HIV and Infant Feeding: Tables 1 & 2

HIV and Infant Feeding:
Tables

Publication Text
References

Table 2: Major Mathematical Models of HIV Transmission and Breastfeeding

Table 1: Major Studies on HIV and Breastfeeding in Developing Countries

Site, Author, Pub.
Date, Sample Size,
Type of Study Objective HIV Test(s) Used
for Infants/ Children Breastfeeding
Definition(s) Used Definition(s) of
Timing of Transmission Major Findings Interpretation

Côte d’Ivoire
Ekpini et al.
Date:
9/90–10/94
Sample size for this analysis:

82 children born to HIV-1 positive mothers
57 infants born to dually infected mothers (HIV-1 & HIV-2)
30 infants born to HIV-2 positive mothers
all were breastfed
39 were HIV-infected at the end of the follow-up

Study design:
Longitudinal, observational
To estimate the risk of HIV transmission through breastfeeding. EIA (HIV-1 and HIV-2) and PCR. Serology obtained at 1, 2, and 3 months of age, and every 3 months after that. Authors suggest that PCR techniques have sensitivity and specificity of 100%. No specific definitions used or practices measured.
The median duration of breast-feeding among children followed for at least 24 months was 20 months.
Early HIV infection (in-utero or intrapartum) was defined as a positive HIV-1 PCR obtained in first 6 months of life.
Late postnatal infection was attributed to infants with a negative PCR at 3 or 6 months, followed by either or both positive HIV-1 PCR at 9 months or older, or persistently positive HIV-1 serology at 15 months or older.
Analysis adjusted for losses to follow-up and for weaning.
Total vertical transmission rate was about 25%.
20% of children acquired HIV infection in first 6 months of life.
12% of all children of HIV-1 positive mothers who escape infection during first 6 months of life will become HIV-positive by 24 months.

Late postnatal transmission is greater, 20%, among children breastfed for at least 24 months.
No instances of late postnatal transmission were observed in infants of mothers infected with HIV-2 only.
Rates of late postnatal transmission were 9.2 per 100 child years of breastfeeding among mothers infected with HIV-1 only, and 4.8 per 100 child years of breastfeed-ing for dually infected women.
Breastfed children born to HIV-1 positive or dually infected mothers in Abidjan are at substantial risk of late postnatal transmission.
Most postnatal transmission is likely to have occurred through breastfeed-ing and not transfusions, injections, or other exposures to HIV-infected blood.
Postnatal transmission may have been underestimated since some children with a positive PCR in first 6 months may have acquired the infection through breastfeeding.
Breastfed children are at a higher risk of infection if mothers seroconvert during the postpartum period.
Weaning at 6 months could reduce risk of HIV transmission through breastfeeding.

Kenya
Datta et al.
Date:
12/90–12/92
Sample size for this analysis:

220 children surviving for at least 12 months
all born to HIV-1 positive mothers
90 children were HIV positive at the end of follow-up

Study design:
Longitudinal; observational

To study the frequency of, and risk factors for, mother- to-child transmission of HIV-1.
ELISA with confirmatory immunoblot.
Blood for serologic assays was collected at 6, 14, and 24 weeks; 9, 12, 15, and 18 months; and every 6 months thereafter.

Breastfeeding initiation was universal in the study population.
Infants who survived for at least 12 months were classified according to whether they were breastfed beyond 15 months (yes/no).
Risk associated with breastfeeding duration (in months) was also measured using Cox proportional hazards modeling.
This study could not distinguish in-utero, intrapartum, from postpartum infection.
However, infants were classified as HIV-1 (negative); persistently HIV-1 positive, and seroconverting (if HIV-negative for at least 3 months and then seroconverting).
N=130 (59%); 50 (23%); and 40 (18%); respectively for these groups.

Total vertical transmission was >40% in this study population.
Breastfeeding duration was longer among surviving HIV-infected infants (p<0.05). Average duration was 14 months in HIV negative, 15.9 months in persistently positive, and 17.5 months in seroconverting children.
HIV positive children were 1.9 times more likely to breastfeed for 15 months or longer (95% CI: 1.1–3.5). Odds were 2.5 among seroconverting children (p<0.05).
A substantial proportion of HIV infection occurred postnatally, possibly through breastfeeding.

Investigators hypothesize that the high rates of vertical transmission in this study may result from longer periods of follow-up. They also suggest that some children may be protected from HIV-1 during first few months of breastfeeding, and studies with shorter follow-up periods may result in lower estimates.
Mouth ulcerations and teething, which increase with age, may increase transmission risks over time.
The increased risk of HIV transmission associated with prolonged breastfeeding (15 months or longer) may exceed the benefits for infants of HIV-1 infected mothers.

Zaire
Bertolli et al.
Date:
10/89–4/90
Sample size for this analysis:

261 children of HIV-1 infected mothers
69 children were HIV-positive at the end of the follow-up period
Average follow-up was 18 months

Study design:
Longitudinal; observational
To estimate HIV vertical transmission rates, and risks attributable to gestation; labor and delivery; the early postpartum period; and the late postpartum period (through breast-feeding).
Venous blood drawn at 0–2 days postpartum, and every 3 months thereafter for PCR and HIV culture. Breastfeeding practices were not specifically measured.
Investigators report a median duration of breastfeeding of 12 months in study children.
Eleven children were either not breastfed (N=9) or weaned before 3 months of age (N=2).
In-utero transmission was assumed if the virus was detected in infant blood within 48 hours of birth.
Intrapartum/early postnatal transmission was assigned if there was a negative PCR in the first 2 days of life, followed by positive result between 3–5 months of ages.
Late postnatal transmission was assumed if there was a negative PCR result between 3–5 months, followed by (positive) PCR, antibody test, or HIV culture at a later age.
Only 59% of children had sufficient data to determine the timing of transmission.

Overall vertical transmission rate was about 25%.
Among the 69 HIV-positive children:

23% were infected in-utero [95% CI: 14–35%];
65% were infected intrapartum or early postnatal [CI: 53–76%];
12% were infected late postnatally [CI: 5%–22%].

The risk of HIV transmission was:

6% in-utero;
18% intrapartum or early postnatal;
4% last postnatal.
Risk of vertical transmission is greater during labor and delivery than during gestation.
The risk of infection attributable to breastfeeding is “significant.”
The study’s inability to distinguish between intrapartum and early postnatal transmission through breastfeeding most likely resulted in an underestimate of the risk of transmission associated with breastfeeding.
A trial of early weaning or formula feeding is suggested by the investigators.
Note: Children with missing test results were assigned to groups (to estimate transmission risks and proportions) according to the distribution of children without missing results.
Children with indeterminant HIV status (N=25) were not included in transmission estimates.

Tanzania
Karlsson et al.
Date:
1991-not reported
Sample size for this analysis:

139 children born to HIV-1 positive mothers who were uninfected at 6 months of age
8 children became infected, all after 11 months of age.

Study design:
Longitudinal; observational

To study the rates of late postnatal transmission of HIV-1 in children born to HIV-1 infected mothers. PCR complemented with p24 antigen and HIV antibody tests.
Infant blood samples were taken every 3 months.
First sample was taken between 4–8 weeks.
HIV infections status was determined when a child had at least 2 negative samples (test results) followed by 2 HIV-positive samples or 1 HIV-negative sample and subsequent death due to HIV-related disease.

Specific breastfeeding practices were not measured.
Investigators report that “all women in the cohort were breastfeeding.”

Late postnatal transmission was assumed if an infant born to an HIV positive mother was PCR negative at 6 months, and PCR positive after 6 months. The overall vertical transmission rate in the study population was about 30%.
Eight infants who were uninfected at 6 months became infected during the follow-up. Seven of these were breastfed at the time of infection (positive test result).
Transmitting mothers had lower CD4 cell count counts than nontransmitting mothers.
Not all infants were followed for the same period of time.
Overall late postnatal transmission rate was about 6 per 100 child years of observation.
No seroconversions occurred between 6–11 months.

Investigators suggest that 6–9 months may be the age after which the advantages of breastfeeding no longer exceed the risk of HIV transmission, but it is unclear how this recommendation emerged from the study.

Rwanda
Simonon et al.
Dates:
1988-not reported
Sample size used in the analysis:

188 children born to HIV-1 infected mothers
47 children who survived to 15 months were HIV positive by 24 months
Follow-up was for 24 months
40 children (not included in the 188) died in the first 24 months of life

Study design:
Longitudinal; observational
To estimate the proportion of HIV-1 transmission occurring in-utero, intrapartum, and postnatally.
To evaluate PCR as a diagnostic tool for pediatric HIV infection.
EIA and double PCR at birth (cord blood), and at 3, 6–12, and 13–24 months of age.
HIV infection in infants was diagnosed as follows:
AIDS diagnosed according to clinical criteria;
child died of AIDS-related disease before 15 months; or
HIV antibody positive at 15 months.

Child was considered uninfected if:
HIV antibody negative at 15 months; or
lost to follow up or died of an unrelated cause and was HIV antibody negative at 9 months or older.

All other children had indeterminant status (N=32), including those who died during the neonatal period.
Breastfeeding practices were not measured or described.
The median duration of breastfeeding was 579 days (range: 0–1302 days).

Timing was based on two different sets of assumptions:
Under the first assumption, all HIV-1 positive children with a (negative) PCR on cord blood were thought to be infected during delivery or birth.
Under the second assumption, only children with a positive PCR obtained after 3 months of age were thought to be infected postnatally.

Total vertical transmission rate (at 24 months) was about 25%.
Using the first assumption, the in-utero transmission rate was 7.7% and the intrapartum transmission rate was 17.6%.
Using the second assumption, the in-utero and intrapartum transmission rate was 20.4% and the late postnatal transmission rate was 4.9%.
5% of cord blood samples produced false positive readings suggesting contamination with maternal blood.
8% false positive PCR result found in children from uninfected mothers.
At 24 months, all HIV-infected children had at least 1 positive PCR result.

Investigators concluded that breastfeeding should continue to be recommended and promoted; however known HIV-1 positive mothers who can afford and are able to safely artificial feed may be counseled individually against breastfeeding.
Results also suggest that the double PCR method is extremely sensitive.
Cord blood is probably not suitable for early diagnosis of HIV-1 infection in newborns due to probable contamination with maternal blood.

CI= confidence interval
EIA = enzyme immunoassay
ELISA = enzyme linked immunosorbent
PCR = polymerase chain reaction

Table 1: Major Studies on HIV and Breastfeeding in Developing Countries

Table 2: Major Mathematical Models of HIV Transmission and Breastfeeding

Author, Pub. Date Objective Unique Features And Major Assumptions Major Findings and Interpretation

Ross/LINKAGES
(unpublished)
To examine the effects on overall infant mortality of breastfeeding versus artificial feeding among mothers with and without HIV under different conditions/assumptions.
Allows sensitivity analysis and identification of critical values for all important variables (breastfeeding rates, baseline infant mortality rate, transmission rate, and relative risk of death due to artificial feeding). By redefining age categories, the model can be used to predict outcomes for specific age periods within infancy. Where infectious diseases are the main cause of infant mortality, the risk of death is lower if mothers with HIV breastfeed their newborn infants.
A switch to artificial feeding sometime during infancy would reduce the infant's risk of death.
Optimal timing of this switch depends on the conditions assumed.

Kuhn and Stein
1997 To examine the effects of optimal breastfeed-ing, complete avoidance of breastfeeding, and early cessation of breastfeeding, in the context of HIV. Considers issue of duration of breastfeeding, and three different feeding practices.
Assumes IMR <100/1000 and relative risks of dying set at 2.5 for non-breastfed compared with optimally breastfed infants.
Avoidance of all breastfeeding by the whole population always produces the worst outcome. The lowest frequency of adverse outcomes occurs if no HIV-seropositive women breastfeed and all HIV-seronegative women breastfeed optimally.
When it is not possible to distinguish individual from community risk (in the absence of HIV testing), sustained promotion of breastfeeding is most desirable.
Early cessation of breastfeeding at 3 months of age for known HIV-infected mothers could be advantageous.

Nagelkerke
1995 To compare the age-specific risks of mother-to-child HIV transmission versus the excess mortality due to not breastfeeding. Considers the issue of duration of breastfeeding.
Assumes that both risk of mother-to-child transmission of HIV through breastfeeding and the relative risk of not breastfeeding do not vary with age. However, the benefits of breastfeeding decrease with age.
In HIV-1 seropositive mothers, the decrease in child mortality afforded by breastfeeding may exceed the risk of mother-to-child HIV-1 transmission only during the first 3–7 months of life (in many African settings). Thereafter, the risk of HIV-1 transmission probably exceeds the mortality reduction benefit of breastfeeding.
Experimental studies on early weaning should be considered.
Supports the WHO/UNICEF 1992 recommendation that in the absence of safe alternatives, HIV-1 infected women should be encouraged to breastfeed.

Del Fante
1993 To evaluate the impact of HIV transmission and breastfeeding practices on under-5 mortality among HIV(+) and HIV(-) mothers living in urban and rural areas. Model takes no account of the duration of breastfeeding.
Assumptions of conditions were those of a hypothetical country in East Africa.
Childhood mortality would increase substantially if breastfeeding ceased. In urban settings, <5 mortality would increase by 27% among children born to HIV-1-infected mothers, and 108% among those born to uninfected mothers. In rural areas, mortality increases would be even more substantial. Adverse effects on mortality occur even if it is possible to restrict cessation of breastfeeding to HIV-infected mothers. Promotion of breastfeeding should continue regardless of HIV prevalence rates, as per WHO/UNICEF (1992) recommendation.

Hu
1992 To compare the mortality associated with HIV transmission through breastfeed-ing with the mortality expected from not breast-feeding in different populations, and to perform sensitivity analyses to illustrate critical boundaries for guiding research and policy. Assumed there were no benefits of breastfeeding after one year.
upports the 1992 WHO/UNICEF recommendation that breastfeeding should continue to be encouraged since, in most cases, it protects the majority of infants from mortality related to infectious diseases and malnutrition. Also supports the U.S. and U.K. recommendations that HIV-infected women in the U.S. and U.K. should not breastfeed.

Kennedy
1990 To compare the infant mortality associated with breastfeeding with that of not breastfeeding in a hypothetical population of 100,000 uninfected infants born to infected mothers Assumes:
probability of HIV transmission by breastfeeding alone is probably quite low (<1%).
95% of infected babies will die before age 5.
four different relative risks of death due to diseases of infancy for bottle-fed babies, compared to breastfed babies, are used.

The breastmilk transmission rate must be 20% before the expected number of deaths among breastfed babies approaches that associated with bottle feeding. However the modelers assume that the transmission rate is likely <1%.
Encourages infected mothers to breastfeed until some basic questions about HIV and breastfeeding are answered.

Heymann
1990 To compare the survival outcomes of children born to HIV-infected women who are breastfed, bottle fed, and wet nursed. Wet nursing is included as an "alternative feeding practice," although potential problems with this practice were mentioned.
This models looks at outcomes of a range of values for HIV transmission through breastfeeding, HIV seroprevalence, positive predictive value of a screening test, the child mortality rate from non-HIV-related causes, and the HIV child mortality rate.
The probability of HIV transmission via breastfeeding would need to be at least 0.12 in a community with an <5 mortality rate from non-HIV causes of 100/1000 live births and at least 0.27 in a community with a rate of 210/1000 before alternative feeding practices should be recommended even to the known HIV-infected mother who has an available feeding alternative with a relative risk of 2:1.

Publication Text | References

Author, Pub. Date	Objective	Unique Features And Major Assumptions	Major Findings and Interpretation
Ross/LINKAGES (unpublished)	To examine the effects on overall infant mortality of breastfeeding versus artificial feeding among mothers with and without HIV under different conditions/assumptions.	Allows sensitivity analysis and identification of critical values for all important variables (breastfeeding rates, baseline infant mortality rate, transmission rate, and relative risk of death due to artificial feeding). By redefining age categories, the model can be used to predict outcomes for specific age periods within infancy.	Where infectious diseases are the main cause of infant mortality, the risk of death is lower if mothers with HIV breastfeed their newborn infants. A switch to artificial feeding sometime during infancy would reduce the infant's risk of death. Optimal timing of this switch depends on the conditions assumed.
Kuhn and Stein 1997	To examine the effects of optimal breastfeed-ing, complete avoidance of breastfeeding, and early cessation of breastfeeding, in the context of HIV.	Considers issue of duration of breastfeeding, and three different feeding practices. Assumes IMR <100/1000 and relative risks of dying set at 2.5 for non-breastfed compared with optimally breastfed infants.	Avoidance of all breastfeeding by the whole population always produces the worst outcome. The lowest frequency of adverse outcomes occurs if no HIV-seropositive women breastfeed and all HIV-seronegative women breastfeed optimally. When it is not possible to distinguish individual from community risk (in the absence of HIV testing), sustained promotion of breastfeeding is most desirable. Early cessation of breastfeeding at 3 months of age for known HIV-infected mothers could be advantageous.
Nagelkerke 1995	To compare the age-specific risks of mother-to-child HIV transmission versus the excess mortality due to not breastfeeding.	Considers the issue of duration of breastfeeding. Assumes that both risk of mother-to-child transmission of HIV through breastfeeding and the relative risk of not breastfeeding do not vary with age. However, the benefits of breastfeeding decrease with age.	In HIV-1 seropositive mothers, the decrease in child mortality afforded by breastfeeding may exceed the risk of mother-to-child HIV-1 transmission only during the first 3–7 months of life (in many African settings). Thereafter, the risk of HIV-1 transmission probably exceeds the mortality reduction benefit of breastfeeding. Experimental studies on early weaning should be considered. Supports the WHO/UNICEF 1992 recommendation that in the absence of safe alternatives, HIV-1 infected women should be encouraged to breastfeed.
Del Fante 1993	To evaluate the impact of HIV transmission and breastfeeding practices on under-5 mortality among HIV(+) and HIV(-) mothers living in urban and rural areas.	Model takes no account of the duration of breastfeeding. Assumptions of conditions were those of a hypothetical country in East Africa.	Childhood mortality would increase substantially if breastfeeding ceased. In urban settings, <5 mortality would increase by 27% among children born to HIV-1-infected mothers, and 108% among those born to uninfected mothers. In rural areas, mortality increases would be even more substantial. Adverse effects on mortality occur even if it is possible to restrict cessation of breastfeeding to HIV-infected mothers. Promotion of breastfeeding should continue regardless of HIV prevalence rates, as per WHO/UNICEF (1992) recommendation.
Hu 1992	To compare the mortality associated with HIV transmission through breastfeed-ing with the mortality expected from not breast-feeding in different populations, and to perform sensitivity analyses to illustrate critical boundaries for guiding research and policy.	Assumed there were no benefits of breastfeeding after one year.	upports the 1992 WHO/UNICEF recommendation that breastfeeding should continue to be encouraged since, in most cases, it protects the majority of infants from mortality related to infectious diseases and malnutrition. Also supports the U.S. and U.K. recommendations that HIV-infected women in the U.S. and U.K. should not breastfeed.
Kennedy 1990	To compare the infant mortality associated with breastfeeding with that of not breastfeeding in a hypothetical population of 100,000 uninfected infants born to infected mothers	Assumes: probability of HIV transmission by breastfeeding alone is probably quite low (<1%). 95% of infected babies will die before age 5. four different relative risks of death due to diseases of infancy for bottle-fed babies, compared to breastfed babies, are used.	The breastmilk transmission rate must be 20% before the expected number of deaths among breastfed babies approaches that associated with bottle feeding. However the modelers assume that the transmission rate is likely <1%. Encourages infected mothers to breastfeed until some basic questions about HIV and breastfeeding are answered.
Heymann 1990	To compare the survival outcomes of children born to HIV-infected women who are breastfed, bottle fed, and wet nursed.	Wet nursing is included as an "alternative feeding practice," although potential problems with this practice were mentioned. This models looks at outcomes of a range of values for HIV transmission through breastfeeding, HIV seroprevalence, positive predictive value of a screening test, the child mortality rate from non-HIV-related causes, and the HIV child mortality rate.	The probability of HIV transmission via breastfeeding would need to be at least 0.12 in a community with an <5 mortality rate from non-HIV causes of 100/1000 live births and at least 0.27 in a community with a rate of 210/1000 before alternative feeding practices should be recommended even to the known HIV-infected mother who has an available feeding alternative with a relative risk of 2:1.