PEDIATRICS Vol. 117 No. 5 May 2006, pp. 1857-1858 (doi:10.1542/peds.2006-0409)
Ampicillin and Cefotaxime as a Risk Factor of Neonatal Death: In Reply
Reese H. Clark, MDBarry T. Bloom, MD
Dale R. Gerstmann, MD
Alan R. Spitzer, MD
Pediatrix Medical Group
Research and Education
Sunrise, FL 33323-2825
We appreciate the comments by Dr Lee. We discussed in our article1 and agree that unmeasured confounding processes of care or unmeasured variables may ultimately be found to explain our results. In the following we try to address Dr Lee's specific concerns.
We combined the reports of cerebrospinal fluid (CSF) and blood cultures to increase our statistical power to detect meaningful differences between the 2 treatment approaches in neonates who had documented infection. That is, were neonates who were treated with ampicillin/cefotaxime infected more often than neonates who were treated with ampicillin/gentamicin? Positive CSF cultures were rare (<2 in 1000 patients; Table 1).
|
In addition, most of the positive cultures were for coagulase-negative Staphylococcus (data are shown in Table 1 of our article1). We therefore evaluated the occurrence of specific pathogens in each treatment group and found that the distribution was similar (data are shown in Table 1 of our article1). Furthermore, we have now also evaluated the effect of a positive CSF culture on mortality in our multivariate analysis, and this addition did not change our results.
In Table 2 we report the mortality rate in 2 subgroups of neonates: those who did not have any report of a positive (blood or CSF) culture result and those neonates in whom there was a report of a positive blood or CSF culture. In both subgroups, mortality is higher for neonates who were treated with ampicillin and cefotaxime compared to neonates who were treated with ampicillin and gentamicin.
|
Dr Lee reinforces our observations that several diagnoses made in the first 21 days after birth occurred more frequently in the ampicillin/cefotaxime group than in the ampicillin/gentamicin group, which is why we included Table 2 in the original article.1
It is reasonable to include diagnoses that precede the decision to treat a patient with antibiotics as confounding variables. Thus, we created the variable "report of depression" to capture neonates who had a diagnosis of perinatal depression and asphyxia, and these types of diagnoses were included in our model. In contrast, liver dysfunction, renal dysfunction, and cardiac arrest may be complications of infection and/or an adverse effect of a drug. We evaluated these events as "complications" of therapy with the hope that we would identify the cause for the association between mortality and ampicillin/cefotaxime use. We were not able to find a cause, and therefore we have tried to be very cautious in our conclusions. Most of these "complications" were rare (occurring in <1% of the study cohort), and none explained the differences in mortality.
An administrative data set is not the ideal tool for definitively finding cause-and-effect relationships. Instead, the data generate an important hypothesis, namely, that the empiric use of ampicillin/cefotaxime may increase mortality when compared to the empiric use of ampicillin/gentamicin. The reason for publishing our work was to stimulate investigators to evaluate this hypothesis.
REFERENCE
- Clark RH, Bloom BT, Spitzer AR, Gerstmann DR. Empiric use of ampicillin and cefotaxime, compared with ampicillin and gentamicin, for neonates at risk for sepsis is associated with an increased risk of neonatal death.
Pediatrics. 2006;117
:67
–74
[Abstract/Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2006 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
Related articles in Pediatrics:
- Ampicillin and Cefotaxime as a Risk Factor of Neonatal Death
- Ping-Ing Lee
Pediatrics 2006 117: 1857.[Extract] [Full Text]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
