P82 NUn score validation at a district general hospital

INTRODUCTION: The NUn score was created to try and predict the risk of anastomotic leak or major complications (using the Clavien- Dindo classification) from upper GI resections with an oesophageal anastomosis. A score of > 10 was used to predict an increased risk. In this study we attempt validation. METHODS: A database of 101 patients was studied, who underwent an Oesophagectomy for cancer between March 2017 and 2020. 72 patients had complete Post-operative day 4 bloods, needed to calculate the score. These patients were then studied for post-operative complications. RESULTS: A total of 12 patients had a NUn score of > 10 (16.67%). There was 1 death (1.37%) and 11 anastomotic leaks (15.28%). Of these the NUn score did not predict the death and predicted 8 of the 11 anastomotic leaks. FROM OUR DATA: Sensitivity = 66.67%, 95% CI [34.89%, 90.08%]: Specificity = 55.00%, 95% CI [41.61%, 67.88%]: PPV = 22.86%, 95% CI [15.39%, 32.56%]: NPV = 89.19%, 95% CI [78.21%, 94.99%]: Mean NUNs score for complications = 10.29, 95% CI [9.59, 10.99]: Mean NUNs score for no complications = 9.87, 95% CI [9.65, 10.10]. CONCLUSION: From our analysis the NUNs score cannot be shown to be sensitive, specific or have useful positive predictive value. The average Nun score was not reliable, with confidence intervals crossing 10. There may be some merit in using the test for its negative predictive value, but further analysis into this is needed. The results of this audit are consistent with previous efforts at external validation.