We administered a standard questionnaire to the mother to assess risk factors associated with HIV perinatal transmission. The variables collected included age of mother and infant, breastfeeding status, history of sexually transmitted diseases, number of sexual partners, and condom use. Breast-feeding status was measured by self-report.

We also asked the mothers about prior history of exposure to ARV and whether the father of the infant had used ARV therapy. To assign the stage of maternal HIV disease, we asked the mothers about history of symptoms associated with HIV/ AIDS disease and based the staging on the World Health Organization criteria for staging ⁶. The mothers in stage 3 and 4 were combined into one group since there were only 3 mothers who fell into stage 4 category.

A series of rapid test kits were administered in series to screen mothers for HIV-1 antibodies. The health centers were already using a similar rapid HIV testing protocol in their voluntary counseling and testing program. In a series of 3 rapid kits, the Determine ^® (Abbot Laboratories, Abbot Park, IL) test kit was used as the initial test for HIV-1 antibodies. If a mother tested negative on Determine^®, she was considered as being negative for HIV-1 antibodies and no further HIV tests were performed. However, If the sample test was positive on the Determine^® test, the Unigold ^® (Trinity Biotech) test was done as a confirmatory test. The third rapid test available was Serocard ^{® (}Trinity Biotech) and was done only if the results of the first two tests were indeterminate. The results were available to the mothers within an hour after the blood draw was done.

The results were given to the mother after a posttest counseling session. Experienced counselors at each clinic carried out the counseling. We tested the infants for HIV using qualitative Ribonucleic acid polymerase chain reaction (RNA PCR) with the Roche AMPLICOR MONITOR assay (Roche Diagnostics, Indianapolis, IN, US). The infant HIV testing was done at the Makerere University-Johns Hopkins-Case Western Reserve University Core laboratory at Mulago Hospital in Uganda.

We determined the descriptive statistics for all the variables and then compared the baseline characteristics in the two treatment categories. HIV transmission rates were estimated separately for the naïve and treated mothers as a proportion of infants with HIV infection. We used unconditional logistic regression to determine the odds ratio (95% confidence interval) for HIV infection in the child according to treatment status of the mother, maternal stage of disease, and breastfeeding status of the infant. We used a stratified analysis and multivariable logistic regression to adjust for confounding and to detect effect modifiers. Data was entered into Epi-Info Version 6 (CDC). Analysis was done using Statistical package for Social Scientists, SPSS ^® (Version 10) and Statistical Analysis Software, SAS ^® (Version 8)

RESULTS

Study population

Between June and October 2000, some 1003 naïve mothers were screened for HIV-1. Of these mothers, 129 (13%) mothers tested positive for HIV-1 antibodies. Of the 129 mothers, 90 infant-mother pairs were enrolled in the study (Figure 1). The rest of the mothers (n=39) were not enrolled because they did not give consent to have their infant in the study, or there was failure to draw blood from the infant, in some (n=8) infants the amount of blood was insufficient to perform the RNA PCR.

During the same time period, 115 HIV-infected mothers treated with either nevirapine or zidovudine as part of the implementation program were screened and all were enrolled in the study. A total of 109 infant-mother pairs were enrolled from the treated group. Of these 109, 48 mothers had received AZT while 61 mothers had received nevirapine. For 6 mother-infant pairs, an HIV status of the infant was not determined for reasons similar to those among naïve mothers.

Among the 45 mother-infant pairs who were excluded or declined to participate, infants were more likely to be born to naïve mothers and were more likely to be breastfed. There were otherwise no significant differences between the mother-infant pairs with and without HIV results of the child in terms of birth weight, gender, and age of infants, or history of maternal sexually transmitted diseases. Because of the absence of HIV results for a proportion of infants, further analysis is based on infants with HIV results available. The final study population comprises 90 naïve mother-infant pairs and 109 treated mother-infant pairs.

The treated mothers were slightly older than the naïve mothers, had older husbands and tended to have had more previous pregnancies compared to the naïve mothers (Table 1). The two groups of mothers were, however, comparable in terms of marital status, place of residence and the number of other wives that their husbands had (Table 1). The naïve mothers were less likely to use condoms than the treated mothers but had had fewer sex partners over the past four years compared to the treated mothers (Table 1). Naïve mothers also had a more advanced stage of HIV disease. The two groups of mothers were comparable in terms of history of symptoms associated with sexually transmitted diseases STDs .

There was no significant difference in the birth weights among infants born to treated or naïve mothers (Table 2). The infants of naïve mothers were on average older than those whose mothers were treated (38.5 vs. 8.7 weeks p< 0.001) and were more likely to have breastfed (Table 2). Some infants in the naïve mothers group (4.5%) received antiretroviral therapy postpartum whereas most of the infants in the treated mothers group (90%) received antiretroviral therapy. There were 4 infants who were treated but had mothers who had not received antiretroviral therapy because of lack of drug supply at the time of delivery or mother refusing to swallow the medications before labor.

Overall transmission rates of HIV

Among the 90 naïve mother-infant pairs, 43 infants (47.8%) were HIV-1 infected whereas among the 109 treated mother-infant pairs, only 18 infants (16.5%) were HIV positive. This gives an unadjusted odds ratio in the treated compared to the naïve mothers of 0.22 (95% CI 0.1-0.4). When the treated mothers were stratified according to the medication received, the proportion of infected infants was comparable between the AZT-treated and nevirapine treated mothers (8/48 or 16.7% and 10/61 or 16.4% in the AZT and nevirapine groups respectively.

HIV infection rates within each infant-age category were calculated among infants born to naïve mothers to show the trends in the prevalence of HIV (Table 3). The results showed a trend towards increasing proportion of HIV infection rates with increasing infant age (chi square for trend P<0.0002). Among children younger than 20 weeks, 37.5% of infants were HIV infected; this proportion rose to 50% among infants aged between 20-40 weeks. A similar observation was made among children of treated mothers (Table 3). In naïve mothers the proportion of infected children did not increase with a more advanced stage of HIV infection. In contrast, in treated mothers the proportion of infected children rose with the stage of disease. The naïve mothers were more likely to have breast fed their infants compared to the treated mothers (Table 3).

In a univariate logistic regression analysis, maternal antiretroviral therapy was significantly protective of HIV in the infant (OR= 0.21, 95% CI 0.1-0.4) The univariate analysis tended to show an increased risk for HIV transmission to the infant with presence of genital ulcer in the last 4 years (OR=1.71; 95% CI 0.91-3.2), history of an STD in the past four years (OR=1.67; 95% CI 0.89-3.14), and among male infants (OR=1.47; 95% CI 0.79-2.74) but none was statistically significant. The age of infants was associated with HIV infection as older infants and children were more likely to be infected compared to the younger ones. Breastfeeding was associated with an increased risk for transmission (OR=1.5, 95% CI 0.78-3.0), but this risk was not statistically significant.

A stratified analysis was done according to the breastfeeding status of the mothers and showed a protective effect of ARV in each stratum. Among the women who were breastfeeding, the odds ratio (OR) for infection among treated mothers compared with naïve mothers was 0.18 (95% CI 0.08-0.44) and among the mothers who did not breastfeed their infants, the OR was 0.24 (95 % CI 0.06-0.9). There was an increasing risk of HIV transmission with the advancing stage of HIV disease with the risk increasing from 1.27 times (95% CI 0.61-2.64) among mothers in stage 2 to 1.76 times (95% CI 0.85-3.68) among mothers in stage 3/4. In a multivariable logistic regression analysis adjusting for stage of disease, breastfeeding and age of the infant, the protective effect of antiretroviral therapy in the prevention of HIV infection in the child persisted (OR= 0.22; 95%CI 0.09, 0.54).

This observational study shows the effectiveness of antiretroviral prophylaxis in the prevention of perinatal transmission of HIV in a government-sponsored program in a developing country. The results show that a brief course of oral ARV therapy can successfully reduce mother-to-child transmission of HIV. In a pilot program that implements ARV in pregnant mothers in Uganda, antiretroviral therapy reduced the risk of perinatal transmission of HIV by 78%. The benefit of ARV remains even after adjusting for the confounding effect of breastfeeding, maternal stage of HIV disease and age of infant. Furthermore, the level of reduction in transmission was similar for nevirapine and zidovudine treatment groups. The degree of protection found in this observational study was greater than that found in the HIVNET 012 and Thai CDC clinical trials. The observational nature of the study and also the significant baseline differences observed between the treated and naïve groups in our study could explain this difference. In particular, the naïve group in this study had a high level of transmission in part because of the older age of infants and the associated longer duration of breastfeeding. Therefore, the protective effect may be exaggerated because of the high level of transmission in the naïve group through breastfeeding.

As was observed in the HIVNET 012 study from Uganda , ARV therapy reduces but does not eliminate the risk of perinatal transmission of HIV. Indeed, the transmission of HIV still remains high (17%) despite administration of ARV therapy in this observational study. What are the potential reasons for this treatment failure? First, transmission may occur in-utero prior to the administration of ARV ^7-11.

The frequency of early in-utero HIV infection appears to be low, compared with perinatal transmission rates in infants born to HIV-1-infected mothers ¹². The strategy of peripartum therapy targets the high-risk period of transmission during labor and delivery. Second, in more advanced stages of HIV disease, brief courses of ARV may not suppress viral replication sufficiently to prevent transmission. Third, after delivery breastfeeding may promote transmission of HIV ^13-19. Finally, in the absence of these factors, intrinsic drug resistance to antiretroviral therapy may account for the continued transmission^20-24. There are no large scale studies done to-date to accurately estimate the proportion of HIV transmission to the infant attributable to de novo resistance in the mother despite ART therapy. However some case studies and other small studies have shown evidence of mother-to-child transmission of HIV drug resistant strains ^25-26. In the first case ever reported of mother-to-child transmission of drug resistant HIV ²⁵, the AZT resistance was not attributable to AZT given at the end of pregnancy or during labor rather AZT resistance was present before gestation. In the French Perinatal Cohort study, sequence analysis was carried out on 34 infected children and mutations related to AZT resistance were found in 7 of the 34 children ²⁷. Evidence of mother to child transmission of AZT resistant HIV-1 was found in 4 cases. However, in the assessment of mothers and infants from the HIVNET 012 study, the authors found little evidence for transmission of nevirapine-resistant variants from women to infants who were infected by 6-8 weeks of age ⁴⁰.