed the 2022 Healthcare Cost and Utilization Project Kids' Inpatient Database. Discharges with Clinical Classifications Software Refined category RSP017 were restricted to those with a major operating-room procedure. AHRQ Pediatric Quality Indicator 09 timing criteria anchored postoperative mechanical ventilation and intubation. Survey-weighted logistic regression accounted for KID strata, hospital clustering, and discharge weights, and adjusted for age, sex, multisystem complication involvement, AHRQ nonweighted comorbidity burden, and procedure domain. Results: Of 7378 operative RSP017 discharges, 6358 were aged $17 years and 6334 entered the final model, with 216 deaths. Survey-weighted mortality was 35.46% with ECMO support, 7.35% with postoperative 24-96 h ventilation/intubation, 4.71% with postoperative ventilation > 96 h, 4.11% with other invasive support, and 0.73% in the reference phenotype. Corresponding adjusted odds ratios were 60.12, 8.76, 5.28, and 4.85. Before hierarchical assignment, 350/6358 discharges (5.50%) met more than one candidate phenotype definition. Alternative hierarchy and operative-anchor analyses yielded similar associations. When the combined phenotype was disaggregated, qualifying postoperative intubation had a stronger association with mortality than 24-96 h ventilation alone. Conclusions: Administrative respiratory-support phenotypes were associated with markedly different in-hospital mortality. Findings were robust to alternative classification and timing rules, but the framework is hypothesis-generating and requires external clinical validation.

Copyright

See the publisher's copyright and reuse policies.

Link to Full Text

Share

COinS