The Impact of Surgeon Volume on Hysterectomy Risk: A Concentration Curve Analysis

Abstract

<h3>Study Objective</h3> To evaluate the relationship between surgeon volume and the outcomes of hospital readmission and return to the operating room in women who underwent hysterectomy. <h3>Design</h3> Retrospective cohort study <h3>Setting</h3> Multicenter community healthcare system <h3>Patients or Participants</h3> 2,730 women who underwent hysterectomy from July 1, 2015, to June 30, 2019. Patients of the resident practice and hysterectomies for obstetric indications were excluded. <h3>Interventions</h3> Surgeon volume greater than 25 hysterectomies per year. <h3>Measurements and Main Results</h3> A concentration curve was used to evaluate the impact of annual surgeon hysterectomy volume on 30-day all-cause readmission or return to the operating room. The concentration curve suggested a disproportionate number of complications attributed to lower volume surgeons. The peak difference between the curve and the line of equality was used to identify a surgeon volume of 25 hysterectomies per year as the cutoff between low- and high-volume surgeons. Demographics and outcomes were compared using Chi-squared and Mann-Whitney U tests where appropriate. Logistic regression was used to correct for confounders. Low-volume surgeons had a higher rate of reoperation or readmission within 30 days compared to those who underwent surgery by high-volume surgeons (4.9% vs 2.9%, <i>P</i>=0.02). After adjusting for confounders, the adjusted odds ratio was 1.67 (95% CI: 1.03 - 2.81). High-volume surgeons performed a greater percentage of laparoscopic hysterectomies (75% vs 45%), had a shorter length of stay (0.4 days vs 1.2 days, <i>P<</i>0.00001), and a greater total operating room cost ($9,827 vs $9,551, <i>P</i>=0.005). <h3>Conclusion</h3> High-volume surgeons performing more than 25 hysterectomies per year had lower rates of reoperation or readmission within 30 days compared to low-volume surgeons. Concentration curve analysis can be useful in determining if outcome differences can be seen based on volume, and helpful in selecting potential cutoffs to analyze surgeon groups by volume. Surgeon grouping is likely not bimodal, and differences are likely to be detectable at many different volume levels.