Bellwether Procedures for Monitoring Subnational Variation of All-cause Perioperative Mortality in Brazil

All-cause perioperative mortality rate (POMR) is a commonly reported metric to assess surgical quality. Benchmarking POMR remains difficult due to differences in surgical volume and case mix combined with the burden of reporting and leveraging this complex and high-volume data. We seek to determine whether the pooled and individual procedure POMR of each bellwether (cesarean section, laparotomy, management of open fracture) correlate with state-level all-cause POMR in the interest of identifying benchmark procedures that can be used to make standardized regional comparisons of surgical quality.

The Brazilian National Healthcare Database (DATASUS) was queried to identify unadjusted all-cause POMR for all patient admissions among public hospitals in Brazil in 2018. Bellwether procedures were identified as any procedure involving laparotomy, cesarean section, or treatment of open long bone fracture and then classified as emergent or elective. The pooled POMR of all bellwether procedures as well as for each individual bellwether procedure was compared with the all-cause POMR in each of the 26 states, and one federal district and correlations were calculated. Funnel plots were used to compare surgical volume to perioperative mortality for each bellwether procedure.

4,756,642 surgical procedures were reported to DATASUS in 2018: 237,727 emergent procedures requiring laparotomy, 852,821 emergent cesarean sections, and 210,657 open, long bone fracture repairs. Pooled perioperative mortality for all of the bellwether procedures was correlated with all-procedure POMR among states (r = 0.77, p < 0.001). POMR for emergency procedures (2.4%) correlated with the all-procedure (emergent and elective) POMR (1.6%, r = 0.93, p < .001), while POMR for elective procedures (0.4%) did not (p = .247). POMR for emergency laparotomy (4.4%) correlated with all-procedure POMR (1.6%, r = 0.52, p = .005), as did the POMR for open, long bone fractures (0.8%, r = 0.61, p < .001). POMR for emergency cesarean section (0.05%) did not correlate with all-procedure POMR (p = 0.400). There was a correlation between surgical volume and emergency laparotomy POMR (r = - 0.53, p = .004), but not for emergency cesarean section or open, long bone fractures POMR.

Procedure-specific POMR for laparotomy and open long bone fracture correlates modestly with all-procedure POMR among Brazilian states which is primarily driven by emergency procedure POMR. Selective reporting of emergency laparotomy and open fracture POMR may be a useful surrogate to guide subnational surgical policy decisions.