I wonder if it came from here?
www.isisonline.org.uk/hcp/where_babies_sleep/parents_bed/how_parents_bedshare/bedsharing_and_nonbreastfeeders/
They state that the findings are ambiguous but
Bed-sharing babies of breastfeeding mothers appear to avoid the presumed hazards of sleeping in adult beds (e.g., suffocation, overlaying, wedging, entrapment) (Nakamura et al 1999), due to the presence and behaviour of their mothers. Interestingly, however, differences have been observed in bed-sharing behaviour between breastfeeding and formula-feeding mothers and babies (Ball 2006). In a study comparing families videoed sleeping in their home environment, formula-fed infants were generally placed high in the bed, with babies at parental face-height, and positioned between, or on top of, parental pillows. In contrast, breastfed babies were always positioned flat on the mattress, below pillow height and level with the mother's chest. Formula-feeding mothers spent significantly less time facing their baby and in mutual face-to-face orientation than did breastfeeding mother-baby pairs, and they did not adopt the "protective" sleep position with the same degree of consistency. Breastfeeding mothers and babies experienced a significantly greater frequency of arousals from sleep, and significantly more of these were synchronous (mother and baby waking together) than among formula-feeding mothers and babies (Ball 2006)
The patterning of these differences is consistent with an understanding of the physiological mechanisms mediating maternal and infant behaviour, in that breastfeeding mothers experience a hormonal feedback cycle, which promotes close contact with, heightened responsiveness towards, and bonding with infants in a way that is absent or greatly diminished among mothers who do not breastfeed (Uvnas Moberg 2003). The implication here-that breastfeeding mothers and babies sleep together in qualitatively and significantly quantitatively different ways than do non-breastfeeding mothers and babies-suggest that epidemiological studies of bed-sharing that have not considered feeding type as a variable for matching cases and controls may have drawn inappropriate results in assessing risk factors associated with bed-sharing. Hopefully epidemiologists will re-examine these issues.