To the Editor: We are pleased that Drs. Wisner and colleagues consider our report on paroxetine in breast milk to be a significant contribution to the literature. They raise three issues for discussion: 1) the clinical interpretation of 0 ng/ml in previous serum measures in nursing infants (1, 2), 2) the theoretical value of measuring inactive metabolites in nursing infants, and 3 ) the policy of informing the pediatrician of maternal paroxetine use.

First, our concern over the potential misinterpretation of 0 ng/ml by clinicians is not negated by the detailed description of how Dr. Wisner and colleagues handled their data set. The interpretation of serum concentrations in nursing infants, in the absence of treatment-emergent side effects, should be both conservative and scientific. To report that an infant’s serum concentration was 0 ng/ml on the basis of either visual or computer interpretation of a chromatograph of ultraviolet absorption would be a less conservative interpretation. To date, all antidepressants studied are found in the breast milk of women taking the medication; the infant is exposed, regardless of infant serum concentration. Further, Dr, Wisner and colleagues reported infant serum concentrations as whole numbers (e.g., not as decimals). This implies that 0.1–0.4 ng/ml could be regarded as 0 ng/ml; they acknowledge as much with the statement "within instrumental limitations." We agree that the limits of detection and quantification are not the same, but they are the lowest value that can be reported. In analytic chemistry, "0" does not exist, and the limits of the assay (be it detection or quantification) convey the most scientific representation of the data (3). The description of categorical division on the basis of infant serum concentration is interesting but fails to address our concern that clinicians may misinterpret the data.

Second, the article cited (Findling et al., 1999) was not available at the time we submitted the manuscript of our paroxetine study. To consider the analysis of a nonactive metabolite more significant that accurate assay reporting is overstated. The only mechanism for combining the burgeoning data is to standardize data reporting and, ideally, the methodology used. The potential use of an inactive metabolite concentration (when the parent compound was detectable in the majority of mothers and in all breast milk samples) as a measure of treatment adherence appears to be of limited value. The pharmacokinetic study by Findling and colleagues (Findling et al., 1999) is of great interest but is not easily extrapolated to neonates since the youngest subject in that study was 6 years old.

Finally, Dr, Wisner and colleagues have stated that the pediatrician must be informed of the clinical treatment decision (Wisner et al., 1996). The complexity of this issue extends beyond the use of psychotropic medications in lactation to include the management of other medical conditions during pregnancy and lactation. Collaborative communication with obstetricians, pediatricians, and neonatologists represents the ideal and is strongly encouraged. We agree that communication across subspecialties is preferable, although such clinician-to-clinician contact is not mandatory in our programs as it is secondary to potential infringement on the mother’s confidentiality.

We appreciate the opportunity to discuss the issues raised by Dr. Wisner et al. We look forward to seeing how the enhanced scientific rigor being applied to this area ultimately affects clinical practice.