Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Infect Dis. Author manuscript; available in PMC 2009 June 15.
Published in final edited form as:
PMCID: PMC2696335

Risk Factors for Early and Late Transmission of HIV via Breast-Feeding among Infants Born to HIV-Infected Women in a Randomized Clinical Trial in Botswana


Risk factors for mother-to-child transmission (MTCT) of human immunodeficiency virus (HIV) via breast-feeding were evaluated in a randomized trial. HIV-infected women and their infants received zidovudine as well as single-dose nevirapine or placebo. Infants were randomized to formula-feed (FF) or breast-feed (BF) in combination with zidovudine prophylaxis. Of 1116 at-risk infants, 6 (1.1%) in the FF group and 7 (1.3%) in the BF group were infected between birth and 1 month)P = .99). Maternal receipt of nevirapine did not predict early MTCT in the BF group (P = .45). Of 547 infants in the BF group at risk for late MTCT, 24 (4.4%) were infected. Maternal HIV-1 RNA levels in plasma (P<.001) and breast milk (P<.001) predicted late MTCT. These findings support the safety of 1 month of breast-feeding in combination with maternal and infant antiretroviral prophylaxis.

The optimal infant feeding strategy for HIV-infected women in Botswana remains unknown. Breast-feeding has been associated with 4%-12% absolute risk of mother-to-child transmission of HIV (MTCT) through 6 months [1- 4], yet replacement feeding with infant formula carries an excess risk of infant mortality [2, 5]. We performed a randomized clinical trial (the Mashi Study) comparing 6 months of breast-feeding in combination with zidovudine (ZDV) prophylaxis with formula feeding among infants born to HIV-infected women in Botswana. At 18 months, there was no difference in the rate of HIV-free infant survival [2].

A closer analysis of these data can define the specific risks for early and late MTCT via breast-feeding and help determine groups of women for whom breast-feeding may be recommended. We therefore studied MTCT via breast-feeding in the Mashi Study to determine the following: (1) the contribution of early breast-feeding to MTCT identified between birth and 1 month, (2) the impact of maternal receipt of single-dose nevirapine (NVP) on early and late MTCT via breast-feeding, and (3) the risk factors for late MTCT.


The Mashi Study enrolled 1200 HIV-infected pregnant women at 4 sites in Botswana. The study employed a randomized, factorial design to determine the following: (1) whether single-dose NVP given to mothers and infants provided additional prevention of MTCT if mothers and infants were also receiving ZDV [6] and (2) whether prophylactic ZDV given to breast-fed infants for 6 months prevented MTCT via breast-feeding [2]. Mothers received antenatal ZDV from 34 weeks' gestation and intrapartum ZDV plus either single-dose NVP or placebo. In October 2002, highly active antiretroviral therapy (HAART) became available through the Botswana Government Antiretroviral Treatment Program and was offered to women with CD4 cell counts <200 cells/mm3 or AIDS, either during the antenatal or postnatal period. Infants initially received single dose NVP or placebo at birth and ZDV prophylaxis for 1 month (formula-fed ff arm) or 6 months (breast-fed [BF] arm) (era 1). After 17 months of enrollment, the study was modified and all infants received single-dose NVP (era 2). Informed consent was obtained from all participants, approval for the study was granted by human subjects committees in Botswana and at the Harvard School of Public Health, and the human experimentation guidelines of the US Department of Health and Human Services were followed in this trial.

We considered infants who had a negative DNA polymerase chain reaction (PCR) result at birth (defined as a test result obtained ≤15 days after birth) and their first positive DNA PCR result at the 1 month visit (defined as a test result obtained ≤45 days after birth) as potentially infected intrapartum or during early breast-feeding. Late MTCT was considered to have occurred in Infants who had a negative DNA PCR result at the 1 month visit (defined as above) and a first positive test result at any point thereafter. Infant DNA PCR testing was performed at birth, at 1, 4, 7, 9, and 12 months, and by ELISA at 18 months. Maternal demographic characteristics, baseline CD4 cell count, and baseline plasma HIV-1 RNA level were assessed at 34 weeks' gestation. Expressed breast milk samples were collected at 2 weeks, 2 months, and 5 months postpartum. The HIV-1 RNA level in breast milk was determined for available samples.

Breast-feeding women were counseled at each visit to breast-feed exclusively, and the actual feeding method as well as adverse events were evaluated each month for 6 months and at least every 3 months thereafter. Women were encouraged to wean infants from breast milk between months 5 and 6 and were provided formula and weaning foods. Women with HIV-infected infants were encouraged to continue breast-feeding.

Laboratory testing was performed at the Botswana-Harvard Partnership HIV Reference Laboratory in Botswana. Infant HIV testing was performed by PCR DNA assay (Amplicor HIV-1 Test; Roche Diagnostic Systems). ELISA for HIV antibodies was used at 18 months (Ortho AB-Capture, Ortho Clinical Diagnostics; Murex HIV2.0, Abbott-Murex). The HIV-1 RNA copy number in plasma was quantified by using the automated standard protocol, with a lower limit of detection of 400 copies/mL, employing the COBAS Amplicor/Ampli Prep HIV-1 Monitor Test V1.5 (Roche Molecular Systems). The HIV-1 RNA level in breast milk supernatant was quantified by using the automated ultrasensitive protocol, with a lower limit of detection of 40 copies/mL, employing the COBAS Amplicor/Ampli Prep HIV-1 Monitor Test.

Statistical analyses were performed by using SAS (version 9.1, SAS Institute). Infant HIV-1 infection rates between birth and 1 month were compared by use of Fisher's exact test. After 1 month of age, infection rates were estimated by the Kaplan-Meier method, and risk factors were evaluated by using Cox proportional hazards modeling. Univariable models were used to estimate associations, with some covariates (diarrhea, pneumonia, and anemia) formulated as time varying to use the most recent status (presence vs. absence) of these comorbidities at each risk-set time in the model. To reduce reverse causality, we excluded time-varying events that occurred during the month before the at-risk time estimation. A stepwise, backward elimination procedure was used to fit multivariable Cox models, retain effect modifiers, and pick the covariate with the fewest missing values and most clinical relevance among colinear covariates (namely, the HIV-1 RNA level in plasma vs. that in breast milk).


Baseline characteristics were well-matched between feeding arms (data not shown). Infant PCR testing occurred within 4 days of birth for 96% of infants, and later PCR testing occurred within 4 days of the scheduled 1-, 4-, and 7-month visits for 92% of infants. Of 1116 infants alive and HIV-free at birth, 6 (1.1%) in the FF arm and 7 (1.3%) in the BF arm were identified as infected between birth and 1 month (P = .99), representing transmission either intrapartum or during early breast-feeding. In the FF arm, no breast-feeding was reported to have occurred among the 6 infants who were infected during the first month.

Maternal receipt of single-dose NVP was not a significant predictor for MTCT between birth and 1 month in the BF arm (P = .45). There was a nonsignificant trend for early protection from maternal receipt of NVP in the FF arm (P = .12); this trend was only apparent during the first study era, before all infants received prophylactic NVP and before HAART became available for women with advanced disease. Table 1 displays the breakdown of transmission between birth and 1 month by maternal and infant receipt of NVP or placebo in each feeding arm. Pharmacologic testing supported maternal self-report of NVP receipt for 95 of 96 randomly selected women [6]. Because 937 (80%) of 1179 live-born infants in the study were randomized to receive single-dose NVP, our ability to assess the importance of the infant dose for preventing transmission of HIV during early breast-feeding was diminished. In the latter half of the study, 71 women who were receiving HAART at the time of delivery (i.e., women with a CD4 cell count <200 cells/mm3 or AIDS) did not receive single-dose NVP or placebo, which limited our ability to evaluate its effect at lower CD4 cell counts.

Table 1
Timing of mother-to-child transmission of HIV (MTCT) in the Mashi Study, by randomized feeding strategy.

Of 547 infants in the BF arm who were alive and HIV-uninfected at 1 month of age, late MTCT occurred in 24 (4.4%); it occurred in 15 infants before 4 months of age, in 6 during the period from month 4 to 6, and in 3 during the period from month 7 to 24 (table 1). An additional 4 transmissions occurred during the first 2 months of life, but the timing could not be determined to be either early or late because of missing PCR results at birth (for 1 infant) or at 1 month (for 3 infants). In the FF arm, 2 infants became infected after the 1 month visit, and 2 became infected at an undetermined time point.

Among the at-risk infants in the BF arm, univariable associations with late MTCT included higher baseline maternal plasma HIV-1 RNA level (P < .001), higher breast milk HIV-1 RNA level (P < .001), lower baseline maternal CD4 cell count (P = .004), and maternal death (P = .006). In multivariable analysis (which excluded breast milk HIV-1 RNA level, because it was colinear with plasma HIV-1 RNA level), maternal HIV-1 RNA level (P = .005), maternal CD4 cell count (P = .06), and lack of electricity in the home (P = .05) predicted late MTCT (table 2). The duration of infant ZDV prophylaxis did not predict late MTCT (P = .98). Four (16.7%) of the 24 late transmissions occurred among infants whose prophylactic ZDV had been stopped prior to their first positive HIV test result. Maternal receipt of single-dose NVP did not predict late MTCT. Although not significant, no MTCT occurred among 34 breast-fed infants whose mothers started HAART before delivery.

Table 2
Selected risk factors for mother-to-child transmission of HIV via breast-feeding after >1 month of breast-feeding.

Among infants with complete feeding information at each time point, 420 (77%) of 546 were exclusively breast-feeding at 1 month, but only 109 (21%) of 526 were exclusively breast-feeding by 5 months. Feeding method was not a significant predictor of late MTCT (P = .55, by Cox proportional hazards modeling). The progression to mixed feeding over time did not differ when late-infected and uninfected infants were compared.

The median baseline maternal CD4 cell count among women who transmitted HIV during late breast-feeding was 225 cells/ mm3 (interquartile range [IQR], 122-399 cells/mm3). The median baseline maternal HIV-1 RNA level among women who transmitted HIV during late breast-feeding was 4.9 log10 copies/mL (IQR, 4.5-5.5 log10 copies/mL). No late MTCT occurred when the HIV-1 RNA level was <3500 copies/mL.


In the Mashi Study, in which mothers and infants received antiretroviral prophylaxis and where HAART became available during the study period, MTCT during early breast-feeding was rare, and maternal receipt of single-dose NVP did not impact the rate of MTCT during early or late breast-feeding. Higher HIV-1 RNA levels in plasma and breast milk were predictors for late MTCT.

The randomized feeding design allowed isolation of risk of MTCT via breast-feeding during the first month of life, when early transmission via breast-feeding is otherwise difficult to separate from intrapartum transmission. Only 1 previous study, performed in Nairobi, randomized women to either breast-feed or formula-feed their infants [7]. That study enrolled 425 women at a time when no antiretroviral interventions were available; HIV transmission was detected in 28 breast-fed infants and 13 formula-fed infants between birth and 6 weeks of age. In contrast, our study—performed in a setting where extensive prophylaxis was possible— detected few early transmissions in either arm. Although we had low power to detect a true difference in the rate of MTCT according to feeding arm because the study was not designed to evaluate transmission during early breast-feeding, it appears that prophylaxis and the availability of HAART blunted the risk of early transmission overall. The similarity between feeding arms suggests that most new infections that developed during this early period may have represented transmission that occurred intrapartum (rather than during breast-feeding).

The World Health Organization recommends 1 week of infant ZDV prophylaxis when mothers receive a full antenatal ZDV course [8], but supporting data for this recommendation were derived from a study population that was not breast-feeding [9]. The low MTCT rate in our BF arm during the first month suggests that the full month of infant ZDV, and possibly the additional receipt of single-dose NVP for infant prophylaxis, may be beneficial. These data are consistent with recent studies suggesting benefit from extended NVP prophylaxis for breast-feeding infants [10]. The availability of HAART for women with advanced HIV infection also may have reduced risk of MTCT during early breast-feeding.

Maternal receipt of single-dose NVP was not significantly associated with MTCT during early or late breast-feeding. The extended presence of NVP in breast milk has been demonstrated to reduce the level of HIV-1 RNA in breast milk [11, 12], but this may provide little additional protection against MTCT when full maternal and infant prophylaxis regimens are used. It should be noted that our study was not powered to detect a significant early difference for maternal receipt of single-dose NVP in the BF arm, but given the low transmission rate of 1.3% between birth and 1 month, a larger study to confirm these findings may not be feasible.

We identified a higher plasma HIV-1 RNA level as the most significant risk factor for MTCT during late breast-feeding in multivariable analyses. The level of HIV-1 RNA in breast milk was excluded from the multivariable analysis, but was a significant univariable predictor (and is a known predictor for late MTCT [13, 14]). Of the 24 late transmissions observed in our study, none occurred when baseline maternal HIV-1 RNA level was <3500 copies/mL. The range of CD4 counts at which transmission occurred was large, although the median baseline CD4 cell count among women who transmitted HIV was only 225 cells/mm3.

In summary, the main risk factors for MTCT via breastfeeding appear to be higher breast milk and plasma levels of HIV-1 RNA, and probably lower CD4 cell count. These factors may be modified by maternal use of HAART. In the setting of extensive maternal and infant prophylaxis, the rate of MTCT was low during the first month of life and neither breastfeeding nor maternal receipt of single-dose NVP predicted transmissions. These findings may have policy implications in locations where early formula-feeding has been associated with high rates of infant mortality (and where a short period of breast-feeding in combination with infant prophylaxis might be a viable recommendation) and where NVP-based treatment and MTCT prevention programs coexist.


We are indebted to the patients who participated in this study, and thank the Mashi Study team, which included

Chuka Anude, Jonas Chanda, Lillian Makori, Janet Banno Moorad, Taolo Agnes Modise, Tholakele Moyo, Dipotso Arbi, Matseo Malamba, Kgomotso Koloi, Lerato Dube, Tumisang Mmolotsi, Setho Babitseng, and Dorcus Mere (Molepolole Site); Jenny Boyle, Jane Magetse, Tumalano Sekoto, Venice Modikwa, Lebopo Garebatho, Margaret Tsuro, Merriam Sesinyi, and Kabo Kelebalekgosi (Mochudi Site); Zegabriel Tedla, Gloria Mayondi, Ntukunu Makubate, Keatile Sebinang, Lesedi Tsalaile, Boniny Tsule, Irene Thebeetsile, Jessica Setswalo, Irene Leteane, and Oarabile Makgabana (Lobatse Site); Mpho Mogodi, Anchilla Owor, Innocent Hove, Aida Asmelash, Tebogo Kakhu, Phana Ramalepa, Joyce Lubinda, S'khatele Ndebele, Florence Modise, Chenesani Bohule, Kebong Motshabi, and Malegogo Ntshimane (Gaborone Site). We also wish to thank Chris Rowley, Shabnam Zavahir, Ria Madison, Tlhongbotho Masoloko, Mary Fran McLane, Edward Garmey, Sui Yuan Chang, Enoch Sepako, Gaseene Sebetso, Nketso Monametsi, Ying Wu, Tapologo Nkoane, Betsi Naledi Bome, Fatima Chand, Ann Marie Reich, Molly Pretorius Holme, Kenneth Onyait, Sandi Johnson, Sarah Yates, Mpho Mmelesi, Kelesitsi Phiri, Carol Suckow, Onalenna Ntogwa, Reginah Molefe, Tlhalefo Maotwe, Best Mafoko, Amelia Sanders, Keafiwa Maswere, Julia Setimela, Lynne Mofenson, Jack Moye, and Anne Willoughby. We thank the Botswana Ministry of Health for providing nursing and driver support, as well as the antiretroviral drugs and monitoring for women and infants on highly active antiretroviral therapy; and the administration, and the maternity ward and Maternal Child Health staff at the Scottish Livingstone Hospital, Deborah Retief Memorial Hospital, Athlone Hospital, and Princess Marina Hospital. We thank the District Health Team Clinics in Molepolole, Mochudi, Lobatse, and Gaborone.

Financial support: National Institute of Child Health and Human Development (R01-HD37793 to M.E., K23-HD01330 to R.L.S., and K23-HD40451 to S.L.).


Potential conflicts of interest: none reported.


1. The Breastfeeding and HIV International Transmission Study Group Late postnatal transmission of HIV-1 in breast-fed children: an individual patient data meta-analysis. J Infect Dis. 2004;189:2154–66. [PubMed]
2. Thior I, Lockman S, Smeaton LM, et al. Breastfeeding plus infant zidovudine prophylaxis for 6 months vs formula feeding plus infant zidovudine for 1 month to reduce mother-to-child HIV transmission in Botswana: a randomized trial: the Mashi Study. JAMA. 2006;296:794–805. [PubMed]
3. Miotti PG, Taha TE, Kumwenda NI, et al. HIV transmission through breastfeeding: a study in Malawi. JAMA. 1999;282:744–9. [PubMed]
4. Coutsoudis A, Pillay K, Kuhn L, Spooner E, Tsai WY, Coovadia HM. Method of feeding and transmission of HIV-1 from mothers to children by 15 months of age: prospective cohort study from Durban, South Africa. AIDS. 2001;15:379–87. [PubMed]
5. Coovadia HM, Rollins NC, Bland RM, et al. Mother-to-child transmission of HIV-1 infection during exclusive breastfeeding in the first 6 months of life: an intervention cohort study. Lancet. 2007;369:1107–16. [PubMed]
6. Shapiro RL, Thior I, Gilbert PB, et al. Maternal single-dose nevirapine versus placebo as part of an antiretroviral strategy to prevent mother-to-child HIV transmission in Botswana. AIDS. 2006;20:1281–8. [PubMed]
7. Nduati R, John G, Mbori-Ngacha D, et al. Effect of breastfeeding and formula feeding on transmission of HIV-1: a randomized clinical trial. JAMA. 2000;283:1167–74. [PubMed]
8. World Health Organization (WHO) Recommendations for a public health approach. WHO; Geneva, Switzerland: 2006. Antiretroviral drugs for treating pregnant women and preventing HIV infection in infants: towards universal access.
9. Lallemant M, Jourdain G, LeCoeur S, et al. A trial of shortened zidovudine regimens to prevent mother-to-child transmission of human immunodeficiency virus type 1. N Engl J Med. 2000;343:982–91. [PubMed]
10. Kumwenda NI, Hoover DR, Mofenson LM, et al. Extended antiretroviral prophylaxis to reduce breast-milk HIV-1 transmission. N Engl J Med. 2008;359:119–29. [PubMed]
11. Chung MH, Kiarie JN, Richardson BA, Lehman DA, Overbaugh J, John-Stewart GC. Breast milk HIV-1 suppression and decreased transmission: a randomized trial comparing HIVNET 012 nevirapine versus short-course zidovudine. AIDS. 2005;19:1415–22. [PMC free article] [PubMed]
12. Rossenkhan R, Ndung'u T, Sebunya TK, et al. Receipt of single dose nevirapine in labor may defer rather than blunt the postpartum breast milk HIV viral load rebound seen after the discontinuation of ZDV [abstract TUPEC029]. Program and abstracts of the 4th International AIDS Society Conference on HIV Pathogenesis, Treatment, and Prevention; Sydney, Australia. 2007. Available at: Accessed 8 December 2008.
13. Taha TE, Hoover DR, Kumwenda NI, et al. Late postnatal transmission of HIV-1 and associated factors. J Infect Dis. 2007;196:10–4. [PubMed]
14. Rousseau CM, Nduati RW, Richardson BA, et al. Longitudinal analysis of human immunodeficiency virus type 1 RNA in breast milk and of its relationship to infant infection and maternal disease. J Infect Dis. 2003;187:741–7. [PMC free article] [PubMed]