Is ingestion of milk-associated bacteria by premature infants fed raw human milk controlled by routine bacteriologic screening?

Expressed human milk is often used to feed premature infants. Raw milk contains bacteria which may be a source of infection. Milk banks have developed screening programs which combine periodic quantitative milk cultures with arbitrary rules specifying limits of bacterial concentration. It is unknown whether such programs succeed in preventing infants from being fed milk containing bacteria. At the Health Sciences Centre (Winnipeg, Manitoba, Canada), milk is screened once weekly. When a woman's milk is found to have excess bacteria, it is discarded only if she is an unrelated donor (as opposed to an infant's mother). To assess the effectiveness of this screening program, we determined the frequency at which infants fed raw human milk were exposed to milk-associated bacteria and compared the bacterial contents of donor and maternal milk. From February 1986 to April 1987, all human milk fed to 98 premature infants during the first 2 weeks of feeding (n = 10,128 feeds) was cultured quantitatively. Among study infants, 100% were exposed at least once to coagulase-negative staphylococci, 41% were exposed to Staphylococcus aureus, and 64% were exposed to gram-negative bacilli. The proportions of feeds containing bacteria and the quantities (log10 CFU [mean +/- standard deviation]) ingested per positive feed were: 39% and 5.9 +/- 0.5 for coagulase-negative staphylococci; 2.4% and 5.1 +/- 1.0 for S. aureus; and 5.2% and 4.8 +/- 1.1 for gram-negative bacilli. There were no adverse events attributable to ingestion of milk-associated bacteria. Milk coagulase-negative staphylococcal isolates were multiply antibiotic susceptible, whereas infant isolates were antibiotic resistant. Donor milk was significantly less likely than maternal milk to contain coagulase-negative staphylococcal species in any quantity (40 versus 93% of samples, respectively [P < 0.001]) or in concentrations exceeding 10(8) CFU/liter (3 versus 27% of samples, respectively [P < 0.0001]). There was no difference between milk from either source in terms of S. aureus or gram-negative bacterial content (4 to 6%). These results suggest that the Health Sciences Centre screening program is effective in limiting the number of harmless coagulase-negative staphylococcal species but has no impact on the quantity of potentially pathogenic bacteria ingested by premature infants. Implications for screening donor milk are discussed.