Abstract

Background

We analysed outcomes of Pancreaticoduodenectomy (PD) (Whipple’s operation) performed across Queensland hospitals during the period 2003 to 2012.

Methods

Thirty day, ninety day and in-hospital postoperative mortality was modelled using multivariate Cox proportional hazards regression controlling for annual surgical volume of hospital, gender, age, remoteness of residence, socioeconomic status, comorbidities, anaesthetic score, emergency surgery and surgery in a public facility.

Results

Of the 6,058 Queensland patients with pancreatic, biliary tract and duodenal cancer diagnosed between 2003 and 2012, 727 (12%) underwent a pancreaticoduodenectomy.  These patients were younger than those who did not receive surgery (median age 64 years vs 72 years) and more likely to be male (59% vs 53%).

Overall thirty day mortality was 2.9%, ninety day mortality was 5.0% and inpatient mortality was 3.2%.  Post-operative mortality was significantly associated with annual surgical volume.  Compared to high volume hospitals performing >9 surgeries a year, the hazard ratio for mortality in very low volume hospitals (averaging less than three surgeries a year) was 4.12 for thirty day mortality (95% confidence interval 2.06-8.25, p<0.001), 3.86 for ninety day mortality (CI 2.28-6.54, p<0.001) and 3.07 for inpatient mortality (CI 1.59-5.93, p<0.001). There was no association with mortality for low volume hospitals performing on average between three and nine surgeries a year.  Increasing age and the presence of comorbidities were also significantly associated with increased risk of postoperative mortality.

Conclusions

The overall postoperative mortality rates for PD in Queensland is low, however there is significant variation across hospitals according to the volume of surgery performed at the hospital with postoperative mortality rates being higher among very low-volume hospitals.