Comment and reply on: Are we not over-estimating the prevalence of celiac disease in the general population?

Dear Editor,

Systematic reviews can be very helpful to busy clinicians, providing summaries that explain and clarify variations in findings. They also limit bias, ultimately enhancing evidence-based optimal health care (1).

Biagi and colleagues have fallen short of these targets (2). They twice refer to their feelings. They offer a parenthetic comment suggesting abuses of tissue transglutaminase testing. These comments demonstrate biases and fail to provide objective information on the prevalence of celiac disease.

Biagi and colleagues explain their choice of systematic review. Yet this choice requires an exhaustive exploration of the relevant literature (1) which should have reached further than this small set of search terms. This may explain why they missed several important studies, including an investigation of 989 Saharawi by Catassi et al (3). Such studies may alter perspectives since Catassi et al found a 5.6% prevalence of celiac disease among the African children investigated.

Biagi et al. assert over-estimation due to inclusion of “patients already known to be affected by CD”, but how could an accurate tally of prevalence ignore such patients?

These authors point out that “anemia or insufficient body weight are universally recognized as precluding enrollment as blood donors”. Since more than one quarter (13/41) of the studies reviewed report rates based on case findings through screening of healthy blood donors, the results of these studies must underestimate prevalence.

Since IgA deficiency forms almost 2% of some populations (4) and the prevalence of celiac disease in IgA deficiency is 14% (4), such a significant confounding variable warrants more attention.

Serology misses more than cases of IgA deficiency (5). This significant portion of the reports reviewed should, therefore, be expected to under-estimate the prevalence of celiac disease. Their case finding methods will have failed to detect, and the investigators will have failed to biopsy, significant numbers of asymptomatic celiac disease, both because of their anemia and/or malnourished condition and because of seronegativity, whether due to IgA deficiency or other reasons.

Much of the evidence offered for over-estimation of celiac disease is more accurately seen as inducing under-estimations of this common ailment.

Finally, are the authors of 41 separate reports failing so miserably at the diagnostic process that they miss the mark on a majority of marginal cases? This proposition ignores variations due to IgA deficiency, seronegativity, and opposes the use of the Marsh system for evaluating intestinal histology.

Declaration of interest: The authors state no conflict of interest and have received no payment in preparation of this manuscript.

References

• Cook DJ, Mulrow CD, Haynes RB. Systematic reviews: synthesis of best evidence for clinical decisions. Ann Intern Med. 1997 Mar 1;126(5):376–80.
   PubMed Web of Science ®Google Scholar
• Biagi F, Klersy C, Balduzzi D, Corazza GR. Are we not over-estimating the prevalence of coeliac disease in the general population? Ann Med. 2010 Dec;42(8):557–61. Epub 2010 Oct 1.
   Google Scholar
• Catassi C, Rätsch IM, Gandolfi L, Pratesi R, Fabiani E, El Asmar R, Frijia M, Bearzi I, Vizzoni L. Why is coeliac disease endemic in the people of the Sahara? Lancet. 1999 Aug 21;354(9179):647–8.
   Google Scholar
• Bahari A, Karimi M, Sanei-Moghaddam I, Firouzi F, Hashemi M. Prevalence of celiac disease among blood donors in Sistan and Balouchestan Province, Southeastern Iran. Arch Iran Med. 2010 Jul;13(4):301–5.
   Google Scholar
• Sugai E, Hwang HJ, Vázquez H, Smecuol E, Niveloni S, Mazure R, Mauriño E, Aeschlimann P, Binder W, Aeschlimann D, Bai JC. New serology assays can detect gluten sensitivity among enteropathy patients seronegative for anti-tissue transglutaminase. Clin Chem. 2010 Apr;56(4): 661–5. Epub 2009 Dec 18.
   Google Scholar

Author's reply

F. Biagi, C. Klersy, D. Balduzzi and G. R. Corazza

Coeliac Centre/1st Department of Internal Medicine, Fondazione IRCCS Policlinico San Matteo, P.le Golgi, 19, I-27100 Pavia, Italy. Phone: [phone]. Fax: [phone]. E-mail: [email protected]

Dear Editor,

As we were writing our review on the prevalence of coeliac disease (CD) in the general population, we knew that it would not be a popular paper (1). The letter from Dr. Hoggan shows we were right.

One of the first things we wrote is that, due to the considerable heterogeneity of the published papers, a proper meta-analysis could not be performed. So, it is pretty obvious that there is some subjectivity in the interpretation of the results we obtained. However, we tried to be as objective as possible and we presented our results in the simplest and clearest way, to allow all readers to reach their own conclusions.

As far as the width of the confidence interval is concerned, the difference between the upper and lower limit is intended, and not the level of confidence. 95% confidence intervals are conventionally used around estimates (if narrow, they indicate a good reliability of the estimate itself) and we are confident that the true value will fall 95 times out of 100 within the interval.

Dr. Hoggan wrote that we did not take several papers into account. However, he quotes only a very “high-sounding” paper (2). First of all it is very difficult to write a review based on > 500 papers without at least missing one paper. Apart from that, to take into account only the papers with the most extraordinary results and to consider them to be valid for the whole of humanity is exactly what we were criticising in our paper. In his nice letter, Catassi very correctly did not write that the prevalence of CD in Saharawis is 5.6%. In fact, 5.6% is the prevalence of the endomysial antibody. It is debatable to conclude that that paper shows a prevalence of CD of 5.6% and to mention that paper as evidence that we were wrong. Dr. Hoggan also suggests that the paper by Dr Catassi et al. would have had a substantial impact on our results. We repeated the analysis. Overall prevalence is 0.74% (1/135) for analysis 1, 0.62% (1/161) for analysis 2 and 0.66% (1/151) for analysis 3.

Finally, in another very unpopular paper we showed that two thirds of diagnoses based on “minimal intestinal lesions” or negative coeliac antibodies are wrong (3). This appalling result can surely be defined as “most of the time”. Since these diagnostic criteria were used in at least some of the papers we included in our review, they may have certainly skewed the results.

Declaration of interest: The authors state no conflict of interest and have received no payment in preparation of this manuscript.