Back to Annual Meeting Program

Introduction: The surgical management of oesophago-gastric cancer in the UK has recently been centralised to high volume surgical cancer centres. Increasing attention has been paid to measuring surgical outcomes and monitoring performance. The authors demonstrate results from a UK surgical cancer centre and present a novel technique for monitoring outcomes in “real-time”.

Methods: A prospectively collected database was used to identify the first consecutive 100 oesophagectomies performed using a “hybrid” technique (laparoscopic abdominal approach, open thoracic approach) in the Unit. The cumulative sum (CUSUM) technique was applied to examine the incidence of clinically relevant anastomotic leaks and 30-day post-operative deaths.
Cumulative sum (CUSUM) techniques, originally developed as industrial control techniques, have the potential to provide rigorous, "real-time" monitoring of surgical outcomes. Already used in cleft surgery and burns intensive care, CUSUM compares actual to predicted outcome rates on a case-by-case basis. The technique described plots case number on the x-axis versus the actual minus predicted outcome rate on the y-axis. The graph is not allowed to pass below the y=0 point, preventing periods of good performance maskng a period of poor performance. A period of performance "as predicted" would be represented by a horizontal line, a period of worsening performance as a rising line and vice versa. The setting of "alarm points", i.e. the y value at which performance warrants review, means the unit can perform that review at the time of the worsened outcomes and make immediate changes.

Results: Ninety-seven cases were completed laparoscopically, with three converted to open procedures. Mean age was 66.4 years (SD 9.2). Median length of inpatient stay was 15 days. The 30-day mortality was 5%.
The crude anastomotic leak rate was 6%. CUSUM analysis of the incidence of leaks (Figure 1) demonstrated a peak at case 65. This however did not breach the alarm line at two leaks above the expected rate.
CUSUM analysis of mortality revealed two peaks, neither of which breached the alarm line at two point five deaths above predicted.


Conclusions: The early learning curve results from a UK surgical cancer centre show acceptable 30 day mortality and anastomotic leak rates.
The CUSUM technique is a viable method of observing trends in anastomotic leak rates and mortality, highlighting when the rates rise above an unacceptable level which would then trigger internal audit. CUSUM analysis could be further refined with the use of a risk stratification tool, such as O-POSSUM. These tools could be implemented in a prospective fashion, allowing “real-time” assessment of periods of varying performance.