Using Statistical Process Control to Monitor Anastomotic Leak

Abstract

Background Surgery remains a cornerstone in the treatment of bowel diseases, such as those involving cancer or inflammation. In the majority of patients, a section of bowel is resected and the remaining bowel is re-joined surgically using sutures or staples (bowel anastomosis). However, in some cases this anastomosis can break down (Anastomotic Leak (AL)), causing significant complications for the patients including increased mortality, prolonged hospital stay and worse cancer outcomes. Despite the significance of this complication most hospitals do not prospectively measure their leak rate or engage in activities to reduce it. Another key postoperative outcome which can act as a surrogate marker of performance is Postoperative Length of Stay (PLoS) One way to address this is to promote the use quality improvement (QI) methodologies such as Statistical Process Control (SPC). This involves mapping the data points in time order and seeing if the process is stable between a set of upper and lower parameters (i.e. confidence intervals) and observing whether there has been a statistical change. Methods The aim of this study was to retrospectively map AL rates and PLoS using Statistical Process Control at Royal Devon and Exeter Foundation NHS Trust. This was to provide a baseline measurement as part of the first phase of a QI project as well as investigating the suitability for SPC chart analysis for monitoring postoperative outcomes. All patients undergoing colorectal resections with ileo-colonic, colo-colonic colorectal, colo-anal or ileo-anal anastomoses from 01//01/2010 to 30/04/2017 were included in this study. AL was defined as cases where there was subsequent return-to-theatre, radiological drainage or medical management of the AL. SPC charts were used to map AL rates to establish whether variation in the rate over time was due to “common-cause variation” or “special-cause variation.” The G-Chart, a type of SPC chart used to count the number of events between rare incidents was used to map AL. I-Charts were used to map median monthly Postoperative Length of Stay (PLoS). Results The AL rate is relatively low at this hospital with a return-to-theatre rate of 4.3% and an overall rate of 6.1% (once conservatively managed ALs and radiologically drained leaks were included). The overall median PLoS was 6 days. The SPC charts show that there is a reasonable chance of special cause variation for the Elective, Stapled and Right-sided AL charts, with some overlap with the former two categories. SPC charts for Sutured ALs and Left-sided ALs both only exhibited common cause variation. SPC charts for all six sub-groups monitoring PLoS indicated periods of special cause variation. Discussion In terms of the AL rate, 4.3% is a very acceptable return-to-theatre rate in line with other studies. The rate of 6.1% is difficult to interpret given that not all cases of medically managed ALs would have been identified. The overall median PLoS was also consistent with the literature. This is the first phase of a QI project to reduce rates of AL at Royal Devon and Exeter Foundation NHS Trust which can now take place prospectively and an intervention can be planned and implemented. Also, now that the methodology is in place, SPC charts can also be used to ensure patient safety over time, acting within a Quality Assurance context. Despite their ability to identify retrospective periods of SCV, the findings in SPC charts monitoring AL and PLoS will now need to be corroborated with the historial clinical context as SPC charts cannot identify which factors have caused the shift. In summary, this dissertation demonstrates that using SPC charts are a feasible methodology to retrospectively map AL and PLoS rates in a Colorectal Unit.