Published online August 1, 2008
PEDIATRICS Vol. 122 No. 2 August 2008, pp. 474-475 (doi:10.1542/peds.2008-1544)

This Article Extract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Request Permissions Citing Articles Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Kramer, M. S. Search for Related Content PubMed Articles by Kramer, M. S. Related Collections Nutrition & Metabolism Social Bookmarking                         What's this?

LETTER TO THE EDITOR

Effects of Prolonged and Exclusive Breastfeeding on Childhood Behavior and Maternal Adjustment: Evidence From a Large Randomized Trial: In Reply

Michael S. Kramer, MD
Departments of Pediatrics and Epidemiology and Biostatistics,
McGill University Faculty of Medicine,
Montreal Children's Hospital,
Montreal, Quebec, Canada H3H 1P3

My colleagues and I beg to differ with the claim of Poulton and Nanan that we analyzed our data "by cohorts." The Promotion of Breastfeeding Intervention Trial (PROBIT) is a randomized trial, not a cohort study,and as appropriate for a randomized, controlled trial, the analysis was based on intention to treat (ie, on randomized treatment allocation) rather than the (observed) duration or exclusivity of breastfeeding. We strongly disagree with their argument that ignoring randomization and basing our analysis on the observed duration or exclusivity of breastfeeding would be a "more logical approach." In fact, such an approach would completely vitiate the methodologic benefits of randomization in preventing selection bias and residual confounding.

Our large sample size was required precisely because of the attenuation of the treatment effect caused by the (completely expected) overlap in breastfeeding behavior in the experimental and control groups. That such a design can detect true causal effects of the experimental intervention with an intention-to-treat analysis is clearly shown by the statistically significant effects we reported on gastrointestinal infection, atopic eczema, and growth in infancy^1,2 and cognitive development at the age of 6.5 years.³ The absence of any observed effect of the intervention on child behavior, therefore, cannot simply be attributed to an insufficient difference in breastfeeding in the 2 randomized treatment groups.

REFERENCES

1. Kramer MS, Chalmers B, Hodnett ED, et al. Promotion of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. JAMA. 2001;285 (4):413 –420[Abstract/Free Full Text]
2. Kramer MS, Guo T, Platt RW, et al. Breastfeeding and infant growth: biology or bias? Pediatrics. 2002;110 (2 pt 1):343 –347[Abstract/Free Full Text]
3. Kramer MS, Aboud F, Mironova E, et al. Breastfeeding and child cognitive development: new evidence from a large randomized trial. Arch Gen Psychiatry. 2008;65 (5):578 –584[Abstract/Free Full Text]

---

PEDIATRICS (ISSN 1098-4275). ©2008 by the American Academy of Pediatrics

CiteULike    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

This Article Extract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Add to My File Cabinet Download to citation manager Request Permissions Citing Articles Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Kramer, M. S. Search for Related Content PubMed Articles by Kramer, M. S. Related Collections Nutrition & Metabolism Social Bookmarking                         What's this?