Physician and Hospital-level Variation in Hemostatic Agent Use in Benign Gynecologic Procedures

Abstract

To identify recent nationwide trends in hemostatic agent (HA) use and to explore factors associated with HA use in 3 benign gynecologic surgery contexts: isolated hysterectomy, pelvic organ prolapse repair, and anti-incontinence surgery. Retrospective cohort study. Vizient Clinical Database. Three cohorts of female patients of ≥18 years who underwent benign isolated hysterectomy, pelvic organ prolapse repair, or anti-incontinence procedures were identified between October 2015 and December 2019. HAs are topically applied procoagulant products used for surgical hemostasis and use during included encounters was determined by charge codes. Subject-, hospital-, and surgeon-level characteristics and costs were captured. Data were initially analyzed in the aggregate and based on procedure category using the chi-square test or independent samples t tests as appropriate. A bootstrap forest model was used to identify the factors most predictive of HA use. In the final cohort of 184 070 encounters, HAs were used most frequently in hysterectomy (20.7%) and least in anti-incontinence surgery (10.9%). The use of HAs increased from 15.6% in quarter 4 2015 to 19.2% in quarter 4 2019 (p <.001). Encounters using HAs cost more than encounters without HAs ($6271.10 vs $4572.00; p <.001). A bootstrap forest model inclusive of all variables found surgeon and hospital identity cumulatively predictive of 84.9% of HA use, 65.5% and 19.4%, respectively. There was significant variation in HA use among individual surgeons, with 59.9% never using HAs. Of those who did use HAs, 72.8% used HAs more frequently than the mean provider HA use rate (19.4%) and 9.2% used HAs in every case he/she performed. The significant variation in HA use is driven primarily by physician and hospital identity, suggesting that use of HA in these benign gynecologic surgical contexts may be determined more by physician- and hospital-level factors than patient-level factors.