Article Text

Download PDFPDF

Screening for coeliac disease: what evidence is required before population programmes could be considered?
  1. E H Young,
  2. N J Wareham
  1. MRC Epidemiology Unit, Cambridge, UK
  1. Correspondence to:
    Dr N J Wareham
    MRC Epidemiology Unit, Strangeways Research Laboratory, Worts Causeway, Cambridge CB1 8RN, UK;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Commentary on the paper by Tommasini et al

In the traditional doctor-patient consultation, the patient initiates a visit and the physician’s imperative is to respond to the best of his or her ability. However, in a screening programme, the initiation comes from health professionals, and many commentators have argued that this poses different ethical challenges.1 In particular it is argued that the level of evidence concerning the balance of overall benefit and harm from the screening programme must be much greater than that for a traditional therapeutic intervention.

Although screening may intuitively be logical for a particular condition which is prevalent and detectable in the preclinical stage, intuition is not an appropriate basis for commencing a screening programme. A number of criteria have been described that allow the evidence for screening programmes to be more formally evaluated.2 Calls are regularly made for the introduction of population screening programmes, such as that for coeliac disease (CD) in the paper by Alberto et al in this issue.3 These screening criteria form the basis for evaluating the strength of the evidence supporting that call. They also highlight key uncertainties, which could be the subject of future research. This process has been undertaken for screening for type 2 diabetes in adults,4 and the history of the debate on this topic may shed some light on how to reach a more speedy decision for CD.


The effectiveness of a screening programme is closely related to the characteristics of the disease for which screening is proposed. CD is an important health problem as it is common, affecting 1 in 120–300 people in Europe,5 and its associated morbidity and excess mortality are well recognised.6,7 However, the natural history of asymptomatic CD is less well understood, although some patients with unrecognised disease have extra-intestinal manifestations such as growth restriction8 or associated autoimmune disease.6 The prevalence of associated autoimmune disorders in CD is related to the duration of exposure to gluten,9 providing a strong imperative for screening and early treatment.


As with type 2 diabetes screening, there is considerable debate around the relative merits of different screening modalities for CD. The performance of a potential screening test is dependent on diagnostic thresholds for the disease and the availability of a simple, precise, and validated test that is acceptable to the population. Serological tests are easy to administer and have an important role in screening for CD. A number of highly sensitive and specific markers are available.5,6 In this issue, Alberto and colleagues show the high positive predictive value of one autoantibody, serum anti-h-tTG-ab.3 For any screening test, there should be an agreed policy on the further diagnostic investigation of individuals with a positive test result. In this instance, individuals with a positive antibody result underwent the gold standard diagnostic test of small intestine biopsy. Screen negative children were not investigated further. However, without biopsying screen negative children the sensitivity or specificity of the test cannot be established.


A key issue in evaluating any screening test is to consider whether screening and subsequent treatment result in health gain. For patients with symptom detected CD, delay in diagnosis unfavourably affects patients’ prognosis.7 Treatment with a gluten-free diet for life provides good relief of symptoms and is effective in reversing abnormal gut pathology. Prompt treatment also decreases mortality.7 For adults, quality of life is improved on a gluten-free diet.10 Children develop normally and report a quality of life similar to that of children in the general population.11 However, not all adults with CD attain the same degree of subjective health as the general population.12 Provided that patients with CD adhere to a strict diet, the prognosis is excellent and there are no data to suggest that there is an excess mortality in this group.

Yet the benefits of early detection and treatment of screen detected asymptomatic CD have not been proved. With treatment, quality of life improves for those whose disease was screen detected.13 The impact of treatment on mortality and morbidity for asymptomatic individuals is less clear. Asymptomatic CD is often associated with other pathology including reduced bone density, and iron and folate deficiencies, which if left untreated would account for a number of chronic health issues. Silent disease may also account for the worsening of other serious medical conditions,14 although this link is by no means established.

It is the evaluation of evidence concerning the costs and benefits of treatment for individuals with screen detected disease that is often lacking. One cannot necessarily translate results from treatment trials in people with established disease and apply the evidence to the population with screen detected, possibly asymptomatic, disease. In type 2 diabetes even though patients may believe their illness to be serious, many can reconcile such beliefs with poor adherence to medical advice about lifestyle.15 Many patients assume that the presence of symptoms is a necessary precursor of complications and equate freedom from symptoms with freedom from risk. This is highly relevant to CD, since screening would result in asymptomatic individuals being advised a lifelong gluten-free diet. Although the diet may be efficacious, its effectiveness in real life is a function of the degree to which patients adhere to advice.


Population screening is always a major undertaking, and the merits of universal population screening must be weighed up against targeting high risk subgroups. The health benefits of mass screening for CD are not known, and in particular no good trial evidence exists to indicate that screening for CD would achieve significant health gain or that it would outweigh the disbenefits of screening. Ideally, evidence of effectiveness in reducing mortality and morbidity should come from randomised trials. A trial would also make it possible to calculate more accurately the costs of screening.


In the case of type 2 diabetes, the argument surrounding the benefits of screening has continued since the 1960s. Although the disease and the test criteria are easily met, as with CD the major uncertainties concern the benefits and disbenefits of early detection and treatment. It becomes increasingly difficult to undertake a randomised controlled trial of screening as evidence mounts, and there is probably a window of opportunity in which people are sufficiently uncertain to tolerate randomisation. The lesson for researchers into disorders such as CD is not to miss that window, otherwise the debate about whether or not to screen may run for decades.


EHY is an MRC Career Development Fellow.

Commentary on the paper by Tommasini et al


Linked Articles

  • Atoms
    Howard Bauchner