Dear Editors:
In reviewing Poole J et al, Timing of Initial Exposure to Cereal
Grains and the Risk of Wheat Allergy, Peds, June 2006 117(6):2175-2182,
the reader appreciates the academic skills demonstrated, and honest
presentation, such as the authors’ upfront declaration that the study
population was selected for a family history of diabetes. Unfortunately,
this is only one of the many factors that lend doubt to the findings, in
spite of the efforts of the authors to include all possible confounders
available from this data set in their analyses. While it is interesting to
explore this issue in an existing data set, there are simply too many
issues with these data to allow the conclusions presented.
The authors start with the recognition that they are using a
population of known propensities – therefore not genetically nor probably
behaviorally necessarily representative of the general population. Then,
due to data constraints, they include a variable for family history of
allergy that does not specify severity, type of allergy, or maternal vs
paternal. This is further problematic in that such loose familial
experience could bias maternal interpretation of a symptom as allergy vs
other. Additionally, they rely on maternal diagnosis of allergy as the
major outcome variable, without systematic recording of symptoms and with
no testing to confirm the negatives. Only 4 of the 16 actually tested
positive for wheat-specific IgE. While all four began wheat after 6
months, the authors correctly acknowledge that the timing of the testing
may have been a factor increasing the attribution to later onset.
Further, nearly half of the reported cases are eliminated for
legitimate reasons, but that are of a sufficient number to mask many other
factors. Also, there is control for history of food allergy before six
months, eliminating those with an immediate reaction to wheat feeding at 4
-5 months from direct inclusion. The authors note the problems of lack of
control for cow’s milk sensitivity, known to have some relationship with
maternal perception of wheat sensitivity, and the table shows that the
strongest association with the occurrence of wheat allergy was allergic
reaction before 6 months.
Finally, and perhaps most importantly, duration of breastfeeding,
even after controlling for what factors they could, was not statistically
significantly associated with the outcome.
Again, my congratulations to the authors for their careful
documentation and their effort to fully utilize an existing data set; it
is only the high quality of the authors’ methods section and presentation
of the findings that allow for this critique. However, the many points
listed above lend significant doubt concerning the reliability of the
findings and the conclusions.
In sum, the findings from this secondary analysis do not warrant the
epidemiological or clinical interpretation offered in the conclusions. It
could be considered very inappropriate, and possibly unconscionable, to
offer general clinical recommendations from these data and this analysis,
let alone a recommendation that may have a negative impact on support for
sustained exclusive breastfeeding.
Thank you for your consideration.
Miriam H Labbok, MD, MPH
FACPM, IBCLC, FABM
Professor of the Practice of Public Health
Director, Center for Infant and Young Child Feeding and Care
Department of Maternal and Child Health
School of Public Health, CB#7445
The University of North Carolina at Chapel Hill
Chapel Hill, NC 27599-7445
Tel: 919-966-0928
Fax: 919-966-0458
labbok@unc.edu
www.sph.unc.edu/mhch/ciycfc
Conflict of Interest:
None declared
eLetters: ![[Read eLetters]](/icons/misc/sectionDOWN.gif){kind=icon}
