Improving surveillance for Barrett's oesophagusBMJ 2006; 332 doi: http://dx.doi.org/10.1136/bmj.332.7553.1320 (Published 01 June 2006) Cite this as: BMJ 2006;332:1320
- Peter A Bampton (), head of gastrointestinal endoscopy1,
- Anne Schloithe, technical officer2,
- Jeff Bull, clinical nurse4,
- Robert J Fraser, associate professor5,
- Rob T A Padbury, divisional director3,
- David I Watson, professor2
- 1 Department of Gastroenterology and Hepatology, Flinders Medical Centre, Bedford Park, SA 5042, Australia,
- 2 Department of Surgery, Flinders University of South Australia, Bedford Park,
- 3 Division of Surgery and Specialty Services, Flinders Medical Centre,
- 4 Investigations and Procedure Unit, Repatriation General Hospital, Daw Park, SA 5042, Australia,
- 5 Department of Gastroenterology, Repatriation General Hospital
- Correspondence to: P Bampton
- Accepted 21 April 2006
Problem A retrospective audit of surveillance for Barrett's oesophagus 1996-2001 identified the need to improve adherence to guidelines for the endoscopic surveillance of patients with Barrett's oesophagus.
Design Prospective audit of the effect of disseminating guidelines in 2002. Prospective audit of the effect of introducing local guidelines and Barrett's oesophagus surveillance officers, 2003-2005.
Setting Two general hospitals in Australia, 2002-5. All adult patients diagnosed with Barrett's oesophagus were included.
Key measures for improvement Proportions of patients in a Barrett's oesophagus surveillance programme who had appropriate time intervals between follow-up endoscopies and who had appropriate numbers of biopsies collected at endoscopy.
Strategies for change Local guidelines were laid down. Surveillance coordinators for Barrett's oesophagus were introduced to manage the process according to a clinical protocol designed for each patient.
Effects of change Disseminating guidelines had little effect on practice. Six months after surveillance coordinators were introduced, adherence to the planned surveillance interval increased from 17% to 92% and the number of endoscopies at which sufficient biopsies were collected increased from 45% to 83%. These changes have been maintained.
Lessons learnt Disseminating guidelines and results of an audit on endoscopic surveillance in Barrett's oesophagus had no effect on practice. Introducing coordinators who proactively managed the process greatly improved adherence to guidelines.
Barrett's oesophagus is a premalignant condition in which damage to the oesophageal mucosa from gastro-oesophageal reflux leads to replacement of normal squamous oesophageal epithelium by columnar epithelium. The presence of columnar epithelium increases the risk of oesophageal adenocarcinoma, which occurs at a rate of 0.5-1% each year.12 The best marker of potential malignancy is dysplasia found on endoscopic biopsy. No prospective randomised trials have shown that endoscopic surveillance for dysplasia in Barrett's oesophagus reduces mortality from adenocarcinoma of the oesophagus. However, other studies support a role for surveillance: patients in whom malignancy is detected at surveillance endoscopy have increased survival, and cancers at surveillance are at an earlier stage than those not detected at surveillance.3–5 Endoscopic surveillance is therefore recommended for patients with Barrett's oesophagus who are fit to undergo oesophagectomy.6–8 Although the optimal interval between surveillance endoscopies is unclear, the British Society of Gastroenterology has recommended two years, and quadrantic biopsies taken every 2 cm in patients without dysplasia.
The outcomes of surveillance depend on the rate of participation of patients and the effectiveness of the surveillance modality. Effectiveness is reduced by patients “dropping out” and by inappropriate endoscopic biopsy practice. In a survey of members of the British Society of Gastroenterology, only 41% of respondents routinely performed quadrantic biopsies as recommended.9 Conversely, excessive screening is a waste of resources, as reported for endoscopic screening of colorectal cancer and in audits of Barrett's epithelium databases.10 11
Outline of the problem
We performed a retrospective audit of surveillance for Barrett's oesophagus from 1996 to 2001 at two general hospitals.12 We reviewed 9732 endoscopy reports and found 253 endoscopies that were performed to screen for Barrett's oesophagus. Quadrantic biopsies had been taken every 2 cm in only 26% of endoscopies. For ongoing screening, the mean interval between endoscopies was 12 months. These data indicated that screening practice was ineffective and wasteful of endoscopic resources.
We formally presented the results of this review to the medical staff who undertook these procedures. Endoscopic practice showed no great improvement after this intervention.
Assessment of problems
Key measures for improvement
The risk of adenocarcinoma developing in Barrett's oesophagus is estimated at 1 per 96 patient years in Australia and 1 per 100 in the United Kingdom.213 On the basis of clinical course and past recommendations, we decided that surveillance endoscopy at two year intervals would be appropriate for patients without dysplasia, who would probably be fit for oesophagectomy. This agrees with recently released guidelines of the British Society of Gastroenterology.2 The guidelines also recommend a shorter surveillance interval of six months in patients with low grade dysplasia on biopsy. Quadrantic biopsies every 2 cm are the current standard during surveillance endoscopy; this was recommended as appropriate practice in our institutions.
We wanted to see improvements in:
The proportion of patients under surveillance who were scheduled for and who underwent endoscopy—two yearly if no dysplasia was found on endoscopy or six monthly if low grade dysplasia was identified
The proportion of patients in whom adequate biopsies were taken at surveillance endoscopy.
In 2002 we performed a prospective audit over 12 months; a research officer examined all the case notes and endoscopy records of patients who presented to the endoscopy services at Flinders Medical Centre. We recorded the length of Barrett's epithelium, the number of biopsies taken, and the histology, together with any evidence of a planned follow-up procedure (and its timing).
Analysis and interpretation
The 2002 prospective audit identified 71 endoscopies in 57 patients with Barrett's oesophagus as a new diagnosis or at screening. Appropriate biopsies were taken in only 32 (45%) of these endoscopies (none in 24 (34%), too few in 12 (17%), too many in three (4%)). We then reviewed follow-up decisions in patients who were being followed up in the public hospital (46 endoscopies in 41 patients; the other 16 patients (25 endoscopies) were followed up in the private sector).
A follow-up endoscopy had been organised in an appropriate manner in only seven (17%) of the 41 patients. Endoscopic follow-up had been planned at too frequent an interval in nine (22%) patients, whereas no apparent decision had been made in 18 (43%), either because no biopsies were taken (12, 28%) or owing to failure of planned clinical review (six, 15%). Five patients should not have been followed up because the diagnosis of Barrett's oesophagus was not confirmed, and two patients had a repeat endoscopy organised, but at an interval longer than that recommended by the guidelines.
Thus, the improvement in surveillance practice for Barrett's oesophagus was small even after we disseminated the results of the retrospective audit and provided published protocols. Several reasons became apparent when we discussed the results with the clinicians involved:
Doctors were unsure about the usefulness of screening for Barrett's oesophagus;
Doctors had divergent views about appropriate screening intervals. This led to confusion, especially among junior staff and general practitioners referring patients to our hospital, who were unsure what recommendations they should make to patients;
Endoscopies are often performed on patients referred by other doctors. The clinician carrying out the procedure may have different ideas about the need for surveillance (reflected in the two points above) from those of the referring doctor, especially if a clear surveillance plan was not present in the clinical record;
Taking quadrantic biopsies for long segment (> 3 cm) Barrett's oesophagus is time consuming. Endoscopists have a busy schedule and no extra time is allowed for such cases.
Strategy for change
We developed guidelines to remove confusion and ensure that the intended surveillance protocol would be performed even if a patient was placed on another doctor's list. After discussion, we established local guidelines similar to those of the British Society of Gastroenterology—two yearly endoscopies that were repeated after six months if low grade dysplasia was found. Our guidelines, however, recommended a repeat endoscopy one year after the initial diagnosis of Barrett's oesophagus.
Barrett's oesophagus surveillance officers
We introduced the role of surveillance coordinators for Barrett's oesophagus into our two hospitals. Patients in whom Barrett's oesophagus was identified on endoscopy were flagged by the surveillance coordinators, who then constructed a case summary for each patient, including histology. The coordinator then drew up a provisional recommendation for surveillance on the basis of this summary. The case record was then reviewed by the gastroenterologist or upper gastrointestinal surgeon who had referred that patient, and if surveillance was thought to be appropriate, the patient was enrolled in the surveillance programme. In instances of doubt about surveillance or ongoing management, the case was reviewed by an upper gastrointestinal surgeon with expertise in managing Barrett's oesophagus.
The surveillance coordinators then communicated the surveillance plan to the patient and their general practitioner and placed a copy of the plan in the medical record. All patients were entered onto a computerised database with a facility for automatic call back.
Coordinators also informed endoscopists if sufficient numbers of biopsies had not been taken. A diagram of best practice was posted in all procedure rooms, and endoscopy nurses were educated about the importance of adequate biopsy technique.
Effects of change
We developed the surveillance programme over the first six months of 2003. The prospective audit was repeated for the first year of the programme (September 2003 to August 2004). During this time, adherence to the agreed guidelines increased greatly for biopsy practice (from 45% to 83%) and planned surveillance interval (from 17% to 92%).
The benefit of screening needs to be weighed against risks and costs. We routinely audit the outcome of endoscopy, and no endoscopic perforations or readmissions with pneumonia (suggesting aspiration) were noted within 28 days of the procedures. Although our study was not designed to evaluate costs, we calculated a cost per screening endoscopy of A$359.20 (£149.50; €216.30; US$273.00); this comprised procedure room costs of A$175, medical and nursing staff costs (a 30 minute procedure given the biopsy requirements, with a staff specialist and two nurses in the procedure room and one in recovery) of A$88.50, and a pathology fee of A$95.70. Although these costs will vary greatly according to the institution and the health funding models used, the cost of A$38 340 for each high grade dysplasia and early adenocarcinoma detected is reasonable (359 times the number of procedures done divided by the number of cases of high grade dysplasia and adenocarcinoma during the surveillance programme follow up (last column of table)).
Our prospective audit is ongoing, and this improved performance has been sustained through 2004 and into 2005 (figure). Adherence to the guidelines was not complete, partly because of anxiety by patients or clinicians about the timing of repeat endoscopy, and because of the continual need to educate new medical staff.
Lessons learnt and messages
In 2001, after an initial retrospective audit, we recognised that the surveillance of Barrett's oesophagus needed improvement. Dissemination of results along with existing published protocols did not produce significant improvement. This reflected a lack of agreement by endoscopists on the target population for surveillance, and a similar lack of consensus between endoscopists and referring doctors.
Key learning points
Passive dissemination of guidelines and results of an audit on endoscopic surveillance in Barrett's oesophagus did not improve practice
Developing a clear protocol and introducing coordinators who proactively managed the process significantly improved adherence to guidelines
The development of local guidelines that were agreed by all specialists within the division (which spans two hospitals) removed this confusion. The surveillance coordinators ensured that the surveillance plan was clearly documented in the notes (and also explained the plan to the patient and their general practitioner), so that the timing of future procedures and the adequacy of screening were more likely to comply with the surveillance plan. Compliance was also higher after education of endoscopy nurses. Although developing the role of the surveillance coordinator and setting up the process was labour intensive, the coordinators currently spend only one day a fortnight carrying out this role.
The benefits of surveillance for Barrett's oesophagus are unknown (the yield in our study was low, two early stage oesophageal tumours were identified and the screen detected oesophageal carcinoma has had a good outcome, table). Because of the high prevalence of gastro-oesphageal reflux disease in the ageing population (the major risk factor for Barrett's oesophagus), the demand for surveillance is unlikely to abate. A consistent approach to managing this demand will ensure that meaningful surveillance is undertaken while precise data on the benefits of screening are established. Such an approach minimises waste of endoscopy resources (such as too frequent endoscopies or taking inadequate biopsies).
Passive dissemination of published protocols did not change practice. To achieve change, locally agreed guidelines and ongoing feedback to clinicians are needed. We believe that the development of a clear protocol, and a designated member of staff to manage this protocol, allowed the appropriate use of resources according to accepted best practice.
We thank the gastroenterologists and gastrointestinal surgeons at Flinders Medical Centre and Repatriation General Hospital. Thanks for the support of the clinical practice improvement programme of gastrointestinal services within the Division of Surgery and Speciality Services.
Contributors: PAB and Jayne Sandford developed the protocol. AS and JB collected the data and enacted the protocol. PAB, RJF, and DIW monitored the protocol. PAB, RJF, DIW, and RTAP prepared and reviewed the manuscript. PAB is guarantor.
Funding: AS was funded by the Department of Surgery, Flinders University of South Australia, as part of the clinical practice improvement programme of gastrointestinal services, within the Division of Surgery and Speciality Services at Flinders Medical Centre. This programme receives funds from the Quality Development Fund of the South Australian Department of Health.
Competing interests: None declared.
Ethical approval: Clinical ethics and research committees of the respective ethics committees of Flinders Medical Centre and Repatriation General Hospital.