P-OGC97 Development and piloting of surgical quality assurance methods for randomised controlled trials (RCTs): an example from a trial comparing laparoscopically assisted oesophagectomy vs. open oesophagectomy

Abstract

Abstract Background RCTs in surgery are frequently criticised because the standard to which operations are performed (quality assurance - QA) is not considered during study design and delivery, risking performance bias. Lack of clarity about surgical QA may also influence the successful implementation of RCT results into routine practice, because it is unclear how procedures were undertaken. We developed QA measures for an RCT comparing laparoscopically assisted and open oesophagectomy (LAO and OO). Methods Five QA categories were developed during the pilot and applied to the main trial, using data from patients receiving their randomized allocation in each group: i) entry criteria for centres; ii) entry criteria for surgeons; surgical protocols for key components of LAO and OO with mandated, prohibited and flexible components, monitored using iii) case report forms (CRFs) to record protocol adherence; and iv) intra-operative photographs to demonstrate protocol adherence (using the visible anatomical structures to determine if the component had been fully completed); and v) lymph node count and length of oesophagus. Results 8 centres and 39 surgeons participated and met entry criteria. 145 (LAO) and 149 (OO) patients underwent their randomized surgical procedure. Key procedural components were reported as complete in CRFs at similar rates in both groups, with >70% undergoing mandated components. However, adherence assessed using photographs was consistently lower than the CRFs. For example, left gastric artery lymphadenectomies were reported as complete in > 98% CRFs (LAO and OO) whereas photographs found this to be complete in 42% (OO) and 54% (LAO). Median nodal count was similar in both groups (145 LAO=24.7, SD = 10.6 and 149 OO = 26.4, SD = 10.2) as was length of resected oesophagus. Conclusions Assessing surgical QA in a multi-centre trial is logistically challenging but feasible. Whilst video data from laparoscopic cases could be collected and assessed, it was not possible with open surgery. Understanding adherence to the study protocol using photographs in addition to CRFs was important because of marked differences between what surgeons reported had been undertaken and images of what had been achieved. It is recommended that surgical trials include QA processes to understand protocol adherence and examine performance bias between groups.