Maternal–infant HIV transmission may take place in utero, during labor and delivery, or postnatally through breastfeeding. Clarifying the risk of infection through each of these routes is important in order to develop interventions to prevent transmission and to monitor their effectiveness.
In 1992, Bryson et al.  proposed that in the absence of breastfeeding, intrauterine infection should be presumed if an infected child had HIV detected by polymerase chain reaction (PCR) or culture within 48 h of birth, and that intrapartum infection be presumed if an infected child had diagnostic studies (PCR, culture, p24 antigen assays) that were negative during the first week of life, but which became positive within the next 3 months. Several studies have found that when tested close to birth, less than one-half of the infants later known to be HIV-infected had HIV detected by virological tests [2–7]. It is generally inferred from these results that less than one-half of maternal–infant HIV transmission in non-breastfed infants occurs in the intrauterine period, and that the rest occurs at the time of labour and delivery.
Following the observation that first-born twins are at greater risk of HIV infection than second-born twins [8,9], several observational studies have sought associations between the risk of maternal–infant HIV transmission and circumstances of delivery [10–14]. Although there is emerging consensus on the significance of intrapartum experiences in maternal–infant HIV transmission, there are still questions about whether specific intrapartum conditions and events, including mode of delivery, duration of labour, duration of membrane rupture (DMR) prior to delivery, invasive obstetric procedures, and other delivery complications have independent roles in transmission .
If labour and delivery events are true risk factors for transmission and not markers of other unknown processes, they can only be associated with intrapartum HIV transmission, and not with intrauterine transmission. This temporal sequence provides an opportunity to use the definition of the timing of maternal–infant HIV transmission based on the results of early virological tests to study the timing of transmission. In this study, we tested the a priori hypothesis that labour and delivery events are only associated with HIV transmission presumed (according to the definition) to have occurred intrapartum, and not with transmission presumed to have occurred intrauterine. Also investigated was a related question of whether other perinatal factors (infant's gestational age and intrauterine growth retardation), maternal immunological functioning (T- lymphocyte subsets), and zidovudine therapy during pregnancy are risk factors specific to either intrauterine or intrapartum HIV transmission.
Since 1986, the New York City Perinatal HIV Transmission Collaborative Study has recruited pregnant and parturient HIV-infected women from six hospitals in New York City, and has followed the women and their infants prospectively. The study is described in detail elsewhere [16,17]. The transmission rate has been previously reported as between 19 and 25% .
As part of the study protocol, infants have blood drawn during the neonatal period for laboratory analysis. Heparin-treated blood samples are shipped overnight to the Centers for Disease Control and Prevention (Atlanta, Georgia, USA) for separation and storage at −70°C until testing. Specimens are tested for HIV proviral DNA by PCR using methods previously described . Infants have blood drawn for repeated tests at 1 month, at 2-month intervals until 6 months of age, and at 3-month intervals thereafter.
Among children later confirmed to be HIV-infected, intrauterine infection was presumed to have occurred if HIV DNA was detected by PCR in samples collected from those aged ≤ 3 days, and infection was presumed to have occurred intrapartum if HIV DNA was not detected by PCR in these early samples. The age cutoff of ≤ 3 days was chosen as an ‘early’ test because an age of up to 3 days encompasses the 48-h interval after birth of the proposed working definition , depending on the exact time the child was born. Children were considered infected if they were HIV-seropositive, confirmed by Western blot at 15 months or older, had two sequential PCR results positive for HIV proviral DNA, had an AIDS-defining illness, or if they had died with an HIV-related condition .
Maternal clinical and interview data were collected during and after pregnancy, and blood specimens were drawn. Lymphocyte subset determination was performed by standard flow cytometry procedures by the New York City Department of Health Bureau of Laboratories. For this analysis, we used the lymphocyte data collected during pregnancy or within 1 month of delivery. Detailed intrapartum information was abstracted for the study.
All liveborn infants of HIV-infected women born through January 1995 or with known HIV status were included. Not all of these infants had samples drawn within 3 days of birth for PCR tests. In the early years of the study, specific diagnostic tests for use close to birth were not available, and in later years some offsite deliveries and postpartum enrollments precluded samples within this timeframe. Because a few women enrolled in the study had more than one livebirth over the study period, we selected one child per woman (the child with the most complete neonatal PCR data, or the first child enrolled in the study) for inclusion in the analysis.
We tested whether there were differences between infants of HIV-infected women with and without early PCR results. The analysis was then restricted to those for whom PCR results were obtained within the first 3 days. We compared the proportion of infants with presumed intrauterine infection, and also the proportion with presumed intrapartum infection by specific intrapartum, perinatal and maternal characteristics, using Fisher's exact test when any cell size was less than five, or using χ2 test otherwise. Multivariate analysis was performed using logistic regression models.
Out of 650 infants of HIV-infected women followed as part of the study, 276 (42%) had the results of PCR tests conducted on samples collected within 3 days of birth. Among those with early PCR results, 48 (17%) were confirmed to be HIV-infected, compared with 75 (20%) of those without (P = 0.39). Those without early test results were more likely than those with early results to be small for gestational age (18 versus 11%, P = 0.02), to be delivered by Cesarean section (22 versus 15%, P = 0.03), and to have mothers with low CD4+ T-lymphocyte counts (≤ 200 × 106/l; 9 versus 5%, P = 0.03), and were less likely to have mothers prescribed zidovudine during pregnancy (39 versus 12%, P < 0.01), but were equally likely to have been delivered following prolonged DMR, to be preterm and to be of low birth weight (P > 0.05). HIV-infected children with and without early tests had a similar survival distribution.
Among the 276 children with early PCR results, 12 [4%; 95% confidence interval (CI), 2–7] had positive PCR at ≤ 3 days of age, and 36 (13%; 95% CI, 9–18) had negative PCR at ≤ 3 days of age but were confirmed to be HIV-infected. Therefore, among HIV- infected children, 25% (95% CI, 14–40) of infections were presumed to be due to intrauterine transmission, and the remaining 75% of infections were presumed to be due to intrapartum transmission. Among the children with presumed intrapartum infection (i.e., a negative PCR ≤ 3 days of age) 87% of the 15 retested within 45 days of birth, and 100% of the 31 retested within 5 months of birth had positive PCR results.
There were no differences in the proportion of infants with presumed intrauterine HIV infection by mode of delivery or, among vaginal deliveries, by longer DMR (Table 1). In contrast, differences were observed in the proportion of infants with presumed intrapartum HIV infection by selected intrapartum characteristics. Among infants delivered vaginally, those with longer DMR (>4 h) were significantly more likely to have presumed intrapartum HIV infection (22%) than those with shorter duration (9%, P = 0.02). If mode of delivery is ignored, 22% of infants with longer DMR had presumed intrapartum HIV infection, compared with 8% of those with shorter DMR (P < 0.01). Infants delivered by Cesarean section had a slightly lower proportion with presumed intrapartum infection (7%) than vaginal deliveries (14%), but this difference did not reach significance (P = 0.32; Table 1). Of 15 infants delivered by elective Cesarean section, two were infected: both had presumed intrauterine infection, and both were delivered by Cesarean section undertaken prior to spontaneous membrane rupture. Of 25 infants delivered by emergency Cesarean delivery, two were infected: both had presumed intrapartum infection, and both were born after prolonged DMR. The circumstances of the other two Cesarean deliveries were unknown.
Among only HIV-infected children, 40% (two out of five) of infections in infants delivered by Cesarean section and 23% (10 out of 43) in infants delivered vaginally were presumed to be due to intrauterine infection. Among HIV-infected infants delivered vaginally, 5% (one out of 19) of infections among those with longer DMR (>4 h), and 36% (five out of 14) of infections among those with shorter DMR were presumed to be due to intrauterine transmission.
Other intrapartum variables investigated (prolonged duration of labour, episiotomy or perineal lacerations, and internal fetal monitoring) were not associated with either presumed intrauterine or with presumed intrapartum HIV infection (Table 1). These results were unchanged with adjustment for DMR, or when mode of delivery was not considered.
Infants born preterm or those of low birth weight did not have significantly higher proportions of presumed intrauterine infection (Table 2). Of note, only one preterm infant confirmed to be HIV-infected was PCR-positive within 3 days of birth. This infant was delivered very preterm (24 weeks), weighed 790 g, and was delivered vaginally 11 days after membrane rupture. No preterm infant delivered ≤ 4 h after membrane rupture was infected through presumed intrauterine transmission. Infants who were small for gestational age were significantly more likely to have presumed intrauterine infection.
Infants born preterm, small for gestational age, or of low birth weight had a significantly higher proportion of presumed intrapartum infection (Table 2). Among infants delivered ≤ 4 h after membrane rupture, 8% (one out of 13) of preterm infants and 8% (eight out of 103) of term infants were infected through presumed intrapartum transmission. Among infants delivered > 4 h after membrane rupture, 32% (nine out of 28) of preterm and 17% (11 out of 64) of term infants were infected through presumed intrapartum transmission.
Presumed intrapartum HIV transmission was lower when mothers used zidovudine during pregnancy (8%) than when they did not (16%), but this difference did not reach statistical significance (Table 2). Among women with CD4 T-lymphocyte cell counts ≤ 500 × 106/l, presumed intrapartum infection occurred among 23% who used zidovudine and among 9% who did not (P = 0.06); among women with CD4 counts > 500 × 106/l, presumed intrapartum infection occurred among 5% who used zidovudine and among 14% of those who did not (P = 0.24). In univariate analysis, maternal CD4 T-lymphocyte cell count (≤ 200 × 106/l), CD4 percentage (≤ 14%), and CD4/CD8 ratio (≤ 0.5) were not associated with presumed intrauterine or intrapartum infection.
In a logistic regression model including all of the following variables, presumed intrapartum infection was less likely if zidovudine was used during pregnancy [odds ratio (OR), 0.31; 95% CI, 0.11–0.86], and more likely with maternal CD4 T-lymphocyte cell count ≤ 500 × 106/l (OR, 2.68; 95% CI, 1.0–7.21), longer DMR (OR, 4.41; 95% CI, 1.61–12.08), and infant birth weight below 2500 g (OR, 4.26; 95% CI, 1.57–11.56).
By applying a definition of the timing of maternal–infant HIV transmission on the basis of the results of early PCR to our cohort of 276 infants of HIV-infected women, we have observed that infants with greater intrapartum exposure, measured by longer DMR with vaginal mode of delivery, are more likely to acquire presumed intrapartum HIV infection than those with less intrapartum exposure, but are equally likely to acquire presumed intrauterine infection. If we assume the validity of the working definition , then the results support the notion that longer DMR among vaginal deliveries is an independent risk factor for transmission rather than simply a marker for other unknown processes. Our results are consistent with recent findings  that DMR, rather than other intrapartum parameters, is a critical risk factor for maternal–infant HIV transmission.
We did not observe associations consistent with our hypothesis between other intrapartum variables (prolonged labour, episiotomy or perineal lacerations, internal fetal monitoring) and the presumed timing of HIV transmission. Although some studies have observed associations between invasive obstetric practices [11,20], intrapartum events, and practices that increase fetal exposure to maternal blood  and maternal–infant HIV transmission, other studies have not found such associations. It is probable that these factors are not, or are only weakly, associated with HIV transmission.
The proportion of infected infants in our study with early positive PCR results (25%) is apparently lower than the approximately 40–50% of infected infants observed in other studies [2–5]. However, in those reports, the timing of an ‘early’ test comprised a wide range of ages, often extending through the first month of life. In our cohort overall, the sensitivity of HIV DNA PCR in the neonatal period was estimated to be 22% between birth and day 3, 42% by day 4, and 73% by day 14 (estimated using non-parametric methods for interval-censored data) . Thus, the exact age at which infants are tested is important when evaluating diagnostic studies on the sensitivity of early PCR tests. There were some differences, albeit inconsistent, in perinatal, maternal and infant characteristics among those children with and those without early tests. It is unclear whether this may have introduced selection bias. The magnitude of associations observed in the subset of children with early PCR tests included in this analysis between HIV transmission (both presumed intrauterine and intrapartum combined) and the intrapartum, perinatal and maternal characteristics were highly similar to that observed in the overall cohort. This argues against selection bias, although we cannot entirely exclude the possibility.
Perinatal characteristics, including preterm delivery, low birth weight, and intrauterine growth retardation, have been associated with transmission in this  and other studies [11,22]. There has been some debate over whether any association between preterm delivery and HIV transmission, if independent of confounders, is a consequence of early intrauterine infection or whether it is a cause of increased vulnerability to intrapartum infection among preterm infants . Our observation of an increased risk of presumed intrapartum infection in preterm infants tends to support the latter explanation. Furthermore, HIV was detected by PCR within the first 3 days in only one preterm infant, and this infant may well have acquired infection at the time of membrane rupture, which occurred more than 1 week before delivery. This finding is especially intriguing, and argues against early intrauterine infection as a cause of preterm delivery. Preterm infants may be more vulnerable to HIV transmission through such factors as immature immune function, incompetent mucosal barriers or lower levels of acquired maternal antibody. However, it is possible that a third factor, such as maternal virological characteristics, increases both the likelihood of preterm delivery and intrapartum HIV transmission. The finding also suggests that intrauterine transmission may occur late in pregnancy, an inference that has also been made on the basis of statistical models of early virological test results . However, the small numbers preclude any strong inference from this observation.
It has been previously reported from this cohort that women prescribed zidovudine during pregnancy are less likely to transmit HIV to their infants . Our analysis here observes that prenatal antiretroviral drug therapy is associated with HIV infection presumed to have been acquired intrapartum. Thus, treatment with zidovudine during pregnancy appears to have effects that persist to labour and delivery. The biological mechanisms still need to be elucidated.
Indicators of maternal immunosuppression, including CD4+ T-lymphocyte count, CD4+ percentage, and CD4/CD8 ratio did not appear to be preferentially associated with either presumed intrauterine or with presumed intrapartum infection in this analysis. This is in contrast to one report which found associations between detection of p24 antigen in the mother at delivery and the risk of in utero transmission . These results need to be confirmed with larger numbers of subjects, since our analysis did not have sufficient statistical power to detect small associations.
In conclusion, our results support the notion that selected intrapartum conditions, specifically long DMR prior to delivery, are independent risk factors for maternal-infant transmission, and suggest that preterm infants may be especially vulnerable to intrapartum HIV exposure. Because a substantial proportion of HIV transmission appears to occur around the time of delivery, the intrapartum period is a critical window of opportunity for HIV prevention.
1. Bryson YJ, Luzuriaga K, Sullivan JL, Wara DW: Proposed definitions for in utero versus intrapartum transmission of HIV-1 [letter]
. N Engl J Med
2. Rogers MF, Ou CY, Rayfield M, et al.
: Use of the polymerase chain reaction for early detection of the proviral sequences of human immunodeficiency virus in infants born to seropositive mothers
. N Engl J Med
3. Burgard M, Mayaux MJ, Blanche S, et al.
: The use of viral culture and p24 antigen testing to diagnose human immunodeficiency virus infection in neonates
. N Engl J Med
4. Krivine A, Firtion G, Cao L, Francoual C, Henrion R, Lebon P: HIV replication during the first weeks of life
5. Comeau AM, Hsu HW, Schwerzler M, et al.
: Identifying human immunodeficiency virus infection at birth: application of polymerase chain reaction to Guthrie cards
. J Pediatr
6. Dunn DT, Brandt CD, Krivine A, et al.
: The sensitivity of HIV-1 DNA polymerase chain reaction in the neonatal period and the relative contributions of intra-uterine and intra-partum transmission
7. Minkoff H, Burns DN, Landesman S, et al.
: The relationship of the duration of ruptured membranes to vertical transmission of human immunodeficiency virus
. Am J Obstet Gynecol
8. Goedert JJ, Duliege AM, Amos CI, Felton S, Biggar RJ: High risk of HIV-1 infection for first-born twins
9. Duliege AM, Amos CI, Felton S, Biggar RJ, Goedert JJ: Birth order, delivery route, and concordance in the transmission of human immunodeficiency virus type 1 from mothers to twins
. J Pediatr
10. European Collaborative Study: Caesarean section and risk of vertical transmission of HIV-1 infection
11. European Collaborative Study: Risk factors for mother-to-child transmission of HIV-1
12. Boyer PJ, Dillon M, Navaie M, et al.
: Factors predictive of maternal–fetal transmission of HIV-1. Preliminary analysis of zidovudine given during pregnancy and/or delivery
13. Burns DN, Landesman S, Muenz LR, et al.
: Cigarette smoking, premature rupture of membranes, and vertical transmission of HIV-1 among women with low CD4+ levels
. J Acquir Immune Defic Syndr
14. Kuhn L, Stein ZA, Thomas PA, Singh T, Tsai WY: Maternal–infant HIV transmission and circumstances of delivery
. Am J Public Health
15. Mofenson LM: A critical review of studies evaluating the relationship of mode of delivery to perinatal transmission of human immunodeficiency virus
. Pediatr Infect Dis J
16. Abrams EJ, Matheson PB, Thomas PA, et al.
: Neonatal predictors of infection status and early death among 332 infants at risk of HIV-1 infection monitored prospectively from birth
17. Thomas PA, Weedon J, Krasinski K, et al.
: Maternal predictors of perinatal human immunodeficiency virus transmission
. Pediatr Infect Dis J
18. Matheson PB, Weedon J, Cappelli M, et al.
: Comparison of methods of estimating the mother-to-child transmission rate of human immunodeficiency virus type 1 (HIV-1)
. Am J Epidemiol
19. Landesman SH, Kalish LA, Burns DN, et al.
: Obstetrical factors and the transmission of human immunodeficiency virus type 1 from mother to child
. N Engl J Med
20. Viscarello RR, Copperman AB, DeGennaro NJ: Is the risk of perinatal transmission of human immunodeficiency virus increased by the intrapartum use of spiral electrodes or fetal scalp pH sampling
? Am J Obstet Gynecol
21. Kuhn L, Abrams EJ, Chinchilla M, Tsai WY, Thea DM: Sensitivity of HIV-1 DNA polymerase chain reaction in the neonatal period [letter]
22. Goedert JJ, Mendez H, Drummond JE, et al.
: Mother-to-infant transmission of human immunodeficiency virus type 1: association with prematurity or low anti-gp120
23. Rouzioux C, Costagliola D, Burgard M, et al.
: Estimated timing of mother-to-child human immunodeficiency virus type 1 (HIV-1) transmission by use of a markov model
. Am J Epidemiol
24. Matheson PB, Abrams EJ, Thomas PA, et al.
: Efficacy of antenatal zidovudine in reducing perinatal transmission of human immunodeficiency virus type 1
. J Infect Dis
Members of the New York City Perinatal HIV Transmission Collaborative Study
New York City Department Of Health: Sarah T. Beatrice, Mary Ann Chiasson, Erica DeBernardo, Sylvia Hutchison, Katherine McVeigh, William Oleszko, Amado Punsalang.
Medical And Health Research Association Of New York City, Inc.: Tina Alford, Abraham Betre, Mark Cappelli, Nilda Carrasquillio, Nancy Cruz, Julia Floyd, Virginia Foye-Sousou, Dorothy Jones Jessop, Luis Macias, Debbie Ng, Katherine Nelson, Jeanette Rios, Vadim Pliner, Lucille Rosenbluth, Roxanne Hodge, Hany Tadros, Jeremy Weedon, Sadarryle Young, Zhong–Ren Zhang.
NYU Medical Center–Bellevue Hospital Center: Renee Courtland, Mitzi Daligadu, William Hoover, Dora Lopez, Henry Pollack, Kieth Krasinski.
Harlem Hospital Center: Ameritha Belmore, Susan Champion, Julia Floyd, Cynthia Freedland, Susan Lovich, Pamela Prince, Adrienne Rogers, Maria Suarez.
Lincoln Hospital: Jean Chow, Aditya Kaul, Sharon Nachman, Kiran Shah.
Metropolitan Hospital Center: S. Ahmed, Elmer Agustin, Roger Henriquez, Lynn Jackson, Nancy Cruz, Eileen Sacharzky, Sylvia I. Losub.
Center For Comprehensive Health Practice: Richard Brotman, Stanley Blanch, Jesse Brutus, Carol Day, David Hutson, Wendy Rhinehart, Raymond Simon, Victor Turkell.
Mount Sinai Hospital: Katherine T. Grimm.
Bronx Lebanon Hospital: Marilyn Crane, Patricia Campbell, Sarah Davila, Joanna Dobrosycki, Delia Grant, Ivan Hand, Ziv Harish, Adell Harris, Marilyn Nieves, Laurie Soloman, Aileen Steiner, Andrew Wiznia.
Montefiore Hosptital: Ellie Schoenbaum, Malathy Sivapalasingham, Marcella Naccarato, Graveola Brooks, Marjorie Nicholson, Marquerite Mayers.
Centers for Disease Control and Prevention: Sayed Garba, Chin-Yih Ou, Jennifer Moore, Martha Rogers, Charles Schable, Nathan Shaffer, R.J. Simonds, Walter Straus.
25. Rouzioux C, Costagliola D, Burgard M, et al.
: Timing of mother-to-child HIV-1 transmission depends on maternal status