Dummies and SIDS: Causality has not been established

BMJ 2006; 332 doi: http://dx.doi.org/10.1136/bmj.332.7534.178-a (Published 19 January 2006) Cite this as: BMJ 2006;332:178
  1. Peter S Blair, research fellow (p.s.blair{at}bris.ac.uk),
  2. Peter J Fleming, professor of infant health and developmental physiology
  1. Institute of Child Life and Health, University of Bristol, Bristol BS2 8AE
  2. Institute of Child Life and Health, University of Bristol, Bristol BS2 8AE

    Editor—The relative risk for associated factors should not be confused with the attributable risk for causative factors. Li et al express the risk associated with sudden infant death syndrome (SIDS) and use of a dummy (pacifier) during sleep as a 90% reduction in risk, which could be potentially misunderstood and brings unwarranted media attention to what otherwise is a very interesting paper.1


    Any calculation of attributable risk would be limited to the prevalence of exposure, which for dummy users in the Californian infant population would be 23% (based on the control data). However, projecting risk reduction or quantifying attributable risk in this study is inappropriate. As the authors point out in the discussion, a causal effect between dummy use and SIDS has not been established. The discrepancy in dummy use has already been reported in previous larger studies,24 with better ascertainment levels providing a more cautious interpretation of the data.

    Researchers in SIDS still need to determine what the protective mechanism might be, if it exists, or whether dummy use is a marker for something else. The discrepancy in dummy use between SIDS cases and age matched control infants for the last sleep does not seem to be as marked among routine users; this may indicate that infants are at greater risk of SIDS if they routinely use a dummy but have not been given their dummy on a particular night. Any discussion of potentially modifiable factors and impact on infant care practices is better served by a review on the subject taking into account a more detailed study of infant sleep behaviour and any detrimental effect of the exposure outside this research area.5


    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    View Abstract