Introduction The aim of this study is to describe the clinical outcomes of those patients in UK tertiary referral centre referred with a diagnosis of refractory coeliac disease (RCD). This is compared to the grim multicentre survival data1 ,2 in current literature- 5-year survival RCD type 1 96% and type 2 44%.

Methods Patients were entered into a database from clinic letters. Notes were reviewed for patient demographics, past medical history with dates, drug history and dietetic input and clinical course. The hospital electronic computer system was trawled for clinical parameters at time of diagnosis and during subsequent follow-up.

Results Review of 32 of the 44 patient notes was able to be done and 12 patients were found to have true RCD. 4 patients had RCD type 1 and 8 patients had RCD type 2. The follow-up range (mths) was as follows type 1 3–56, type 2 3–96 and the age range (years) type 1 28–74, type 2 44–79. There is one primary enteropathy associated T-Cell lymphoma in the type 2 group who has undergone resection and CHOP chemotherapy in 1997. The HLA type II status is known for 9 of the 12 RCD patients. Type1 RCD have 1 DQ2 and 1 heterogeneous DQ2/DQ8 whereas type 2 RCD have 4 DQ2 and 3 DQ8. All RCD patients had normal albumen and haemoglobin at time of presentation. Four of the RCD type 2 patients changed to type 1 with steroids and azathioprine, however none of the type 1 RCD improved histologically. All RCD patients are alive to date.

20 non-responsive coeliac disease patients were found to have the following diagnoses: normal treated coeliac disease 4, not on gluten free diet (GFD) 6, small bowel bacterial overgrowth 10, lactose intolerance 2, Crohn's disease 1.

Conclusion Compared to the current literature available with grim prognosis for type 2 RCD, our patient data would not support this. Part of the discrepancy is perhaps to do with the use of T-cell receptor in the diagnosis of RCD which was not used in reference 2. Interestingly there is a high DQ8 penetrance which may implicate a different mechanism driving RCD. More numbers are required to be analysed however our evidence supports treatment with steroids and azathioprine.