We read with interest the recent article on this
subject(1). In the backdrop of the information already available from
sub-Saharan Africa, Thailand and Europe, the conclusions of the
authors are not surprising. However, there are several limitations
that are relevant to the interpretation of the results of the study in
question.
Firstly, the two groups (intervention versus
non-intervention) are non-comparable, because this was a non-blinded
and non-randomized intervention trial. There is an obvious bias in
patient selection, because the women in the intervention group
belonged to a higher socio-economic class, who could afford therapy;
and also had a more desirable health seeking behavior, having come for
antenatal visits from before 32 weeks gestation. This in itself could
confound the results because severity of underlying HIV disease,
nutritional status, vitamin deficiency status, co-infections with
other sexually trans-mitted disease, etc. could all be related
to economic class and health seeking behavior, and these factors could
independently influ-ence the outcome. This is further com-pounded by
the fact that in the non-interven-tion group, the dropout rate is
enormous and is far higher than in the intervention group.
Secondly, we have no information whether there was
any cross-contamination between the two groups. Despite being in the
non-intervention group, it is possible that many patients in this
group would have also undergone Caesarean sections beyond 38 weeks for
reasons other than HIV, thus getting an inadvertent benefit. Since the
mothers were counseled about the pros and cons of breastfeeding and
the final decision was left to them, it is quite likely that there
would be some mothers in the intervention group who opted to
breastfeed and vice versa in the non-intervention group. It is
not clear from the article what were the numbers in each of these
sub-groups and how they were tackled in analysis. Since the authors
are testing a package of interventions, rather than individual
interventions, it is desirable that there should be homogeneity within
each group.
Thirdly, the authors say that the ELISA and Western
blot tests can be falsely positive till 15 to 18 months of age. In
that case why did they test at 15 to 18 months by ELISA to confirm
"infection status" and use this parameter as their key
outcome. They should have either chosen a time period that was clearly
beyond the grey zone, e.g., 21 to 24 months, or they should
have chosen a test that unequivocally demonstrate presence of virus,
such as PCR or p24 antigen. The test at 15 to 18 months continues to
leave one in doubt whether these were genuine infections or passive
transfer of antibodies.
Fourthly, only the cost of zidovudine treatment of
the mothers has been mentioned. The cost of treating the babies for 6
weeks (roughly Rs 100 to Rs 150), the cost of feeding formula milk for
many months, and the cost related to increased infections in the
formula fed babies are qually relevant. Only then can the cost-benefit
ratio of the package of interventions be ascertained.
Saurabh Dutta,
Anil Narang,
Neonatology Unit, Department of Pediatrics,
Postgraduate Institute of Medical Education and
Research,
Chandigarh 160 012, India
Reference
1. Merchant RH, Damania K, Gilada IS, Bhagwat RV,
Karkare JS, Oswal JS, et al. Strategy for preventing vertical
transmission of HIV: Bombay experience. Indian Pedia-trics 2001; 38:
132-138.