NEW YORK - The number of baby boys conceived by a fertility treatment known as ICSI may be lower than what is produced by Mother Nature, a new study suggests.

On average, there are 105 baby boys born for every 100 girls -- a natural advantage that helps balance out the higher number of deaths among male fetuses and infants. But in the new study, researchers found that this male-to-female birth ratio seems to be reversed when infants are conceived through intracytoplasmic sperm injection, or ICSI.

Among more than 15,000 U.S. babies born in 2005 via assisted reproduction, the investigators found that a particular ICSI approach appeared to result in a smaller-than-average number of boys.

The effect was seen when ICSI was performed using blastocyst-stage embryos -- where embryos are allowed to mature a couple days longer than the traditional norm before they are transferred to the mother. This allows doctors to transfer fewer embryos, reducing the odds of couples having triplets or higher-order births.

Among couples undergoing this procedure, just under 50 percent of births were boys. That compared with a U.S. norm of 52.5 percent for 2005, according to findings published in the journal Fertility & Sterility.

ICSI involves injecting sperm from the father directly into eggs taken from the mother; if one or more embryos develop over the next few days, they are transferred to the mother's uterus.

ICSI is typically used to treat male fertility problems, such as a low sperm count or poor-quality sperm. However, it is also sometimes used when the cause of a couple's infertility is unclear, and some fertility clinics opt to use ICSI for all patients.

The full implications of the current findings are not clear, according to the researchers, led by Dr. Barbara Luke of Michigan State University in East Lansing.

About 1 percent of U.S. births result from all assisted reproductive techniques combined, meaning ICSI accounts for only a small number of births. So it is unlikely that the small effects on male-to-female birth ratio seen in this study would have "any major implications for public health," Luke and her colleagues write.

Still, they conclude, "because our findings suggest that ICSI may reduce the sex ratio, we recommend that ICSI only be done if medically necessary, in an effort to prevent this potential side effect."

It is not clear why ICSI might reduce the proportion of male births. However, Luke and her colleagues point out, the study found no evidence that male infertility itself was related to a lower sex ratio -- supporting the idea that something about the ICSI process is to blame.