Article Text

PDF

Control your controls and conclusions
  1. SIBYLLE KOLETZKO,
  2. NIKOLAOS KONSTANTOPOULOS
  1. Kinderklinik und Kinderpoliklinik
  2. Dr v. Haunersches Kinderspital
  3. Ludwig-Maximillians-University
  4. Pettenkoferstrasse 8a
  5. D-80336 Munich, Germany
  6. koletzko{at}pk-i.med.uni-muenchen.de
  7. Institute of Med. Microbiology and Hygiene
  8. University Regensburg
  9. Regensburg, Germany
  10. Unit of Epidemiology and Health Service Research
  11. The Medical School
  12. University of Leeds, Leeds, UK
    1. NORBERT LEHN
    1. Kinderklinik und Kinderpoliklinik
    2. Dr v. Haunersches Kinderspital
    3. Ludwig-Maximillians-University
    4. Pettenkoferstrasse 8a
    5. D-80336 Munich, Germany
    6. koletzko{at}pk-i.med.uni-muenchen.de
    7. Institute of Med. Microbiology and Hygiene
    8. University Regensburg
    9. Regensburg, Germany
    10. Unit of Epidemiology and Health Service Research
    11. The Medical School
    12. University of Leeds, Leeds, UK
      1. DAVID FORMAN
      1. Kinderklinik und Kinderpoliklinik
      2. Dr v. Haunersches Kinderspital
      3. Ludwig-Maximillians-University
      4. Pettenkoferstrasse 8a
      5. D-80336 Munich, Germany
      6. koletzko{at}pk-i.med.uni-muenchen.de
      7. Institute of Med. Microbiology and Hygiene
      8. University Regensburg
      9. Regensburg, Germany
      10. Unit of Epidemiology and Health Service Research
      11. The Medical School
      12. University of Leeds, Leeds, UK

        Statistics from Altmetric.com

        Editor,—In a retrospective study, Kerr and coworkers investigated formalin fixed, paraffin embedded tissues (stomach, trachea, and lung) of 32 infants who had died of SIDS, and eight control cases, with nested polymerase chain reaction (PCR) and ELISA of the amplicons. A child was considered as infected withH pylori if the optical density of the ELISA was above the mean value plus 2 SD obtained in the tissue of control infants. The authors found that 28 of the 32 SIDS cases, but only one of the eight control cases fulfilled these criteria. They conclude from their results that H pylori infection may play a causative role in SIDS. We have serious doubts about their results and conclusions.

        The control group was extremely small in size and we would expect most, if not all, of these eight infants to have received one or more antibiotics in high doses intravenously over several days before death, as the causes of death were bacterial meningitis, septicaemia, pneumonia, necrotising enterocolitis, ileal perforation, and prematurity. In contrast, few if any of the SIDS victims would have received intravenous antibiotics. Therefore, if control children had been colonised with H pylori, the bacteria may have been suppressed. These eight infants are certainly not appropriate controls for this kind of study.

        Nested PCR is a very sensitive method with a high risk of false positive results caused by contamination. The applied ELISA is yet another amplifying method which also increases the risks of unspecific binding. Although the authors stated that they tried to minimise contamination, no precautions have been performed at the time of autopsy and preservation of the tissue due to the retrospective character of the study. Because of the low specificity of the methods used, it is mandatory to prove the identity of the PCR amplicons asH pylori specific by sequencing the products. Such confirmation is not reported in the paper. To show the specificity of their method, the authors could have also performed analyses on control tissues—for example, brain, which are unlikely to be H pylori infected even when other tissues were assessed as “positive”.

        The fact that H pylori was not shown in the stomach, trachea, or lung by histology in any of the children must raise major concerns that the applied methods were not specific. Other methods for detection of H pylori infection like fluorescence in situ hybridisation (FISH) have not been applied.1 The authors do not report whether any of the children had histological signs of acute or chronic gastritis, which is found even in young children with H pyloriinfection.2 If the bacterial load was so small that neither the bacteria nor the associated inflammation could be detected by histology, it seems questionable that metabolic products produced byH pylori—for example, ammonia, may play a causative role as a cause of SIDS as suggested by the authors.

        Finally, the authors mention that both H pylori infection and SIDS are more common in poor socioeconomic populations but fail to provide any information on the ethnic and socioeconomic background of their cases and control infants. From many epidemiologic studies and our own experience, it seems extremely unlikely that 28 of 32 infants (87%) under 28 weeks of age are infected by H pylori in a country such as the UK, unless these children are from immigrant groups. We are, for example, following a cohort of German children from birth with regular testing for H pylori infection by two non-invasive tests: the detection of H pylori antigen in stool (HpSA, Meridian Diagnostics, Cincinnati, USA) and the 13C-urea breath test corrected for estimated individual CO2 production rate.3 Although a quarter of the children have at least one H pylori infected parent (positive serology and/or a positive 13C-urea breath test) only 1.5% of the children have positive tests during the first three years of age.

        On publication, this paper was widely reported by the media, a process actively assisted by the authors. This is likely to result in considerable anxiety among young parents and pregnant women, feelings of guilt in parents of SIDS children and unjustifiedH pylori eradication therapy in asymptomatic children. Since neither the selection of the control group nor the methodology used is fully robust, this study does not, however, permit valid conclusions on the association of H pylori infection with SIDS. We believe it is irresponsible to promote inconclusive results in the light of such inadequate data.

        References

        View Abstract

        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.