European and North American populations should be screened for coeliac disease

As European and North American societies become more opulent, the focus of attention moves inevitably from diagnosis and treatment of diseases to screening of populations. Successful screening would require an increased detection of the disease, improvement in the health of that population, and prevention of known complications. Population screening involves testing of otherwise healthy people following an educational programme inviting them to participate with informed consent. If the uptake of this were small, the alternative would be the compulsory testing of body samples allowing little choice to the individual—an action hardly conducive to current societal sentiments!

For a screening test to be useful, it needs to be simple, cheap, have a high sensitivity and specificity, and be easy to perform on large numbers of the population. In coeliac disease the endomysial antibody test¹ is a possible screening tool. The specificity of this test is approaching 100%² and sensitivities of 86% or greater have been observed.^3,4 Measurement of anti-tissue transglutaminase antibodies (transglutaminase being the principal component of the endomysium autoantigen)⁵ may become a more objective method for testing.

So would screening the population detect more disease in the population? The prevalence of coeliac disease in several recent population studies appears now to be approximately 1 in 200.^6,7 The common model used to explain the epidemiology has been the “iceberg”, showing that only a minority of individuals actually have clinical symptoms of the disease. If these subjects have no symptoms, then the benefits of detecting “occult” disease need to be demonstrated. There are, of course, certain “associated diseases” that occur with an increased frequency⁸ and these patients are being screened in some centres. These conditions include type 1 diabetes mellitus,⁹ autoimmune thyroid disease,¹⁰ primary biliary cirrhosis,¹¹ dermatitis herpetiformis, Down's syndrome,¹² osteoporosis,¹³ epilepsy,¹⁴ and various undiagnosed neurological disorders.¹⁵ There are also a possible 10–15% of first degree relatives of patients with occult coeliac disease.¹⁶ Thus it would seem reasonable to screen these disease groups with this high incidence, but this is not screening the population.

Would screening the population make a difference to the health of the populace? There are no data on this as it would be extremely difficult to avoid the placebo effect of a special diet. Paperback books on “diets” of any sort have always sold well in Western communities and it is easy to fall prey to fad diets. A small search on the net produced over 1.25 million references! An extremely small study comparing screen detected and symptom detected patients and quality of life on a gluten free diet showed a similar improvement but the authors were not sure how compliance would be in a long term study.¹⁷ However, compliance has always been a problem, even in symptom detected coeliac disease,¹⁸ and an Italian study showed poor compliance among screen detected adolescents.¹⁹ It could also be argued that screening would prevent the development of osteoporosis. However, as all older females are being advised to have DXA scans and are also using HRT, screening the population for coeliac disease becomes unnecessary.

Whether screening tests would reduce mortality is a highly debatable point. A large amount of public money has been spent on disease screening but inevitably most of these demonstrate lead time and length time biases. To demonstrate a reduction in malignancy in coeliac disease would statistically require screening many thousands and there is no evidence as yet to suggest that a gluten free diet would prevent malignancy in asymptomatic/occult cases.

Workers have tried to make the burden of screening populations less by using a gradation of tests.⁶ However, the cost of this could well be prohibitive in the context of a national screening programme. There is also a question of ethical justification. Would one seriously consider imposing a strict gluten free diet on a well asymptomatic member of the general population? There is no doubt that in the future there will come a time when one's odds ratio for developing coeliac disease could be given at birth following genetic testing.²⁰ This would of course raise further more complicated ethical issues as well as giving fodder to insurance companies and actuaries. I would therefore submit that European and North American populations should not be screened for coeliac disease unless we can produce proper evidence for a reduction in morbidity and mortality, an increase in survival, and a good cost effectiveness-benefit ratio. We should not be the harbingers of yet another source of anxiety for the general population (often helped by the media), into making us an even more neurotic population! Pragmatism must win over misguided purism.

Summary.

• No evidence that screening the general population will lead to a reduction in morbidity.

• No evidence that instituting a gluten free diet in asymptomatic individuals will reduce mortality.

• Asymptomatic screen detected patients have shown poor compliance with a gluten free diet.

• No evidence that screening will have a good cost effective-benefit ratio.

• Screening will only demonstrate lead time bias.