Clinical phenotypes of children undergoing adenotonsillectomy: a cluster analysis based on comorbidities and Its association with surgical strategy and length of stay.

OBJECTIVE

To identify clinical phenotypes among children undergoing surgery for adenotonsillar hypertrophy using unsupervised cluster analysis based on comorbidities, and to explore the association of phenotypes with obstructive sleep apnea (OSA) severity, surgical procedures, and length of hospital stay (LOS).

METHODS

This retrospective cohort study included 199 consecutive children who underwent adenoidectomy or adenotonsillectomy between August 2023 and April 2026. Nine binary comorbidity variables were extracted. K-modes clustering was used to partition patients into phenotypic groups. Demographic characteristics, OSA severity, surgical procedures, and LOS were compared. Multivariable negative binomial regression identified independent predictors of LOS.

RESULTS

Four phenotypes emerged: allergic-inflammatory (n = 48), sinusitis-dominant (n = 90), otitis-prone (n = 22), and hypertrophy-only (n = 39). Age, OSA prevalence, and severity varied significantly among clusters (P < 0.05). Tympanostomy tube insertion was concentrated in the otitis-prone cluster (86.4%, P < 0.001). LOS differed modestly (P = 0.043), with the otitis-prone cluster staying longest. Only surgical extent independently predicted LOS: standard T&A (IRR 1.27, 95% CI 1.07-1.51) and extended surgery (IRR 1.45, 95% CI 1.01-2.03).

CONCLUSIONS

Comorbidity-based clustering reveals four clinically meaningful phenotypes differing in age, OSA, and otologic surgery needs. LOS is driven by surgical complexity, yet phenotyping aids preoperative planning.