Abstract
Background:
Lifelong ART use is recommended for pregnant and breastfeeding women living with HIV (WLWH) to prevent perinatal HIV transmission and improve maternal health. We address two objectives in this analysis: (1) determine timing and factors associated with breastfeeding cessation, and (2) assess the impact of breastfeeding on health of women living with HIV on ART.
Setting:
Multi-country study of eight sites in Uganda, Malawi, Zimbabwe and South Africa.
Methods:
Prospective study of WLWH on lifelong ART. These women initially participated from 2011 to 2016 in a randomized clinical trial (PROMISE) to prevent perinatal HIV transmission, and subsequently re-enrolled in an observational study (PROMOTE, 2016–2021) to assess ART adherence, safety, and impact.
Results:
The PROMOTE cohort included 1987 women on ART. Of these, 752 breastfed and were included in analyses of objective 1; all women were included in analyses of objective 2. Median time to breastfeeding cessation varied by country (11.2–19.7 months). Country of residence, age and health status of women were significantly associated with time to breastfeeding cessation (compared to Zimbabwe: Malawi, adjusted hazard ratio [aHR] 0.50, 95% CI 0.40–0.62, p<0.001; South Africa, aHR 1.49, 95% CI 1.11–2.00, p=0.008; Uganda, aHR 1.77, 95% CI 1.37–2.29, p<0.001). Women who breastfed had lower risk of being “unwell” compared to women who never breastfed (adjusted rate ratio [aRR] 0.87,95% CI 0.81–0.95 p=0.030).
Conclusion:
Women on lifelong ART should be encouraged to continue breastfeeding with no concern for their health. Time to breastfeeding cessation should be monitored for proper counseling in each country.
Keywords: Africa, ART, breastfeeding, HIV, maternal Health, women living with HIV
INTRODUCTION
Breastfeeding (BF) protects against diarrheal and respiratory diseases and undernutrition in African children. Exclusive BF for the first six months of life, continued with complementary foods for up to two years or beyond is recommended.1 In southern and eastern Africa where child mortality rates are among the highest in the world,2 substantial progress has been made in prevention of perinatal HIV transmission through use of antiretroviral treatment (ART).3,4 In a recent report from a multi-country study from Africa we showed high adherence to ART and better health status among women living with HIV (WLWH) who had been followed for approximately 10 years.5 ART is now the standard of care and all pregnant and BF women are recommended to receive lifelong ART.1 Although the WHO recommendations for BF are generally followed within Africa, local guidelines may influence timing to BF cessation.
Multiple studies have provided rich data on the historical evolutions of HIV and BF, associated risk factors, and benefits of BF for both mothers and children. Some studies predating the introduction of universal ART in Africa reported interesting but inconsistent associations of BF with maternal HIV disease progression.6–10 Other studies reported on the impact of HIV and BF on maternal postpartum weight.7,11,12 The impact of prophylactic and therapeutic ART use was extensively evaluated in sub-Saharan Africa and were the basis for implementing programs that have led to successful prevention of perinatal HIV transmission.13–16 A more recent study conducted in four African countries showed that BF was not associated with mother’s HIV disease progression; however, only women who were not on ART per prevailing recommendations at the time were included.17
Following the implementation of successful ART programs in sub-Saharan Africa, studies of HIV/ART impact on BF duration have become scarce. It is important to evaluate the potential impact of BF among women using lifelong ART. Additionally, some studies have shown that obese women (a potential consequence of ART)18 breastfeed less and for shorter durations than non-obese women.19 Maternal health has also been described as a predictor of BF self-efficacy: healthier women may feel confident in their BF ability, and women with illnesses are more likely to wean infants sooner.20,21 Therefore, monitoring BF trends in the era of universal ART use is essential. The current analysis has to two main objectives: 1) determine time to BF cessation among WLWH on lifelong ART in four African countries with diverse populations and practices, and identify factors associated with cessation of BF; and 2) determine the impact of BF on the general health status of WLWH on lifelong ART.
METHODS
Study Design and Population
We analyzed data from the large observational PROMOTE study. WLWH on ART were enrolled and followed from 2016–2021 at eight sites in four African countries: Uganda (1), Malawi (2), Zimbabwe (3), and South Africa (2).5,22 Women were originally enrolled in IMPAACT PROMISE, a randomized clinical trial conducted between 2011 and 2016 that investigated regimens to prevent perinatal HIV transmission (NCT01061151).3 Briefly, inclusion criteria for women taking part in the PROMISE study were a CD4 count of over 350 cells/mm3 (or higher according to each country’s threshold for initiating ART at the time), hemoglobin >7.0 grams/dL, white blood cell count >1,500 cells/mm3, absolute neutrophil count >750 cells/mm3, platelets >50,000 cells/mm3, alanine aminotransferase (ALT) ≤ 2.5 × upper limit of normal, and estimated creatinine clearance of ≥60 mL/min. For the current study, we included all women who provided consent to re-enroll in PROMOTE. Re-enrollment criteria for PROMOTE were mothers willing to provide informed consent to enroll and continue follow-up in the PROMOTE study for herself and for her children, living within the study catchment area; and having no plans to move during the study period.22 The eight PROMOTE sites represented the main enrolling sites in the previous PROMISE trial. Level of CD4 count was not a criterion for enrollment in the PROMOTE study.
Study Procedures
Maternal follow-up evaluations were scheduled every six months and included questionnaires on clinical and reproductive history, physical examinations, anthropometry, and ART adherence history. Women were counseled on ART adherence and received ART supplied locally per each country’s national guidelines. The ART regimens used at the start of PROMOTE in 2016 consisted of efavirenz/nevirapine (EFV/NVP)-based for all sites except for the Kampala, Uganda site which used lopinavir/ritonavir (Lpv/r)-based. All sites switched to using the new ART regimen dolutegravir (DTG) before the PROMOTE study closed in 2021 (~37% of women reported using DTG at the last recorded visit5). Laboratory investigations included maternal complete blood count, CD4 cell count, and HIV viral load. The reproductive history questionnaires collected data on new pregnancies, BF history, and health status.
Statistical Analysis
We first described time to BF cessation and characterized factors associated with earlier BF cessation. We included data only on time to first pregnancy, and do not consider time to cessation of BF in subsequent pregnancies. For this analysis, population included women who were pregnant at enrollment into PROMOTE or became pregnant during follow-up, had a live birth, and at least one postnatal follow-up visit. We then evaluated impact of BF on maternal health, using data from all WLWH who enrolled in the PROMOTE study.
Time to BF cessation was defined as time from date of delivery to reported date of last BF; if this date was not available, the midpoint between the first visit when mothers reported no longer BF and the date of the immediate prior visit was used. Participants that never reported BF cessation were censored at the last follow-up visit of their first pregnancy. For participants who had a live birth but did not report BF at their first follow-up visit, their data were censored at the midpoint between birth and first postnatal visit (to minimize bias).
In objective 2 of this analysis were considered two outcomes: 1) Body-mass index (BMI) (weight (kilogram) / height2 [meters]) categorized as normal (<25), overweight (≥25 – <30), and obese (≥ 30); and 2) Maternal health status defined as “well” if participants self-reported their health as “excellent” or “very well”, had no hospitalizations and no unscheduled visits for adverse effects since last study visit; or as “unwell” if they self-reported their health as worse than “very well”, had been hospitalized, or had had unscheduled visits for adverse effects. The information on reported maternal health (well/unwell), including hospital admissions, unscheduled visits to the clinic, and adverse events was obtained through structured questionnaires. Review of structured case report forms, medical records and laboratory tests to assess adverse events (e.g., abnormalities of complete blood count or liver enzymes) was conducted by trained clinical staff at each site. BF was considered the main predictor variable.
We first performed descriptive analyses to compare different characteristics across countries using Pearson’s Chi-squared test for categorical data, and Kruskal-Wallis rank sum test for continuous data. In assessment of time to BF cessation, we conducted time-to-event analysis using the Kaplan-Meier (K-M) method to estimate median time to BF cessation and the corresponding interquartile range (IQR). We compared these estimates with the log-rank test. We used multivariable Cox proportional hazards regression to determine risk factors associated with BF cessation; we present hazards ratios (HRs) and 95% confidence intervals (CIs). Selection of risk factors included in these models was based on biological, epidemiological, and statistical considerations. Both time-fixed (country, work status, marital status, availability of electricity or tap water in the dwelling) and time-varying (age, BMI, health status) variables were included in the models. We also performed separate analyses to determine the impact of BF on the outcomes of BMI and reported health status over time in women who breastfed and the rest of the cohort of women in the PROMOTE study who never breastfed. For this longitudinal, repeated measures correlated data, we used generalized estimating equation (GEE) models to assess association of BF and other factors with the outcomes of interest. For the GEE models, we estimated rate ratios (RRs) using the log link, and an exchangeable correlation structure clustered on the participant level. P-values <0.05 were considered statistically significant. All analyses were performed using R (Version 4.1.3, R Core Team, Vienna, Austria) via RStudio (Version 2022.02.01 Build 461, R Studio Team, Boston, MA, USA).
Ethical Approvals
The PROMOTE study was approved by all relevant Institutional Review Boards in the U.S. and at each collaborating African research site. All women provided written informed consent to all aspects of the study.
RESULTS
A total of 1987 WLWH on lifelong ART were included in the PROMOTE study (Figure 1). Of these, 752 BF women were included in the analysis to asses timing and factors associated to BF cessation (1032 women who never became pregnant in the PROMOTE study, 96 women with BF missing data, and 107 women with non-viable births were excluded). All 1987 women who enrolled in PROMOTE study were included in the second objective analyses. Table 1 shows selected characteristics of the 752 BF women included in these analyses, stratified by country. The median age at delivery varied by country (30.8–32.3 years). Other characteristics also varied by country: in Malawi, only 16% of women completed secondary schooling, while in South Africa 51% of women completed secondary school or more. Similarly, in Zimbabwe, only 11% of women had formal employment, compared to 37% in South Africa. A lower proportion of women in Malawi reported having electricity in their household (43%) compared to 60% in Zimbabwe, 80% in Uganda, and 99% in South Africa. There were no significant differences in proportions of women married or with a regular partner by country. At enrollment (PROMOTE baseline), 84% of women had an undetectable viral load; this was highest in South Africa (93%) and lowest in Malawi (78%). At enrollment, 90% of women had a CD4 cell count of over 500 cells/mm3; highest in Uganda (96%) and Zimbabwe (95%), and lowest in Malawi (85%).
Figure 1. Flowchart showing distribution of women living HIV who participated in the PROMOTE study.
shows a flowchart for the distribution of women living with HIV on lifelong antiretroviral therapy who participated in the PROMOTE study in the countries of Uganda, Malawi, Zimbabwe and South Africa.
Table 1:
Characteristics of women living with HIV on long-term ART who became pregnant during the PROMOTE study by country (N= 752)
| Country | |||||
|---|---|---|---|---|---|
| Characteristic | Malawi N = 279 | South Africa N = 140 | Uganda N = 134 | Zimbabwe N = 199 | p-value |
| Median (IQR) age at delivery, years 1 | 31.7 (27.3, 34.9) | 31.9 (29.2, 35.4) | 30.8 (27.6, 34.0) | 32.3 (28.7, 36.1) | 0.009a |
| Median BMI (IQR) at enrollment, kg/m 2 | 23.1 (20.5, 26.5) | 27.8 (23.1, 33.0) | 25.4 (22.3, 29.4) | 23.7 (21.4, 28.1) | <0.001a |
| Highest education 2 | <0.001b | ||||
| Secondary or more | 44 (16%) | 72 (51%) | 22 (16%) | 111 (56%) | |
| Up to primary | 235 (84%) | 68 (49%) | 112 (84%) | 88 (44%) | |
| Has regular partner/married 2 | 233 (84%) | 118 (84%) | 114 (85%) | 181 (91%) | 0.120b |
| Work status 2 | <0.001b | ||||
| Formal employment | 36 (13%) | 52 (37%) | 29 (22%) | 22 (11%) | |
| Self-employed | 97 (35%) | 6 (4.3%) | 51 (38%) | 79 (40%) | |
| Unemployed/Housewife | 146 (52%) | 82 (59%) | 54 (40%) | 98 (49%) | |
| Has electricity at home 2 | 120 (43%) | 138 (99%) | 107 (80%) | 120 (60%) | <0.001b |
| HIV viral load at enrollment 2 | <0.001b | ||||
| Detectable | 62 (22%) | 10 (7.1%) | 23 (17%) | 22 (11%) | |
| Undetectable3 | 216 (78%) | 130 (93%) | 111 (83%) | 177 (89%) | |
| Unknown | 1 | 0 | 0 | 0 | |
| CD4 cell count at enrollment 2 | <0.001b | ||||
| <500 cells/mm3 | 41 (15%) | 15 (11%) | 6 (4.5%) | 10 (5.0%) | |
| ≥500 cells/mm3 | 235 (85%) | 125 (89%) | 128 (96%) | 189 (95%) | |
| Unknown | 3 | 0 | 0 | 0 | |
Median (Interquartile range)
n (%)
No RNA detected, or below limit of detection (<20 copies/mL in Uganda and <40 copies/mL in Malawi, Zimbabwe, and South Africa)
Kruskal-Wallis rank sum test
Pearson’s Chi-squared test
Table 2 summarizes median times to BF cessation among these 752 women, obtained from the K-M analyses; the accompanying survival curves are shown by country, maternal BMI, and health status categories (see Figure S1, Supplemental Digital Content 1 showing time to breastfeeding cessation). Overall, BF cessation occurred at a median of 16 months (IQR 12.2 – 20.1). These results differed by country and were statistically significant (P<0.001): South African women stopped BF earliest (median 11.2, IQR 6.2 – 16.2), followed by women in Uganda (median 12.2, IQR 10.2 – 13.2), Zimbabwe (median 15.5, IQR 13.3 – 17.7), and Malawi (median 19.7, IQR 16.9 – 21.9). The differences in median time to cessation of BF by BMI category were statistically significant (P=0.004) and showed an interesting trend: “obese” (14.9 month), “overweight” (16.0 months) and “normal” (16.3 months). There were no differences in median time to BF cessation by health status category (“unwell” 16.9 vs “well” 16.0 months). In a sub-analysis stratified by country (data not shown), no statistically significant differences were noted for time to BF cessation by health status category at delivery. The only statistically significant difference was an association between time to BF cessation by BMI category in Malawi; as is in the overall results, cessation of BF was early (18.2 months) in obese women; intermediate (19.2 months) in overweight women; and late (20.2 months) in women with normal weight (P=0.022).
Table 2:
Median time in months to breastfeeding cessation among women living with HIV on long-term ART (N=752)
| Median (months) | Interquartile range | P-valuea | |
|---|---|---|---|
| Overall (N=752) | 16.0 | 12.2 – 20.1 | |
| By Country | |||
| Malawi (N=279) | 19.7 | 16.9 – 21.9 | <0.001 |
| South Africa (N=140) | 11.2 | 6.2 – 16.2 | |
| Uganda (N=134) | 12.2 | 10.2 – 13.2 | |
| Zimbabwe (N=199) | 15.5 | 13.3 – 17.7 | |
| By body mass index at delivery b | |||
| Normal (N=349) | 16.3 | 12.3 – 20.9 | 0.004 |
| Overweight (N= 222) | 16.0 | 12.2 – 20.0 | |
| Obese (N=179) | 14.9 | 11.9 – 18.4 | |
| By health status after delivery | |||
| Unwell (N=168) | 16.9 | 12.8 – 20.9 | 0.270 |
| Well (N=584) | 16.0 | 12.2 – 19.9 | |
Log-rank test
normal (<25 kg/m2); overweight (≥25 – <30 kg/m2); obese (≥ 30 kg/m2)
Table 3 shows the results of the Cox proportional hazards analysis to assess the association of BF cessation with selected risk factors. In the multivariable model, a five-year increase in age was associated with a lower hazard of BF cessation (adjusted HR 0.90, 95% CI 0.81–0.98, P=0.023). There was a strong and statistically significant association between BF cessation and country of residence: compared to WLWH on ART from Zimbabwe, women from Malawi had lower hazard of BF cessation (HR=0.50, 95% CI 0.40–0.62, P<0.001), while women from Uganda (HR 1.77, 95% CI 1.37–2.29, P<0.001) and South Africa (HR 1.49, 95% CI 1.11–2.00, P=0.008) had higher hazards of BF cessation. Additionally, women who reported their health status as “unwell” had lower hazards of BF cessation compared to those who reported their health as “well” (HR 0.79, 95% CI 0.61–0.99, P=0.049). BMI and having electricity at home (indicator of socioeconomic status) were significantly associated with BF cessation in the univariate analyses but not after adjusting for other factors. Similarly, having tap water in the house, employment of the participant, or marital status were not significantly associated with BF cessation.
Table 3:
Cox proportional hazards model for breastfeeding cessation with selected risk factors (N=752)
| Variable | Unadjusted hazard ratio (HR) | Adjusted hazard ratio (HR) | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P | HR | 95% CI | P | |
| Maternal Age, 5-year increments | 0.92 | 0.85 – 1.01 | 0.095 | 0.90 | 0.81 – 0.98 | 0.023 |
| Country, Malawi† | 0.48 | 0.39 – 0.59 | <0.001 | 0.50 | 0.40 – 0.62 | <0.001 |
| Country, South Africa† | 1.47 | 1.11 – 1.93 | 0.006 | 1.49 | 1.11 – 2.00 | 0.008 |
| Country, Uganda† | 1.93 | 1.52 – 2.47 | <0.001 | 1.77 | 1.37 – 2.29 | <0.001 |
| Has electricity, yes vs. no | 1.57 | 1.32 – 1.86 | <0.001 | 1.05 | 0.86 – 1.28 | 0.620 |
| Has tap water, yes vs. no | 1.03 | 0.87 – 1.22 | 0.690 | ‡ | ||
| Work status, self-employed vs. unemployed | 0.99 | 0.82 – 1.20 | 0.945 | 1.04 | 0.87 – 1.27 | 0.622 |
| Work status, formally employed vs. unemployed | 1.21 | 0.96 – 1.52 | 0.113 | 1.06 | 0.84 – 1.35 | 0.609 |
| Marital status, married/coupled vs. not | 1.18 | 0.90 – 1.53 | 0.229 | 1.05 | 0.81 – 1.37 | 0.70 |
| BMI, <lower quartile vs. ≥lower quartile[1] | 0.75 | 0.62 – 0.88 | 0.007 | 0.88 | 0.71 – 1.03 | 0.098 |
| Health status, Unwell vs. well[2] | 0.68 | 0.54 – 0.86 | 0.001 | 0.79 | 0.61 – 0.99 | 0.049 |
Reference is Zimbabwe
Not included in adjusted model because this variable is highly correlated with electricity variable
Lowest quartile of BMI versus upper three quartiles of BMI
Well: Self-reported health status “excellent” or “very well”, no hospitalizations and no unscheduled visits due to sickness; Unwell: Self-reported health status worse than “very well”, hospitalizations or unscheduled visits due to sickness.
Table 4 shows the results of two GEE models for the associations of BF (exposure) with two outcomes: reported health status and BMI, after adjusting for other covariates. These models examine the impact of BF on the health status of WLWH on lifelong ART. In the health status model, women who breastfed had lower risk of being “unwell” compared to women who never breastfed (adjusted RR= 0.87, 95% CI 0.81–0.95, P=0.030). Other variables significantly associated with “unwell” compared to “well” health status included: increase in age had lower risk; and being from Malawi, South Africa, or Uganda (all compared to being from Zimbabwe) was associated with increased risk. Additionally, women with an undetectable viral load had lower risk of being “unwell” compared to women with detectable viral load (aRR 0.91, 95% CI 0.84–0.98, P=0.015) and women in the lowest quartile of BMI (compared to women in upper quartiles) had a statistically significant increased risk of being “unwell”. When BMI was the outcome measure, BF was not associated with lower BMI. However, increase in age was associated with lower risk of being in the lowest quartile of BMI (aRR=0.93, 95% CI 0.91–0.98, P=0.027); and being from South Africa (aRR=0.55, 95% CI 0.43–0.68, P<0.001) or Uganda (aRR 0.72, 95% CI 0.57–0.90, P=0.005) was significantly associated with lower risk of being in the lower quartile of the BMI compared to those from Zimbabwe. Also, women who had electricity in their household (indicator of higher socioeconomic status) had lower risk of being in the lower BMI quartile (aRR 0.78, 95% CI 0.67–0.90, P <0.001) compared to women without household electricity. Additionally, women with “unwell” health status (compared to women with “well” health status) had a statistically significant increased risk of being in the lowest BMI quartile (aRR 1.07, 95% CI 1.03–1.10, P<0.001). We also examined impact of BF with a more objective measure of maternal health status – detectability of viral load – by calendar time (to account for variabilities of testing during the COVID-19 restrictions) during conduct of the study (See Table S1, Supplemental Digital Content 2, showing viral load detectability by breastfeeding status) and did not observe differences between those who breastfed and did not breastfeed throughout the study.
Table 4:
Generalized estimating equation models for the association of breastfeeding with health status and BMI (N=1945)
| Variable | Association with health status (Unwell vs. Well) [3] | Association with BMI (<1st quart. vs. ≥1st quar.) [2] | ||||
|---|---|---|---|---|---|---|
| aRR * | 95% CI | P | aRR * | 95% CI | P | |
| Breastfed, yes vs. no | 0.87 | 0.81 – 0.95 | 0.030 | 1.02 | 0.98– 1.07 | 0.448 |
| Age, 5-year increments | 0.90 | 0.87 – 0.92 | 0.010 | 0.93 | 0.91 – 0.98 | 0.027 |
| Country, Malawi† | 1.71 | 1.54 – 1.93 | <0.001 | 1.12 | 0.94 – 1.32 | 0.214 |
| Country, South Africa† | 2.24 | 1.98 – 2.47 | <0.001 | 0.55 | 0.43 – 0.68 | <0.001 |
| Country, Uganda† | 1.41 | 1.23 – 1.59 | <0.001 | 0.72 | 0.57 – 0.90 | 0.005 |
| Electricity, yes vs. no | 0.99 | 0.90 – 1.07 | 0.859 | 0.78 | 0.67 – 0.90 | <0.001 |
| Highest education, up to primary vs. secondary | 1.09 | 1.01 – 1.17 | 0.022 | 0.99 | 0.85 – 1.16 | 0.888 |
| Viral load, undetectable vs. detectable[1] | 0.91 | 0.84 – 0.98 | 0.015 | 1.03 | 0.96 – 1.09 | 0.474 |
| BMI, <1st quart. vs. ≥1st quar.[2] | 1.16 | 1.09 – 1.24 | <0.001 | - | - | - |
| Health status: Unwell vs Well[3] | - | - | - | 1.07 | 1.03–1.10 | <0.001 |
Reference is Zimbabwe
Adjusted risk ratio
Undetectable: No RNA detected, or below limit of detection (<20 copies/mL in Uganda and <40 copies/mL in Malawi, Zimbabwe, and South Africa)
Lowest quartile of BMI versus upper three quartiles of BMI
Well: Self-reported health status “excellent” or “very well”, no hospitalizations and no unscheduled visits due to sickness; Unwell: Self-reported health status worse than “very well”, hospitalizations or unscheduled visits due to sickness
DISCUSSION
This muti-country study focused on BF in WLWH on long-term ART in eastern and southern Africa. We showed that median BF cessation varied by country. For example, in Malawi, the median time to cessation of BF extended to 22 months (close to the WHO recommendation of 24 months), compared to South Africa where the median duration of BF was relatively short – less than 12 months. The observed variability in BF duration by country is most-likely due to differences in local BF guidelines and underlying cultural and contextual differences, such as mothers’ return to formal employment among mothers in South Africa.24
The observation that cessation of BF is associated with women’s BMI category postnatally is of interest and may have implications in the context of ART regimens that are currently being recommended throughout Africa. Recently, WLWH were advised to switch to a dolutegravir-based (DTG) regimen.25 In several studies DTG was reported to increase weight and potentially non-communicable disease complications.26 A systematic review found that obese women have lower intent to breastfeed, and out of those who do, it was associated with shorter BF duration for any BF mode (exclusive or complementary)27; the authors suggested difficulties with BF for obese women, perceived lower quality of BF and association with socioeconomic status as potential explanations.19 This is an area that needs further monitoring and assessment.
In the multivariate analysis, after adjusting for other factors, there were strong and statistically significant associations between BF cessation and country of residence confirming our univariate observations; the risk of earlier cessation of BF was lower in Malawi where median BF was long, and the risk of BF cessation was higher in countries such as South Africa and Uganda where median BF was short. Other interesting associations with BF cessation in this cohort included: less healthy (unwell) women had a lower hazard of BF cessation, possibly these women cannot afford an alternative to feed the baby; older women also had lower hazard, likely an intergenerational or cultural tradition. The other contextual factors such as availability of water, electricity, maternal employment, and marital status we included in the model for their link with safer alternatives to BF were not statistically significant after adjusting for other factors. However, these factors may still be important within each country (a factor strongly associated with BF cessation – i.e., these factors may be highly correlated).
The PROMOTE study data we have from a large cohort of WLWH on long-term ART provided an opportunity to assess the impact of BF on HIV disease progression. In the era predating use of ART some studies suggested that BF may influence HIV disease progression or even increase mortality.6 Although these observations were not confirmed, examination of these hypotheses in women using lifelong ART is of interest. Our data show that BF is not associated with lower (unwell) health status or lower BMI (wasting). Additionally, a more objective measure of HIV disease progression, detectability of viral load, was also not associated with BF in this cohort. The factors that showed statistically significant associations (GEE models) with adverse health status and lower BMI in this study were explainable, and epidemiologically and biologically plausible. These findings are encouraging, and women should be counseled to continue BF for as long as they can.28 However, our findings showing no association of BF with adverse health status or BMI should be interpreted with caution. The challenge of reverse causality in determining the directionality of a relationship between exposure and outcome (e.g., women being healthier or with higher BMI may be the reason for BF) has been frequently raised in several areas of research – including BF and HIV studies.29 In our analysis we controlled for health status and BMI categories in the respective models (Table 4) to minimize this issue of reverse causality when assessing the impact of BF on health status and BMI over time. These findings support our statement that BF is not associated with adverse health outcomes independent of health status or BMI at an earlier visit. However, we acknowledge post-hoc analytic techniques alone are not adequate to address a ‘common-sense’ issue of reverse causality.
A limitation we had in this study is that we did not have data on BF modalities (exclusive, predominant, or mixed). Also, misclassification of data in determining time of BF cessation and biases with reported data are inevitable and unavoidable. Nonetheless, the diversity of the populations and settings we studied provided us with a broader perspective about the status of BF in several countries in eastern and southern Africa at a time when use of ART has become the standard of care. The PROMOTE data showed that women have become experienced in use of ART and are maintaining high adherence despite changes in ART regimens being used.5 We did not collect data on BF from women who were not living with HIV at these sites; such data could have provided useful comparisons regarding duration of breastfeeding in the overall population, enhanced generalizability, and an assessment of impact of ART. An important follow-up we were not able to complete in the PROMOTE study is evaluation of BF, weight gain, and ART (especially with the recently introduced regimen of dolutegravir).
In conclusion, this multi-country study of WLWH on lifelong ART showed that cessation of BF varied by country and appeared to be driven by local guidelines; this may need regular monitoring in each country to avoid unnecessarily shortening of BF. Additionally, BF was not associated with adverse maternal health effects. This encouraging result complements the benefits of ART in improving health of WLWH and preventing perinatal HIV transmission. Clinicians and health care providers attending to the needs of WLWH should emphasize these benefits.
Supplementary Material
Supplemental Digital Content 1: Figure that shows time in months to breastfeeding cessation by country, pregnancy order, maternal BMI, and maternal health status. doc
Supplemental Digital Content 2: Table that shows viral load detectability in women by breastfeeding status by calendar time. doc
ACKNOWLEDGEMENTS
We thank the women and children who participated in the PROMOTE study at each of the research sites. We acknowledge the research teams at each of the following sites: MU-JHU, Kampala, Uganda; UNC Project Clinical Research Site, Lilongwe, Malawi; Johns Hopkins-College of Medicine Research Project, Blantyre, Malawi; University of Zimbabwe College of Health Sciences Clinical Trials Research Centre (UZCHS-CTRC), Zimbabwe; Perinatal HIV Research Unit (PHRU), Soweto, South Africa; Centre for the AIDS Programme of Research in South Africa (CAPRISA), uMlazi Clinical Research Site, Durban, South Africa.
Conflict of Interest and Sources of Funding
No conflict of interest for all authors.
The PROMOTE study was funded by the President’s Emergency Plan for AIDS Relief (PEPFAR) through DAIDS/NIAID/NIH grants to each of the following Clinical Trials Units (CTUs): JHU-Uganda CTU Makerere University-Johns Hopkins University (MU-JHU) Research Collaboration, grant # UM1 AI069530-11; The Johns Hopkins University-Blantyre Clinical Trials Unit, grant # UM1AI069518-12; The University of North Carolina Global HIV Prevention and Treatment Clinical Trials Unit, grant # 5UM1AI069423-12; University of Zimbabwe College of Health Sciences Clinical Trials Research Centre, grant # 5UM1AI069436-12; PHRU KARABELO Clinical Trials Unit for NIAID Networks grant # 5UM1AI069453; Clinical Trials Unit for AIDS/Tuberculosis Prevention and Treatment - grant Number: 5 UM1AI069469-11; and CAPRISA Clinical Trials Unit for AIDS/Tuberculosis Prevention and Treatment, grant # 5UM1 AI069469.
Footnotes
Previous presentation
A limited version of this work was presented at the 24th International AIDS Society (IAS) conference in Montreal, Canada July 24-August 3, 2022.
References
- 1.World Health Organization, United Nation Children’s Fund. Updates on HIV and Infant Feeding: The Duration of Breastfeeding, and Support from Health Services to Improve Feeding Practices among Mothers Living with HIV.; 2016. [PubMed]
- 2.United Nations Inter-agency Group for Child Mortality Estimation. Levels & Trends in Child Mortality: Report 2021, Estimates Developed by the United Nations Inter-Agency Group for Child Mortality Estimation.; 2021.
- 3.Fowler MG, Qin M, Fiscus SA, et al. Benefits and Risks of Antiretroviral Therapy for Perinatal HIV Prevention. N Engl J Med. 2016;375(18):1726–1737. doi: 10.1056/NEJMoa1511691. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Flynn PM, Taha TE, Cababasay M, et al. Prevention of HIV-1 transmission through breastfeeding: Efficacy and safety of maternal antiretroviral therapy versus infant nevirapine prophylaxis for duration of breastfeeding in HIV-1-infected women with high CD4 cell count (IMPAACT PROMISE): a randomized, open label, clinical trial. J Acquir Immune Defic Syndr. 2018;77(4):383. doi: 10.1097/QAI.0000000000001612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Taha TE, Yende-Zuma N, Brummel SS, et al. Effects of long-term antiretroviral therapy in reproductive-age women in sub-Saharan Africa (the PEPFAR PROMOTE study): a multi-country observational cohort study. Lancet HIV. 2022;9(6):e394–e403. doi: 10.1016/S2352-3018(22)00037-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nduati R, Richardson BA, John G, et al. Effect of breastfeeding on mortality among HIV-1 infected women: a randomised trial. Lancet. 2001;357(9269):1651–1655. doi: 10.1016/S0140-6736(00)04820-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Otieno PA, Brown ER, Mbori-Ngacha DA, et al. HIV-1 disease progression in breast-feeding and formula-feeding mothers: A prospective 2-year comparison of T cell subsets, HIV-1 RNA levels, and mortality. J Infect Dis. 2007;195(2):220–229. doi: 10.1086/510245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kuhn L, Kasonde P, Sinkala M, et al. Prolonged breast-feeding and mortality up to two years post-partum among HIV-positive women in Zambia. AIDS. 2005;19(15):1677–1681. doi: 10.1097/01.AIDS.0000186817.38112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sedgh G, Spiegelman D, Larsen U, Msamanga G, Fawzi WW. Breastfeeding and maternal HIV-1 disease progression and mortality. Aids. 2004;18(7):1043–1049. doi: 10.1097/00002030-200404300-00013. [DOI] [PubMed] [Google Scholar]
- 10.Taha TE, Kumwenda NI, Hoover DR, et al. The impact of breastfeeding on the health of HIV-positive mothers and their children in sub-Saharan Africa. Bull World Health Organ. 2006;84(7):546–554. doi: 10.2471/BLT.05.027664. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cames C, Cournil A, De Vincenzi I, et al. Postpartum weight change among HIV-infected mothers by antiretroviral prophylaxis and infant feeding modality in a research setting. AIDS. 2014;28(1):85–94. doi: 10.1097/01.AIDS.0000433243.24481. [DOI] [PubMed] [Google Scholar]
- 12.Papathakis PC, Loan MDV, Rollins NC, Chantry CJ, Bennish ML, Brown KH. Body composition changes during lactation in HIV-infected and HIV-uninfected South African women. J Acquir Immune Defic Syndr. 2006;43(4):467–474. doi: 10.1097/01.QAI.0000243094.42276.92. [DOI] [PubMed] [Google Scholar]
- 13.Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet (London, England). 1999;354(9181):795–802. doi: 10.1016/S0140-6736(99)80008-7. [DOI] [PubMed] [Google Scholar]
- 14.Taha TE, Kumwenda NI, Gibbons A, et al. Short postexposure prophylaxis in newborn babies to reduce mother-to-child transmission of HIV-1: NVAZ randomised clinical trial. Lancet. 2003;362(9391):1171–1177. doi: 10.1016/S0140-6736(03)14538-2. [DOI] [PubMed] [Google Scholar]
- 15.Kumwenda NI, Hoover DR, Mofenson LM, et al. Extended Antiretroviral Prophylaxis to Reduce Breast-Milk HIV-1 Transmission. N Engl J Med. 2008;359(2):119–129. doi: 10.1056/nejmoa0801941. [DOI] [PubMed] [Google Scholar]
- 16.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(7):794–805. doi: 10.1001/JAMA.296.7.794. [DOI] [PubMed] [Google Scholar]
- 17.Somé EN, Engebretsen IMS, Nagot N, et al. HIV-1 disease progression in immune-competent HIV-1-infected and breastfeeding mothers participating in the ANRS 12174 clinical trial in Burkina Faso, South Africa, Uganda and Zambia: a cohort study. BMJ Open. 2018;8(4). doi: 10.1136/BMJOPEN-2017-019239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Venter WDF, Moorhouse M, Sokhela S, et al. Dolutegravir plus Two Different Prodrugs of Tenofovir to Treat HIV. N Engl J Med. 2019;381(9):803–815. doi: 10.1056/NEJMOA1902824. [DOI] [PubMed] [Google Scholar]
- 19.Amir LH, Donath S. A systematic review of maternal obesity and breastfeeding intention, initiation and duration. BMC Pregnancy Childbirth. 2007;7. doi: 10.1186/1471-2393-7-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zubaran C, Foresti K. Correlation between breastfeeding and maternal health status. Einstein. 2013;11(2):180. doi: 10.1590/S1679-45082013000200008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Jakobsen MS, Sodemann M, Mølbak K, Aaby P. Reason for termination of breastfeeding and the length of breastfeeding. Int J Epidemiol. 1996;25(1):115–121. doi: 10.1093/IJE/25.1.115. [DOI] [PubMed] [Google Scholar]
- 22.Taha TE, Yende-Zuma N, Aizire J, et al. The multi-country PROMOTE HIV antiretroviral treatment observational cohort in Sub-Saharan Africa: Objectives, design, and baseline findings. PLoS One. 2018;13(12):e0208805. doi: 10.1371/JOURNAL.PONE.0208805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zhao Q, Zhang B, LaValley MP, Massaro JM, Lunetta KL, Chang M. Extended Rank Tests for Analyzing Recurrent Event Data. Stat Biopharm Res. 2019;12(1):90–98. doi: 10.1080/19466315.2019.1601596. [DOI] [Google Scholar]
- 24.Goon D Ter, Ajayi AI, Adeniyi OV. Sociodemographic and lifestyle correlates of exclusive breastfeeding practices among mothers on antiretroviral therapy in the Eastern Cape, South Africa. Int Breastfeed J. 2021;16(1). doi: 10.1186/S13006-021-00366-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.WHO. Consolidated Guidelines on HIV Prevention, Testing, Treatment, Service Delivery and Monitoring: Recommendations for a Public Health Approach; 2021. [PubMed]
- 26.Bourgi K, Ofner S, Musick B, et al. Weight Gain Among Treatment-Naïve Persons With HIV Receiving Dolutegravir in Kenya. J Acquir Immune Defic Syndr. 2022;91(5). doi: 10.1097/QAI.0000000000003087. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Turcksin R, Bel S, Galjaard S, Devlieger R. Maternal obesity and breastfeeding intention, initiation, intensity and duration: a systematic review. Matern Child Nutr. 2014;10(2):166–183. doi: 10.1111/J.1740-8709.2012.00439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Evans C, Humphrey JH. Optimal breastfeeding for children born to mothers living with HIV. Lancet Child Adolesc Heal. 2020;4(3):172–174. doi: 10.1016/S2352-4642(19)30399-2. [DOI] [PubMed] [Google Scholar]
- 29.Taha TE, Hoover DR, Chen S, Kumwenda NI, Mipando L, Nkanaunena K, Thigpen M, Taylor A, Fowler MG. Effects of Cessation of Breastfeeding in HIV-1-Exposed, Uninfected Children in Malawi. Clinical Infectious Diseases, 2011, 53 (4): 388–395. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Digital Content 1: Figure that shows time in months to breastfeeding cessation by country, pregnancy order, maternal BMI, and maternal health status. doc
Supplemental Digital Content 2: Table that shows viral load detectability in women by breastfeeding status by calendar time. doc