Validation of post-colonoscopy colorectal cancer cases in english population data

Introduction

The World Endoscopy Organisation (WEO) has recommended a method of calculating PCCRC rates to enable valid comparison. The accuracy of this rate depends on complete and correct capture of both colonoscopies and cancers. This includes the correct date of diagnosis, because the timing of diagnosis of cancer can affect whether a cancer is labelled as a PCCRC. There have been reports of PCCRCs not fulfilling the criteria,¹ but single site studies may not be representative of the wider system. This study aimed to quantify the extent of incorrect labelling of PCCRC in a national audit in England.

Method

PCCRC-4yr (6–48 months after a colonoscopy) were identified by linking hospital episode statistics and cancer registry databases. The cancer cases were the most recent and may not have undergone the usual validation process. The PCCRCs were loaded into a secure audit template portal. Hospitals could reject the case on the basis it was not felt to be a PCCRC. The auditors were encouraged to elaborate on the reason for rejection with free text. All rejected cases were reviewed independently by two members of the study group (NB and RV). Areas of disagreement were discussed with a third expert in classification of PCCRC (DB).

Results

2722 PCCRC-4yr were uploaded into the portal. The final dataset included 1724 completed templates, 539 incomplete and 459 rejected cases. Review of the rejected cases reached agreement on all but 25 which had further review. of the rejected cases 168/459 (37%) were incorrectly rejected. The overall rate of correct designation of PCCRC was (1724+168)/(1724+459)% = 87%. of the incorrectly rejected PCCRCs 11(7%) were WEO category A, 12(7%) B, 109(65%) C and 15(9%) D.² There was insufficient information to categorise the remainder. of the correctly rejected cases, 147 (62%) were incorrect CRC diagnoses, 28 (12%) were colonoscopy data errors, and 63 (26%) related to the timing of the colonoscopy or cancer.

Conclusions

13% of PCCRC-4yr were incorrectly classified as PCCRCs. Error relating to CRC diagnosis accounted for nearly 2/3 of these cases. This high proportion due to incorrect CRC diagnosis was expected because the most recent cancers were included. Thus, the 13% estimate of error is greater than would be expected when benchmarking PCCRC rates because only CRC that have been fully validated would be included in the calculation. Nevertheless, this study confirms a substantial minority of cases included in benchmarking rates may not be true PCCRCs as defined by the WEO. The high rate of WEO category C in the incorrectly rejected sample indicates that endoscopy teams do not fully understand this category. WEO category C describes the most plausible explanation as ‘detected lesion, not resected’. Reviewers might not have classified these cases as PCCRCs because they were not technically missed lesions – they were just not resected. of note, 30% of WEO category C were deemed unavoidable because of patient morbidity, patient choice or other factors beyond the control of the endoscopy team. In these circumstances the team may not regard the cancer to be a PCCRC because they were not at fault for the delay in diagnosis. In summary, this study provides an estimate of the extent of misclassification of PCCRCs when national datasets are used. Future reports of benchmarking PCCRC rates need to take account of misclassification. The finding of a high rate of WEO category C in the incorrectly rejected sample indicates that investigators cannot assume frontline endoscopy teams understand how PCCRCs in this category are defined.

References

Ahmad A, Dhillon A, Saunders BP, et al. Validation of post-colonoscopy colorectal cancer (PCCRC) cases reported at national level following local root cause analysis: REFLECT study. Frontline Gastroenterology 2022;13:374–380. Anderson R, Burr N, Valori R. Gastroenterology 2020;158:1287–1299.