31.   Although the interaction is again not significant the differences between exemplars are more marked for the high risk junctions than the low risk ones.

32.   For the low risk junctions it does not appear that high risk exemplars actually contain more risk related information than other exemplars of the same junction.

33.   For high risk junctions increasing exemplar risk is associated with an increase in the amount of information described.

34.   For the low risk junctions, however, the opposite is the case, high risk exemplars actually have significantly less information described.

35.   HBIG should be given when the source is known or suspected to be at high risk of HBV infection.

36.   Endoscopic sclerotherapy, shunt surgery, and staple transection of the oesophagus or liver transplantation should all be considered for this high risk group.

37.   Alternatively, it may be important to demonstrate the presence of unsuspected brain metastases in high risk patients when planning management.

38.   Because of the high risk of Gram positive infections and in particular pneumococcal infection there may be an indication to give this group of patients pneumococcal vaccination.

39.   Others prefer primary surgical intervention, assuming a high risk of recurrence, and see primary endoscopic treatment as an alternative in patients with a high operative risk only.

40.   Can molecular biology help in the screening of other high risk groups?