Coeliac disease (CD) is an autoimmune disease affecting the digestive system that causes an adverse reaction to gluten. Untreated CD can increase the risk of anaemia, osteoporosis, cancer, and infertility. Around 1 in 100 people in the UK have CD, although many are not diagnosed. There is a need for clear, evidence-based guidance for identifying adults and children with CD to speed up the diagnostic process and improve patient outcomes.
We want to establish who should be tested for CD, what tests should be offered, and whether a biopsy is necessary in all patients. We also want to investigate the cost-effectiveness of different strategies. We will do this by reviewing existing studies on tests for CD and studies on what symptoms, signs, or other factors are associated with CD; by analysing routinely collected data to see which factors may predict CD; by working with patients to help us work out which strategies are best for patients; and by economic evaluation of each strategy to find the most cost-effective strategy.
Coeliac disease (CD) is an immune-mediated disorder, triggered by the protein gluten, estimated to affect 1% of the UK population.
Guidelines recommend that adults and children at high risk of CD should be offered testing. However, it is not clear which groups are at sufficiently high risk to justify routine testing, which symptoms should prompt testing, which tests should be offered, and whether confirmatory biopsy is necessary.
We are currently conducting a systematic review on the accuracy of diagnostic indicators (such as risk factors and symptoms) for CD. Diagnostic indicators that are associated with an increased risk of CD will be selected to inform the development of prediction models using CPRD GOLD data. We will fit logistic regression models with CD as the outcome and multiple predictors. From the results, we will produce estimates of the probability of CD diagnosis for each combination of diagnostic indicators available. We will validate the model developed in CPRD GOLD using bootstrapping (internal validation) and in CPRD Aurum (external validation). We will also validate any existing prediction model identified by the systematic review in CPRD Aurum. We will carry out additional analyses to explore the unexpected findings of the relationship between fertility and coeliac disease identified in the prediction model.
Finally, the cost-effectiveness of CD testing of patients with pre-test probabilities of CD above certain thresholds will be evaluated with a long-term economic model. We will use CPRD Aurum linked to HES to estimate the risks of adverse outcomes in patients with CD (such as lymphoma or infertility) and the risk of adverse events, including death, associated with biopsy. These estimates will directly inform the economic model.
We will also use CPRD GOLD data to quantify medical costs associated with a diagnosis of coeliac disease. These estimates will directly inform the economic model.
Prediction modelling (case control) and validation of existing prediction models: the primary outcome is diagnosed CD (see Appendix A for Read codes).
Estimating risk of adverse long-term outcomes associated with CD and adverse events associated with biopsy (cohort): key outcomes are all-cause mortality; any malignancy; Hodgkin-lymphoma; non-Hodgkin-lymphoma; small intestinal cancer; colon cancer; splenic hypofunction; osteoporosis; iron deficiency anaemia; vitamin B12 and folate deficiency anaemia; pregnancy-related complications, including unexplained infertility, recurrent miscarriage or intrauterine growth restriction (see Appendix B for ICD-10 codes); and perforation, infection, bleeding, sepsis, myocardial infarction, and death associated with biopsy (see Appendix C for ICD-10 and Read codes).
Estimating medical costs associated with CD: key outcomes are number of primary care consultations, tests, referrals to out-patient hospital care and prescriptions.
Penny Whiting - Chief Investigator - University of Bristol
Martha Elwenspoek - Corresponding Applicant - University of Bristol
Alastair Hay - Collaborator - University of Bristol
Edna Keeney - Collaborator - University of Bristol
Gerry Robins - Collaborator - York Teaching Hospital NHS Foundation Trust
Hayley Jones - Collaborator - University of Bristol
Hazel Everitt - Collaborator - University of Southampton
Helen Russell - Collaborator - University of Bristol
Howard Thom - Collaborator - University of Bristol
Jessica Watson - Collaborator - University of Bristol
Joni Jackson - Collaborator - University of Bristol
Peter Gillett - Collaborator - NHS Scotland
Rachel O'Donnell - Collaborator - University of Bristol
Susan Mallett - Collaborator - University of Birmingham
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation